CN118915566A - Heating ventilation equipment abnormity on-line monitoring system based on Internet of things - Google Patents

Abstract

The application discloses an abnormal on-line monitoring system of heating and ventilation equipment based on the Internet of things, and belongs to the technical field of heating and ventilation control. The system comprises: the data acquisition module is used for acquiring multidimensional operation data of the heating and ventilation equipment in real time; the edge computing module is responsible for executing real-time data processing, preliminary abnormality detection and local caching at the equipment end; the communication module ensures the reliability, safety and efficiency of data transmission; the central processing module is used for carrying out depth data analysis by utilizing cloud computing resources, maintaining a multidimensional dynamic reference model and carrying out comprehensive abnormality diagnosis by combining a knowledge graph; the self-adaptive threshold management module is used for dynamically optimizing an abnormal detection threshold and balancing detection accuracy and efficiency; the predictive maintenance module predicts the performance trend and potential faults of the equipment based on the historical fault data and the real-time data, and makes a personalized maintenance plan and optimizes resource scheduling; and the man-machine interaction module is used for providing a multi-platform visual user interface.

Description

Heating ventilation equipment abnormity on-line monitoring system based on Internet of things

Technical Field

The application relates to the technical field of heating and ventilation control, in particular to an abnormal on-line heating and ventilation equipment monitoring system based on the Internet of things.

Background

With the rapid development of internet of things (IoT) technology, traditional approaches to management of heating and ventilation devices are gradually replaced by intelligent, networked solutions. Heating and ventilation equipment plays a vital role in modern buildings, and the running state of the heating and ventilation equipment directly influences the comfort and energy consumption of the building. However, the heating and ventilation equipment has complex operating environment and large workload, and various faults and anomalies often occur, which not only affect the service life of the equipment, but also can cause serious economic loss and user dissatisfaction. Therefore, how to effectively monitor and maintain the heating and ventilation equipment has become a problem to be solved in the industry.

The traditional heating and ventilation equipment monitoring system mainly depends on periodic inspection and passive fault response, and the method is not only low in efficiency, but also difficult to discover potential problems of equipment in time, so that fault occurrence and high maintenance cost are caused. With the continuous progress of sensor technology, communication technology and data processing technology, intelligent monitoring systems based on the internet of things are beginning to be applied to the field of heating and ventilation equipment. The system can realize on-line monitoring and intelligent diagnosis of the state of the equipment by collecting the running data of the equipment in real time and combining big data analysis and a machine learning algorithm, thereby improving the running reliability and maintenance efficiency of the equipment.

The existing heating and ventilation equipment monitoring system based on the Internet of things has the defects although the monitoring and diagnosis capability is improved to a certain extent. For example, the accuracy and stability of data acquisition are insufficient, the problems of data loss and delay in the transmission process are prominent, the adaptive management of an abnormal detection threshold value is lacking, and the accuracy and effect of predictive maintenance are also to be improved. Therefore, a more sophisticated system is needed to solve the above-mentioned problems and further improve the intelligent management level of the heating and ventilation equipment.

The application provides an on-line monitoring system for abnormal heating and ventilation equipment based on the Internet of things, which is innovated from multiple aspects such as data acquisition, edge calculation, communication optimization, depth data analysis, self-adaptive threshold management, predictive maintenance and the like by comprehensively applying multiple advanced technologies, and aims to provide an efficient, reliable and intelligent heating and ventilation equipment monitoring solution, and comprehensively improve the efficiency and effect of equipment operation management.

Disclosure of Invention

In order to overcome a series of defects in the prior art, the application aims to provide an abnormal on-line monitoring system of heating and ventilation equipment based on the Internet of things, aiming at the problems, comprising the following modules:

The data acquisition module is used for acquiring multidimensional operation data of the heating and ventilation equipment in real time;

The edge computing module is responsible for executing real-time data processing, preliminary abnormality detection and local caching at the equipment end, reducing network load and improving response speed;

The communication module adopts a hybrid communication architecture and a dynamic routing algorithm to ensure the reliability, safety and efficiency of data transmission;

The central processing module is used for carrying out depth data analysis by utilizing cloud computing resources, maintaining a multidimensional dynamic reference model and carrying out comprehensive abnormality diagnosis by combining a knowledge graph;

the self-adaptive threshold management module is used for dynamically optimizing an abnormal detection threshold and balancing detection accuracy and efficiency;

The predictive maintenance module predicts the performance trend and potential faults of the equipment based on the historical fault data and the real-time data, and makes a personalized maintenance plan and optimizes resource scheduling;

And the man-machine interaction module provides a multi-platform visual user interface, integrates AR and natural language processing, and realizes visual and convenient system operation and information acquisition.

Further, the data acquisition module comprises the following units:

the multi-parameter sensor array integrates various sensors and comprehensively collects the operation data of heating and ventilation equipment;

The self-adaptive sampling control unit dynamically adjusts the sampling frequency according to the running state of the equipment and the data change rate, and optimizes the energy consumption while guaranteeing the data precision;

The self-calibration unit is used for regularly executing the self-checking and calibrating procedures of the sensor and ensuring the accuracy and stability of long-term data acquisition;

And the environment compensation unit is used for monitoring and compensating the influence of environmental factors on the reading of the sensor and improving the accuracy of the data.

Further, the central processing module comprises the following units:

The data preprocessing unit is used for cleaning, normalizing and integrating data from different devices and sensors and providing high-quality input for subsequent analysis;

The multidimensional reference performance model library is used for storing theoretical performance models of various heating and ventilation devices under different environmental conditions;

the initial reference performance curve generating unit extracts corresponding models from the multi-dimensional reference performance model library according to the equipment information and the initial environmental conditions, and generates an initial reference performance curve of the equipment;

The dynamic reference model updating unit is used for continuously optimizing and updating a reference model of the equipment performance by combining the actual operation data, the environmental change and the historical performance of the equipment based on the initial reference performance curve;

the deep learning analysis unit is used for analyzing the time sequence data by using a deep learning algorithm and identifying complex performance modes and trends;

The knowledge graph engine integrates expert knowledge, historical cases and equipment association information, and builds a comprehensive knowledge base to support intelligent diagnosis;

The abnormality diagnosis unit is used for accurately classifying the detected abnormality and analyzing root cause by combining a reference model, a deep learning analysis result and a knowledge map;

and the real-time decision support unit is used for generating operation suggestions and early warning information based on the diagnosis result and supporting quick decisions of operation and maintenance personnel.

Further, combining the reference model, the deep learning analysis result and the knowledge graph, accurately classifying the detected abnormality and analyzing the root cause, comprising the following steps:

Comparing the real-time monitoring data with a dynamic reference model, identifying abnormal data points deviating from a normal range and performing preliminary classification;

Inputting the detected abnormal data points into a trained deep learning model, identifying a complex mode in time sequence data, and outputting the severity, duration and potential fault type of the abnormality;

Based on the preliminary classification and the deep learning analysis result, carrying out knowledge map query and matching so as to quickly find a historical case similar to the current abnormality;

comprehensively analyzing the reference model comparison result, the deep learning analysis output and the knowledge-graph matching information to generate a comprehensive and accurate abnormal description and a preliminary diagnosis result;

starting a root cause analysis reasoning engine based on comprehensive analysis information, automatically checking related sensor data, verifying a fault path of each root cause, and calculating probabilities of various root causes;

based on the reasoning and verification results, a diagnostic report is generated that includes a plurality of root causes.

