A nuclear power plant rotating machinery health management method, system and medium

Abstract

The embodiment of the present invention provides a method, system and medium for health management of rotating machinery in a nuclear power plant, including: obtaining monitoring data and DCS data of rotating machinery; performing data preprocessing on the monitoring data and DCS data; extracting feature data from the monitoring data and DCS data after preprocessing to generate state monitoring feature data, fault diagnosis feature data and fault prediction feature data; performing state monitoring, fault diagnosis and fault prediction on the rotating machinery according to the state monitoring feature data, fault diagnosis feature data and fault prediction feature data; judging the health status of the rotating machinery in the nuclear power plant according to the state monitoring data, fault diagnosis data, fault prediction data, the historical situation of the rotating machinery and prior knowledge. The embodiment of the present invention solves the problem of low efficiency of operation and maintenance management caused by poor real-time and incomplete monitoring of rotating machinery in the prior art.

Description

Technical Field

[0001] The invention relates to a nuclear power plant rotating machinery health management method, system and medium. Background Art

[0002] Nuclear power is a very complex system, and once an accident occurs, it will bring immeasurable consequences. Rotating machinery accounts for a considerable proportion in nuclear power plants, and is prone to failure, which seriously affects the safety of nuclear power. Therefore, the research on its condition monitoring and fault diagnosis has important practical application value.

[0003] In China, in addition to the imported vibration monitoring system installed on the main pumps of nuclear power plants, the common means of monitoring the status of many other rotating machinery is to use the BENTLY vibration data acquisition device to collect data on site; and the main pump vibration monitoring system of the domestic M310 unit is not fully functional at the cabinet end, and offline data acquisition and analysis are also required during overhaul. Each nuclear power unit has hundreds of rotating machinery. The existing monitoring and operation and maintenance management mode of rotating machinery has the following problems: manual data collection and offline analysis are inefficient and have poor real-time performance, which cannot detect faults in time and prolongs the maintenance time of the main pump; the on-site noise is extremely high, which poses a health hazard to the collection personnel; the existing data analysis is only for the vibration data of rotating machinery, and the analysis of vibration data is also limited to the comparison and evaluation of vibration intensity values ​​and thresholds (warning values, shutdown values), and it is impossible to carry out more comprehensive equipment health management. Summary of the invention

[0004] In order to solve the problem of low operation and maintenance management efficiency caused by poor real-time and incomplete monitoring of rotating machinery in the prior art, an embodiment of the present invention provides a nuclear power plant rotating machinery health management method, system and medium.

[0005] The embodiment of the present invention is achieved through the following technical solutions:

[0006] In a first aspect, an embodiment of the present invention provides a method for health management of rotating machinery in a nuclear power plant, comprising:

[0007] Obtain monitoring data and DCS data of rotating machinery;

[0008] Performing data preprocessing on the monitoring data and DCS data;

[0009] Extract feature data from the pre-processed monitoring data and DCS data to generate condition monitoring feature data, fault diagnosis feature data, and fault prediction feature data;

[0010] Performing condition monitoring, fault diagnosis and fault prediction on rotating machinery based on condition monitoring characteristic data, fault diagnosis characteristic data and fault prediction characteristic data;

[0011] The health status of rotating machinery in nuclear power plants is determined based on condition monitoring data, fault diagnosis data, fault prediction data, historical conditions of rotating machinery, and prior knowledge.

[0012] Furthermore, the monitoring data includes: vibration monitoring data and acoustic emission data of the rotating machinery; and the DCS data includes the operating conditions of the unit and the operating parameters of the rotating machinery.

[0013] Further, the monitoring data and DCS data are preprocessed; including:

[0014] According to the data structure and data length of the monitoring data and DCS data information, the monitoring data and DCS data are screened and filtered, and the filtered monitoring data and DCS data are fused to obtain the pre-processed monitoring data and DCS data.

[0015] Further, the state monitoring, fault diagnosis and fault prediction of the rotating machinery are performed according to the state monitoring characteristic data, the fault diagnosis characteristic data and the fault prediction characteristic data; including:

[0016] Determine whether to perform a threshold alarm, trend alarm and / or comprehensive alarm based on the condition monitoring characteristic data; if so, visualize the alarm information and alarm log in three-dimensional results; if a comprehensive alarm is determined based on the condition monitoring characteristic data, determine whether to start fault diagnosis and fault prediction; if so, perform fault diagnosis and fault prediction.

