CN112784501A - Modeling system and method for residual life prediction model of equipment and prediction system - Google Patents
Modeling system and method for residual life prediction model of equipment and prediction system Download PDFInfo
- Publication number
- CN112784501A CN112784501A CN202110308044.0A CN202110308044A CN112784501A CN 112784501 A CN112784501 A CN 112784501A CN 202110308044 A CN202110308044 A CN 202110308044A CN 112784501 A CN112784501 A CN 112784501A
- Authority
- CN
- China
- Prior art keywords
- algorithm
- basic
- residual life
- life prediction
- modeling
- 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 32
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 207
- 238000012360 testing method Methods 0.000 claims abstract description 57
- 238000011156 evaluation Methods 0.000 claims abstract description 34
- 230000008569 process Effects 0.000 claims abstract description 19
- 238000004806 packaging method and process Methods 0.000 claims abstract description 16
- 238000012549 training Methods 0.000 claims description 60
- 238000007781 pre-processing Methods 0.000 claims description 31
- 238000000605 extraction Methods 0.000 claims description 26
- 238000005457 optimization Methods 0.000 claims description 21
- 230000036541 health Effects 0.000 claims description 6
- 230000010354 integration Effects 0.000 claims description 6
- 238000005516 engineering process Methods 0.000 description 11
- 230000006870 function Effects 0.000 description 7
- 238000012423 maintenance Methods 0.000 description 5
- 238000011161 development Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000010835 comparative analysis Methods 0.000 description 3
- 230000007613 environmental effect Effects 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000002068 genetic effect Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 208000013407 communication difficulty Diseases 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000000513 principal component analysis Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000010183 spectrum analysis Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000013076 uncertainty analysis Methods 0.000 description 1
Images
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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/231—Hierarchical techniques, i.e. dividing or merging pattern sets so as to obtain a dendrogram
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/04—Ageing analysis or optimisation against ageing
-
- 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
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Software Systems (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Medical Informatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Monitoring And Testing Of Nuclear Reactors (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention discloses a residual life prediction model modeling system of equipment, which comprises a basic algorithm library, a database and a general platform. And the basic algorithm library is used for storing basic algorithms required in the residual life prediction model modeling process. And the database is used for storing a data set required in the residual life prediction model modeling process. And the universal platform is respectively connected with the basic algorithm library and the database and is used for calling the basic algorithm and the data set to construct, train and test a basic model, and packaging the basic model which passes the test evaluation into a residual life prediction model of a preset file type, wherein the residual life prediction model is used for predicting the residual life of the equipment. A corresponding modeling method and a residual life prediction system of nuclear power equipment are also provided. The modeling system can help researchers in nuclear power major to conveniently and quickly establish a residual life prediction model of specific equipment, and reduce the requirements on IT technical capability of the modeling personnel.
Description
Technical Field
The invention belongs to the technical field of nuclear power, and particularly relates to a residual life prediction model modeling system and method of equipment and a residual life prediction system of nuclear power equipment.
Background
The predictive maintenance of nuclear power equipment means that the prediction and the accurate decision of maintenance opportunity and maintenance objects are realized by continuously monitoring and predicting the system/equipment state and faults of a nuclear power plant, the unplanned shutdown caused by sudden faults of the equipment is avoided, the service life of the equipment is utilized to the maximum extent, and the storage scale of spare parts required by maintenance is reduced. Residual life prediction is a key technology at the core of implementing predictive maintenance of equipment. The data-driven prediction method mainly depends on information mined from data to predict the service life, and for large and complex equipment or systems of which failure mechanisms are difficult to accurately describe, as long as relevant data can be obtained, a service life prediction result with application value can be obtained by using a data-driven model, so that the safety and the economy of nuclear power equipment are greatly improved by applying the data-driven prediction method to the nuclear industry.
However, data-driven remaining life prediction requires analysis processing capability of mass data, and therefore, the degree of dependence on a big data processing technology and an intelligent algorithm is high, so that research on the remaining life prediction technology in the current nuclear industry is greatly limited by understanding and programming development capability of researchers on the intelligent algorithm, and development and application of the technology in the nuclear industry are limited.
Disclosure of Invention
The invention aims to solve the technical problems of the prior art, and provides a residual life prediction model modeling system and method of equipment and a residual life prediction system of nuclear power equipment, so that researchers in nuclear power major can conveniently and quickly establish a specific residual life prediction model of the nuclear power equipment, and the requirement on IT technical capability of the modeling personnel is reduced.