Further, based on the comprehensive analysis information, a root cause analysis reasoning engine is started, related sensor data is automatically checked, a fault path of each root cause is verified, and probabilities of various root causes are calculated, comprising the following steps:

Collecting sensor data D and a preliminary abnormality detection result A, wherein D= { D ₁,d₂,…,d_n }, and the preliminary abnormality detection result A comprises abnormal equipment states and possible fault modes;

Reasoning the preliminary abnormal detection result A by applying the root cause fault model F, and calculating the likelihood P (A|f _i) of each fault mode F _i;

Acquiring the prior probability P (f _i) of each fault mode f _i;

And calculating the posterior probability of each fault mode according to the Bayesian theorem, wherein the formula is as follows: p (f _i∣A)＝P(A∣f_i)·P(f_i)/P (a), P (a) is the total probability of observing the preliminary anomaly detection result a, calculated by the weighted probability of all possible failure modes: p (a) = Σ _k＝1 ^mP(A∣f_k)·P(f_k), where m represents the total number of failure modes; p (f _k) represents the prior probability of the failure mode f _k, that is, the occurrence probability of the failure mode f _k in the case where the preliminary abnormality detection result a is not observed; p (a|f _k) represents the probability that the preliminary abnormality detection result a is observed in the case of the failure mode f _k;

Based on the calculated posterior probabilities, it is determined whether each failure mode is the root cause of A and the validity of the failure path is verified.

Further, the adaptive threshold management module includes the following units:

The fuzzy logic controller converts quantitative input of the equipment state and the environmental condition into qualitative rules, and realizes smooth adjustment of the threshold value;

The genetic algorithm optimization engine continuously optimizes the threshold parameter combination through simulating the evolution process so as to adapt to different running conditions;

the multi-target evaluation unit is used for comprehensively evaluating the detection accuracy, the false alarm rate and the system efficiency and providing comprehensive performance indexes for the optimization process;

The dynamic threshold calculation unit is used for calculating and updating the abnormal detection threshold of each monitoring parameter in real time based on the output of the fuzzy logic controller and the genetic algorithm optimization engine;

The historical abnormal data analysis unit is used for analyzing the historical abnormal data and operation and maintenance feedback and providing long-term trend and seasonal change information for threshold adjustment;

a rapid response unit for rapidly adjusting a threshold value to maintain detection sensitivity of the system for an emergency or an abrupt change of operation condition;

And the threshold effect verification unit is used for continuously monitoring the detection effect after the threshold adjustment and feeding back the result to the genetic algorithm optimization engine to form closed-loop optimization.

Further, the anomaly detection threshold value of each monitoring parameter is calculated and updated in real time by the following formula: t _j(t)＝T_j(t-1)+ΔT_j(t)+Δθ_j (T), wherein T _j (T) is the threshold value of the jth monitoring parameter at time tset; t _j (T-1) is the threshold value of the jth monitoring parameter at time T-1; deltaT _j (T) is the threshold adjustment amount output by the fuzzy logic controller, representing the adjustment amount based on the current equipment state and environmental conditions; Δθ _j (t) is a threshold adjustment amount of the genetic algorithm optimization engine output, and represents a parameter adjustment amount obtained based on the optimization algorithm.

Further, the detection effect after the threshold adjustment is continuously monitored, and the result is fed back to the optimization engine, so that the specific steps of forming closed loop optimization are as follows:

Continuously collecting and analyzing key indexes of abnormal detection, including indexes of detection rate, false alarm rate and false alarm rate, so as to evaluate the effectiveness of the current threshold value setting;

performing multidimensional evaluation on the detection effect after the threshold adjustment at regular intervals, wherein the multidimensional evaluation comprises analysis of detection accuracy, response time and resource utilization efficiency dimension of different types of anomalies;

performing difference analysis and reason inference on detection effects before and after threshold adjustment, identifying significant difference change, and automatically analyzing possible reasons;

based on the effect evaluation and the difference analysis results, generating a series of optimization suggestions, each suggestion accompanied by a quantitative estimate of the expected improvement effect;

Inputting the evaluation result and the optimization suggestion into a genetic algorithm optimization engine, and sequencing the optimization suggestion according to the expected effect, implementation difficulty and current state of the system;

And the genetic algorithm optimization engine gradually executes threshold adjustment and other optimization measures according to feedback and priority, and immediately enters a new monitoring period after each adjustment, so that the actual effect of adjustment is rapidly estimated.

Further, it is judged whether the difference is significant by the following formula: wherein h represents a normalized measure of the difference in effect relative to the standard error before and after threshold adjustment; Δx is the difference in mean before and after threshold adjustment; σΔx is the standard error of the mean difference; e is the sample size of the threshold.

Further, the predictive maintenance module includes the following elements:

The data integration engine is used for collecting historical fault data and equipment real-time operation data and preprocessing the data, so that a comprehensive data base is provided for predictive analysis;

A time series analysis unit for analyzing the equipment performance data by using an advanced time series algorithm and identifying a long-term trend, a periodic pattern and an abnormal change;

a machine learning prediction model that is trained based on historical fault data, predicting a type and time of fault that may occur to the device;

the health degree evaluation unit comprehensively considers the running state, the prediction result and the history maintenance record of the equipment, and calculates and dynamically updates the health index of the equipment;

The maintenance strategy generation unit is used for making personalized maintenance plans and suggestions for each device according to the health degree evaluation and prediction results;

The resource optimization scheduling unit is used for optimizing the overall scheduling and resource allocation of maintenance work by considering the factors such as maintenance requirements, human resources, spare part inventory and the like;

And the effect evaluation and feedback unit is used for tracking the actual effect of the maintenance activity and feeding back the result to the prediction model and the strategy generator so as to continuously improve the system performance.

Compared with the prior art, the application has the following beneficial effects:

According to the application, through technologies such as data acquisition, edge calculation, mixed communication, cloud calculation, knowledge graph and the like, the running state of the heating and ventilation equipment is monitored in real time, anomaly detection and diagnosis are realized, and the system performance and resource utilization efficiency are optimized through self-adaptive threshold management and predictive maintenance, so that a multi-platform, visual and intelligent man-machine interaction interface is provided, and the management and maintenance efficiency of the heating and ventilation equipment is remarkably improved.

Drawings

Fig. 1 is a schematic structural diagram of an on-line monitoring system for abnormal heating ventilation equipment based on the internet of things, which is disclosed by the embodiment of the application.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention become more apparent, the technical solutions in the embodiments of the present invention will be described in more detail below with reference to the accompanying drawings in the embodiments of the present invention. In the drawings, the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The described embodiments are some, but not all, embodiments of the invention.

All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

The embodiments described below, together with the words of orientation, are exemplary and intended to explain the invention and should not be taken as limiting the invention.

As shown in FIG. 1, an on-line monitoring system for heating ventilation equipment abnormality based on the Internet of things comprises the following modules:

The data acquisition module is used for acquiring multidimensional operation data of the heating and ventilation equipment in real time;

The edge computing module is responsible for executing real-time data processing, preliminary abnormality detection and local caching at the equipment end, reducing network load and improving response speed;

The communication module adopts a hybrid communication architecture and a dynamic routing algorithm to ensure the reliability, safety and efficiency of data transmission;

The central processing module is used for carrying out depth data analysis by utilizing cloud computing resources, maintaining a multidimensional dynamic reference model and carrying out comprehensive abnormality diagnosis by combining a knowledge graph;

the self-adaptive threshold management module is used for dynamically optimizing an abnormal detection threshold and balancing detection accuracy and efficiency;

The predictive maintenance module predicts the performance trend and potential faults of the equipment based on the historical fault data and the real-time data, and makes a personalized maintenance plan and optimizes resource scheduling;

And the man-machine interaction module provides a multi-platform visual user interface, integrates AR and natural language processing, and realizes visual and convenient system operation and information acquisition.

The data acquisition module plays a key role in an abnormal on-line monitoring system of heating and ventilation equipment based on the Internet of things. The main task is to obtain multidimensional operation data of heating and ventilation equipment in real time, wherein the multidimensional operation data comprise a plurality of parameters such as temperature, humidity, pressure, wind speed and the like. The technical effect of the module is firstly reflected on the accuracy and timeliness of data, and the real state of equipment operation can be obtained. In real-time data acquisition, the accuracy and acquisition frequency of the sensor is critical to the quality of the final data. In addition, the data acquisition module also needs to have certain anti-interference capability, and can maintain the stability of data acquisition under the conditions of environmental noise, equipment failure and the like. Through high-efficiency data acquisition, the overall monitoring effect of the system and the timeliness of anomaly detection can be ensured, so that a foundation is laid for subsequent processing and analysis.

