CN110689171A - Turbine health state prediction method based on E-LSTM - Google Patents
Turbine health state prediction method based on E-LSTM Download PDFInfo
- Publication number
- CN110689171A CN110689171A CN201910837861.8A CN201910837861A CN110689171A CN 110689171 A CN110689171 A CN 110689171A CN 201910837861 A CN201910837861 A CN 201910837861A CN 110689171 A CN110689171 A CN 110689171A
- Authority
- CN
- China
- Prior art keywords
- data
- training
- lstm
- model
- prediction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 28
- 238000012549 training Methods 0.000 claims abstract description 40
- 230000002068 genetic effect Effects 0.000 claims abstract description 13
- 230000000694 effects Effects 0.000 claims abstract description 8
- 238000007781 pre-processing Methods 0.000 claims abstract description 7
- 238000012360 testing method Methods 0.000 claims abstract description 7
- 238000012795 verification Methods 0.000 claims abstract description 5
- 238000013528 artificial neural network Methods 0.000 claims description 20
- 230000006870 function Effects 0.000 claims description 13
- 238000005457 optimization Methods 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 5
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 3
- 230000004913 activation Effects 0.000 claims description 3
- 238000012790 confirmation Methods 0.000 claims description 3
- 230000008014 freezing Effects 0.000 claims description 3
- 238000007710 freezing Methods 0.000 claims description 3
- 210000002569 neuron Anatomy 0.000 claims description 3
- 238000013210 evaluation model Methods 0.000 claims description 2
- 238000012544 monitoring process Methods 0.000 abstract description 8
- 238000003745 diagnosis Methods 0.000 abstract description 7
- 238000012417 linear regression Methods 0.000 abstract description 2
- 238000005516 engineering process Methods 0.000 description 5
- 238000010248 power generation Methods 0.000 description 5
- 238000011160 research Methods 0.000 description 5
- 238000013473 artificial intelligence Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 241000282414 Homo sapiens Species 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 239000002131 composite material Substances 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 230000035515 penetration Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Human Resources & Organizations (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Strategic Management (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Data Mining & Analysis (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- Control Of Turbines (AREA)
Abstract
The invention provides a steam turbine health state prediction method based on E-LSTM. Collecting the operation data of the steam turbine from a sensor and preprocessing the operation data; feeding the preprocessed data into an LSTM network, and performing repeated iterative training; inputting a plurality of trained model parameters into a genetic algorithm to serve as an initial population, operating the genetic algorithm, and selecting model parameters with optimal effects; performing generalization performance verification on the optimal model by using more turbine operation data; and predicting the test data set according to the optimal model parameters, and evaluating the model error. The method can improve the accuracy of model prediction, avoid overfitting, realize multivariate linear regression prediction, ensure that the prediction model has better fitting effect on real data, greatly reduce errors of manpower monitoring, improve the fault diagnosis efficiency and make the fault be known and precedent. The method can be widely applied to the state management of various thermal power plants and nuclear power plants even steam turbines of ships.
Description
Technical Field
The invention relates to a health state prediction method, in particular to a health state prediction method of a steam turbine generator of a nuclear power and thermal power plant.
Background
According to data, the annual thermal power generation and nuclear power generation capacity of China accounts for nearly 80% of the total power generation capacity, and a steam turbine generator is one of core equipment in a thermal power generation and nuclear power generation system. Ensuring safe and stable operation of the turbonator has been one of the most important links in the power supply system. However, in the industrial 4.0 era, the conventional sensor and manual monitoring mode has many problems of high cost, low efficiency and the like, and an intelligent and efficient power supply system state prediction scheme is urgently needed.