[0017] Further, the fault diagnosis includes:

[0018] Diagnose the equipment fault type, location and degree according to the fault diagnosis characteristic data to obtain diagnostic data;

[0019] Determine the time until equipment failure becomes serious and whether secondary failures will occur based on diagnostic data, fault-related data and historical trends.

[0020] Furthermore, fault prediction includes:

[0021] Compare the trend of fault prediction feature data with the standard trend to determine whether the rotating machinery will have potential faults, the fault type, the fault location, and the time when the fault will occur;

[0022] Use the fault anomaly database to obtain the historical average time before failure to predict the time when the next fault will occur;

[0023] By comparing the trend of the characteristics of potential faults with the standard trend, potential fault information, next fault time and remaining life of the rotating machinery are generated.

[0024] Furthermore, the health status of the rotating machinery of the nuclear power plant is judged based on the condition monitoring data, fault diagnosis data, fault prediction data, the historical situation of the rotating machinery and prior knowledge; including:

[0025] The health status of rotating machinery is comprehensively evaluated using condition monitoring results, fault diagnosis results, fault prediction results, maintenance results and prior knowledge.

[0026] In a second aspect, an embodiment of the present invention provides a nuclear power plant rotating machinery health management system, including:

[0027] An acquisition unit, used for acquiring monitoring data and DCS data of rotating machinery;

[0028] A preprocessing unit, used for performing data preprocessing on the monitoring data and DCS data;

[0029] An extraction unit, used to extract feature data from the monitoring data and DCS data after preprocessing, and generate condition monitoring feature data, fault diagnosis feature data and fault prediction feature data;

[0030] A diagnosis and prediction unit, used for performing condition monitoring, fault diagnosis and fault prediction on the rotating machinery according to the condition monitoring characteristic data, the fault diagnosis characteristic data and the fault prediction characteristic data; and

[0031] The judgment unit is used to judge the health status of the rotating machinery of the nuclear power plant based on the status monitoring data, fault diagnosis data, fault prediction data, the historical conditions of the rotating machinery and prior knowledge.

[0032] In a third aspect, an embodiment of the present invention provides a nuclear power plant rotating machinery health management system, including:

[0033] The edge end is used to be arranged in the rotating machinery, and is used to obtain the monitoring data and DCS data of the rotating machinery; perform data preprocessing on the monitoring data and DCS data; perform feature data extraction on the monitoring data and DCS data after preprocessing to generate state monitoring feature data, fault diagnosis feature data and fault prediction feature data; perform state monitoring, fault diagnosis and fault prediction on the rotating machinery according to the state monitoring feature data, fault diagnosis feature data and fault prediction feature data; and is used to judge the health status of the rotating machinery of the nuclear power plant according to the state monitoring data, fault diagnosis data, fault prediction data, the historical situation of the rotating machinery and prior knowledge; and

[0034] The cloud is used to utilize key data and processing results sent back by the edge to establish, maintain, and push deployment of various algorithms, and to evaluate and make decisions on the comprehensive health status of the equipment.

[0035] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which instructions are stored. When the instructions are run on a computer, the method for health management of rotating machinery in a nuclear power plant is executed.

[0036] Compared with the prior art, the embodiments of the present invention have the following advantages and beneficial effects:

[0037] A method, system and medium for health management of rotating machinery in a nuclear power plant according to an embodiment of the present invention obtains monitoring data and DCS data of the rotating machinery; performs data preprocessing on the monitoring data and DCS data; performs feature data extraction on the monitoring data and DCS data after preprocessing to generate state monitoring feature data, fault diagnosis feature data and fault prediction feature data; performs state monitoring, fault diagnosis and fault prediction on the rotating machinery according to the state monitoring feature data, fault diagnosis feature data and fault prediction feature data; and judges the health status of the rotating machinery in the nuclear power plant according to the state monitoring data, fault diagnosis data, fault prediction data, the historical conditions of the rotating machinery and prior knowledge, thereby solving the problem of low operation and maintenance management efficiency caused by poor real-time and incomplete monitoring of rotating machinery in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0039] Figure 1 The figure is a flow chart of the health management method for rotating machinery in nuclear power plants.

[0040] Figure 2 This is a schematic diagram of the structure of the rotating machinery health management system in a nuclear power plant.