The edge computing module is responsible for carrying out real-time data processing, preliminary anomaly detection and local caching at the equipment end, and the technical effects of the edge computing module are mainly reflected in reducing network load and improving response speed. The edge calculation transfers the data processing task from the central server to the edge equipment closer to the data source, so that the delay of data transmission can be remarkably reduced, and real-time data change can be responded quickly. The preliminary anomaly detection can be completed at the place where the data are generated, the possible anomaly situation can be identified in time, and a large amount of data are prevented from being transmitted to the central server, so that the burden of a network is reduced, and the pressure of a central processing module is reduced. The edge calculation also has a local cache function, so that data can be stored when the network fails or delays, and the integrity and consistency of the data are ensured. The method improves the robustness of the system, and can continuously monitor and analyze the foundation when the network condition is poor, thereby improving the reliability of the system.

The communication module plays an important role in connecting different modules and devices in the whole system. The method adopts a hybrid communication architecture and a dynamic routing algorithm to ensure the reliability, the safety and the efficiency of data transmission. The hybrid communication architecture combines both wired and wireless communication technologies so that the system can maintain good communication quality in various environments. For example, wired communication may be preferred in a stable environment, while wireless communication may be utilized in a wide distribution of devices or in a complex environment. The dynamic routing algorithm is used for adjusting the data transmission path in real time according to the network state so as to avoid data loss or delay caused by network congestion or failure. The communication module also needs to realize data encryption and authentication functions, ensures the safety of data in the transmission process and prevents data leakage or tampering. Together, these technical measures ensure the communication stability of the system and the efficiency of data transmission.

The central processing module utilizes cloud computing resources to perform depth data analysis and anomaly diagnosis, and the technical effect is represented in the processing capacity and analysis depth of a large amount of data. Cloud computing resources have a high degree of computing and storage power, and can process and analyze large amounts of data collected from various devices. By establishing a multidimensional dynamic reference model, the current running state and the normal running state of the equipment can be compared in real time, and abnormal conditions can be accurately identified. Comprehensive abnormality diagnosis is carried out by combining the knowledge graph, valuable information can be extracted from the data, and a more accurate diagnosis result is provided. The knowledge graph organizes the operation knowledge and abnormal mode of the equipment in a graph form, and can carry out complex reasoning and association analysis. The powerful computing and analyzing capacity of the central processing module enables the system to find potential problems in time and provide comprehensive fault diagnosis and processing suggestions, so that the intelligent level of the system is improved.

The self-adaptive threshold management module is responsible for dynamically optimizing an anomaly detection threshold value to balance detection accuracy and efficiency. Conventional anomaly detection methods typically use a fixed threshold to determine if an anomaly has occurred in the device, but such methods can result in false positives or false negatives. The self-adaptive threshold management module dynamically adjusts the detection threshold by analyzing historical data and real-time data so as to adapt to the change of the running state of the equipment. The module automatically identifies the normal fluctuation range of the equipment in operation by utilizing a machine learning algorithm and a data mining technology, and adjusts according to real-time data. The dynamic adjustment mechanism can effectively improve the accuracy of anomaly detection, reduce false alarm and missing report, and improve the monitoring efficiency. The self-adaptive threshold management module has the technical effects of being capable of flexibly adapting to the change of the state of equipment and realizing more intelligent and efficient abnormality detection.

The technical effect of the predictive maintenance module is mainly reflected in fault prevention and resource optimization. Based on the large amount of historical data and real-time data, the predictive maintenance module uses predictive algorithms (such as regression analysis, time series analysis, and machine learning models) to predict future states and possible faults of the device. This predictive capability enables the system to be maintained before a failure occurs, reducing downtime and maintenance costs of the equipment. The module can also formulate a personalized maintenance plan according to the prediction result, optimize the scheduling of maintenance resources and improve the efficiency of maintenance work. The predictive maintenance not only improves the reliability of the equipment, but also can effectively reduce the overall operation cost and realize more efficient equipment management.

The man-machine interaction module realizes intuitive and convenient system operation and information acquisition. The technical effect of the module is mainly reflected in the improvement of user experience. Through the visual interface, a user can intuitively check the running state, the historical data and the abnormal information of the equipment, so that the monitoring and the operation are more convenient. The augmented reality technology can display the device information in the form of three-dimensional graphics in the field of view of the user, and provides a more intuitive operation experience. Natural language processing techniques allow users to interact with the system, query information, or issue instructions via voice or text. The man-machine interaction mode enables the system to operate more naturally and conveniently, reduces the operation threshold and improves the working efficiency of users. By integrating the advanced technologies, the man-machine interaction module greatly improves the usability and the intelligent level of the system.

Further, the data acquisition module comprises the following units:

the multi-parameter sensor array integrates various sensors and comprehensively collects the operation data of heating and ventilation equipment;

The self-adaptive sampling control unit dynamically adjusts the sampling frequency according to the running state of the equipment and the data change rate, and optimizes the energy consumption while guaranteeing the data precision;

The self-calibration unit is used for regularly executing the self-checking and calibrating procedures of the sensor and ensuring the accuracy and stability of long-term data acquisition;

And the environment compensation unit is used for monitoring and compensating the influence of environmental factors on the reading of the sensor and improving the accuracy of the data.

The multi-parameter sensor array is a core component of the data acquisition module and is responsible for comprehensively acquiring various operation data of heating and ventilation equipment, including a plurality of key parameters such as temperature, humidity, pressure, wind speed and the like. These sensors enable the overall monitoring of the operating state of the device by integrating different types of sensors, such as temperature sensors, humidity sensors, pressure sensors and wind speed sensors. The technical effect of the multi-parameter sensor array is manifested in several aspects. First, integrating multiple sensors can provide comprehensive equipment state information, so that the operation condition of equipment can be comprehensively known, and potential abnormality or fault can be found. Secondly, through integrated design, a plurality of sensors can work on same platform, have reduced wiring complexity and installation space, have improved the overall stability and the reliability of system. In addition, the integrated sensor array can also improve the synchronism and consistency of data acquisition and ensure the time stamp and accuracy of the data. This comprehensive monitoring capability enables the system to perform more accurate data analysis and anomaly detection, providing a solid data basis for subsequent processing and decision making.

The adaptive sampling control unit is designed to optimize the efficiency of data acquisition and the energy consumption. The unit dynamically adjusts the sampling frequency based on the operating state of the device and the rate of change of data. The technical effect is manifested in two main aspects. Firstly, by dynamically adjusting the sampling frequency, the sampling frequency can be increased when the state of the equipment changes rapidly, so that the timeliness and the accuracy of data are ensured; and when the equipment runs stably, the sampling frequency is reduced, so that the energy consumption and the data storage requirement are reduced. The self-adaptive sampling control unit analyzes the trend and the amplitude of the data change by monitoring the running state of the equipment in real time and intelligently adjusts the sampling strategy. Therefore, the accuracy of the data is ensured, the burden of data acquisition and processing is effectively reduced, the service life of equipment is prolonged, and the energy consumption of a system is reduced. The design of the unit makes the data acquisition system more efficient and economical and can adapt to the requirements of different equipment and application scenes.

The self-calibration unit is responsible for periodically executing self-checking and calibration procedures of the sensor to ensure long-term accuracy and stability of data acquisition. The technical effect of the unit is mainly to improve the data quality and the system reliability. The sensor may experience deviations in readings during prolonged use due to environmental changes, aging, or other factors. The self-calibration unit automatically adjusts the measurement parameters of the sensor by periodically running a self-checking and calibrating program so as to maintain the accuracy thereof. The process of self-calibration typically involves comparing the differences between the sensor readings and known standard values and correcting for these differences by adjusting the calibration coefficients of the sensor. The design ensures that the reading of the sensor is still reliable even in the long-term operation process, reduces the need of human intervention, and improves the automation level and maintenance efficiency of the system. In addition, the auto-calibration function also reduces maintenance costs and operational complexity, enabling the system to operate more stably and permanently.