From the current research results, the traditional understanding of the health state of the steam turbine by observing the sensor data has considerable subjectivity and sidedness, and the interpretation of the data is completely dependent on the experience of people. In the past decades, a rule-based expert system has been established by cumulatively summarizing a large number of experience with steam turbine operations. However, the expert system has the following significant disadvantages: (1) the relationships between the rules are opaque. The logical relationships between a large number of rules may be opaque and lack a hierarchical knowledge representation. (2) An inefficient search strategy. The inference engine searches for all rules in each cycle. When there are many rules, the system runs very slowly and large expert systems based on rules are not suitable for real-time applications. (3) There is no learning ability. General rule-based expert systems do not have the ability to learn from experience and are difficult to handle in special or emergency situations.
For the steam turbine generator unit, if the steam turbine generator unit is maintained regularly, the economic benefit is low, and if the steam turbine generator unit is maintained again after a fault occurs, the opportunity for preventing the fault loss from being further expanded is often missed and the fault loss is not paid. The technology based on the knowledge and experience of field experts in the past can not meet the requirement of safe and economic operation of a unit. The development of artificial intelligence technologies such as neural networks and the like and the rapid penetration of the artificial intelligence technologies into the engineering field bring new vitality to the fault state prediction technology, so that the modern diagnosis technology enters a brand-new stage. The artificial intelligence algorithm is realized without the need of rich priori knowledge of users, and fault characteristics can be directly mined from data, so that fault classification and state prediction can be carried out. The model obtained based on the artificial intelligence algorithm has the characteristics of small volume and strong mobility, is suitable for industrial fault diagnosis, and becomes an important research subject in the technical field of current fault diagnosis.
Summarizing the existing research results, the existing steam turbine health state monitoring system has the following problems to be solved:
(1) the manpower monitoring is with high costs, inefficient, and can't avoid human error.
(2) The fault is judged manually according to the sensor data, the fault is subjective, and the judgment result depends on the experience of a person. Moreover, it is difficult for human beings to fully discover the intrinsic relationship among various parameters, so that failure information cannot be fully interpreted. And the expert system is rigid, lacks real-time performance, does not have the ability of learning newly-discovered fault characteristics, and is difficult to deal with complicated and variable production environments.
(3) The existing fault monitoring mode can not give a sense of 'after-knowing' about the impending fault, and can not deal with the fault in enough time when the fault is found. And the mode of avoiding the fault occurrence by excessive maintenance and replacement in advance has low economic benefit.
Disclosure of Invention
The invention aims to provide the method for predicting the health state of the steam turbine based on the E-LSTM, which has high prediction accuracy, small error and high diagnosis efficiency.
The purpose of the invention is realized as follows:
step one, collecting the operation data of a steam turbine from a sensor and preprocessing the operation data;
feeding the preprocessed data into an LSTM network for repeated iterative training;
inputting a plurality of trained model parameters into a genetic algorithm as an initial population, operating the genetic algorithm, and selecting model parameters with optimal effects;
step four, using more turbine operation data to carry out generalization performance verification on the optimal model;
and step five, predicting the test data set according to the optimal model parameters, and evaluating the model error.
The present invention may further comprise:
1. preprocessing the sampled data, normalizing the sequence and expressing the sequence as Y, Y ═ Y0,y1,y2,…,yr,yr-1Inputting Y as training data into an initialized LSTM network to complete parameter learning, using an actual value as the input of the next step in each step of prediction of training stage time T epsilon (0, T), updating the state of the neuron, circulating the residual prediction,
let ht＝ytThe prediction method comprises the following steps:
in the formula, h is the value of an output gate at the previous layer, y is the input value of the current node, f is the weight of the forgetting gate output by the sigmoid activation function, and C is the output value of the forgetting gate and the input gate after confirmation updating and forgetting.
2. The training is divided into the following 3 training methods:
① for initial training, all neural networks are trained, and the networks are trained by setting a variable learning rate optimization cross-loss function through an Adam gradient descent algorithm and an SGD gradient descent algorithm;
② when new category is needed to be added as training data, on the premise of ① training result, setting small learning rate for main structure of LSTM network to learn, then freezing all neural network layers except the fully-connected layer, retraining the last fully-connected layer;
③ when new stations need to be controlled, using ① training result as pre-training model to activate all neural networks and set variable learning rate optimization cross loss function to train the network.