[0041] Figure 3 The schematic diagram of the structure of the rotating machinery health management system of a nuclear power plant is shown as an example.

[0042] Figure 4 Condition monitoring logic diagram for health management of rotating machinery in nuclear power plants.

[0043] Figure 5 Fault prediction logic diagram for rotating machinery health management in nuclear power plants.

[0044] Figure 6 Intelligent decision logic diagram for health management and maintenance of rotating machinery in nuclear power plants. DETAILED DESCRIPTION

[0045] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0046] In the following description, a large number of specific details are set forth in order to provide a thorough understanding of the present invention. However, it is apparent to one of ordinary skill in the art that these specific details are not necessarily employed to practice the present invention. In other embodiments, in order to avoid obscuring the present invention, well-known structures, circuits, materials, or methods are not specifically described.

[0047] Throughout the specification, references to "one embodiment," "an embodiment," "an example," or "an example" mean that the specific feature data, structure, or characteristic described in conjunction with the embodiment or example is included in at least one embodiment of the present invention. Therefore, the phrases "one embodiment," "an embodiment," "an example," or "an example" appearing in various places throughout the specification do not necessarily all refer to the same embodiment or example. In addition, specific feature data, structures, or characteristics may be combined in one or more embodiments or examples in any suitable combination and / or sub-combination. In addition, it will be understood by those of ordinary skill in the art that the illustrations provided herein are for illustrative purposes and that the illustrations are not necessarily drawn to scale. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0048] In the description of the present invention, the directions or positional relationships indicated by terms such as “front”, “rear”, “left”, “right”, “up”, “down”, “vertical”, “horizontal”, “high”, “low”, “inside” and “outside” are based on the directions or positional relationships shown in the accompanying drawings and are only for the convenience of describing the present invention and simplifying the description. They do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction. Therefore, they should not be understood as limiting the scope of protection of the present invention.

[0049] Example

[0050] In order to solve the problem of low operation and maintenance management efficiency caused by poor real-time monitoring of rotating machinery in the prior art, the embodiments of the present invention provide a nuclear power plant rotating machinery health management method, system and medium. In the first aspect, the embodiments of the present invention provide a nuclear power plant rotating machinery health management method, system and medium. Figure 1 As shown, including:

[0051] S1. Obtain monitoring data and DCS data of rotating machinery;

[0052] S2. Preprocessing the monitoring data and DCS data;

[0053] S3. Extract feature data from the pre-processed monitoring data and DCS data to generate condition monitoring feature data, fault diagnosis feature data, and fault prediction feature data;

[0054] For feature data extraction, a deep learning network can be specifically designed according to the vibration signal characteristics of the rotating bearing. For example, an embedded recurrent network architecture can be designed based on the time-frequency transformation and quasi-periodicity of the signal. Residual block stacking can be used to extract deep local features, and recurrent units can be used to extract time domain features, so as to extract weak fault features under low signal-to-noise ratio conditions.

[0055] S4. Perform condition monitoring, fault diagnosis and fault prediction on rotating machinery according to condition monitoring characteristic data, fault diagnosis characteristic data and fault prediction characteristic data;

[0056] In actual engineering, the model, size, load, speed, etc. of rotating machinery equipment are different from the test conditions, and the equipment on site is in normal state for a long time, so the amount of labeled fault data is small. In view of the difficulty in extracting features from time domain data, the difference in data under different working conditions, and the problem of too much data and few labels, the transfer learning method can be used to build a fault diagnosis or classification recognition algorithm.

[0057] The premise of rotating machinery fault prediction is based on a large amount of historical monitoring data, and is supported by prior knowledge and experience, so as to predict the life and equipment operation trend. Historical monitoring data is used as the data source to build comprehensive monitoring indicators. Big data algorithms (neural networks, random forests, SVM, Bayesian, etc.) are used to analyze and model key monitoring indicators to form a fault diagnosis model, and the consistency with the results is continuously compared to improve the accuracy of the model. The real-time data of rotating machinery status monitoring is connected to the prediction model for mining and analysis, and the model results are obtained and evaluated. The final results of the evaluation are saved to continue to optimize the model.