The environment compensation unit is used for monitoring and compensating the influence of environment factors on the sensor reading, so that the accuracy of the data is improved. The readings of the sensor may be affected by environmental temperature, humidity, electromagnetic interference, etc., which may cause deviations in the measurement results. The environmental compensation unit adjusts the sensor readings by monitoring environmental conditions in real time using a compensation algorithm to eliminate these environmental effects. The technical effect is manifested in several aspects. Firstly, the environmental compensation can significantly improve the accuracy of the data, so that the sensor reading more accords with the running state of the actual equipment. Secondly, the compensation mechanism can reduce data noise generated by environmental change and improve the reliability of data. The environmental compensation unit typically includes an environmental sensor that monitors environmental changes in real time and a compensation algorithm that adjusts sensor readings based on the monitored data. In this way, more stable and accurate data can be provided, supporting high quality data analysis and decision making. This compensation mechanism is critical to maintaining data quality under a variety of environmental conditions, especially for devices operating in complex or extreme environments.

Further, the central processing module comprises the following units:

The data preprocessing unit is used for cleaning, normalizing and integrating data from different devices and sensors and providing high-quality input for subsequent analysis;

The multidimensional reference performance model library is used for storing theoretical performance models of various heating and ventilation devices under different environmental conditions;

the initial reference performance curve generating unit extracts corresponding models from the multi-dimensional reference performance model library according to the equipment information and the initial environmental conditions, and generates an initial reference performance curve of the equipment;

The dynamic reference model updating unit is used for continuously optimizing and updating a reference model of the equipment performance by combining the actual operation data, the environmental change and the historical performance of the equipment based on the initial reference performance curve;

the deep learning analysis unit is used for analyzing the time sequence data by using a deep learning algorithm and identifying complex performance modes and trends;

The knowledge graph engine integrates expert knowledge, historical cases and equipment association information, and builds a comprehensive knowledge base to support intelligent diagnosis;

The abnormality diagnosis unit is used for accurately classifying the detected abnormality and analyzing root cause by combining a reference model, a deep learning analysis result and a knowledge map;

and the real-time decision support unit is used for generating operation suggestions and early warning information based on the diagnosis result and supporting quick decisions of operation and maintenance personnel.

The data preprocessing unit is an important component part in the central processing module and is responsible for cleaning, normalizing and integrating data from different devices and sensors, and providing high-quality input for subsequent analysis. The technical effects thereof are mainly represented in the following aspects. First, data cleansing can remove noise, errors, and missing values in the data, which is critical to improving the accuracy and reliability of the data. Common data cleansing operations include removing outliers, filling in missing data, and correcting erroneous data. Second, data normalization ensures that data from different sensors and devices has a uniform format and dimension, facilitating subsequent processing and analysis. For example, all data is uniformly converted to the same unit or scale so that there is comparability between the data. The integration operation combines the data from different sources to form a comprehensive data set. The comprehensive data set can reflect the actual operation condition of the equipment more comprehensively, and provides a more accurate basis for subsequent analysis. Through the operation of the data preprocessing unit, the data quality can be obviously improved, errors in analysis are reduced, and further the analysis precision and reliability of the whole system are improved.

The multidimensional reference performance model library is a key component in the central processing module and is used for storing theoretical performance models of various heating and ventilation equipment under different environmental conditions. These performance models cover the theoretical behavior of the device under different operating conditions, environmental conditions and loads. The technical effects are mainly manifested in the following aspects. First, the multi-dimensional baseline performance model library provides a systematic frame of reference that can help analyze and compare differences between actual operating data and theoretical models. By comparing the actual data with the theoretical performance model, the operational state of the device can be evaluated to find possible anomalies or performance deviations. Second, the baseline performance model library can support performance prediction under different environmental conditions. For example, in extreme environments such as high temperature, high humidity, or low temperature, the model library may provide the expected performance of the device. Through the multidimensional models, the performance of the equipment can be more accurately evaluated, and more targeted diagnosis and maintenance suggestions are provided. The efficient management and updating of the multi-dimensional baseline performance model library can promote the understanding and predicting capabilities of the system to the performance of the equipment, so that the maintenance and operation strategies of the equipment are optimized.

The initial reference performance curve generating unit extracts corresponding models from the multi-dimensional reference performance model library according to the basic information and the initial environmental conditions of the equipment, and generates an initial reference performance curve of the equipment. The technical effect is embodied in generating a performance reference curve that best matches the current state of the device. The initial baseline performance curve serves as an initial reference point for device performance and is critical for subsequent performance monitoring and anomaly detection. The unit first selects a theoretical performance model to match to based on the model, specification, installation location and initial environmental conditions of the device, and generates an initial performance curve for the device based on these models. The initial performance curve provides a benchmark for the normal operating range and performance level of the device, and actual data can be compared and analyzed in real time during the operation of the device. In this way, the system can more accurately identify performance changes and potential anomalies, providing a reliable reference for long-term monitoring and maintenance of equipment.

The dynamic reference model updating unit is responsible for continuously optimizing and updating the reference model of the equipment performance based on the initial reference performance curve and combined with the actual operation data, the environmental change and the historical performance of the equipment. The technical effect is mainly reflected in improving the adaptability and accuracy of the reference model. The performance of the device may be affected by various factors such as environmental changes, device aging, load changes, etc. The dynamic reference model updating unit is used for adjusting and optimizing the initial performance curve in real time by continuously monitoring the actual operation data of the equipment and combining the data. The unit uses a data-driven approach to continually update the reference model to reflect the actual operating state and performance trends of the device. Through the dynamic adjustment mechanism, the device can be better adapted to the running change of the device, and the accuracy and the practicability of the performance benchmark model are improved. In addition, the dynamic reference model updating unit can identify and correct long-term trend and periodical change through analysis of historical data, so that the reference model is more stable and reliable, and the overall diagnosis and prediction capability of the system is improved.

The deep learning analysis unit analyzes the time series data by using a deep learning algorithm and identifies complex performance modes and trends. The deep learning technology can automatically extract features from a large amount of time sequence data by constructing a complex neural network model, and identify potential modes and anomalies. The technical effects thereof are manifested in the following aspects. First, the deep learning algorithm has strong pattern recognition capability, and can find complex patterns and trends that are difficult to perceive by the conventional algorithm. For example, by using Long and Short Term Memory (LSTM) or convolutional neural (CNN) models, the operational data of the device can be effectively analyzed to identify long term trends, periodic variations, and transient anomalies. And secondly, deep learning analysis can process a large amount of time sequence data, so that efficient feature extraction and anomaly detection are realized. Through training the deep learning model, the recognition capability of the equipment performance change can be improved, and the requirement of manual intervention is reduced. The application of the deep learning analysis unit enables the system to realize more intelligent data analysis and provides more accurate fault diagnosis and performance evaluation.

The knowledge graph engine integrates expert knowledge, historical cases and equipment association information, and constructs a comprehensive knowledge base for supporting intelligent diagnosis. The technical effect is mainly reflected in the aspects of organization and utilization of knowledge. Knowledge graph organizes information such as domain knowledge, equipment operation history, failure modes and the like in a graph form, so that the system can perform complex reasoning and association analysis. By constructing the knowledge graph, the experience knowledge and the historical data of the expert can be integrated, and support is provided for abnormality diagnosis. For example, knowledge maps may help the system identify common failure modes, the cause of associated equipment failure, and apply such knowledge to real-time diagnostics. The associated information in the atlas can also help the system to find hidden failure modes, providing more comprehensive diagnostic results. The knowledge graph engine has the technical effects of improving the intelligent level and the diagnosis capability of the system, so that the fault analysis of the equipment is more accurate and comprehensive, and the maintenance and management strategies of the equipment are optimized.

The technical effect of the abnormality diagnosis unit is to improve the accuracy of abnormality detection and diagnosis. First, the abnormality diagnosis unit can identify an abnormality in the operation of the apparatus by comparing the actual data with the reference model. Based on the results of the deep learning analysis, further classification and assessment of anomalies can be performed. The knowledge graph provides rich fault modes and cause information, and helps the system to conduct root cause analysis. In combination with this information, the anomaly diagnostic unit may generate detailed anomaly reports providing accurate fault classification and root cause analysis. Thus, the accuracy of abnormality detection is improved, and a targeted solution and maintenance advice can be provided, so that the overall diagnosis capability and maintenance efficiency of the system are improved.