The embodiment of the invention provides a residual life prediction model modeling system of equipment, which comprises a basic algorithm library, a database and a general platform. And the basic algorithm library is used for storing basic algorithms required in the residual life prediction model modeling process. And the database is used for storing a data set required in the residual life prediction model modeling process. And the universal platform is respectively connected with the basic algorithm library and the database and is used for calling the basic algorithm and the data set to construct, train and test a basic model and packaging the basic model which passes the test evaluation into a residual life prediction model of a preset file type.
Preferably, the types of the basic algorithm include a data preprocessing algorithm, a feature extraction algorithm, a data driving algorithm and an optimization algorithm.
The general platform comprises a modeling module, a training module and a test evaluation module. And the modeling module is connected with the basic algorithm library and the database, and is used for dividing the called data set into a training set and a test set according to the division ratio, calling a data preprocessing algorithm to preprocess the training set, calling a feature extraction algorithm to extract features of the training set after data preprocessing and generate a feature set, and calling a data driving algorithm and an optimization algorithm to obtain the basic model. And the training module is connected with the modeling module and used for performing model training on the generated feature set by applying the basic model. And the test evaluation module is connected with the modeling module and used for calling the test set to test and evaluate the trained basic model and comparing and analyzing the evaluation result. And the modeling module is also used for packaging the basic model with the evaluation result exceeding the preset threshold value into a residual life prediction model of a preset file type so as to meet the integration requirement of an equipment health management platform of the nuclear power plant.
Preferably, the modelling module comprises an algorithm correlation unit. And the algorithm association unit is used for defining interfaces among various basic algorithms so as to realize association among the data preprocessing algorithm, the feature extraction algorithm, the data driving algorithm and the optimization algorithm.
Preferably, the generic platform further comprises an algorithm management module. And the algorithm management module is connected with the basic algorithm library and is used for deleting, modifying or checking the basic algorithms in the basic algorithm library and adding new algorithms to the basic algorithm library in a programming mode or a script file importing mode.
Preferably, the residual life prediction model modeling system of the equipment further comprises a typical use case library. And the typical use case library is connected with the general platform and used for storing a residual life prediction model of a typical failure mode of a typical device packaged by the general platform so as to be convenient for a user to view or integrate the call.
Preferably, the general platform further comprises a use case management module. And the use case management module is connected with the modeling module and used for viewing, calling, adding, deleting or modifying the basic model obtained by the modeling module.
Preferably, the modeling module further comprises a parameter adjusting unit. And the parameter adjusting unit is used for receiving parameters input by a user when a data preprocessing algorithm, a feature extraction algorithm, a data driving algorithm and an optimization algorithm are called.
In a second aspect, an embodiment of the present invention further provides a system for predicting remaining life of nuclear power equipment, including an input module and the system for modeling a model for predicting remaining life of equipment in the first aspect. The input module is used for receiving a data set of the nuclear power equipment input by a user. And the residual life prediction model modeling system of the equipment is connected with the input module and is used for outputting a residual life prediction model and a residual life prediction result of the nuclear power equipment according to the data set of the nuclear power equipment input by a user.
In a third aspect, an embodiment of the present invention further provides a modeling method for a residual life prediction model of a device, including: acquiring a basic algorithm required in the residual life prediction model modeling process and acquiring a data set of equipment for which the residual life prediction model is to be established; dividing the data set into a training set and a test set according to a division ratio; and constructing and training a basic model according to the basic algorithm and the training set, testing, evaluating and training the basic model according to the test set, and packaging the basic model which passes the test evaluation into a residual life prediction model of a preset file type.
The residual life prediction model modeling system of the equipment comprises a basic algorithm library, a database and a general platform. And the basic algorithm library is used for storing basic algorithms required in the residual life prediction model modeling process. And the database is used for storing a data set required in the residual life prediction model modeling process. And the universal platform is respectively connected with the basic algorithm library and the database and is used for calling the basic algorithm and the data set to construct, train and test a basic model and packaging the basic model which passes the test evaluation into a residual life prediction model of a preset file type. Nuclear power researchers can utilize the modeling system to develop a modeling method of a flow standardized and building block type residual life prediction model, and the residual life prediction model of specific nuclear power equipment can be conveniently and quickly established.
