CN114861522B - Precision manufacturing quality monitoring method and device based on artificial intelligence element learning technology - Google Patents
Precision manufacturing quality monitoring method and device based on artificial intelligence element learning technology Download PDFInfo
- Publication number
- CN114861522B CN114861522B CN202210371923.2A CN202210371923A CN114861522B CN 114861522 B CN114861522 B CN 114861522B CN 202210371923 A CN202210371923 A CN 202210371923A CN 114861522 B CN114861522 B CN 114861522B
- Authority
- CN
- China
- Prior art keywords
- meta
- learning
- optimization
- quality monitoring
- data
- 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.)
- Active
Links
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 165
- 238000000034 method Methods 0.000 title claims abstract description 153
- 238000012544 monitoring process Methods 0.000 title claims abstract description 144
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 61
- 238000005516 engineering process Methods 0.000 title claims abstract description 55
- 238000005457 optimization Methods 0.000 claims abstract description 106
- 238000012549 training Methods 0.000 claims abstract description 97
- 238000012360 testing method Methods 0.000 claims abstract description 56
- 238000012795 verification Methods 0.000 claims abstract description 52
- 238000012545 processing Methods 0.000 claims abstract description 28
- 238000004140 cleaning Methods 0.000 claims abstract description 17
- 230000003044 adaptive effect Effects 0.000 claims abstract description 13
- 238000010606 normalization Methods 0.000 claims abstract description 6
- 238000007781 pre-processing Methods 0.000 claims description 26
- 238000012806 monitoring device Methods 0.000 claims description 16
- 230000006870 function Effects 0.000 claims description 13
- 238000012512 characterization method Methods 0.000 claims description 9
- 238000001514 detection method Methods 0.000 claims description 7
- 230000005856 abnormality Effects 0.000 claims description 6
- 238000010924 continuous production Methods 0.000 claims description 4
- 238000009826 distribution Methods 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 10
- 238000003860 storage Methods 0.000 description 7
- 238000003754 machining Methods 0.000 description 6
- 238000004590 computer program Methods 0.000 description 5
- 239000000047 product Substances 0.000 description 5
- 238000007689 inspection Methods 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000000638 solvent extraction Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000010923 batch production Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 238000003908 quality control method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- 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
- G06N3/08—Learning methods
- G06N3/088—Non-supervised learning, e.g. competitive learning
-
- 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
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06395—Quality analysis or management
-
- 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Evolutionary Computation (AREA)
- Economics (AREA)
- Tourism & Hospitality (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- General Health & Medical Sciences (AREA)
- Development Economics (AREA)
- Educational Administration (AREA)
- Entrepreneurship & Innovation (AREA)
- Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Game Theory and Decision Science (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Operations Research (AREA)
- Medical Informatics (AREA)
- Primary Health Care (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Manufacturing & Machinery (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- General Factory Administration (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The application relates to the technical field of precision manufacturing, and provides a precision manufacturing quality monitoring method and device based on an artificial intelligence element learning technology, wherein the method comprises the following steps: acquiring time sequence data of a plurality of sensors under different process parameters in the precision manufacturing process, and performing data cleaning and normalization processing; dividing the data into a meta training set, a meta verification set and a meta test set; constructing a precision manufacturing process time sequence quality monitoring model based on meta learning; performing internal optimization and external optimization on a plurality of tasks to learn meta knowledge, and completing meta training and meta verification; using a model trained in the meta-training stage, and updating internal optimization parameters on new data by adopting a small amount of samples, so that the model is rapidly adapted to a new quality monitoring task; setting an adaptive threshold updating method, dynamically adapting to different scenes, and executing monitoring alarm processing. Can learn on a small number of samples rapidly, adapt to new tasks, improve the effect of quality monitoring.
Description
Technical Field
The application relates to the technical field of precision manufacturing, in particular to a precision manufacturing quality monitoring method and device based on an artificial intelligence element learning technology.
Background
Modern manufacturing technology is one of the important means for developing economy in a country, and precision manufacturing technology is an important branch of modern manufacturing technology, and development and research of the precision manufacturing technology are also highly valued by the country. The development and popularization of precision manufacturing and ultra-precision machining technology improves the machining precision and technical level of the whole manufacturing industry, and generally improves the quality, performance and competitiveness of mechanical products. The precision manufacturing technology is widely applied in the fields of ultra-precision part machining, ultra-precision special-shaped part machining, ultra-precision optical part machining, micro-electromechanical system device machining and the like.
Precision manufacturing processes are a typical multi-variable, multi-process, multi-stage operating industrial process, and quality control problems have been a focus of academic and industrial attention. The precision manufacturing process is a typical batch production process, and is composed of a plurality of stages, and is repeated in accordance with the stages. The process state changes with time and does not have a steady-state operating point. Resulting in a process that exhibits strong non-linear and time-varying characteristics, linear time-varying models have not adequately described the process.
The main aim of the monitoring of the precise manufacturing process is to improve the production quality of products, discover abnormality in time in the production stage and adjust equipment. This process involves a number of factors, during the production process, monitoring the variables in the production process by means of sensors, which reflect the production process, the anomalies of the variables and the quality of the product having a certain correlation. The quality of the product is judged by analyzing the monitoring variable, so that a large amount of manpower and material resources can be saved in the quality inspection link.
However, the existing linear or supervised algorithm is poor in precision manufacturing time sequence quality monitoring effect, especially precision manufacturing equipment is high in data dimension because process parameters are frequently adjusted, the collected data cannot be generalized well after the process parameters are adjusted, a large amount of data is required to be retrained, real-time monitoring cannot be achieved, all sampling inspection cannot be achieved in a quality inspection link, a large amount of monitoring variable data has no label, and quality monitoring cannot be achieved by the supervised algorithm.
Disclosure of Invention
In order to solve the technical problems in the prior art, the main purpose of the application is to provide a precision manufacturing quality monitoring method and device based on an artificial intelligence element learning technology, so as to solve the defects in the prior art.
In order to achieve the above object, the present application provides a precision manufacturing quality monitoring method based on artificial intelligence meta learning technology, comprising:
Collecting time sequence data in the precise manufacturing process, and cleaning and normalizing the time sequence data to obtain preprocessing data;
Dividing the preprocessing data into a meta training set, a meta verification set and a meta test set;
Constructing a precision manufacturing process time sequence quality monitoring model based on meta learning, and learning meta knowledge through two optimization processes of internal optimization and external optimization on a plurality of tasks through a meta training set and a meta verification set to complete meta training and meta verification;
updating internal optimization parameters of the model after meta training by adopting the samples on the meta test set to obtain a quality monitoring model adapting to a new quality monitoring task;
and setting a self-adaptive threshold updating method based on the quality monitoring model, dynamically adapting to different scenes, and executing monitoring alarm processing.