The technical effect of the real-time decision support unit is embodied in improving the decision efficiency and response speed. By analyzing the operating data and abnormal conditions of the device in real time, the decision support unit can generate specific operation suggestions, such as adjusting device parameters, performing maintenance operations or replacing components. In addition, the unit can also send early warning information to remind operation and maintenance personnel of potential faults or equipment anomalies. Real-time decision support units typically include a visual interface and an automated notification mechanism that enable the operator to quickly obtain information and respond. By the mode, the operation and maintenance work efficiency can be remarkably improved, the downtime of equipment faults is reduced, the resource scheduling is optimized, and the operation reliability of the equipment and the management level of the whole system are finally improved.

Further, combining the reference model, the deep learning analysis result and the knowledge graph, accurately classifying the detected abnormality and analyzing the root cause, comprising the following steps:

Comparing the real-time monitoring data with a dynamic reference model, identifying abnormal data points deviating from a normal range and performing preliminary classification;

Inputting the detected abnormal data points into a trained deep learning model, identifying a complex mode in time sequence data, and outputting the severity, duration and potential fault type of the abnormality;

Based on the preliminary classification and the deep learning analysis result, carrying out knowledge map query and matching so as to quickly find a historical case similar to the current abnormality;

comprehensively analyzing the reference model comparison result, the deep learning analysis output and the knowledge-graph matching information to generate a comprehensive and accurate abnormal description and a preliminary diagnosis result;

starting a root cause analysis reasoning engine based on comprehensive analysis information, automatically checking related sensor data, verifying a fault path of each root cause, and calculating probabilities of various root causes;

based on the reasoning and verification results, a diagnostic report is generated that includes a plurality of root causes.

Further, based on the comprehensive analysis information, a root cause analysis reasoning engine is started, related sensor data is automatically checked, a fault path of each root cause is verified, and probabilities of various root causes are calculated, comprising the following steps:

Collecting sensor data D and a preliminary abnormality detection result A, wherein D= { D ₁,d₂,…,d_n }, and the preliminary abnormality detection result A comprises abnormal equipment states and possible fault modes;

Reasoning the preliminary abnormal detection result A by applying the root cause fault model F, and calculating the likelihood P (A|f _i) of each fault mode F _i;

Acquiring the prior probability P (f _i) of each fault mode f _i;

And calculating the posterior probability of each fault mode according to the Bayesian theorem, wherein the formula is as follows: p (f _i∣A)＝P(A∣f_i)·P(f_i)/P (a), P (a) is the total probability of observing the preliminary anomaly detection result a, calculated by the weighted probability of all possible failure modes: p (a) = Σ _k＝1 ^mP(A∣f_k)·P(f_k), where m represents the total number of failure modes; p (f _k) represents the prior probability of the failure mode f _k, that is, the occurrence probability of the failure mode f _k in the case where the preliminary abnormality detection result a is not observed; p (a|f _k) represents the probability that the preliminary abnormality detection result a is observed in the case of the failure mode f _k;

Based on the calculated posterior probabilities, it is determined whether each failure mode is the root cause of A and the validity of the failure path is verified.

Further, the adaptive threshold management module includes the following units:

The fuzzy logic controller converts quantitative input of the equipment state and the environmental condition into qualitative rules, and realizes smooth adjustment of the threshold value;

The genetic algorithm optimization engine continuously optimizes the threshold parameter combination through simulating the evolution process so as to adapt to different running conditions;

the multi-target evaluation unit is used for comprehensively evaluating the detection accuracy, the false alarm rate and the system efficiency and providing comprehensive performance indexes for the optimization process;

The dynamic threshold calculation unit is used for calculating and updating the abnormal detection threshold of each monitoring parameter in real time based on the output of the fuzzy logic controller and the genetic algorithm optimization engine;

The historical abnormal data analysis unit is used for analyzing the historical abnormal data and operation and maintenance feedback and providing long-term trend and seasonal change information for threshold adjustment;

a rapid response unit for rapidly adjusting a threshold value to maintain detection sensitivity of the system for an emergency or an abrupt change of operation condition;

And the threshold effect verification unit is used for continuously monitoring the detection effect after the threshold adjustment and feeding back the result to the genetic algorithm optimization engine to form closed-loop optimization.

The fuzzy logic controller is responsible for converting quantitative inputs of the device state and the environmental condition into qualitative rules in the adaptive threshold management module so as to realize smooth adjustment of the threshold. The technical effects thereof are mainly represented in the following aspects. The fuzzy logic controller is capable of handling the complexity and uncertainty of the device state and environmental conditions, and by mapping these inputs to a set of fuzzy rules, smooth control of the threshold is achieved. Unlike conventional precise logic control, fuzzy logic controllers are capable of coping with real-world ambiguities and uncertainties, for example, the operating state of a device may be affected by a number of factors, the variation of which is not always linear. By defining fuzzy rules (such as 'high' temperature, 'medium' humidity, etc.), the fuzzy logic controller can dynamically adjust the abnormality detection threshold according to actual conditions. The method enables the threshold value to be adjusted more flexibly and adapt to environmental changes, and avoids the problem of false alarm or missing report possibly caused by hard setting of the threshold value. In addition, the fuzzy logic controller can also model complex equipment behaviors, comprehensively consider a plurality of input factors, and enable the threshold value adjusting process to be smoother and accord with actual conditions. By the mode, more intelligent abnormality detection can be realized, and the accuracy of overall monitoring and response is improved.

The genetic algorithm optimization engine continuously optimizes the threshold parameter combination in the self-adaptive threshold management module through simulating the evolution process so as to adapt to different running conditions. The technical effect is that the accuracy and the adaptability of the threshold setting are improved. Genetic algorithm is an optimization method based on natural selection and genetic principle, and different threshold combinations are continuously generated and evaluated through simulating the evolution process (including selection, crossover, mutation and other operations) so as to find the optimal parameter setting. The genetic algorithm optimization engine initially sets the current threshold value, then generates a plurality of threshold value combinations through an iterative process, and selects the optimal threshold value parameters by evaluating the performances (such as detection accuracy, false alarm rate and the like) of the combinations. Compared with the traditional optimization method, the genetic algorithm can effectively solve the problem of complex and multidimensional threshold optimization, and the accuracy of threshold setting is improved through the multi-generation evolutionary process. In addition, the genetic algorithm has global optimization capability, can jump out of a local optimal solution, and finds out a threshold combination which is more in line with the actual situation. Through the optimization mechanism, the threshold value can be automatically adjusted under different running conditions, and the accuracy and the efficiency of detection are ensured.

The multi-target evaluation unit is responsible for comprehensively evaluating the detection accuracy, the false alarm rate and the system efficiency in the self-adaptive threshold management module, and provides comprehensive performance indexes for the threshold optimization process. The technical effect is mainly reflected in the comprehensiveness and scientificity of evaluating and optimizing threshold setting. The multi-objective evaluation unit can provide a comprehensive performance evaluation system by comprehensively evaluating the detection performances set by different thresholds. For example, the detection accuracy index reflects the ability of the threshold setting to identify the actual anomaly, the false positive rate index measures the false positive rate of the threshold setting to the normal data, and the system efficiency index evaluates the impact of the threshold setting to the overall system performance. By comprehensively considering these metrics, the threshold settings can be optimized to balance detection accuracy and system efficiency. The multi-objective evaluation unit typically employs a weighted scoring system or a multi-dimensional evaluation method to quantify various performance metrics and to perform comprehensive optimization based on these metrics. Such an evaluation method can provide a more comprehensive and scientific threshold optimization scheme, ensuring that optimal performance and efficiency are maintained under different operating conditions.