3. The method for selecting the optimal effect model parameters comprises the following steps: training a plurality of models according to data sampled at different time intervals, taking parameters of the models as initial populations, performing genetic algorithm iterative optimization, and selecting a parameter sequence of an optimal offspring as an optimal model.
4. The error of the evaluation model is specifically as follows: substituting the optimal model parameters into the LSTM, inputting a test set, and calculating the error between a predicted value and a true value; the error is calculated in two ways:
mean square error:
root mean square error:
wherein N is the number of data sets, YiIs a real data set, Yi *Is a set of predicted data that is,
and (4) according to the error calculation result, checking whether the model precision meets the requirement, and if not, continuing training and optimizing.
The invention provides a method for predicting the health state of a steam turbine based on a long-short term memory neural network (LSTM) aiming at the problem of monitoring the health state of the steam turbine, and the method is a method for predicting the health state of the steam turbine based on an improved long-short term memory neural network and combined with an evolutionary algorithm to optimize a model. The neural network is trained for multiple times to obtain a plurality of models, genetic algorithm optimization is carried out on model parameters obtained by multiple training, and the model with excellent prediction effect and best generalization capability is selected, so that the accuracy of model prediction is improved, and overfitting is avoided. By the LSTM neural network, the intrinsic relation among all parameters (pressure, vibration, temperature, rotating speed and the like) of the steam turbine system is fully explored, and the multiple linear regression prediction is realized. And then, by using a genetic algorithm, optimizing a plurality of trained LSTM model parameters, so that the prediction model has a better fitting effect on real data. The health condition of the turbonator is predicted by using the preferred model, so that the error of manpower monitoring can be greatly reduced, the fault diagnosis efficiency is improved, and the fault is known first and perceived. The method can be widely applied to the state management of various thermal power plants and nuclear power plants even steam turbines of ships.
In order to overcome the defects in the prior art, the invention provides a steam turbine health state prediction model E-LSTM on the basis of the research of the predecessor, namely model optimization is carried out by using a long-short term memory neural network (LSTM) training model and combining an Evolutionary algorithm (evolution algorithms) so as to achieve the purposes of improving the prediction accuracy and avoiding overfitting.
Drawings
FIG. 1 is a functional block diagram of a steam turbine condition prediction system.
Fig. 2 model training and preferred flow chart.
FIG. 3 is a flow chart of an implementation of steam turbine health prediction.
FIG. 4 is a diagram of the structure of E-LSTM.
The predicted value and actual value errors of the optimal model of fig. 5.
Detailed Description
The invention is described in more detail below by way of example.
The invention relates to a steam turbine health degree prediction method based on E-LSTM, the structure diagram of which is shown in figure 1, and the method comprises the steps of collecting steam turbine operation data and obtaining the distribution characteristics of the health condition of the steam turbine;
in order to overcome the defects in the prior art, the invention provides a steam turbine health state prediction model E-LSTM on the basis of the research of predecessors, namely model optimization is carried out by using a long-short term memory neural network (LSTM) training model and combining an Evolutionary algorithm (evolution algorithms) so as to achieve the purposes of improving the prediction accuracy and reducing overfitting, and the invention adopts the following steps to realize the state prediction of a steam turbine:
and step 01, collecting the operation data of the steam turbine from the sensor, and preprocessing the data.
And step 02, feeding the processed data into an LSTM network, and performing repeated iterative training.
And 03, inputting the trained multiple models into a genetic algorithm to serve as an initial population, operating the genetic algorithm, and selecting a model with the optimal effect.
And step 04, performing generalization performance verification on the optimal model by using more turbine operation data.
And step 05, predicting the test data set according to the optimal model, and evaluating the error of the model.
The step 01 is specifically as follows:
0101, arranging sensors at each monitoring point of the steam turbine, and performing verification and sensor data fusion on the data of the sensors of multiple types to obtain data which effectively and reliably reflects the operation condition of the steam turbine.