[0058] You can try to use high-order spectrum analysis and typical time-frequency transformation techniques, such as high-order moment spectrum, high-order cumulant spectrum, fractional Fourier transform, wavelet transform, Lipschitz index, fractal dimension, Cohen-type time-frequency transform, fuzzy function, etc. to comprehensively extract features from the signal, and combine PCA and popular learning to screen the features so that the obtained features can maximize the information of the monitoring data and minimize the size of the features. Then use neural networks, SVM, KNN, decision trees, combined with integrated algorithms such as Boosting and Bagging, to identify the operating status of rotating machinery and promptly discover its early faults.

[0059] S5. Determine the health status of the rotating machinery of the nuclear power plant based on the condition monitoring data, fault diagnosis data, fault prediction data, the historical conditions of the rotating machinery and prior knowledge.

[0060] The embodiment of the present invention obtains monitoring data and DCS data of rotating machinery; performs data preprocessing on the monitoring data and DCS data; performs feature data extraction on the preprocessed monitoring data and DCS data to generate state monitoring feature data, fault diagnosis feature data and fault prediction feature data; performs state monitoring, fault diagnosis and fault prediction on the rotating machinery according to the state monitoring feature data, fault diagnosis feature data and fault prediction feature data; and judges the health status of the rotating machinery of a nuclear power plant according to the state monitoring data, fault diagnosis data, fault prediction data, the historical conditions of the rotating machinery and prior knowledge, thereby solving the problem of low operation and maintenance management efficiency caused by poor real-time and incomplete monitoring of rotating machinery in the prior art.

[0061] Furthermore, the monitoring data includes: vibration monitoring data and acoustic emission data of the rotating machinery; and the DCS data includes the operating conditions of the unit and the operating parameters of the rotating machinery.

[0062] Optionally, the DCS data includes the operating conditions of the unit (such as inlet and outlet temperatures and pressures of important equipment / pipe sections, unit operating power, etc.) and important operating parameters of rotating machinery (such as coil voltage / current, lubricating oil temperature, sealing pressures at various levels, etc.).

[0063] Further, the monitoring data and DCS data are preprocessed; including:

[0064] According to the data structure and data length of the monitoring data and DCS data information, the monitoring data and DCS data are screened and filtered, and the filtered monitoring data and DCS data are fused to obtain the pre-processed monitoring data and DCS data.

[0065] Further, the state monitoring, fault diagnosis and fault prediction of the rotating machinery are performed according to the state monitoring characteristic data, the fault diagnosis characteristic data and the fault prediction characteristic data; including:

[0066] Determine whether to perform a threshold alarm, trend alarm and / or comprehensive alarm based on the condition monitoring characteristic data; if so, visualize the alarm information and alarm log in three-dimensional results; if a comprehensive alarm is determined based on the condition monitoring characteristic data, determine whether to start fault diagnosis and fault prediction; if so, perform fault diagnosis and fault prediction.

[0067] Further, the fault diagnosis includes:

[0068] Diagnose the equipment fault type, location and degree according to the fault diagnosis characteristic data to obtain diagnostic data;

[0069] Determine the time until equipment failure becomes serious and whether secondary failures will occur based on diagnostic data, fault-related data and historical trends.

[0070] Furthermore, fault prediction includes:

[0071] Compare the trend of fault prediction feature data with the standard trend to determine whether the rotating machinery will have potential faults, the fault type, the fault location, and the time when the fault will occur;

[0072] Use the fault anomaly database to obtain the historical average time before failure to predict the time when the next fault will occur;

[0073] By comparing the trend of the characteristics of potential faults with the standard trend, potential fault information, next fault time and remaining life of the rotating machinery are generated.

[0074] Furthermore, the health status of the rotating machinery of the nuclear power plant is judged based on the condition monitoring data, fault diagnosis data, fault prediction data, the historical situation of the rotating machinery and prior knowledge; including:

[0075] The health status of rotating machinery is comprehensively evaluated using condition monitoring results, fault diagnosis results, fault prediction results, maintenance results and prior knowledge.

[0076] An exemplary method for health management of rotating machinery in a nuclear power plant comprises:

[0077] (1) Receiving and storing data and information

[0078] The data sources of the rotating machinery health management system include data obtained from additional sensors (stored in binary files), data obtained from DCS, offline test data, and result data after feature extraction, condition monitoring, fault diagnosis, and fault prediction, as well as intermediate data that need to be visualized. The main information sources of the system include equipment design, manufacturing, installation, and retirement information (from files or manual input), abnormal equipment operating status information (from historical diagnosis, prediction, and evaluation results), maintenance work order information in the maintenance information management system (from the database of the maintenance information management system), and equipment maintenance plan in the calculation management module (from the equipment maintenance outline or production plan).