Optionally, the collecting the time sequence data in the precision manufacturing process, and cleaning and normalizing the time sequence data to obtain the preprocessing data includes:
Collecting the data of a plurality of monitoring sensors in the precise manufacturing process;
The method comprises the steps that collection of a plurality of variables under different process parameters is synchronously carried out, and data sampled by each variable in each complete precision manufacturing process period are uploaded in real time to obtain time sequence data;
and cleaning the uploaded time sequence data, removing data which do not meet the requirements, and separating and processing each period of data according to the acquired parameters into standard data with equal length according to the stages.
Optionally, the dividing the preprocessing data into a meta training set, a meta verification set and a meta test set further includes:
The preprocessing data is divided into three parts, namely a meta training set, a meta verification set and a meta testing set, wherein the meta training set, the meta verification set and the meta testing set are in a set form, each set is divided into a supporting set and a query set, the supporting set is trained, and the query set is tested;
the meta training set consists of a plurality of precision manufacturing process time sequence quality detection tasks, a meta learner learns on a plurality of support set tasks for quality monitoring of each task, learns to process meta knowledge of precision manufacturing problems, and verifies on the verification set; the meta-test set contains precision manufacturing process data.
Optionally, the building of the precision manufacturing process time sequence quality monitoring model based on meta-learning includes:
Constructing a meta learning model based on an unsupervised time sequence quality monitoring algorithm;
The meta learning model is divided into an external network comprising learning meta knowledge and an internal network sensitive to tasks based on an unsupervised meta learning algorithm;
The external network is used as an external circulation parameter of meta-learning and is used for learning initial characterization of time sequence data in a precision manufacturing process, and has the parameter which is most sensitive to the distribution of new task learning domains;
the internal network is used as an internal circulation parameter of meta learning, and based on the characterization of the external network, gradient adjustment based on samples is used.
Optionally, the learning of the meta-knowledge by performing two optimization processes, i.e., internal optimization and external optimization, on a plurality of tasks, completes meta-training and meta-verification, including:
The meta training stage comprises an internal optimization process and an external optimization process, wherein the external optimization process optimizes the parameters theta 1 of the element-learning meta-knowledge external network, and the internal optimization process optimizes the parameters theta 1 of the internal network;
In the internal optimization stage, the external network parameter theta is fixed, the internal network parameter theta 'is optimized by adopting a learning rate alpha and a gradient descent algorithm on a time sequence monitoring task T i in the precise manufacturing process, and after training on all the tasks is completed, the internal network parameter theta' is fixed, and primary knowledge, namely the external network parameter theta, is updated by the learning rate beta.
Optionally, the learning of meta-knowledge by performing two optimization processes, i.e. internal optimization and external optimization, on a plurality of tasks, completes meta-training and meta-verification, and further includes:
f θ is a model trained in an unsupervised mode, a loss function is defined on a training set of one task T i, and an external network parameter theta is updated to an internal network parameter theta' by adopting a learning rate alpha in a gradient updating mode; calculating a difference between the reconstructed sample and the actual sample by L (f θ(xj),yj);
L(fθ(xj),yj)＝λ1L1(fθ(xj),yj)+λ2LSSIM(fθ(xj),yj)
The goal of meta-learning is to learn the optimization parameter θ' over different tasks by equation 1, minimizing the overall loss over all tasks, meta-learning overall loss function is defined as:
Optionally, updating internal optimization parameters of the model after meta training by using the samples on the meta test set to obtain a quality monitoring model adapting to a new quality monitoring task, including:
And (3) updating the internal optimization parameters theta by adopting K samples on the new abnormality detection task S new in the meta-test set by using the initial parameters of the model trained in the meta-training stage, so that the model is quickly adapted to the new task, and testing on the test set of S new.
Optionally, based on the quality monitoring model, an adaptive threshold updating method is set, different scenes are dynamically adapted, and monitoring alarm processing is executed, including:
And executing a monitoring task on the data set by adopting the trained meta-learning model, calculating the error score of each sample, and setting a self-adaptive threshold updating method to monitor new data so that the threshold is adapted to the offset condition generated along with continuous production of precision manufacturing.
In addition, in order to achieve the above object, the present application also provides a precision manufacturing quality monitoring device based on an artificial intelligence meta-learning technique, the precision manufacturing quality monitoring device based on the artificial intelligence meta-learning technique comprising: the data acquisition module is used for acquiring time sequence data in the precise manufacturing process, and cleaning and normalizing the time sequence data to obtain preprocessing data; the data dividing module is used for dividing the preprocessing data into a meta training set, a meta verification set and a meta test set; the modeling training module is used for constructing a precision manufacturing process time sequence quality monitoring model based on meta-learning, learning meta-knowledge through two optimization processes of internal optimization and external optimization on a plurality of tasks through the meta-training set and the meta-verification set, and finishing meta-training and meta-verification; the meta-test module is used for updating internal optimization parameters of the model after meta-training by adopting the samples on the meta-test set to obtain a quality monitoring model adapting to a new quality monitoring task; and the quality monitoring module is used for setting a self-adaptive threshold updating method based on the quality monitoring model, dynamically adapting to different scenes and executing monitoring alarm processing.
Optionally, the precision manufacturing quality monitoring device based on the artificial intelligence meta-learning technology further comprises a model construction module for constructing a precision manufacturing process time sequence quality monitoring model based on meta-learning.
Optionally, the precision manufacturing quality monitoring device based on the artificial intelligence meta-learning technology further comprises a meta-training module, wherein the meta-training module is used for learning meta-knowledge in two optimization processes of internal optimization and external optimization on a plurality of tasks, and meta-training and meta-verification are completed.
In addition, in order to achieve the above object, the present application also provides a precision manufacturing quality monitoring apparatus based on an artificial intelligence meta-learning technology, which includes a processor, a memory, and a precision manufacturing quality monitoring program based on an artificial intelligence meta-learning technology stored on the memory and executable by the processor, wherein the precision manufacturing quality monitoring program based on an artificial intelligence meta-learning technology implements the steps of the precision manufacturing quality monitoring method based on an artificial intelligence meta-learning technology as described above when being executed by the processor.