0102, sampling is carried out on the data, and the sampling time intervals are 5 minutes, 10 minutes, 15 minutes, 30 minutes and 60 minutes.
0103, taking 70% of the sampled data as a training set and 30% as a test set according to each time interval.
The step 02 specifically comprises:
following the data preprocessing described above, the sequences are normalized and denoted as Y, Y ═ Y0,y1,y2,…,yr,yr-1L. And inputting Y as training data into the initialized LSTM network to complete parameter learning. To predict the values of the L time steps, the conventional LSTM network bases each prediction on the prediction value of the previous step. The method is improved in that an actual value is used as the input of the next step in each step of prediction of the training stage time T epsilon (0, T), the state of a neuron is updated, gradient propagation of errors is reduced, and residual prediction is circulated.
Let ht＝ytAfter improvementThe prediction method comprises the following steps:
in the formula, h is the value of an output gate at the previous layer, y is the input value of the current node, f is the weight of the forgetting gate output by the sigmoid activation function, and C is the output value of the forgetting gate and the input gate after confirmation updating and forgetting.
The practical problems associated with steam turbine fault diagnosis can be classified into the following 3 training methods:
① for initial training, all neural networks are trained, and the networks are trained by setting a variable learning rate optimization cross-loss function through an Adam gradient descent algorithm and an SGD gradient descent algorithm;
② when new category is needed to be added as training data, on the premise of ① training result, setting small learning rate for main structure of LSTM network to learn, then freezing all neural network layers except the fully-connected layer, retraining the last fully-connected layer;
③ when a new measuring point needs to be controlled, using the training result of ① as a pre-training model to activate all neural networks and setting a variable learning rate optimization cross loss function to train the networks;
the step 03 is specifically as follows:
the parameters required to be optimized by the LSTM neural network prediction model comprise: the LSTM neural network hiding layer number, the time window step length, the training times and the forgetting rate Dropout. The model for optimizing the LSTM neural network by the genetic algorithm is to perform parameter combination optimization in a parameter search space by taking the minimum prediction error and the strongest generalization capability as objective functions to form a composite E-LSTM, and comprises the following steps:
step 0301: step S21, initializing and decoding the population;
0302, taking the mean square error of the LSTM neural network as a fitness function;
0303, carrying out selective cross mutation operation on the solved individuals;
0304, if the fitness function target value reaches the optimal value, carrying out the next step; otherwise, returning to the step 0303;
0305, obtaining a fitness function target value and an optimal parameter;
0306, calculating a prediction mean square error based on the optimal parameters;
0307, judging termination conditions, if the number of times of population iteration is satisfied, stopping calculation, and at the moment, combining the LSTM network global optimal parameters; otherwise, returning to the step 0306;
further, the step 04 specifically includes:
and (3) taking data sampled at different time intervals in the step (01), inputting the data into the optimal model in the step (03), obtaining the error between the predicted value and the actual value of the model, and turning to the step (02) if the error is larger than the threshold allowed by the system. The error is calculated as follows:
mean square error:
in the formula, N is the number of data sets, is a true data set, and is a predicted data set.
Finally, the step 05 specifically comprises:
using an optimal model to predict the health degree of the steam turbine on a prediction data set, and performing error calculation on the prediction data and actual data, wherein the error calculation adopts two indexes of mean square error and root mean square error to restore the prediction data for output, and in the prediction, the smaller the values of the mean square error and the root mean square error are, the higher the representative prediction precision is, wherein:
mean square error:
root mean square error:
wherein N is the number of data sets, YiIs a real data set, Yi *Is a predictive data set.
Claims (5)
1. A steam turbine health state prediction method based on E-LSTM is characterized by comprising the following steps:
step one, collecting the operation data of a steam turbine from a sensor and preprocessing the operation data;
feeding the preprocessed data into an LSTM network for repeated iterative training;
inputting a plurality of trained model parameters into a genetic algorithm as an initial population, operating the genetic algorithm, and selecting model parameters with optimal effects;
step four, using more turbine operation data to carry out generalization performance verification on the optimal model;
and step five, predicting the test data set according to the optimal model parameters, and evaluating the model error.