The dynamic threshold calculation unit calculates and updates the abnormality detection threshold of each monitoring parameter in real time based on the outputs of the fuzzy logic controller and the genetic algorithm optimization engine. The technical effect is that the real-time performance and the accuracy of threshold adjustment are improved. The main task of the dynamic threshold calculation unit is to apply the optimal threshold combination generated by the rules provided by the fuzzy logic controller and the genetic algorithm optimization engine to actual data monitoring. The detection threshold can be dynamically adjusted according to the current state and environmental change of the equipment by calculating and updating the threshold in real time. For example, the dynamic threshold calculation unit may adjust the threshold value instantaneously when a significant change occurs in the operating state of the device to maintain the sensitivity and accuracy of the detection. The design of the unit also comprises the processing of the real-time data stream, so that the updating of the threshold value can reflect the actual condition of the equipment in time. Through the dynamic adjustment mechanism, the system can adapt to continuously changing operation conditions, optimize the performance of anomaly detection, reduce false alarm and missing report, and improve the effectiveness of the whole monitoring system.

The technical effect of the historical abnormal data analysis unit is mainly reflected in the long-term stability and adaptability of threshold adjustment. The historical anomaly data analysis unit identifies long-term trends and periodic changes by performing in-depth analysis of past anomaly events and equipment operation and maintenance records. For example, by analyzing the historical data, abnormal modes of the device under different seasons, running periods or workload can be found, so that basis is provided for adjustment of the threshold value. The unit performs pattern recognition and trend prediction on the historical data by using a statistical analysis method or a machine learning technology. Based on these analysis results, the threshold can be adjusted to accommodate long-term trends and seasonal changes, avoiding false positives or false negatives due to environmental changes. The effective application of the historical abnormal data analysis unit can carry out more accurate threshold adjustment based on long-term data accumulation, and stability and reliability of the system are improved.

The technical effect of the quick response unit is mainly reflected in improving the response speed and the sensitivity of the system to sudden changes. Sudden events, such as equipment faults, environmental changes, etc., may occur during the operation of the equipment, and these events may cause the state of the equipment to change sharply. The rapid response unit can rapidly adjust the detection threshold value when abnormal conditions are found by monitoring the running state and environmental changes of the equipment in real time. For example, when a sudden fluctuation in the plant operating parameters is detected, the fast response unit may adjust the threshold value on the fly to maintain the sensitivity and accuracy of the detection. The unit comprises an efficient data processing and adjusting mechanism, can finish resetting of the threshold value in a short time, ensures that the system can respond to changes in time, and avoids missing report or false report. Through the quick response mechanism, stable detection performance can be maintained under various emergency conditions, and the robustness and the coping ability of the whole system are enhanced.

The threshold effect verification unit is responsible for continuously monitoring the detection effect after threshold adjustment and feeding back the result to the genetic algorithm optimization engine to form closed loop optimization. The technical effect is mainly reflected in closed loop optimization and continuous improvement of threshold adjustment. The threshold effect verification unit monitors and evaluates the actual detection effect after the threshold adjustment, and can judge whether the adjustment achieves the expected effect. For example, the validity of the threshold adjustment is verified by monitoring indexes such as detection accuracy, false alarm rate and the like. If the adjusted threshold results in an improvement or degradation of the detection performance, the verification unit will feed back these results to the genetic algorithm optimization engine. The genetic algorithm optimization engine then further optimizes the threshold settings based on the feedback information. Through the closed loop optimization mechanism, the threshold setting can be continuously adjusted and improved, and the accuracy and efficiency of anomaly detection are improved. The design of the threshold effect verification unit ensures continuous optimization and improvement of the system, can adapt to continuously changing equipment states and environmental conditions, and realizes intelligent and self-adaptive threshold management.

Further, the anomaly detection threshold value of each monitoring parameter is calculated and updated in real time by the following formula: t _j(t)＝T_j(t-1)+ΔT_j(t)+Δθ_j (T), wherein T _j (T) is the threshold value of the jth monitoring parameter at time tset; t _j (T-1) is the threshold value of the jth monitoring parameter at time T-1; deltaT _j (T) is the threshold adjustment amount output by the fuzzy logic controller, representing the adjustment amount based on the current equipment state and environmental conditions; Δθ _j (t) is a threshold adjustment amount of the genetic algorithm optimization engine output, and represents a parameter adjustment amount obtained based on the optimization algorithm.

Further, the detection effect after the threshold adjustment is continuously monitored, and the result is fed back to the optimization engine, so that the specific steps of forming closed loop optimization are as follows:

Continuously collecting and analyzing key indexes of abnormal detection, including indexes of detection rate, false alarm rate and false alarm rate, so as to evaluate the effectiveness of the current threshold value setting;

performing multidimensional evaluation on the detection effect after the threshold adjustment at regular intervals, wherein the multidimensional evaluation comprises analysis of detection accuracy, response time and resource utilization efficiency dimension of different types of anomalies;

performing difference analysis and reason inference on detection effects before and after threshold adjustment, identifying significant difference change, and automatically analyzing possible reasons;

based on the effect evaluation and the difference analysis results, generating a series of optimization suggestions, each suggestion accompanied by a quantitative estimate of the expected improvement effect;

Inputting the evaluation result and the optimization suggestion into a genetic algorithm optimization engine, and sequencing the optimization suggestion according to the expected effect, implementation difficulty and current state of the system;

And the genetic algorithm optimization engine gradually executes threshold adjustment and other optimization measures according to feedback and priority, and immediately enters a new monitoring period after each adjustment, so that the actual effect of adjustment is rapidly estimated.

Further, it is judged whether the difference is significant by the following formula: wherein h represents a normalized measure of the difference in effect relative to the standard error before and after threshold adjustment; Δx is the difference in mean before and after threshold adjustment; σΔx is the standard error of the mean difference; e is the sample size of the threshold.

Further, the predictive maintenance module includes the following elements:

The data integration engine is used for collecting historical fault data and equipment real-time operation data and preprocessing the data, so that a comprehensive data base is provided for predictive analysis;

A time series analysis unit for analyzing the equipment performance data by using an advanced time series algorithm and identifying a long-term trend, a periodic pattern and an abnormal change;

a machine learning prediction model that is trained based on historical fault data, predicting a type and time of fault that may occur to the device;

the health degree evaluation unit comprehensively considers the running state, the prediction result and the history maintenance record of the equipment, and calculates and dynamically updates the health index of the equipment;

The maintenance strategy generation unit is used for making personalized maintenance plans and suggestions for each device according to the health degree evaluation and prediction results;

The resource optimization scheduling unit is used for optimizing the overall scheduling and resource allocation of maintenance work by considering the factors such as maintenance requirements, human resources, spare part inventory and the like;

And the effect evaluation and feedback unit is used for tracking the actual effect of the maintenance activity and feeding back the result to the prediction model and the strategy generator so as to continuously improve the system performance.

The data integration engine is responsible for collecting historical fault data and equipment real-time operation data in the predictive maintenance module, preprocessing the data, and providing a comprehensive data base for predictive analysis. The technical effect of this process is mainly to build a data-driven predictive analysis platform to ensure that the subsequent analysis model and predictive system can operate on the basis of high quality data. The data consolidation engine first extracts relevant information from a plurality of data sources, including historical fault records of the device, real-time operating data, environmental conditions, and the like. By integrating these data into a unified data platform, a comprehensive database can be formed for supporting subsequent predictive analysis. By establishing an efficient data integration mechanism, the data integration engine not only improves the usability of data, but also lays a solid foundation for accurate predictive analysis.

The time series analysis unit analyzes the device performance data using an advanced time series algorithm to identify long-term trends, periodic patterns, and abnormal changes. The technical effect of this process is mainly represented by the deep analysis of the device operational data to reveal the time-dependent nature of the device performance and the potential for abnormal patterns. The time series analysis unit employs a variety of advanced algorithms such as autoregressive moving average (ARMA), autoregressive integrated moving average (ARIMA), and long-term short-term memory network (LSTM) to analyze the historical operating data of the device. These algorithms can capture long-term trends in data (e.g., gradual decline in device performance), periodic patterns (e.g., fluctuations in device performance over a particular period of time), and sudden abnormal changes. By identifying these patterns, future performance of the device can be predicted and potential signs of failure can be found in time. For example, the identification of periodic patterns may help predict maintenance periods for the device, while the detection of abnormal changes may alert the device in time of possible failures. The deep analysis of the time sequence analysis unit not only improves the accuracy of prediction, but also enhances the understanding of the system to the running state of the equipment, thereby providing a reliable basis for the subsequent predictive maintenance.