In addition, in order to achieve the above object, the present application also provides a computer readable storage medium having stored thereon a precision manufacturing quality monitoring program based on an artificial intelligence element learning technique, wherein the precision manufacturing quality monitoring program based on the artificial intelligence element learning technique implements the steps of the precision manufacturing quality monitoring method based on the artificial intelligence element learning technique as described above when being executed by a processor.
The application provides a precision manufacturing quality monitoring method and device based on an artificial intelligence meta-learning technology, which can conveniently standardize and unify time sequence data generated by various types of precision manufacturing equipment by a data preprocessing method and can directly utilize a meta-learning model to train and test; through the learning of a meta learner, the model learns the meta knowledge of the precision manufacturing industry, faces a new precision manufacturing scene and is quickly generalized to the new scene; by setting the adaptive threshold method, the quality monitoring model is still applicable even if the process time sequence data of the precision manufacturing equipment deviates.
These and other aspects of the application will be more readily apparent from the following description of the embodiments. It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the application as claimed.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following description will briefly introduce the drawings that are needed in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the present application. In the drawings:
FIG. 1 is a flow chart of a precision manufacturing quality monitoring method based on an artificial intelligence element learning technology;
FIG. 2 is a schematic flow chart of a first embodiment of a precision manufacturing quality monitoring method based on artificial intelligence element learning technology according to the present application;
FIG. 3 is a schematic flow chart of the data acquisition and preprocessing stage of the precision manufacturing quality monitoring method based on the artificial intelligence element learning technology in FIG. 2;
FIG. 4 is a schematic diagram of a data partitioning stage of the precision manufacturing quality monitoring method based on the artificial intelligence element learning technique of FIG. 2;
FIG. 5 is a schematic diagram of a model building stage of the precision manufacturing quality monitoring method based on the artificial intelligence element learning technique of FIG. 2;
FIG. 6 is a schematic diagram of a meta-training stage of the method of FIG. 2 for monitoring quality of precision manufacturing based on artificial intelligence meta-learning technique;
FIG. 7 is a schematic diagram of a meta-test stage of the method for monitoring quality of precision manufacturing of the present application in FIG. 2 based on artificial intelligence meta-learning technology;
FIG. 8 is a schematic diagram of a quality monitoring stage of the precision manufacturing quality monitoring method based on the artificial intelligence element learning technique of FIG. 2;
FIG. 9 is a system block diagram of a precision manufacturing quality monitoring device based on artificial intelligence meta-learning technology in a second embodiment of the present application;
FIG. 10 is a schematic diagram of functional blocks of a modeling training module in the precision manufacturing quality monitoring device of FIG. 9 based on artificial intelligence meta-learning technology of the present application;
FIG. 11 is a schematic diagram of a precision manufacturing quality monitoring apparatus based on artificial intelligence meta-learning technology in a third embodiment of the present application.
The achievement of the objects, functional features and advantages of the present application will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
Reference will now be made in detail to the present embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein the accompanying drawings are used to supplement the description of the written description so that one can intuitively and intuitively understand each technical feature and overall technical scheme of the present invention, but not to limit the scope of the present invention.
In the description of the present invention, a number means one or more, a number means two or more, and greater than, less than, exceeding, etc. are understood to not include the present number, and above, below, within, etc. are understood to include the present number.
In the description of the present invention, the continuous reference numerals of the method steps are used for facilitating examination and understanding, and by combining the overall technical scheme of the present invention and the logic relationships between the steps, the implementation sequence between the steps is adjusted without affecting the technical effect achieved by the technical scheme of the present invention.
In the description of the present invention, unless explicitly defined otherwise, terms such as arrangement and the like should be construed broadly, and those skilled in the art can reasonably determine the specific meaning of the terms in the present invention in combination with the specific contents of the technical scheme.
In the embodiment of the application, a precision manufacturing quality monitoring method and a device based on an artificial intelligence element learning technology are provided, wherein the precision manufacturing quality monitoring method based on the artificial intelligence element learning technology can be applied to a precision manufacturing quality monitoring device based on the artificial intelligence element learning technology, and the precision manufacturing quality monitoring device based on the artificial intelligence element learning technology can be a device with display and processing functions such as a PC, a portable computer and a mobile terminal, but is not limited to the method.
Referring to fig. 1, the method for monitoring precision manufacturing quality based on artificial intelligence element learning technology includes the following steps S10-S50:
s10, collecting time sequence data in the precise manufacturing process, and cleaning and normalizing the time sequence data to obtain preprocessing data;
s20, dividing the preprocessing data into a meta training set, a meta verification set and a meta test set;
S30, constructing a precision manufacturing process time sequence quality monitoring model based on meta learning, and learning meta knowledge through two optimization processes of internal optimization and external optimization on a plurality of tasks through the meta training set and the meta verification set to complete meta training and meta verification;
S40, updating internal optimization parameters of the model after meta training by adopting the samples on the meta test set to obtain a quality monitoring model adapting to a new quality monitoring task;
s50, setting an adaptive threshold updating method based on the quality monitoring model, dynamically adapting to different scenes, and executing monitoring alarm processing.
In some embodiments of the present application, the collecting time series data in the precision manufacturing process, and performing cleaning and normalization processing on the time series data to obtain pre-processing data includes:
Collecting the data of a plurality of monitoring sensors in the precise manufacturing process;
The method comprises the steps that collection of a plurality of variables under different process parameters is synchronously carried out, and data sampled by each variable in each complete precision manufacturing process period are uploaded in real time to obtain time sequence data;
and cleaning the uploaded time sequence data, removing data which do not meet the requirements, and separating and processing each period of data according to the acquired parameters into standard data with equal length according to the stages.
In some embodiments of the present application, the partitioning the preprocessing data into a meta training set, a meta verification set, and a meta test set further includes: the preprocessing data is divided into three parts, namely a meta training set, a meta verification set and a meta testing set, wherein the meta training set, the meta verification set and the meta testing set are in a set form, each set is divided into a supporting set and a query set, the supporting set is trained, and the query set is tested;
the meta training set consists of a plurality of precision manufacturing process time sequence quality detection tasks, a meta learner learns on a plurality of support set tasks for quality monitoring of each task, learns to process meta knowledge of precision manufacturing problems, and verifies on the verification set; the meta-test set contains precision manufacturing process data.