Drawings
FIG. 1: the invention provides a structural schematic diagram of a residual life prediction model modeling system of equipment in embodiment 1.
Detailed Description
In order to make the technical solutions of the present invention better understood, the present invention is further described in detail below with reference to the accompanying drawings and examples.
Example 1:
as shown in fig. 1, the present embodiment provides a modeling system for a residual life prediction model of a device, which is suitable for building, training, testing, and packaging a residual life prediction model for a specific device, and is particularly suitable for a nuclear power professional researcher to perform modeling, training, testing, and packaging of a data-driven residual life prediction model for a certain nuclear power device after acquiring massive operation data of the nuclear power device and fault data of the nuclear power device of the same type.
The residual life prediction model modeling system of the device of the present embodiment includes a basic algorithm library 1, a database 2, and a general platform 3.
The basic algorithm library 1 is used for storing basic algorithms required in the residual life prediction model modeling process, wherein the types of the basic algorithms comprise a data preprocessing algorithm, a feature extraction algorithm, a data driving algorithm and an optimization algorithm, and the number of the basic algorithms of each type is multiple. And the database 2 is used for storing a data set required in the residual life prediction model modeling process. And the universal platform 3 is respectively connected with the basic algorithm library 1 and the database 2 and is used for calling a basic algorithm and a data set to construct, train and test a basic model and packaging the basic model which passes the test evaluation into a residual life prediction model of a preset file type.
In this embodiment, the basic algorithm library 1 is used for storing a commonly used basic algorithm, and the basic algorithm in the basic algorithm library 1 may be directly called in the residual life prediction model modeling process, where the types of the basic algorithm at least include, but are not limited to, the following types, and the following algorithms are algorithms known to those skilled in the art:
(1) the data preprocessing algorithms comprise a missing data filling algorithm, an abnormal data detection algorithm, a noise data processing algorithm, a data normalization algorithm and the like.
(2) And the characteristic extraction algorithms comprise algorithms of Wingner-Ville analysis, wavelet analysis, spectrum analysis, blind source separation and the like so as to realize the extraction of characteristic parameters and establish a characteristic parameter set.
(3) And the sensitive characteristic parameter selection algorithm is used for reducing the dimension of the characteristic parameter set and automatically selecting the optimized sensitive characteristic parameters, and comprises a principal component analysis algorithm, a hierarchical clustering algorithm and the like.
(4) The data-driven algorithm for predicting the service life based on the characteristic data comprises a support vector machine, an artificial neural network, a cyclic neural network and the like.
(5) And an optimization algorithm for automatically optimizing the model parameters, such as a genetic algorithm.
(6) And uncertainty analysis algorithms of the model, such as Bootstrap and Gaussian regression.
The data types in the database 2 include: lifetime data, status data, environmental parameters, and the like. Wherein the life data is the duration from the beginning of the operation of the equipment to the failure of the equipment; the state data is a parameter for representing the running state and performance of the equipment, such as the vibration frequency of a pump shaft; the environmental parameters are used to describe the usage conditions of the device, such as the temperature, pressure, humidity, etc. of the system in which the device is located. The data can be real data of the operation of the equipment (including historical data of the operation of the equipment and data of the operation of the same equipment), or data generated by simulating the operation in an experimental or modeling simulation mode, wherein the life data, the state data, the environmental parameters and the like form a data set of the equipment A. In general, when a user plans to model the remaining life prediction model for the device a, the acquired data set of the device a needs to be imported into the database 2. In the dataset table to be imported, a header should describe the attribute of each column of data, so that the storage module can store the data in a structured manner.
Optionally, the generic platform 3 comprises a modeling module 31, a training module 32, a test evaluation module 33.
And the modeling module 31 is connected with the basic algorithm library 1 and the database 2, and is used for dividing the called data set into a training set and a test set according to the division ratio, calling a data preprocessing algorithm to perform data preprocessing on the training set, calling a feature extraction algorithm to perform feature extraction on the training set after data preprocessing and generate a feature set, and calling a data driving algorithm and an optimizing algorithm to obtain a basic model. And the training module 32 is connected with the modeling module 31 and is used for performing model training on the generated feature set by applying the basic model. The modelling module 31 further comprises a parameter tuning unit. And the parameter adjusting unit is used for receiving parameters input by a user when a data preprocessing algorithm, a feature extraction algorithm, a data driving algorithm and an optimization algorithm are called.