Machine learning predictive models machine learning models are trained based on historical fault data to predict the type and time of failure that a device may develop. The technical effect of the process is mainly that accurate fault prediction is provided through a data-driven model, so that the maintenance efficiency of equipment is improved. Machine learning predictive models train various machine learning algorithms (e.g., decision trees, random forests, support vector machines, and deep learning models) to identify potential patterns of equipment failure by using historical failure data. The training process includes feature selection (extraction of key features from historical data), model training (adjustment of model parameters to improve prediction accuracy), and model verification (evaluation of the prediction effect of the model). Once the model training is complete, it can analyze the real-time data and predict the type of failure and time of occurrence that the device may have occurred. For example, by predicting the type of failure of the equipment, a maintenance team may prepare the required spare parts and tools in advance, while predicting the time of failure may help to make a maintenance plan, avoiding sudden downtime of the equipment. The application of the machine learning prediction model can obviously improve the accuracy of fault prediction and reduce the occurrence of unexpected faults, thereby improving the operation reliability and maintenance efficiency of the equipment.

The technical effect of the health degree evaluation unit is mainly that the health state evaluation of various information providing devices is comprehensively analyzed, so that scientific maintenance decision is supported. The health evaluation unit firstly gathers real-time running states (such as temperature, pressure, vibration and the like) of the equipment, fault prediction results of the machine learning prediction model and historical maintenance records (such as past faults and maintenance histories) of the equipment. The information is comprehensively analyzed to calculate a dynamically updated device Health Index (Health Index) which reflects the overall Health status of the device and the risk of possible future faults. The calculation of the health index is typically combined with a weighted scoring model to take into account the effects of different factors, e.g., the operating state of the device may take up the primary weight of the health index, while the historic maintenance record acts as an auxiliary factor. By dynamically updating the health index, the health condition of the equipment can be monitored in real time, and data support is provided for maintenance decisions. The effective application of the health degree evaluation unit improves the scientificity and pertinence of equipment maintenance, reduces the risk of faults and improves the reliability and stability of the equipment.

And the maintenance strategy generating unit makes personalized maintenance plans and suggestions for each device according to the health degree evaluation and prediction results. The technical effect of this process is mainly to provide a targeted maintenance strategy to optimize the operation and maintenance management of the equipment. The maintenance policy generation unit first evaluates the maintenance requirements and priorities of each device based on the health index of the device and the output of the predictive model. Then, a personalized maintenance plan is formulated according to the evaluation results, and the personalized maintenance plan comprises different types of maintenance activities such as periodic inspection, preventive maintenance, predictive maintenance, repairable maintenance and the like. The maintenance strategy generating unit also considers the factors such as the use environment, the operation load, the historical fault record and the like of the equipment so as to ensure the comprehensiveness and the effectiveness of the maintenance plan. For example, for a device with a lower health index, the maintenance schedule may include more frequent inspections and preventative maintenance to reduce the risk of failure; for devices with higher health indices, it may be recommended to reduce maintenance frequency and optimize resource allocation. By making a personalized maintenance plan, the system can improve the maintenance efficiency of the equipment, reduce the maintenance cost, prolong the service life of the equipment and finally realize the optimal running state and maintenance effect of the equipment.

The resource optimization scheduling unit considers factors such as maintenance requirements, human resources, spare part inventory and the like, and optimizes the overall scheduling and resource allocation of maintenance work. The technical effect of the process is mainly that the efficiency and the economy of maintenance work are improved through scientific scheduling and resource management. The resource optimization scheduling unit makes an optimal maintenance plan and resource allocation scheme by analyzing factors such as maintenance requirements of equipment, available maintenance personnel, spare part inventory conditions, maintenance time windows and the like. The unit uses an optimization algorithm (such as linear programming, integer programming, or heuristic) to solve the problem of resource constraint in maintenance scheduling to achieve efficient utilization of maintenance resources. For example, maintenance tasks are arranged reasonably according to skills and available time of maintenance personnel; and optimizing purchasing and allocation of spare parts according to the inventory and the use frequency of the spare parts. The resource optimization scheduling unit also needs to monitor the progress of maintenance work in real time and adjust the resource allocation scheme according to actual conditions. Through the scientific scheduling and resource optimization, the system can reduce the maintenance time and cost, improve the overall efficiency of maintenance work and ensure that the equipment operates in an optimal state.

The effect evaluation and feedback unit tracks the actual effect of the maintenance activity and feeds the result back to the predictive model and the policy generator to continuously improve system performance. The technical effect of this process is mainly represented by the continuous optimization of system performance achieved by a feedback mechanism. The effect evaluation and feedback unit firstly collects actual effect data after maintenance activities, including the running state of equipment, the failure occurrence rate after maintenance, the actual cost and time of maintenance and the like. These data are then analyzed to evaluate the gap between the actual effect and the expected effect of the maintenance activity. For example, by comparing the maintained equipment failure rate with the predicted failure rate, evaluating the accuracy of the prediction model; the economy of the maintenance strategy is assessed by analyzing the actual maintenance cost and the budget cost. After the evaluation result is fed back to the prediction model and the strategy generator, adjustment and optimization can be performed based on the feedback information. For example, the predictive model may be retrained and adjusted based on actual effect data to improve the accuracy of the prediction; the policy generator may optimize the maintenance plan based on the feedback results to improve the effectiveness and efficiency of the maintenance. Through the continuous effect evaluation and feedback, the system can continuously improve the performance, and the scientificity and effectiveness of equipment maintenance are improved.

Finally, it should be pointed out that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting. Although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims (10)

1. Heating ventilation equipment abnormal online monitoring system based on the Internet of things is characterized by comprising the following modules:

The data acquisition module is used for acquiring multidimensional operation data of the heating and ventilation equipment in real time;

The edge computing module is responsible for executing real-time data processing, preliminary abnormality detection and local caching at the equipment end, reducing network load and improving response speed;

The communication module adopts a hybrid communication architecture and a dynamic routing algorithm to ensure the reliability, safety and efficiency of data transmission;

The central processing module is used for carrying out depth data analysis by utilizing cloud computing resources, maintaining a multidimensional dynamic reference model and carrying out comprehensive abnormality diagnosis by combining a knowledge graph;

the self-adaptive threshold management module is used for dynamically optimizing an abnormal detection threshold and balancing detection accuracy and efficiency;

The predictive maintenance module predicts the performance trend and potential faults of the equipment based on the historical fault data and the real-time data, and makes a personalized maintenance plan and optimizes resource scheduling;

And the man-machine interaction module provides a multi-platform visual user interface, integrates AR and natural language processing, and realizes visual and convenient system operation and information acquisition.

2. The internet of things-based heating and ventilation equipment abnormality online monitoring system according to claim 1, wherein the data acquisition module comprises the following units:

the multi-parameter sensor array integrates various sensors and comprehensively collects the operation data of heating and ventilation equipment;

The self-adaptive sampling control unit dynamically adjusts the sampling frequency according to the running state of the equipment and the data change rate, and optimizes the energy consumption while guaranteeing the data precision;

The self-calibration unit is used for regularly executing the self-checking and calibrating procedures of the sensor and ensuring the accuracy and stability of long-term data acquisition;

And the environment compensation unit is used for monitoring and compensating the influence of environmental factors on the reading of the sensor and improving the accuracy of the data.