In some embodiments of the application, the building of the precision manufacturing process time sequence quality monitoring model based on meta-learning comprises:
Constructing a meta learning model based on an unsupervised time sequence quality monitoring algorithm;
The meta learning model is divided into an external network comprising learning meta knowledge and an internal network sensitive to tasks based on an unsupervised meta learning algorithm;
The external network is used as an external circulation parameter of meta-learning and is used for learning initial characterization of time sequence data in a precision manufacturing process, and has the parameter which is most sensitive to the distribution of new task learning domains;
the internal network is used as an internal circulation parameter of meta learning, and based on the characterization of the external network, gradient adjustment based on samples is used.
In some embodiments of the present application, learning meta-knowledge for both internal and external optimization processes across multiple tasks, completing meta-training and meta-verification, includes:
The meta training stage comprises an internal optimization process and an external optimization process, wherein the external optimization process optimizes the parameters theta 'of the element-learning meta-knowledge external network, and the internal optimization process optimizes the parameters theta' of the internal network;
In the internal optimization stage, the external network parameter theta is fixed, the internal network parameter theta 'is optimized by adopting a learning rate alpha and a gradient descent algorithm on a time sequence monitoring task T i in the precise manufacturing process, and after training on all the tasks is completed, the internal network parameter theta' is fixed, and primary knowledge, namely the external network parameter theta, is updated by the learning rate beta.
In some embodiments of the present application, the learning of meta-knowledge by performing both internal optimization and external optimization on a plurality of tasks, completing meta-training and meta-verification, further comprises:
f θ is a model trained in an unsupervised mode, a loss function is defined on a training set of one task T i, and an external network parameter theta is updated to an internal network parameter theta' by adopting a learning rate alpha in a gradient updating mode; calculating a difference between the reconstructed sample and the actual sample by L (f θ(xj),yj);
L(fθ(xj),yj)＝λ1L1(fθ(xj),yj)+λ2LSSIM(fθ(xj),yj)
The goal of meta-learning is to learn the optimization parameter θ' over different tasks by equation 1, minimizing the overall loss over all tasks, meta-learning overall loss function is defined as:
In some embodiments of the present application, updating internal optimization parameters of the meta-trained model using the samples on the meta-test set to obtain a quality monitoring model adapted to a new quality monitoring task, including:
And (3) updating the internal optimization parameters theta' by adopting K samples on the new abnormality detection task S new in the meta-test set by using the initial parameters of the model trained in the meta-training stage, so that the model is quickly adapted to the new task, and testing on the test set of S new.
In some embodiments of the present application, an adaptive threshold updating method is set based on the quality monitoring model, and the method dynamically adapts to different scenes, and performs monitoring alarm processing, including:
And executing a monitoring task on the data set by adopting the trained meta-learning model, calculating the error score of each sample, and setting a self-adaptive threshold updating method to monitor new data so that the threshold is adapted to the offset condition generated along with continuous production of precision manufacturing.
Some embodiments of the present application are described in detail below with reference to the accompanying drawings. The following embodiments and features of the embodiments may be combined with each other without conflict.
Referring to fig. 2, fig. 2 is a flow chart of a precision manufacturing quality monitoring method based on an artificial intelligence element learning technique according to a first embodiment of the application. In the embodiment of the application, the precision manufacturing quality monitoring method based on the artificial intelligence element learning technology comprises the following steps S100 to S600:
S100, data acquisition and preprocessing: acquiring time sequence data of a plurality of sensors under different process parameters in the precision manufacturing process, and performing data cleaning and normalization processing;
S200, data dividing stage: dividing the data into a meta training set, a meta verification set and a meta test set;
s300, model establishment: constructing a precision manufacturing process time sequence quality monitoring model based on meta learning;
S400, meta training: performing internal optimization and external optimization on a plurality of tasks to learn meta knowledge, and completing meta training and meta verification;
S500, a meta-test stage: using a model trained in the meta-training stage, and updating internal optimization parameters on new data by adopting a small amount of samples, so that the model is rapidly adapted to a new quality monitoring task;
s600, quality monitoring: setting an adaptive threshold updating method, dynamically adapting to different scenes, and executing monitoring alarm processing.
In the embodiment of the present application, referring to fig. 3, in the data acquisition and preprocessing stage, steps S110 to S120 are further included:
S110, collecting time series data of a precise manufacturing process through various sensors;
S120, cleaning, grading, storing and preprocessing the acquired data into standard data.
By way of example, time series data of a plurality of sensors under different process parameters in the precision manufacturing process are collected, the number of actually collected variables is determined according to the variables actually required to be monitored, for example, the abnormality of a product is strongly related to some variables, or some variables reflect the quality of the precision manufacturing production. The collection of multiple variables is synchronized, each time a complete precision manufacturing process is passed, referred to as a cycle, or cycle, each variable samples a certain amount of data within each cycle, e.g., 500 points per cycle. Because the precision manufacturing time series data is easy to have the problems of deletion, repetition, multiple stages and the like, the data needs to be cleaned, and the data which does not meet the requirements, such as cycle repetition or cycle interruption, is deleted, and the data of each period is segmented according to the collected parameters and processed into standard data with equal length according to the stages. If the data of each stage is processed into a sequence with equal length, the comparison of different data in the same dimension is convenient.
In the embodiment of the present application, referring to fig. 4, the data dividing stage includes steps S210 to S220, and the flow is as follows:
S210, dividing the data into a meta training set, a meta verification set and a meta test set;
s220, the data of each set is further divided into a support set and a query set.
Illustratively, the data is divided into a meta-training set, a meta-validation set, and a meta-test set according to a task ratio, such as 6:2:2. They are presented in aggregate form. Each part is further divided into a support set and a query set, for example, the ratio of the support set to the query set is 8:2, and the support set is trained on the query set and tested on the query set. The meta-training set is composed of a plurality of precision manufacturing process time sequence quality monitoring tasks, such as different process time sequence data collected on different precision manufacturing equipment and precision manufacturing process time sequence data under different set parameters. Each combined process setting parameter corresponds to a task T, and the meta-learner learns on a plurality of support set tasks in an attempt to achieve each task quality monitoring objective, which aims at learning meta-knowledge that deals with precision manufacturing problems. The meta learner verifies on the verification set. The meta-test set contains new precision manufacturing process data, such as new data collected after tuning, and set parameter changes.
Referring to fig. 5, the model building stage includes steps S310 to S320, and the flow is as follows:
s310, constructing an unsupervised precision manufacturing process time sequence quality monitoring model based on meta learning;
s320, dividing the meta learning model into an inner loop network structure and an outer loop network structure.