[0079] The system combines standard data transmission protocols to establish efficient, safe, accurate and reasonable data transmission logical channels (non-physical channels) to achieve real-time transmission of the entire data flow. At the same time, based on the above data information, the advantages of structured and unstructured databases are used to design a reasonable storage mode for each type of data to adapt to large-capacity and diversified data storage scenarios.

[0080] (2) Data preprocessing and feature extraction

[0081] The system screens the monitoring data according to the data structure and data length of the monitoring data and DCS data information, fuses the monitoring data and the data acquired by the DCS using the preprocessing algorithm in the algorithm library, and saves the data or data information in the database.

[0082] Filter out unqualified data, write monitoring data or related information into the database, and establish corresponding indexes.

[0083] The system uses the state monitoring feature extraction algorithm, fault diagnosis feature extraction algorithm and fault prediction feature extraction algorithm in the algorithm library to transform the pre-processed monitoring data information, generate state monitoring features, fault diagnosis features and fault prediction features, and use them as inputs of the state monitoring node, fault diagnosis node and fault prediction node respectively. At the same time, it provides trend visualization service for the main interface, calculates the features that need to be visualized through the customized visualization feature extraction algorithm, and uses the data and time information as the input of trend visualization.

[0084] The system establishes service deployment interfaces for condition monitoring feature extraction algorithms, fault diagnosis feature extraction algorithms, fault prediction feature extraction algorithms, and visualization feature extraction algorithms, establishes a visualization standard library, connects upstream and downstream data of service scenarios, and establishes a standard information transmission channel.

[0085] (3) Condition monitoring

[0086] It mainly includes threshold alarm, trend alarm and comprehensive alarm, and is also a prerequisite for the start of fault diagnosis and fault prediction execution nodes. Its main logic diagram is referenced Figure 4 shown.

[0087] The system establishes service deployment interfaces for threshold alarms, trend alarms, and comprehensive alarms, and connects all the data flows in the above figure. It establishes an alarm interaction page based on the alarm results. The interactive information includes alarm devices and alarm logs, provides over-limit feature names, and writes the results into the status monitoring result database.

[0088] (4) Fault diagnosis

[0089] It mainly includes two parts: online fault diagnosis and result evaluation. The online fault diagnosis process mainly uses fault diagnosis feature data to diagnose the type, location and degree of equipment faults. The result evaluation process mainly uses the data obtained from the equipment diagnosis results and historical trends to determine the time required for the equipment to become seriously faulty, as well as the fault association database to determine possible secondary faults.

[0090] The system has established service deployment interfaces for empirical formulas, physical simulation models, data-driven models, and result evaluation algorithms, and deployed an incremental learning application framework. In addition, all data flows in the above figure are connected, and an interactive page is established based on the fault diagnosis results. The interactive information includes the fault location, fault severity, fault type, and the time required for the equipment to have a serious fault and secondary fault information. The results are written into the fault diagnosis result database, and the main characteristic trends before the fault are saved in the fault prediction standard database to prepare for subsequent fault prediction.

[0091] (5) Fault prediction

[0092] It mainly includes potential fault prediction and equipment remaining life prediction. Potential fault prediction is mainly based on the trend of fault prediction feature data and the standard trend to determine whether the equipment will have potential faults, the main type of fault, the fault location and the time when the fault may occur. At the same time, the fault anomaly database is used to obtain the historical average time before failure and give the time when the next fault may occur. The logic diagram is as follows Figure 5 shown.

[0093] The system has established service deployment interfaces for mathematical statistics algorithms, data-driven algorithms, and life assessment algorithms, and connected all the data streams in the above figure. An interactive page is established based on the fault prediction results. The interactive information includes the comparison between the actual trend and standard trend of potential fault characteristics, potential fault information, the next failure time, and the remaining life of the equipment. The results are written into the fault prediction result database.

[0094] (6) Intelligent Decision-making

[0095] The system mainly uses the status monitoring results, fault diagnosis results and fault prediction results database to conduct a comprehensive evaluation of the diagnosis results. According to the maintenance information management system, it mines the mapping relationship between historical faults and maintenance strategies. Combined with the plan management module, it formulates maintenance decision suggestions and plans, generates automatic reports, and gives feedback results based on the workers' maintenance results. The feedback results and abnormal information are written into the equipment information database to accumulate experience for subsequent health management. Its logic diagram is as follows: Figure 6 shown.