2. The method of predicting the health of a turbine based on E-LSTM of claim 1, wherein: preprocessing the sampled data, normalizing the sequence and expressing the sequence as Y, Y ═ Y0,y1,y2,…,yr,yr-1Inputting Y as training data into an initialized LSTM network to complete parameter learning, using an actual value as the input of the next step in each step of prediction of training stage time T epsilon (0, T), updating the state of the neuron, circulating the residual prediction,
let ht＝ytThe prediction method comprises the following steps:
inputting: y ═ Yo,y1,…,yT-1,yT}，
Output prediction value of { y'T+1,y′T+2,…,y′T+L}，
for t＝0,t≤T,t++：
Ct←inputgate←Ct-1,ht-1,yt,ft
ft←forgetgate←ht-1,yt
y′t←outputgate←ht-1,yt
Loss(yt,y′t)
end；
for l＝1,l＜＝L,l++：
CT+1←inputgate←CT+L-1,hT+1,y′T+l,fT+
fT+l←forgetgate←hT+L-1,y′T+l
y′T+l←outputgate←CT+l,hT+l-1,y′T+l
end；
In the formula, h is the value of an output gate at the previous layer, y is the input value of the current node, f is the weight of the forgetting gate output by the sigmoid activation function, and C is the output value of the forgetting gate and the input gate after confirmation updating and forgetting.
3. The method of predicting health of a turbine based on E-LSTM as claimed in claim 2, wherein said training is divided into the following 3 training methods:
① for initial training, all neural networks are trained, and the networks are trained by setting a variable learning rate optimization cross-loss function through an Adam gradient descent algorithm and an SGD gradient descent algorithm;
② when new category is needed to be added as training data, on the premise of ① training result, setting small learning rate for main structure of LSTM network to learn, then freezing all neural network layers except the fully-connected layer, retraining the last fully-connected layer;
③ when new stations need to be controlled, using ① training result as pre-training model to activate all neural networks and set variable learning rate optimization cross loss function to train the network.
4. The method of predicting the health of a steam turbine according to claim 3, wherein the method of selecting the optimal model parameters comprises: training a plurality of models according to data sampled at different time intervals, taking parameters of the models as initial populations, performing genetic algorithm iterative optimization, and selecting a parameter sequence of an optimal offspring as an optimal model.
5. The E-LSTM based turbine health prediction method of claim 4, further comprising: the error of the evaluation model is specifically as follows: substituting the optimal model parameters into the LSTM, inputting a test set, and calculating the error between a predicted value and a true value; the error is calculated in two ways:
mean square error:
root mean square error:
wherein N is the number of data sets, YiIs a real data set, YiIs a set of predicted data, and,
and (4) according to the error calculation result, checking whether the model precision meets the requirement, and if not, continuing training and optimizing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910837861.8A CN110689171A (en) | 2019-09-05 | 2019-09-05 | Turbine health state prediction method based on E-LSTM |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910837861.8A CN110689171A (en) | 2019-09-05 | 2019-09-05 | Turbine health state prediction method based on E-LSTM |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110689171A true CN110689171A (en) | 2020-01-14 |
Family
ID=69107835
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910837861.8A Pending CN110689171A (en) | 2019-09-05 | 2019-09-05 | Turbine health state prediction method based on E-LSTM |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110689171A (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111539154A (en) * | 2020-04-16 | 2020-08-14 | 哈尔滨工业大学 | Quantitative evaluation method for operation performance change of steam turbine |