In this embodiment, the modeling module 31 is configured to provide a selection interface of a path stored in a data set to be called to a user, so that the user can select the required data set, and divide the called data set into a training set and a test set according to a division ratio set by the user. For example, for the called data set of the device a, the training set is set to account for 70%, and the test set is set to account for 30%, then the modeling module automatically divides and generates a temporary training set and a temporary test set, where the data amount in the training set accounts for 70% of the data set of the device a, and the data amount in the test set accounts for 30% of the data set of the device a. The proportion of the training set and the test set can be set according to the requirements of users. The modeling module 31 is further configured to provide a selection interface of the data preprocessing algorithm to the user, so that the user can select a specific data preprocessing algorithm to be invoked, and receive an algorithm parameter input by the user. The modeling training module 31 is configured to automatically invoke a data preprocessing algorithm selected by the user to perform data preprocessing on the generated training set. The modeling module 31 is further configured to provide a selection interface of the feature extraction algorithm to the user, and may receive parameters input by the user for the specific feature extraction algorithm selected. For example, the modeling module 31 is configured to perform feature parameter selection on the training set after data preprocessing according to a feature parameter extraction algorithm and a sensitive parameter selection algorithm selected by a user, and generate a plurality of sensitive feature sets. The modeling module 31 is further configured to provide a selection interface of the data-driven algorithm and the optimization algorithm for the user, and receive parameters input by the user for the selected specific data-driven algorithm and optimization algorithm to obtain the basic model. For example, if the user selects a data-driven algorithm based on historical failure records and exponential distribution, parameters such as a failure rate parameter and an initial value in the exponential distribution need to be input, and the user also selects a genetic algorithm to optimize parameters of a basic model. The basic model in the embodiment includes a specific data preprocessing algorithm, a feature extraction algorithm, a data-driven algorithm, and an optimization algorithm selected by a user. The training module 32 is configured to apply the basic model to automatically perform model training on the generated feature set, and before the model training, the training module 32 is further configured to receive a training cutoff condition set by a user, where the training cutoff condition includes training times, a performance index, and the like. When the model training reaches the training cutoff condition set by the user, the training module 32 stops the model training and outputs the result, at this time, there are two situations: firstly, the training is stopped when the training times are reached, the output result is the training failure, and the user is prompted to select the data preprocessing algorithm, the feature extraction algorithm, the data driving algorithm and the optimization algorithm and train the model again. Secondly, the performance index is reached, the output result is that the model training is successful, the performance index of the model is displayed, and whether the user stores the successfully trained basic model is prompted, and the modeling module 31 is used for storing the successfully trained basic model.
And the test evaluation module 33 is connected with the modeling module 31, and is used for calling the test set to test and evaluate the trained basic model, and performing comparative analysis on the evaluation result. The modeling module 31 is further configured to package a basic model with an evaluation result meeting a preset index requirement as a residual life prediction model of a preset file type, so as to meet an integration requirement of an equipment health management platform of the nuclear power plant.
In this embodiment, the test evaluation module 33 is configured to provide a selection interface for a user to perform a test, and if the user selects to perform the test, the test evaluation module 33 is configured to invoke a test set to perform the test and evaluation on the trained basic model. When the basic model is tested, dynamically displaying a prediction result, wherein the prediction result comprises: remaining useful life and probability of failure distribution. When the basic model is evaluated, the evaluation indexes include, but are not limited to, prediction accuracy, uncertainty, reliability, model complexity, response speed, and the like (the specific evaluation indexes are customized according to user requirements and algorithms). The test evaluation module 33 is also used for performing comparative analysis of evaluation results on a plurality of basic models. And if the evaluation result does not meet the performance index set by the user, prompting the user to perform data preprocessing algorithm, feature extraction algorithm, data driving algorithm and optimization algorithm selection, model training and test evaluation again. And if the evaluation result meets the performance index set by the user, prompting the user that the modeling of the residual life prediction model is finished, and further prompting the user whether to package the residual life prediction model. If the user selects the package type (e.g., C language), the modeling module 31 is configured to convert the completed residual life prediction model into a program file of a file type specified by the user, and complete model packaging, so as to meet an integration requirement of an equipment health management platform of a nuclear power plant or facilitate integration on other upper nuclear power platforms.