Illustratively, an unsupervised precision manufacturing process time sequence quality monitoring model based on meta-learning is constructed. The time sequence data of the precise manufacturing process is generally provided with no tags or a very small number of tags, and the meta-learning model adopts an unsupervised time sequence quality monitoring algorithm. The unsupervised meta-learning algorithm is functionally divided into an external network comprising learning meta-knowledge and an internal network sensitive to tasks. The external network is used as an external circulation parameter of meta-learning, learns the initial characterization of time sequence data of the precise manufacturing process, and has the parameter which is most sensitive to the distribution of new task learning domains; the internal network is used as an internal circulation parameter of meta-learning, is based on the characterization of the external network, and can be efficiently adjusted by using gradient update based on a small number of samples, and extremely small parameter variation can bring about great improvement to the model. Because the precision manufacturing process uses sensor time series data, the external network and the internal network structure can use a neural network structure which is more suitable for processing time series, such as LSTM and one-dimensional convolution neural network.
The external cyclic parameters of meta-learning learn an optimal characterization of parameters by learning quality inspection meta-knowledge about the timing of the precision manufacturing process over a large number of tasks. The internal network acts as an internal optimization parameter for meta-learning. In the new scenario, the parameters of the optimal external circulation network are fixed, and the internal circulation parameters are updated rapidly through a small number of sample training, so that the model can monitor new precision manufacturing process data.
Referring to fig. 6, the meta training phase includes steps S410 to S420, and the flow is as follows:
S410, optimizing an external circulation structure of the model, and learning element knowledge;
s420, optimizing the internal circulation structure of the model, and adapting to new tasks.
Illustratively, the meta-knowledge is learned by performing two optimization processes, namely internal optimization and external optimization, on a plurality of tasks, so as to complete meta-training and meta-verification. The meta training stage comprises two optimization processes, namely internal optimization and external optimization, wherein the external optimization forms meta-knowledge external network parameters theta of meta learning, and the internal optimization optimizes parameters theta' of an internal network.
In the internal optimizing stage, the external network parameter theta is fixed, the internal network parameter theta 'is optimized by adopting a learning rate alpha through a gradient descent algorithm on a time sequence monitoring task T i in a certain precision manufacturing process, and after training on all tasks is finished, the internal network parameter theta' is fixed, and primary knowledge, namely the external network parameter theta, is updated through the learning rate beta.
Eq.1：
F θ is a model trained in an unsupervised mode, a loss function is defined on a training set of one task T i, and θ is updated to θ 'by adopting a learning rate α in a gradient update mode, specifically, parameters θ' of an internal network are updated. L (f θ(xj),yj) calculates the difference between the reconstructed sample and the actual sample. Commonly used are L1, L2 and SSIM loss, and a plurality of loss can be combined with different weights. Such as
L(fθ(xj),yj)＝λ1L1(fθ(xj),yj)+λ2LSSIM(fθ(xj),yj)
The goal of meta-learning is to learn the optimization parameter θ' over different tasks by eq.1, minimize the loss of eq.3 over all tasks, and the meta-learning overall loss function is defined as:
Eq.3：
referring to fig. 7, the meta-test stage includes steps S510 to S520, and the flow is as follows:
s510, updating internal optimization parameters by adopting a small amount of samples on a new task;
s520, predicting all cycle data under the set parameters by using the trained model.
Illustratively, using a model trained during the meta-training phase, the meta-learner at this time is considered to have the optimal external circulation parameter θ *. Updating internal optimization parameters with a small number of samples on the new task to adapt the model quickly to the new quality monitoring task includes: the initial parameters of the model trained by the meta-training stage are used, a small number of samples (K) are adopted on a new abnormality detection task S new in the meta-test set, and the internal optimization parameters theta' are updated through Eq.3, so that the model is quickly adapted to a new task. The test is performed on the test set of S new.
Referring to fig. 8, the quality monitoring stage includes steps S610 to S620, and the flow is as follows:
s610, dynamically updating a threshold by adopting an adaptive threshold updating method;
s620, monitoring alarm processing is executed under the current threshold value.
Exemplary, an adaptive threshold updating method is set, different scenes are dynamically adapted, and monitoring alarm processing is executed. The self-adaptive threshold adopts a mode of combining local area with global threshold, and a complex threshold strategy is set to cope with a complex precision manufacturing process, for example, as the precision manufacturing process is continued, the temperature of the charging barrel is continuously increased, temperature drift can be generated, so that the trend of other monitoring variables is influenced, and if a fixed threshold is adopted, the normal condition can be predicted as abnormal. Thus employing a dynamic threshold update strategy. The threshold is calculated using a sliding window, as based on the calculated reconstruction error, and the window size is empirically set to 30C (C represents 30 cycles) and the step size is set to 2C. The sliding window size determines the number of historical anomaly scores that evaluate the current threshold. For each sliding window, a simple static threshold is used, as defined as 3 standard deviations from the window mean. And executing a monitoring task on the data set by adopting the trained meta-learning model, calculating the error score of each sample, and monitoring new data according to the self-adaptive threshold updating method so as to adapt the threshold to the offset condition generated along with continuous production of precision manufacturing.
Referring to fig. 9, fig. 9 is a system block diagram of a precision manufacturing quality monitoring apparatus based on an artificial intelligence meta-learning technique according to a second embodiment of the present application.
In an embodiment of the present application, the precision manufacturing quality monitoring device 500 based on the artificial intelligence meta-learning technology includes:
The data acquisition module 501 is used for acquiring time sequence data in the precision manufacturing process, and cleaning and normalizing the time sequence data to obtain preprocessing data;
The data dividing module 502 is configured to divide the preprocessed data into a meta training set, a meta verification set and a meta test set;
the modeling training module 503 is configured to construct a precision manufacturing process time sequence quality monitoring model based on meta-learning, and perform internal optimization and external optimization on a plurality of tasks through the meta-training set and the meta-verification set to learn meta-knowledge, thereby completing meta-training and meta-verification;
The meta-test module 504 is configured to update internal optimization parameters of the meta-trained model by using the samples on the meta-test set, so as to obtain a quality monitoring model adapted to a new quality monitoring task;
The quality monitoring module 505 is configured to set an adaptive threshold updating method based on the quality monitoring model, dynamically adapt to different scenes, and execute monitoring alarm processing.
In some embodiments, referring to fig. 10, the precision manufacturing quality monitoring device based on the artificial intelligence meta-learning technology further includes a model construction module 5031 for constructing a precision manufacturing process time sequence quality monitoring model based on meta-learning;
the meta training module 5032 is configured to perform two optimization processes, i.e., internal optimization and external optimization, on a plurality of tasks to learn meta knowledge, thereby completing meta training and meta verification.