| CN111804146A (en) * | 2020-06-29 | 2020-10-23 | 远光软件股份有限公司 | Intelligent ammonia injection control method and intelligent ammonia injection control device |
| CN111859807A (en) * | 2020-07-23 | 2020-10-30 | 润电能源科学技术有限公司 | Initial pressure optimizing method, device, equipment and storage medium for steam turbine |
| CN112257893A (en) * | 2020-09-08 | 2021-01-22 | 长春工业大学 | Complex electromechanical system health state prediction method considering monitoring error |
| CN112464563A (en) * | 2020-11-27 | 2021-03-09 | 河北建设投资集团有限责任公司 | Data mining method for steam turbine fault diagnosis |
| CN112667394A (en) * | 2020-12-23 | 2021-04-16 | 中国电子科技集团公司第二十八研究所 | Computer resource utilization rate optimization method |
| CN112836941A (en) * | 2021-01-14 | 2021-05-25 | 哈电发电设备国家工程研究中心有限公司 | Online health condition evaluation method for high-pressure steam turbine system of coal-electric unit |
| CN112990435A (en) * | 2021-03-22 | 2021-06-18 | 华北电力大学 | Long-short-time memory network power station fan fault early warning method and system |
| CN113530921A (en) * | 2020-04-14 | 2021-10-22 | 湖州职业技术学院 | Hydraulic machine fault diagnosis method based on ES-MLSTM |
| CN113761795A (en) * | 2021-08-17 | 2021-12-07 | 浙江工商大学 | Aircraft engine fault detection method and system |
| WO2021261659A1 (en) * | 2020-06-24 | 2021-12-30 | 주식회사 파워인스 | Artificial intelligence-based nondestructive inspection method and system |
| CN113867306A (en) * | 2021-07-30 | 2021-12-31 | 西安建筑科技大学 | Fault detection method and system for air conditioning system of subway station hall |
| CN114154266A (en) * | 2021-12-03 | 2022-03-08 | 合肥工业大学 | Gas turbine fault prediction method based on partial rank correlation flow causal structure learning |
| CN114492158A (en) * | 2021-12-22 | 2022-05-13 | 中译语通科技（青岛）有限公司 | Fault diagnosis method for motor of frequency converter |
-
2019
- 2019-09-05 CN CN201910837861.8A patent/CN110689171A/en active Pending
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113530921A (en) * | 2020-04-14 | 2021-10-22 | 湖州职业技术学院 | Hydraulic machine fault diagnosis method based on ES-MLSTM |
| CN111539154A (en) * | 2020-04-16 | 2020-08-14 | 哈尔滨工业大学 | Quantitative evaluation method for operation performance change of steam turbine |
| WO2021261659A1 (en) * | 2020-06-24 | 2021-12-30 | 주식회사 파워인스 | Artificial intelligence-based nondestructive inspection method and system |
| CN111804146A (en) * | 2020-06-29 | 2020-10-23 | 远光软件股份有限公司 | Intelligent ammonia injection control method and intelligent ammonia injection control device |
| CN111804146B (en) * | 2020-06-29 | 2022-07-01 | 远光软件股份有限公司 | Intelligent ammonia injection control method and intelligent ammonia injection control device |
| CN111859807A (en) * | 2020-07-23 | 2020-10-30 | 润电能源科学技术有限公司 | Initial pressure optimizing method, device, equipment and storage medium for steam turbine |
| CN112257893A (en) * | 2020-09-08 | 2021-01-22 | 长春工业大学 | Complex electromechanical system health state prediction method considering monitoring error |
| CN112464563A (en) * | 2020-11-27 | 2021-03-09 | 河北建设投资集团有限责任公司 | Data mining method for steam turbine fault diagnosis |
| CN112667394A (en) * | 2020-12-23 | 2021-04-16 | 中国电子科技集团公司第二十八研究所 | Computer resource utilization rate optimization method |
| CN112667394B (en) * | 2020-12-23 | 2022-09-30 | 中国电子科技集团公司第二十八研究所 | Computer resource utilization rate optimization method |
| CN112836941A (en) * | 2021-01-14 | 2021-05-25 | 哈电发电设备国家工程研究中心有限公司 | Online health condition evaluation method for high-pressure steam turbine system of coal-electric unit |
| CN112836941B (en) * | 2021-01-14 | 2024-01-09 | 哈电发电设备国家工程研究中心有限公司 | Online health condition assessment method for high-pressure system of steam turbine of coal motor unit |
| CN112990435A (en) * | 2021-03-22 | 2021-06-18 | 华北电力大学 | Long-short-time memory network power station fan fault early warning method and system |