3. The internet of things-based heating and ventilation equipment abnormality online monitoring system according to claim 1, wherein the central processing module comprises the following units:

The data preprocessing unit is used for cleaning, normalizing and integrating data from different devices and sensors and providing high-quality input for subsequent analysis;

The multidimensional reference performance model library is used for storing theoretical performance models of various heating and ventilation devices under different environmental conditions;

the initial reference performance curve generating unit extracts corresponding models from the multi-dimensional reference performance model library according to the equipment information and the initial environmental conditions, and generates an initial reference performance curve of the equipment;

The dynamic reference model updating unit is used for continuously optimizing and updating a reference model of the equipment performance by combining the actual operation data, the environmental change and the historical performance of the equipment based on the initial reference performance curve;

the deep learning analysis unit is used for analyzing the time sequence data by using a deep learning algorithm and identifying complex performance modes and trends;

The knowledge graph engine integrates expert knowledge, historical cases and equipment association information, and builds a comprehensive knowledge base to support intelligent diagnosis;

The abnormality diagnosis unit is used for accurately classifying the detected abnormality and analyzing root cause by combining a reference model, a deep learning analysis result and a knowledge map;

and the real-time decision support unit is used for generating operation suggestions and early warning information based on the diagnosis result and supporting quick decisions of operation and maintenance personnel.

4. The internet of things-based heating and ventilation equipment abnormality online monitoring system according to claim 3, wherein the accurate classification and root cause analysis of the detected abnormality are performed by combining a reference model, a deep learning analysis result and a knowledge graph, and the method comprises the following steps:

Comparing the real-time monitoring data with a dynamic reference model, identifying abnormal data points deviating from a normal range and performing preliminary classification;

Inputting the detected abnormal data points into a trained deep learning model, identifying a complex mode in time sequence data, and outputting the severity, duration and potential fault type of the abnormality;

Based on the preliminary classification and the deep learning analysis result, carrying out knowledge map query and matching so as to quickly find a historical case similar to the current abnormality;

comprehensively analyzing the reference model comparison result, the deep learning analysis output and the knowledge-graph matching information to generate a comprehensive and accurate abnormal description and a preliminary diagnosis result;

starting a root cause analysis reasoning engine based on comprehensive analysis information, automatically checking related sensor data, verifying a fault path of each root cause, and calculating probabilities of various root causes;

based on the reasoning and verification results, a diagnostic report is generated that includes a plurality of root causes.

5. The internet of things-based heating and ventilation equipment abnormality online monitoring system according to claim 4, wherein a root cause analysis inference engine is started based on comprehensive analysis information, related sensor data is automatically checked, a fault path of each root cause is verified, and probabilities of various root causes are calculated, comprising the steps of:

Collecting sensor data D and a preliminary abnormality detection result A, wherein D= { D ₁,d₂,…,d_n }, and the preliminary abnormality detection result A comprises abnormal equipment states and possible fault modes;

Reasoning the preliminary abnormal detection result A by applying the root cause fault model F, and calculating the likelihood P (A|f _i) of each fault mode F _i;

Acquiring the prior probability P (f _i) of each fault mode f _i;

And calculating the posterior probability of each fault mode according to the Bayesian theorem, wherein the formula is as follows: p (f _i∣A)＝P(A∣f_i)·P(f_i)/P (a), P (a) is the total probability of observing the preliminary anomaly detection result a, calculated by the weighted probability of all possible failure modes: p (a) = Σ _k＝1 ^mP(A∣f_k)·P(f_k), where m represents the total number of failure modes; p (f _k) represents the prior probability of the failure mode f _k, that is, the occurrence probability of the failure mode f _k in the case where the preliminary abnormality detection result a is not observed; p (a|f _k) represents the probability that the preliminary abnormality detection result a is observed in the case of the failure mode f _k;

Based on the calculated posterior probabilities, it is determined whether each failure mode is the root cause of A and the validity of the failure path is verified.

6. The internet of things-based heating and ventilation equipment abnormality online monitoring system according to claim 1, wherein the adaptive threshold management module comprises the following units:

The fuzzy logic controller converts quantitative input of the equipment state and the environmental condition into qualitative rules, and realizes smooth adjustment of the threshold value;

The genetic algorithm optimization engine continuously optimizes the threshold parameter combination through simulating the evolution process so as to adapt to different running conditions;

the multi-target evaluation unit is used for comprehensively evaluating the detection accuracy, the false alarm rate and the system efficiency and providing comprehensive performance indexes for the optimization process;

The dynamic threshold calculation unit is used for calculating and updating the abnormal detection threshold of each monitoring parameter in real time based on the output of the fuzzy logic controller and the genetic algorithm optimization engine;

The historical abnormal data analysis unit is used for analyzing the historical abnormal data and operation and maintenance feedback and providing long-term trend and seasonal change information for threshold adjustment;

a rapid response unit for rapidly adjusting a threshold value to maintain detection sensitivity of the system for an emergency or an abrupt change of operation condition;

And the threshold effect verification unit is used for continuously monitoring the detection effect after the threshold adjustment and feeding back the result to the genetic algorithm optimization engine to form closed-loop optimization.

7. The internet of things-based on-line monitoring system for anomalies of heating and ventilation equipment according to claim 6, wherein the anomaly detection threshold value of each monitored parameter is calculated and updated in real time by the following formula: t _j(t)＝T_j(t-1)+ΔT_j(t)+Δθ_j (T), wherein T _j (T) is the threshold value of the jth monitoring parameter at time tset; t _j (T-1) is the threshold value of the jth monitoring parameter at time T-1; deltaT _j (T) is the threshold adjustment amount output by the fuzzy logic controller, representing the adjustment amount based on the current equipment state and environmental conditions; Δθ _j (t) is a threshold adjustment amount of the genetic algorithm optimization engine output, and represents a parameter adjustment amount obtained based on the optimization algorithm.

8. The internet of things-based heating and ventilation equipment abnormality online monitoring system according to claim 6, wherein the specific steps of continuously monitoring the detection effect after the threshold adjustment and feeding back the result to the optimization engine to form closed loop optimization are as follows:

Continuously collecting and analyzing key indexes of abnormal detection, including indexes of detection rate, false alarm rate and false alarm rate, so as to evaluate the effectiveness of the current threshold value setting;

performing multidimensional evaluation on the detection effect after the threshold adjustment at regular intervals, wherein the multidimensional evaluation comprises analysis of detection accuracy, response time and resource utilization efficiency dimension of different types of anomalies;

performing difference analysis and reason inference on detection effects before and after threshold adjustment, identifying significant difference change, and automatically analyzing possible reasons;

based on the effect evaluation and the difference analysis results, generating a series of optimization suggestions, each suggestion accompanied by a quantitative estimate of the expected improvement effect;

Inputting the evaluation result and the optimization suggestion into a genetic algorithm optimization engine, and sequencing the optimization suggestion according to the expected effect, implementation difficulty and current state of the system;

And the genetic algorithm optimization engine gradually executes threshold adjustment and other optimization measures according to feedback and priority, and immediately enters a new monitoring period after each adjustment, so that the actual effect of adjustment is rapidly estimated.

9. The internet of things-based heating and ventilation equipment abnormality online monitoring system according to claim 8, wherein whether the difference is significant is determined by the following formula: wherein h represents a normalized measure of the difference in effect relative to the standard error before and after threshold adjustment; Δx is the difference in mean before and after threshold adjustment; σΔx is the standard error of the mean difference; e is the sample size of the threshold.

10. The internet of things-based heating and ventilation equipment abnormality online monitoring system according to any one of claims 1-9, wherein the predictive maintenance module comprises the following units:

The data integration engine is used for collecting historical fault data and equipment real-time operation data and preprocessing the data, so that a comprehensive data base is provided for predictive analysis;

A time series analysis unit for analyzing the equipment performance data by using an advanced time series algorithm and identifying a long-term trend, a periodic pattern and an abnormal change;

a machine learning prediction model that is trained based on historical fault data, predicting a type and time of fault that may occur to the device;

the health degree evaluation unit comprehensively considers the running state, the prediction result and the history maintenance record of the equipment, and calculates and dynamically updates the health index of the equipment;

The maintenance strategy generation unit is used for making personalized maintenance plans and suggestions for each device according to the health degree evaluation and prediction results;

The resource optimization scheduling unit is used for optimizing the overall scheduling and resource allocation of maintenance work by considering the factors such as maintenance requirements, human resources, spare part inventory and the like;

And the effect evaluation and feedback unit is used for tracking the actual effect of the maintenance activity and feeding back the result to the prediction model and the strategy generator so as to continuously improve the system performance.