The modules in the precision manufacturing quality monitoring device based on the artificial intelligence element learning technology correspond to the steps in the embodiment of the precision manufacturing quality monitoring method based on the artificial intelligence element learning technology, and the functions and the implementation process of the precision manufacturing quality monitoring device based on the artificial intelligence element learning technology are not described in detail herein.
Referring to fig. 11, a device diagram according to an embodiment of the present application is shown. The third embodiment of the present application provides a precision manufacturing quality monitoring device based on an artificial intelligence meta-learning technology, which includes a memory 100 and a processor 200, wherein the processor 200 stores a computer program for executing the steps in the embodiment of the precision manufacturing quality monitoring method based on the artificial intelligence meta-learning technology:
Data acquisition and pretreatment: acquiring time sequence data of a plurality of sensors under different process parameters in the precision manufacturing process, and performing data cleaning and normalization processing;
data dividing stage: dividing the data into a meta training set, a meta verification set and a meta test set;
Model establishment: constructing a precision manufacturing process time sequence quality monitoring model based on meta learning;
Meta training stage: performing internal optimization and external optimization on a plurality of tasks to learn meta knowledge, and completing meta training and meta verification;
meta-test stage: using a model trained in the meta-training stage, and updating internal optimization parameters on new data by adopting a small amount of samples, so that the model is rapidly adapted to a new quality monitoring task;
Quality monitoring: setting an adaptive threshold updating method, dynamically adapting to different scenes, and executing monitoring alarm processing.
It should be appreciated that the Processor may be a central processing unit (Central Processing Unit, CPU), it may also be other general purpose processors, digital signal processors (DIGITAL SIGNAL Processor, DSP), application SPECIFIC INTEGRATED Circuit (ASIC), field-Programmable gate array (Field-Programmable GATE ARRAY, FPGA) or other Programmable logic device, discrete gate or transistor logic device, discrete hardware components, or the like. Wherein the general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
Wherein the processor is configured to execute a computer program stored in the memory to implement various embodiments of the present application of a precision manufacturing quality monitoring method based on artificial intelligence meta-learning techniques.
It should be appreciated that the method steps in embodiments of the present invention may be implemented or carried out by computer hardware, a combination of hardware and software, or by computer instructions stored in non-transitory computer-readable memory. The method may use standard programming techniques. Each program may be implemented in a high level procedural or object oriented programming language to communicate with a computer system. However, the program(s) can be implemented in assembly or machine language, if desired. In any case, the language may be a compiled or interpreted language. Furthermore, the program can be run on a programmed application specific integrated circuit for this purpose.
Furthermore, the operations of the processes described herein may be performed in any suitable order unless otherwise indicated herein or otherwise clearly contradicted by context. The processes (or variations and/or combinations thereof) described herein may be performed under control of one or more computer systems configured with executable instructions, and may be implemented as code (e.g., executable instructions, one or more computer programs, or one or more applications), by hardware, or combinations thereof, collectively executing on one or more processors. The computer program includes a plurality of instructions executable by one or more processors.
Further, a fourth embodiment of the present application also provides a computer-readable storage medium.
The computer readable storage medium of the application stores a precision manufacturing quality monitoring program based on an artificial intelligence element learning technology, wherein the precision manufacturing quality monitoring program based on the artificial intelligence element learning technology is executed by a processor to realize the steps of the precision manufacturing quality monitoring method based on the artificial intelligence element learning technology, as described above:
Data acquisition and pretreatment: acquiring time sequence data of a plurality of sensors under different process parameters in the precision manufacturing process, and performing data cleaning and normalization processing;
data dividing stage: dividing the data into a meta training set, a meta verification set and a meta test set;
Model establishment: constructing a precision manufacturing process time sequence quality monitoring model based on meta learning;
Meta training stage: performing internal optimization and external optimization on a plurality of tasks to learn meta knowledge, and completing meta training and meta verification;
meta-test stage: using a model trained in the meta-training stage, and updating internal optimization parameters on new data by adopting a small amount of samples, so that the model is rapidly adapted to a new quality monitoring task;
Quality monitoring: setting an adaptive threshold updating method, dynamically adapting to different scenes, and executing monitoring alarm processing.
The method implemented when the precision manufacturing quality monitoring program based on the artificial intelligence meta-learning technology is executed may refer to various embodiments of the precision manufacturing quality monitoring method based on the artificial intelligence meta-learning technology of the present application, which are not described herein.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
The foregoing embodiment numbers of the present application are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.
The application is operational with numerous general purpose or special purpose computer device environments or configurations. Further, the method may be implemented in any type of computing platform operatively connected to a suitable computing platform, including, but not limited to, a personal computer, mini-computer, mainframe, workstation, network or distributed computing environment, separate or integrated computer platform, or in communication with a charged particle tool or other imaging device, and so forth. Aspects of the application may be implemented in machine-readable code stored on a non-transitory storage medium or device, whether removable or integrated into a computing platform, such as a hard disk, optical read and/or write storage medium, RAM, ROM, etc., such that it is readable by a programmable computer, which when read by a computer, is operable to configure and operate the computer to perform the processes described herein. Further, the machine readable code, or portions thereof, may be transmitted over a wired or wireless network. When such media includes instructions or programs that, in conjunction with a microprocessor or other data processor, implement the steps described above, the application described herein includes these and other different types of non-transitory computer-readable storage media. The application also includes the computer itself when programmed according to the methods and techniques of the present application.
The computer program can be applied to the input data to perform the functions described herein, thereby converting the input data to generate output data that is stored to the non-volatile memory. The output information may also be applied to one or more output devices such as a display. In a preferred embodiment of the invention, the transformed data represents physical and tangible objects, including specific visual depictions of physical and tangible objects produced on a display.
From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) as described above, comprising instructions for causing a terminal device (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to perform the method according to the embodiments of the present application.
The application provides a precision manufacturing quality monitoring method and device based on an artificial intelligence meta-learning technology, which can conveniently standardize and unify time sequence data generated by various types of precision manufacturing equipment by a data preprocessing method and can directly utilize a meta-learning model to train and test; through the learning of a meta learner, the model learns the meta knowledge of the precision manufacturing industry, faces a new precision manufacturing scene and is quickly generalized to the new scene; by setting the adaptive threshold method, the quality monitoring model is still applicable even if the process time sequence data of the precision manufacturing equipment deviates.