| CN113867306A (en) * | 2021-07-30 | 2021-12-31 | 西安建筑科技大学 | Fault detection method and system for air conditioning system of subway station hall |
| CN113867306B (en) * | 2021-07-30 | 2024-08-16 | 西安建筑科技大学 | Method and system for detecting faults of subway station hall air conditioning system |
| CN113761795A (en) * | 2021-08-17 | 2021-12-07 | 浙江工商大学 | Aircraft engine fault detection method and system |
| CN114154266A (en) * | 2021-12-03 | 2022-03-08 | 合肥工业大学 | Gas turbine fault prediction method based on partial rank correlation flow causal structure learning |
| CN114154266B (en) * | 2021-12-03 | 2024-02-20 | 合肥工业大学 | Gas turbine fault prediction method based on bias rank correlation flow causal structure learning |
| CN114492158A (en) * | 2021-12-22 | 2022-05-13 | 中译语通科技（青岛）有限公司 | Fault diagnosis method for motor of frequency converter |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110689171A (en) | Turbine health state prediction method based on E-LSTM | |
| Ke et al. | Short-term electrical load forecasting method based on stacked auto-encoding and GRU neural network | |
| CN114282443B (en) | Residual service life prediction method based on MLP-LSTM supervised joint model | |
| CN112990556A (en) | User power consumption prediction method based on Prophet-LSTM model | |
| CN106709820A (en) | Power system load prediction method and device based on deep belief network | |
| CN112733417B (en) | Abnormal load data detection and correction method and system based on model optimization | |
| CN110119854A (en) | Voltage-stablizer water level prediction method based on cost-sensitive LSTM Recognition with Recurrent Neural Network | |
| CN112986827A (en) | Fuel cell residual life prediction method based on deep learning | |
| CN114218872B (en) | DBN-LSTM semi-supervised joint model-based residual service life prediction method | |
| CN111736084A (en) | Valve-regulated lead-acid storage battery health state prediction method based on improved LSTM neural network | |
| CN111461463A (en) | Short-term load prediction method, system and equipment based on TCN-BP | |
| CN113705922A (en) | Improved ultra-short-term wind power prediction algorithm and model establishment method | |
| CN116842337A (en) | Transformer fault diagnosis method based on LightGBM (gallium nitride based) optimal characteristics and COA-CNN (chip on board) model | |
| CN110796281B (en) | Wind turbine state parameter prediction method based on improved deep belief network | |
| CN114757330A (en) | Urban instantaneous water consumption prediction method based on LSTM | |
| CN114139783A (en) | Wind power short-term power prediction method and device based on nonlinear weighted combination | |
| CN111582588B (en) | Building energy consumption prediction method based on triple convolution fusion GRU | |
| Chen et al. | Remaining useful life prediction of turbofan engine based on temporal convolutional networks optimized by genetic algorithm | |
| CN113762591B (en) | Short-term electric quantity prediction method and system based on GRU and multi-core SVM countermeasure learning | |
| CN114548701B (en) | Full-measurement-point-oriented coupling structure analysis and estimation process early warning method and system | |
| CN112767692A (en) | Short-term traffic flow prediction system based on SARIMA-GA-Elman combined model | |
| CN117744495A (en) | Method for predicting service life of extra-large bearing driven by multiple models in different degradation stages | |
| CN112232570A (en) | Forward active total electric quantity prediction method and device and readable storage medium | |
| Ge et al. | Research on Temperature and Humidity Prediction Model of Granary Based on RNN-LSTM | |
| CN117034808A (en) | Natural gas pipe network pressure estimation method based on graph attention network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200114 |
|
| RJ01 | Rejection of invention patent application after publication |