[0096] The system has established service deployment interfaces for maintenance work order information mining algorithms, comprehensive health status assessment algorithms, and optimal decision-making algorithms, established real-time interactive interfaces for the planning management module, equipment information database, and maintenance decision knowledge base, and interfaces with the maintenance work order management system. It has also opened up all the data flows in the above figure, and can autonomously generate reports based on maintenance decision results. It can also rate maintenance decision recommendations and plans based on maintenance completion reports, write the results into the equipment information database, and automatically correct the recommendation weights in the maintenance decision knowledge base.

[0097] (7) Planning management module and maintenance

[0098] The system establishes a rotating machinery planning management module according to the equipment maintenance plan, uniformly manages the periodic maintenance time and planned maintenance window of rotating machinery, realizes the input of planned maintenance information through an interactive interface, and uses this to guide the time window for maintenance decisions.

[0099] (8) Knowledge base and maintenance

[0100] The system establishes an equipment operation and maintenance knowledge base for rotating machinery based on professional knowledge and knowledge obtained from the maintenance management system, and uniformly manages expert experience, fault knowledge, maintenance plans, operation and maintenance strategies, etc., to form a core equipment operation and maintenance knowledge base.

[0101] The key equipment failure knowledge base and maintenance knowledge base are regularized in the form of tables and rich texts, and the failure knowledge base, maintenance knowledge base and field work are associated through the relationship of "failure problem - recommended operation - maintenance feedback". The key equipment failure knowledge base is mainly a fault tree stored according to the failure. When a device fails, the fault tree is traversed according to the reasoning rules based on the failure situation.

[0102] The maintenance knowledge base manages maintenance plans for known faults of key equipment, making it easy for maintenance personnel to quickly find corresponding maintenance plans and carry out maintenance work when encountering key equipment maintenance.

[0103] (9) Algorithm platform and visualization standard library

[0104] The algorithm platform serves as an integration center for algorithm files, and also as a programming environment and compiler for algorithm development. The algorithms in the algorithm platform can be published to each execution node in the form of services. The system establishes a suitable algorithm platform, provides mainstream languages ​​(such as java, python (can be connected to python third-party libraries) and C) and database addition, modification, deletion and query algorithm module development interfaces, and provides common standard algorithm modules for each process node (such as signal processing, feature processing, intelligent classification, intelligent regression and other algorithms), algorithm file addition, modification, deletion and query management and publishing functions.

[0105] The visualization standard library serves as an integration center for visualization standard parts. Visualization standard parts can be deployed in the form of controls on a blank page. Suppliers should establish appropriate visualization standard libraries based on demand and provide mainstream visualization standard parts (including but not limited to bar charts, line charts, area charts, pie charts, ring charts, scatter charts, bubble charts, chord charts, tree ring charts, wrench charts, survey charts, radar charts, rose charts, word cloud charts, network charts, text flow charts, migration charts, cluster charts, network node charts, box charts, circle charts, heat maps, 3D charts, etc.), supporting common visualization functions such as multiple series, multiple Y axes, and real-time updates.

[0106] Therefore, the embodiment of the present invention utilizes the basic information, maintenance information, plan information of rotating equipment, additional sensors arranged on the equipment, equipment status information in the DCS system and offline test data, and develops and continuously maintains diagnostic algorithms through the construction of a big data algorithm platform, so as to realize rotating machinery data monitoring and early warning, fault diagnosis and prediction; through the accumulation and management of monitoring data, diagnostic results and operation and maintenance knowledge, operation and maintenance strategy support is provided, decision-making basis is provided for production and maintenance, and the associated management of rotating machinery operation, maintenance and maintenance is realized, so as to continuously improve maintenance technology and maintenance system, and enhance the safety and economy of nuclear power plants.

[0107] The embodiment of the present invention can realize the management of the whole life of nuclear power equipment by building a data center for the whole life of equipment and integrating the operation data of rotating equipment in nuclear power plants; realize four maintenance modes for rotating equipment, namely, post-maintenance, planned maintenance, condition-based maintenance, and predictive maintenance. Continuously improve maintenance technology and maintenance system; realize data monitoring and early warning, fault diagnosis, fault analysis, and early warning of rotating machinery; realize the associated management of operation, maintenance, and maintenance of rotating machinery.