The foregoing description is only of the preferred embodiments of the present application, and is not intended to limit the scope of the application, but rather is intended to cover any equivalents of the structures or equivalent processes disclosed herein or in the alternative, which may be employed directly or indirectly in other related arts.
Claims (8)
1. The precision manufacturing quality monitoring method based on the artificial intelligence element learning technology is characterized by comprising the following steps of:
Collecting time sequence data in the precise manufacturing process, and cleaning and normalizing the time sequence data to obtain preprocessing data;
Dividing the preprocessing data into a meta training set, a meta verification set and a meta test set;
Constructing a precision manufacturing process time sequence quality monitoring model based on meta learning, and learning meta knowledge through two optimization processes of internal optimization and external optimization on a plurality of tasks through a meta training set and a meta verification set to complete meta training and meta verification;
updating internal optimization parameters of the model after meta training by adopting the samples on the meta test set to obtain a quality monitoring model adapting to a new quality monitoring task;
setting a self-adaptive threshold updating method based on the quality monitoring model, dynamically adapting to different scenes, and executing monitoring alarm processing;
the method comprises the steps of performing internal optimization and external optimization on a plurality of tasks to learn meta knowledge, completing meta training and meta verification, and comprises the following steps:
The meta training stage comprises an internal optimization process and an external optimization process, wherein the external optimization process optimizes the parameters theta ′ of the element-learning meta-knowledge external network, and the internal optimization process optimizes the parameters theta ′ of the internal network;
In the internal optimization stage, fixing an external network parameter theta, optimizing the internal network parameter theta ′ by adopting a learning rate alpha and a gradient descent algorithm on a time sequence monitoring task T i in the precise manufacturing process, fixing the internal network parameter theta ′ after training on all tasks is finished, and updating primary element knowledge, namely the external network parameter theta, by using the learning rate beta;
and learning the meta-knowledge in two optimization processes of internal optimization and external optimization on a plurality of tasks to complete meta-training and meta-verification, and further comprising:
F θ is a model trained in an unsupervised mode, a loss function is defined on a training set of one task T i, and an external network parameter theta is updated to an internal network parameter theta' by adopting a learning rate alpha in a gradient updating mode; calculating a difference between the reconstructed sample and the actual sample by L (f θ(xj),yj);
L(fθ(xj),yj)＝λ1L1(fθ(xj),yj)+λ2LSSIM(fθ(xj),yj)
The goal of meta-learning is to learn the optimization parameter θ' over different tasks by equation 1, minimizing the overall loss over all tasks, meta-learning overall loss function is defined as:
2. The method for monitoring precision manufacturing quality based on artificial intelligence meta-learning technology according to claim 1, wherein the steps of collecting time series data in the precision manufacturing process, and performing cleaning and normalization processing on the time series data to obtain preprocessing data comprise:
Collecting the data of a plurality of monitoring sensors in the precise manufacturing process;
The method comprises the steps that collection of a plurality of variables under different process parameters is synchronously carried out, and data sampled by each variable in each complete precision manufacturing process period are uploaded in real time to obtain time sequence data;
and cleaning the uploaded time sequence data, removing data which do not meet the requirements, and separating and processing each period of data according to the acquired parameters into standard data with equal length according to the stages.
3. The method for precision manufacturing quality monitoring based on artificial intelligence meta-learning technology according to claim 2, wherein the dividing the pre-processing data into meta-training sets, meta-verification sets and meta-test sets further comprises:
The preprocessing data is divided into three parts, namely a meta training set, a meta verification set and a meta testing set, wherein the meta training set, the meta verification set and the meta testing set are in a set form, each set is divided into a supporting set and a query set, the supporting set is trained, and the query set is tested;
the meta training set consists of a plurality of precision manufacturing process time sequence quality detection tasks, a meta learner learns on a plurality of support set tasks for quality monitoring of each task, learns to process meta knowledge of precision manufacturing problems, and verifies on the verification set; the meta-test set contains precision manufacturing process data.
4. The method for monitoring precision manufacturing quality based on artificial intelligence meta-learning technology according to claim 1, wherein the constructing a precision manufacturing process time sequence quality monitoring model based on meta-learning comprises:
Constructing a meta learning model based on an unsupervised time sequence quality monitoring algorithm;
The meta learning model is divided into an external network comprising learning meta knowledge and an internal network sensitive to tasks based on an unsupervised meta learning algorithm;
The external network is used as an external circulation parameter of meta-learning and is used for learning initial characterization of time sequence data in a precision manufacturing process, and has the parameter which is most sensitive to the distribution of new task learning domains;
the internal network is used as an internal circulation parameter of meta learning, and based on the characterization of the external network, gradient adjustment based on samples is used.
5. The method for precisely manufacturing quality monitoring based on artificial intelligence meta-learning technology according to claim 4, wherein updating internal optimization parameters of the meta-trained model by using the samples on the meta-test set to obtain a quality monitoring model suitable for new quality monitoring tasks, comprises:
And (3) updating the internal optimization parameters theta' by adopting K samples on the new abnormality detection task S new in the meta-test set by using the initial parameters of the model trained in the meta-training stage, so that the model is quickly adapted to the new task, and testing on the test set of S new.
6. The method for monitoring quality of precision manufacturing based on artificial intelligence meta-learning technology according to claim 1, wherein setting an adaptive threshold updating method based on the quality monitoring model, dynamically adapting to different scenes, performing monitoring alarm processing, comprises:
And executing a monitoring task on the data set by adopting the trained meta-learning model, calculating the error score of each sample, and setting a self-adaptive threshold updating method to monitor new data so that the threshold is adapted to the offset condition generated along with continuous production of precision manufacturing.