Optionally, the modelling module 31 comprises an algorithm association unit. And the algorithm association unit is used for defining interfaces among various basic algorithms so as to realize association among the data preprocessing algorithm, the feature extraction algorithm, the data driving algorithm and the optimization algorithm.
In this embodiment, the types of the basic algorithms are multiple, and the business processes, data, input/output and core algorithms between different algorithms and models may have differences, so how to ensure the universality of the residual life prediction model modeling system of the device in this embodiment is a key technical difficulty. Firstly, combing the service flow of the method and the applicable algorithm of each stage in the service flow aiming at different algorithm categories; then, aiming at the algorithms with the same function, a basic algorithm library (container) is established, the functions of adding, deleting, modifying and searching elements (single algorithm) in the basic algorithm library are supported, and a corresponding algorithm calling function and an interface are developed based on the service flow of a prediction model to guide a user to carry out modeling of residual life prediction; and finally, establishing and testing a passed model for a user, and providing a packaging function so as to be conveniently integrated in the existing equipment health management platform of the nuclear power plant and execute a service life prediction function. In addition, an algorithm association unit is arranged and used for defining and standardizing the form of each type of algorithm input and output interface, visually describing the meaning of the interface, guiding a user to accurately define the input and output interfaces (such as data sources, model storage positions, parameter storage positions and the like) of the selected algorithm in the using process of the residual life prediction model modeling system of the equipment, and ensuring that a plurality of algorithms selected by the user can be correctly connected, so that a complete residual life prediction model is constructed. As shown in table 1, an example of interface normalization for a data-driven class algorithm.
TABLE 1
Optionally, the generic platform 3 further comprises an algorithm management module 34. And the algorithm management module 34 is connected with the basic algorithm library 1 and is used for deleting, modifying or checking the basic algorithms in the basic algorithm library and adding new algorithms to the basic algorithm library through a programming mode or a script file importing mode.
In this embodiment, a function of "adding new algorithm" is provided, and a user can load a newly written algorithm file (matlab in the m file format) to the residual life prediction model modeling system of the device, and expand the content of the basic algorithm library to meet the application requirements of a wider modeling scene. Similarly, the user can standardize and customize the new algorithm interface according to the system direction so as to be capable of being associated with the existing algorithm of the basic algorithm library.
Optionally, the residual life prediction model modeling system of the device further includes a typical use case library 4. And the typical use case library 4 is connected with the general platform 3 and used for storing a residual life prediction model of a typical failure mode of a typical device packaged by the general platform 3 (particularly the modeling module 31) so as to be convenient for a user to view or integrate and call.
Optionally, the general platform 3 further includes a use case management module 35. And the use case management module 35 is connected with the modeling module 31 and used for viewing, calling, adding, deleting or modifying the trained basic model stored by the modeling module 31.
The modeling system of the residual life prediction model of the equipment provided by the embodiment provides a residual life prediction modeling mode of building-block equipment with standardized flow, can help researchers in nuclear power profession to conveniently and quickly select proper data preprocessing technology, feature extraction technology, data-driven algorithm and parameter optimization technology, establishes the residual life prediction model of specific equipment, and avoids the problem that the researchers in nuclear power profession need to program to realize algorithm functions and combination of algorithms in the manual modeling process, so that the requirements of the researchers on understanding and programming technical capability of intelligent algorithms are reduced, and general technologies such as information technology and intelligent algorithms become general tools which can be directly applied or applied after a small amount of development in the professional technical field; meanwhile, an interface for fusing cross-professional technologies (such as a nuclear power technology and a data processing technology) in the development process of the data-driven residual life prediction model is simplified, and communication difficulty and cost are reduced. In addition, in the manual modeling process, a modeling worker learns, optimizes, realizes codes, tests and evaluates the prediction method and each algorithm until a satisfactory prediction model is obtained, the period is at least about two months, and if the residual life prediction model modeling system of the equipment is used, the period of constructing, testing and packaging the prediction model for the same equipment is about ten days. Therefore, the modeling system of the embodiment can save labor cost and improve modeling efficiency.
Example 2:
the embodiment provides a system for predicting the residual life of nuclear power equipment, which comprises an input module 21 and a system for modeling a model for predicting the residual life of the equipment in embodiment 1.
The input module 21 is used for receiving a data set of the nuclear power equipment input by a user.
The residual life prediction model modeling system of the equipment is connected with the input module 21 and is used for outputting a residual life prediction model and a residual life prediction result of the nuclear power equipment according to a data set of the nuclear power equipment input by a user.
Example 3:
the embodiment provides a modeling method of a residual life prediction model of equipment, which comprises the following steps:
step 301, obtaining a basic algorithm required in the residual life prediction model modeling process, and obtaining a data set of the device to be built with the residual life prediction model. The types of the basic algorithm comprise a data preprocessing algorithm, a feature extraction algorithm, a data driving algorithm and an optimization algorithm.
Step 302, dividing the data set into a training set and a test set according to the division ratio.
And 303, constructing and training a basic model according to the basic algorithm and the training set, testing, evaluating and training the basic model according to the test set, and packaging the basic model passing the test evaluation into a residual life prediction model of a preset file type.
Optionally, step 303 specifically includes:
And step 35, calling the test set to test and evaluate the trained basic model, carrying out comparative analysis on the evaluation result, and executing the steps 31-35 again when the evaluation result does not meet the performance index until the evaluation result of the new basic model meets the performance index.
And step 36, packaging the basic model of which the evaluation result meets the requirement of a preset index into a residual life prediction model of a preset file type so as to meet the integration requirement of an equipment health management platform of the nuclear power plant.
Optionally, step 303 further includes defining an interface between various types of basic algorithms to implement an association between a data preprocessing type algorithm, a feature extraction type algorithm, a data driving type algorithm, and an optimization type algorithm.
It will be understood that the above embodiments are merely exemplary embodiments taken to illustrate the principles of the present invention, which is not limited thereto. It will be apparent to those skilled in the art that various modifications and improvements can be made without departing from the spirit and substance of the invention, and these modifications and improvements are also considered to be within the scope of the invention.
Claims (9)
1. The residual life prediction model modeling system of the equipment is characterized by comprising a basic algorithm library, a database and a general platform,
a basic algorithm library used for storing basic algorithms required in the residual life prediction model modeling process,
a database for storing data sets required in the residual life prediction model modeling process,
and the universal platform is respectively connected with the basic algorithm library and the database and is used for calling the basic algorithm and the data set to construct, train and test a basic model and packaging the basic model which passes the test evaluation into a residual life prediction model of a preset file type.
2. The system of claim 1, wherein the types of base algorithms include a data pre-processing class algorithm, a feature extraction class algorithm, a data-driven class algorithm, and an optimization class algorithm,
the universal platform comprises a modeling module, a training module and a test evaluation module,
the modeling module is connected with the basic algorithm library and the database, and is used for dividing the called data set into a training set and a test set according to the division ratio, calling a data preprocessing algorithm to preprocess the training set, calling a feature extraction algorithm to extract the features of the training set after data preprocessing and generate a feature set, calling a data driving algorithm and an optimization algorithm to obtain a basic model,
a training module connected with the modeling module and used for carrying out model training on the generated feature set by applying the basic model,
the test evaluation module is connected with the modeling module and used for calling the test set to test and evaluate the trained basic model and comparing and analyzing the evaluation result,
and the modeling module is also used for packaging the basic model with the evaluation result exceeding the preset threshold value into a residual life prediction model of a preset file type so as to meet the integration requirement of an equipment health management platform of the nuclear power plant.
3. The system of claim 2, wherein the modeling module comprises an algorithm correlation unit,
and the algorithm association unit is used for defining interfaces among various basic algorithms so as to realize association among the data preprocessing algorithm, the feature extraction algorithm, the data driving algorithm and the optimization algorithm.
4. The system of claim 3, wherein the generic platform further comprises an algorithm management module,
and the algorithm management module is connected with the basic algorithm library and is used for deleting, modifying or checking the basic algorithms in the basic algorithm library and adding new algorithms to the basic algorithm library in a programming mode or a script file importing mode.
5. The system of claim 4, further comprising a typical use case library,
and the typical use case library is connected with the general platform and used for storing a residual life prediction model of a typical failure mode of a typical device packaged by the general platform so as to be convenient for a user to view or integrate the call.
6. The system of claim 5, wherein the generic platform further comprises a use case management module,
and the use case management module is connected with the modeling module and used for viewing, calling, adding, deleting or modifying the basic model obtained by the modeling module.
7. The system of claim 6, wherein the modeling module further comprises a parameter adjusting unit,
and the parameter adjusting unit is used for receiving parameters input by a user when a data preprocessing algorithm, a feature extraction algorithm, a data driving algorithm and an optimization algorithm are called.
8. A residual life prediction system of nuclear power equipment, characterized by comprising an input module and a residual life prediction model modeling system of the equipment of any one of claims 1-7,
the input module is used for receiving a data set of nuclear power equipment input by a user,
and the residual life prediction model modeling system of the equipment is connected with the input module and is used for outputting a residual life prediction model and a residual life prediction result of the nuclear power equipment according to the data set of the nuclear power equipment input by a user.
9. A method for modeling a residual life prediction model of a device, comprising:
acquiring a basic algorithm required in the residual life prediction model modeling process and acquiring a data set of equipment for which the residual life prediction model is to be established;
dividing the data set into a training set and a test set according to a division ratio;
and constructing and training a basic model according to the basic algorithm and the training set, testing and evaluating the basic model according to the test set, and packaging the basic model which passes the test evaluation into a residual life prediction model of a preset file type.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110308044.0A CN112784501A (en) | 2021-03-23 | 2021-03-23 | Modeling system and method for residual life prediction model of equipment and prediction system |
| PCT/CN2021/141613 WO2022199179A1 (en) | 2021-03-23 | 2021-12-27 | Remaining life prediction model modeling system and method for equipment, and prediction system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110308044.0A CN112784501A (en) | 2021-03-23 | 2021-03-23 | Modeling system and method for residual life prediction model of equipment and prediction system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112784501A true CN112784501A (en) | 2021-05-11 |
Family
ID=75762890
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110308044.0A Pending CN112784501A (en) | 2021-03-23 | 2021-03-23 | Modeling system and method for residual life prediction model of equipment and prediction system |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN112784501A (en) |
| WO (1) | WO2022199179A1 (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113567858A (en) * | 2021-07-30 | 2021-10-29 | 北京航空航天大学 | Control moment gyro residual life prediction system |
| CN113836806A (en) * | 2021-09-23 | 2021-12-24 | 中国科学院空间应用工程与技术中心 | PHM model construction method, system, storage medium and electronic equipment |
| WO2022199179A1 (en) * | 2021-03-23 | 2022-09-29 | 中国核电工程有限公司 | Remaining life prediction model modeling system and method for equipment, and prediction system |
| CN115729536A (en) * | 2022-11-29 | 2023-03-03 | 金航数码科技有限责任公司 | Generalized equipment fault prediction and health management modeling system |
| CN117009861A (en) * | 2023-10-08 | 2023-11-07 | 湖南国重智联工程机械研究院有限公司 | Hydraulic pump motor life prediction method and system based on deep learning |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115759408B (en) * | 2022-11-21 | 2024-03-08 | 贵州电网有限责任公司 | Power transmission and transformation equipment life prediction method, device, equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109948860A (en) * | 2019-03-26 | 2019-06-28 | 哈工大机器人（合肥）国际创新研究院 | A kind of mechanical system method for predicting residual useful life and system |
| CN110609524A (en) * | 2019-08-14 | 2019-12-24 | 华中科技大学 | Industrial equipment residual life prediction model and construction method and application thereof |
| CN111274737A (en) * | 2020-02-25 | 2020-06-12 | 山东大学 | Method and system for predicting remaining service life of mechanical equipment |
| CN112418277A (en) * | 2020-11-03 | 2021-02-26 | 西安电子科技大学 | Method, system, medium, and apparatus for predicting remaining life of rotating machine component |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017161064A1 (en) * | 2016-03-18 | 2017-09-21 | Schlumberger Technology Corporation | Tracking and estimating tubing fatigue in cycles to failure considering non-destructive evaluation of tubing defects |
| CN107291475B (en) * | 2017-06-27 | 2020-06-12 | 中国电子产品可靠性与环境试验研究所 | Universal PHM application configuration method and device |
| CN108470071B (en) * | 2018-03-29 | 2022-02-18 | 联想(北京)有限公司 | Data processing method and device |
| CN109615265A (en) * | 2018-12-26 | 2019-04-12 | 北京寄云鼎城科技有限公司 | Industrial data analysis method, device and electronic equipment based on integrated development system |
| US11396386B2 (en) * | 2019-05-20 | 2022-07-26 | The Boeing Company | Supporting off-wing maintenance of an engine of an aircraft |
| CN112784501A (en) * | 2021-03-23 | 2021-05-11 | 中国核电工程有限公司 | Modeling system and method for residual life prediction model of equipment and prediction system |
-
2021
- 2021-03-23 CN CN202110308044.0A patent/CN112784501A/en active Pending
- 2021-12-27 WO PCT/CN2021/141613 patent/WO2022199179A1/en active Application Filing
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109948860A (en) * | 2019-03-26 | 2019-06-28 | 哈工大机器人（合肥）国际创新研究院 | A kind of mechanical system method for predicting residual useful life and system |
| CN110609524A (en) * | 2019-08-14 | 2019-12-24 | 华中科技大学 | Industrial equipment residual life prediction model and construction method and application thereof |
| CN111274737A (en) * | 2020-02-25 | 2020-06-12 | 山东大学 | Method and system for predicting remaining service life of mechanical equipment |
| CN112418277A (en) * | 2020-11-03 | 2021-02-26 | 西安电子科技大学 | Method, system, medium, and apparatus for predicting remaining life of rotating machine component |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2022199179A1 (en) * | 2021-03-23 | 2022-09-29 | 中国核电工程有限公司 | Remaining life prediction model modeling system and method for equipment, and prediction system |
| CN113567858A (en) * | 2021-07-30 | 2021-10-29 | 北京航空航天大学 | Control moment gyro residual life prediction system |
| CN113836806A (en) * | 2021-09-23 | 2021-12-24 | 中国科学院空间应用工程与技术中心 | PHM model construction method, system, storage medium and electronic equipment |
| CN115729536A (en) * | 2022-11-29 | 2023-03-03 | 金航数码科技有限责任公司 | Generalized equipment fault prediction and health management modeling system |
| CN115729536B (en) * | 2022-11-29 | 2023-10-03 | 金航数码科技有限责任公司 | Generalized equipment fault prediction and health management modeling system |
| CN117009861A (en) * | 2023-10-08 | 2023-11-07 | 湖南国重智联工程机械研究院有限公司 | Hydraulic pump motor life prediction method and system based on deep learning |
| CN117009861B (en) * | 2023-10-08 | 2023-12-15 | 湖南国重智联工程机械研究院有限公司 | Hydraulic pump motor life prediction method and system based on deep learning |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2022199179A1 (en) | 2022-09-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112784501A (en) | Modeling system and method for residual life prediction model of equipment and prediction system | |
| US11422776B2 (en) | Intelligent assistant for automating recommendations for analytics programs | |
| US20130239098A1 (en) | Source code conversion method and source code conversion program | |
| CN107844424A (en) | Model-based testing system and method | |
| KR20160122452A (en) | Deep learnig framework and image recognition method for content-based visual image recognition | |
| CN110297760A (en) | Building method, device, equipment and the computer readable storage medium of test data | |
| CN101833504A (en) | Timing sequence software quality defect detection method and system based on model detection | |
| CN116882978B (en) | Deep sea submersible operation and maintenance support system based on product information frame | |
| CN107103363A (en) | A kind of construction method of the software fault expert system based on LDA | |
| CN111199469A (en) | User payment model generation method and device and electronic equipment | |
| CN111931936A (en) | Equipment fault diagnosis method based on collaborative case reasoning and semantic model reasoning | |
| CN116771576A (en) | Comprehensive fault diagnosis method for hydroelectric generating set | |
| CN117666546B (en) | Distributed control system fault diagnosis method and device | |
| CN116361147A (en) | Method for positioning root cause of test case, device, equipment, medium and product thereof | |
| CN113377962B (en) | Intelligent process simulation method based on image recognition and natural language processing | |
| CN115470854A (en) | Information system fault classification method and classification system | |
| CN111723580B (en) | Power dispatching station information graph validation method based on voice recognition and image recognition | |
| CN114610590A (en) | Method, device and equipment for determining operation time length and storage medium | |
| CN113095794A (en) | Production problem checking method and device based on Markov chain | |
| CN110928761A (en) | System and method for demand chain and application thereof | |
| CN115810137B (en) | Construction method of interactive artificial intelligence technical evaluation scheme | |
| CN117436444B (en) | Tag-based data processing method, device and computer-readable storage medium | |
| CN117391643B (en) | Knowledge graph-based medical insurance document auditing method and system | |
| US20230281482A1 (en) | Systems and methods for rule-based machine learning model promotion | |
| WO2022254607A1 (en) | Information processing device, difference extraction method, and non-temporary computer-readable medium |
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 |