[0108] In a second aspect, an embodiment of the present invention provides a nuclear power plant rotating machinery health management system, referring to Figure 2 As shown, including:

[0109] An acquisition unit, used for acquiring monitoring data and DCS data of rotating machinery;

[0110] A preprocessing unit, used for performing data preprocessing on the monitoring data and DCS data;

[0111] An extraction unit, used to extract feature data from the monitoring data and DCS data after preprocessing, and generate condition monitoring feature data, fault diagnosis feature data and fault prediction feature data;

[0112] A diagnosis and prediction unit, used for performing condition monitoring, fault diagnosis and fault prediction on the rotating machinery according to the condition monitoring characteristic data, the fault diagnosis characteristic data and the fault prediction characteristic data; and

[0113] The judgment unit is used to judge the health status of the rotating machinery of the nuclear power plant based on the status monitoring data, fault diagnosis data, fault prediction data, the historical conditions of the rotating machinery and prior knowledge.

[0114] In a third aspect, an embodiment of the present invention provides a nuclear power plant rotating machinery health management system, including:

[0115] The edge end is used to be arranged in the rotating machinery, and is used to obtain the monitoring data and DCS data of the rotating machinery; perform data preprocessing on the monitoring data and DCS data; perform feature data extraction on the monitoring data and DCS data after preprocessing to generate state monitoring feature data, fault diagnosis feature data and fault prediction feature data; perform state monitoring, fault diagnosis and fault prediction on the rotating machinery according to the state monitoring feature data, fault diagnosis feature data and fault prediction feature data; and is used to judge the health status of the rotating machinery of the nuclear power plant according to the state monitoring data, fault diagnosis data, fault prediction data, the historical situation of the rotating machinery and prior knowledge; and

[0116] The cloud is used to utilize key data and processing results sent back by the edge to establish, maintain, and push deployment of various algorithms, and to evaluate and make decisions on the comprehensive health status of the equipment.

[0117] Optionally, the cloud is used to utilize key data and processing results sent back by the edge to mainly establish, maintain, and push deployment of various algorithms (alarm, diagnosis, prediction, etc.) to evaluate and make decisions on the comprehensive health status of the equipment.

[0118] The principle and concept of the health management system are the same as the above method.

[0119] Exemplary, reference Figure 3 As shown. The nuclear power plant rotating machinery health management system adopts a health management model that combines the edge and the cloud. By deploying edge hardware near the rotating machinery, it completes most of the work such as data and information reception, data preprocessing and feature extraction, status monitoring, fault diagnosis, and fault prediction, and transmits key data and processing results back to the cloud through the network; in the cloud, it mainly establishes, maintains, and pushes various algorithms (alarm, diagnosis, prediction, etc.) to evaluate and make decisions on the comprehensive health status of the equipment.

[0120] The system uses the basic information, maintenance information, planning information of rotating equipment, additional sensors arranged on the equipment, equipment status information in the DCS system and offline test data to establish a complete rotating machinery health management system through intelligent data preprocessing, feature extraction, status warning, fault diagnosis, fault prediction and intelligent maintenance decision-making.

[0121] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which instructions are stored. When the instructions are run on a computer, the method for health management of rotating machinery in a nuclear power plant is executed.

[0122] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for health management of rotating machinery in a nuclear power plant, characterized in that: include: Obtain monitoring data and DCS data of rotating machinery; The monitoring data includes: vibration monitoring data and acoustic emission data of the rotating machinery; the DCS data includes the operating conditions of the unit and the operating parameters of the rotating machinery; Performing data preprocessing on the monitoring data and DCS data; specifically, including: screening and filtering the monitoring data and DCS data according to the data structure and data length of the monitoring data and DCS data information, fusing the filtered monitoring data and DCS data, and obtaining preprocessed monitoring data and DCS data; Extract feature data from the pre-processed monitoring data and DCS data to generate condition monitoring feature data, fault diagnosis feature data, and fault prediction feature data; For the vibration signals of rotating machinery, an embedded recurrent network architecture is designed based on the time-frequency transformation and quasi-periodicity of vibration signals. Deep local features are extracted based on residual block stacking, and time domain features are extracted based on recurrent units to achieve the extraction of weak fault features under low signal-to-noise ratio conditions. High-order spectrum analysis and typical time-frequency transformation techniques are used to comprehensively extract features from signals; the high-order spectrum analysis and typical time-frequency transformation techniques include: high-order moment spectrum, high-order cumulant spectrum, fractional Fourier transform, wavelet transform, Lipschitz index, fractal dimension, Cohen-type time-frequency transform and fuzzy function; and PCA and popular learning are combined to screen features so that the obtained features can maximize the information of the monitoring data, while minimizing the size of the features; Performing condition monitoring, fault diagnosis and fault prediction on rotating machinery according to condition monitoring characteristic data, fault diagnosis characteristic data and fault prediction characteristic data; specifically including: judging whether to perform threshold alarm, trend alarm and / or comprehensive alarm according to condition monitoring characteristic data; if so, visually displaying alarm information and alarm log in three-dimensional results; if it is judged that comprehensive alarm is performed according to condition monitoring characteristic data, judging whether to start fault diagnosis and fault prediction; if so, performing fault diagnosis and fault prediction; Fault diagnosis includes: diagnosing the equipment fault type, location and degree based on fault diagnosis characteristic data to obtain diagnostic data; determining the time required for the equipment to become seriously faulty and whether a secondary fault has occurred based on the diagnostic data, fault-related data and historical trends; Fault prediction includes: comparing the trend of fault prediction feature data with the standard trend to determine whether the rotating machinery will have a potential fault, the fault type, the fault location and the time of the fault; using the fault anomaly database to obtain the historical average time before failure to predict the time of the next fault; using the trend of the potential fault feature to compare with the standard trend to generate potential fault information, the next fault time and the remaining life of the rotating machinery; The health status of rotating machinery in nuclear power plants is determined based on condition monitoring data, fault diagnosis data, fault prediction data, historical conditions of rotating machinery, and prior knowledge.

2. The method for health management of rotating machinery in a nuclear power plant according to claim 1, characterized in that: Determine the health status of rotating machinery in nuclear power plants based on condition monitoring data, fault diagnosis data, fault prediction data, historical conditions of rotating machinery, and prior knowledge; including: The health status of rotating machinery is comprehensively evaluated using condition monitoring results, fault diagnosis results, fault prediction results, maintenance results and prior knowledge.

3. A nuclear power plant rotating machinery health management system, characterized in that: A method for health management of rotating machinery in a nuclear power plant for implementing claim 1 or 2, comprising: An acquisition unit, used for acquiring monitoring data and DCS data of rotating machinery; A preprocessing unit, used for performing data preprocessing on the monitoring data and DCS data; An extraction unit, used to extract feature data from the monitoring data and DCS data after preprocessing, and generate condition monitoring feature data, fault diagnosis feature data and fault prediction feature data; A diagnosis and prediction unit, used for performing condition monitoring, fault diagnosis and fault prediction on the rotating machinery according to the condition monitoring characteristic data, the fault diagnosis characteristic data and the fault prediction characteristic data; and a judgment unit for judging the health status of the rotating machinery of the nuclear power plant according to the status monitoring data, the fault diagnosis data, the fault prediction data, the historical conditions of the rotating machinery and the prior knowledge.

4. A nuclear power plant rotating machinery health management system, characterized in that: A method for health management of rotating machinery in a nuclear power plant for implementing claim 1 or 2, comprising: The edge end is used to be arranged on the rotating machinery to obtain the monitoring data and DCS data of the rotating machinery; perform data preprocessing on the monitoring data and DCS data; perform feature data extraction on the monitoring data and DCS data after preprocessing to generate state monitoring feature data, fault diagnosis feature data and fault prediction feature data; perform state monitoring, fault diagnosis and fault prediction on the rotating machinery according to the state monitoring feature data, fault diagnosis feature data and fault prediction feature data; judge the health status of the rotating machinery of the nuclear power plant according to the state monitoring data, fault diagnosis data, fault prediction data, the historical situation of the rotating machinery and prior knowledge; and the cloud end is used to use the key data and processing results sent back by the edge end to establish, maintain and push deployment of various algorithms, and evaluate and make decisions on the comprehensive health status of the equipment.

5. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and when the instructions are executed on a computer, the method for health management of rotating machinery in a nuclear power plant as claimed in claim 1 or 2 is executed.

Images