7. A precision manufacturing quality monitoring device based on an artificial intelligence element learning technology, characterized in that the precision manufacturing quality monitoring device based on an artificial intelligence element learning technology comprises the precision manufacturing quality monitoring method based on an artificial intelligence element learning technology as set forth in any one of claims 1 to 6, and is characterized in that the device comprises:
The data acquisition module is used for acquiring time sequence data in the precise manufacturing process, and cleaning and normalizing the time sequence data to obtain preprocessing data;
The data dividing module is used for dividing the preprocessing data into a meta training set, a meta verification set and a meta test set;
The modeling training module is used for constructing a precision manufacturing process time sequence quality monitoring model based on meta-learning, learning meta-knowledge through two optimization processes of internal optimization and external optimization on a plurality of tasks through the meta-training set and the meta-verification set, and finishing meta-training and meta-verification; the method comprises the steps of performing internal optimization and external optimization on a plurality of tasks to learn meta knowledge, completing meta training and meta verification, and comprises the following steps:
The meta training stage comprises an internal optimization process and an external optimization process, wherein the external optimization process optimizes the parameters theta ′ of the element-learning meta-knowledge external network, and the internal optimization process optimizes the parameters theta ′ of the internal network;
In the internal optimization stage, fixing an external network parameter theta, optimizing the internal network parameter theta ′ by adopting a learning rate alpha and a gradient descent algorithm on a time sequence monitoring task T i in the precise manufacturing process, fixing the internal network parameter theta ′ after training on all tasks is finished, and updating primary element knowledge, namely the external network parameter theta, by using the learning rate beta;
and learning the meta-knowledge in two optimization processes of internal optimization and external optimization on a plurality of tasks to complete meta-training and meta-verification, and further comprising:
F θ is a model trained in an unsupervised mode, a loss function is defined on a training set of one task T i, and an external network parameter theta is updated to an internal network parameter theta ′ by adopting a learning rate alpha in a gradient updating mode; calculating a difference between the reconstructed sample and the actual sample by L (f θ(xj),yj);
L(fθ(xj),yj)＝λ％L％(fθ(xj),yj)+λ'L(()*(fθ(xj),yj)
The goal of meta-learning is to learn the optimization parameter θ ′ over different tasks by equation 1, minimizing the overall loss over all tasks, the meta-learning overall loss function being defined as:
the meta-test module is used for updating internal optimization parameters of the model after meta-training by adopting the samples on the meta-test set to obtain a quality monitoring model adapting to a new quality monitoring task;
And the quality monitoring module is used for setting a self-adaptive threshold updating method based on the quality monitoring model, dynamically adapting to different scenes and executing monitoring alarm processing.
8. The artificial intelligence meta-learning technology-based precision manufacturing quality monitoring device of claim 7, further comprising:
the model construction module is used for constructing a precision manufacturing process time sequence quality monitoring model based on meta learning;
And the meta training module is used for learning meta knowledge in two optimization processes of internal optimization and external optimization on a plurality of tasks to complete meta training and meta verification.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210371923.2A CN114861522B (en) | 2022-04-11 | 2022-04-11 | Precision manufacturing quality monitoring method and device based on artificial intelligence element learning technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210371923.2A CN114861522B (en) | 2022-04-11 | 2022-04-11 | Precision manufacturing quality monitoring method and device based on artificial intelligence element learning technology |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114861522A CN114861522A (en) | 2022-08-05 |
| CN114861522B true CN114861522B (en) | 2024-08-23 |
Family
ID=82628946
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210371923.2A Active CN114861522B (en) | 2022-04-11 | 2022-04-11 | Precision manufacturing quality monitoring method and device based on artificial intelligence element learning technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114861522B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116561289B (en) * | 2023-04-13 | 2024-01-19 | 西安天智数字信息科技有限公司 | Real-time monitoring method and system for digital factory |
| CN116502959B (en) * | 2023-06-21 | 2023-09-08 | 南京航空航天大学 | Product manufacturing quality prediction method based on meta learning |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113469219A (en) * | 2021-06-08 | 2021-10-01 | 中国地质大学（武汉） | Rotary machine fault diagnosis method under complex working condition based on element transfer learning |
| EP3965020A1 (en) * | 2020-09-07 | 2022-03-09 | Siemens Aktiengesellschaft | A model for a rapid continual anomaly detector |
-
2022
- 2022-04-11 CN CN202210371923.2A patent/CN114861522B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3965020A1 (en) * | 2020-09-07 | 2022-03-09 | Siemens Aktiengesellschaft | A model for a rapid continual anomaly detector |
| CN113469219A (en) * | 2021-06-08 | 2021-10-01 | 中国地质大学（武汉） | Rotary machine fault diagnosis method under complex working condition based on element transfer learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114861522A (en) | 2022-08-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114861522B (en) | Precision manufacturing quality monitoring method and device based on artificial intelligence element learning technology | |
| EP3361330B1 (en) | Event analyzing device, event analyzing system, event analyzing method, event analyzing program, and non-transitory computer readable storage medium | |
| US20200202508A1 (en) | Image-based maintenance prediction and detection of operating errors | |
| CN113326590B (en) | Rolling bearing service life prediction method and device based on deep learning model | |
| CN115427968A (en) | Robust artificial intelligence reasoning in edge computing devices | |
| CN113609790B (en) | Product virtual measuring method, system, device and medium | |
| Kozjek et al. | Knowledge elicitation for fault diagnostics in plastic injection moulding: A case for machine-to-machine communication | |
| CN115906399A (en) | Improved method for predicting key process quality of product under small sample data | |
| CN113641525B (en) | Variable anomaly repair method, variable anomaly repair device, variable anomaly repair medium, and variable anomaly repair computer program product | |
| Lutska et al. | Forecasting the efficiency of the control system of the technological object on the basis of neural networks | |
| WO2021066194A1 (en) | Learning device, learning method, and learning program | |
| CN117952254B (en) | Water treatment process optimization full-flow model system based on artificial intelligence | |
| CN105787507B (en) | LS SVMs on-line study methods based on budget supporting vector collection | |
| CN111797136A (en) | Big data based air conditioner soft fault analysis method and device and storage medium | |
| Krupp et al. | A hybrid framework for bearing fault diagnosis using physics-guided neural networks | |
| Qi et al. | Remaining useful life prediction of rolling element bearings based on unscented Kalman filter | |
| CN117436848B (en) | Motor energy consumption monitoring method based on artificial intelligence | |
| JP7400064B1 (en) | Learning devices, learning methods and learning programs | |
| CN115599037B (en) | Automatic monitoring method for gene detection laboratory equipment | |
| KR102639655B1 (en) | Method, apparatus and program for classifying patterns of wafer maps | |
| CN118012479B (en) | Digital twin real-time updating method, medium and system based on pre-estimation mode | |
| Wang et al. | On-line modeling and monitoring for multi-operation batch processes with infinite data types | |
| Subbotin et al. | Simulation Model Calibration for Condition Monitoring | |
| CN117933450A (en) | Advanced technology industrialization trend prediction method and system based on deep reinforcement learning | |
| CN116796195A (en) | Sulfur recovery process self-adaptive soft measurement method based on instant learning and twin support vector regression machine |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |