Intelligent fault diagnosis model updating method, processor and machine-readable storage medium
Through the intelligent fault diagnosis model update method, the fault diagnosis model of multiple devices is matched and updated, and the problems of high development difficulty, long cycle and unstable effect in the existing technology are solved, and low-code development, rapid iteration and stable diagnostic effects are achieved. Intelligent fault diagnosis.
Patent Information
- Application Number
- CN202411380663.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-30
- Publication Date
- 2024-12-10
- Estimated Expiration
- 2044-09-30
AI Technical Summary
The development of existing intelligent fault diagnosis methods is difficult, with high technical threshold, long development cycle, unstable diagnostic effects after deployment, and difficult to maintain in the later stage, especially in the case of multi-device deployment.
Through the intelligent fault diagnosis model update method, the training data input by the user is obtained, the adapted target model is matched, the pre-training and target training parameters are processed, the model deployed to multiple devices is retrained, and the update parameters are calculated based on the multi-device data to realize the model automated updates and iterations.
It reduces the development difficulty and technical threshold, shortens the development cycle, accelerates iteration speed, stabilizes the diagnostic effect, reduces the difficulty of later maintenance, and optimizes the diagnostic effect through multi-device data and automatically updates model parameters.
Smart Images
Figure CN118885861B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of fault diagnosis, and in particular to an intelligent fault diagnosis model updating method, a processor and a machine-readable storage medium. Background Art
[0002] A low-code development platform is a software that provides a development environment, allowing users to quickly create applications in a visual way, reducing the development cycle of application software and reducing the development cost of application software. At present, there are many low-code development platforms in my country that have been implemented, but the functions of these platforms are mostly concentrated on workflows, business reports and statistical data analysis, with the focus on the research and development of BPM platforms for internal business process management.
[0003] Fault diagnosis is a technology that determines the overall or local health status of a machine during its operation, detects faults and their causes at an early stage, and predicts the development trend of faults. In recent years, with the rapid development of artificial intelligence, many machine learning algorithms have been applied to fault diagnosis. Intelligent fault diagnosis methods combined with machine learning have high diagnostic efficiency and good diagnostic effects, but these methods are difficult to develop and require developers to have a considerable level of programming and relevant professional knowledge.
[0004] In view of this, there is an urgent need for an intelligent fault diagnosis model updating method that can solve the above problems. Summary of the invention
[0005] In order to solve the above technical problems, the purpose of the embodiments of the present application is to provide an intelligent fault diagnosis model updating method, device and machine-readable storage medium.
[0006] In order to achieve the above-mentioned object, the first aspect of the present application provides a method for updating an intelligent fault diagnosis model, the method comprising:
[0007] Acquire first training data input by a user;
[0008] Matching a target intelligent fault diagnosis model adapted to the first training data in an intelligent model library according to the first training data;
[0009] Matching at least one pre-screened fault diagnosis model in the intelligent model library;
[0010] Processing the pre-training parameters and the target training parameters to obtain the first training parameters of the target intelligent fault diagnosis model, wherein the pre-training parameters are obtained by training the pre-screening fault diagnosis model, and the target training parameters are obtained by training the target intelligent fault diagnosis model;
[0011] When the target intelligent fault diagnosis model deployed in multiple devices triggers a start training instruction, retrain the target fault diagnosis model in each device, and obtain a second training parameter of each target fault diagnosis model according to the training result;
[0012] The third training parameters are recalculated based on the second training parameters of the multiple devices and sent to all devices to obtain a target intelligent fault diagnosis model after the parameters are updated.
[0013] In an embodiment of the present application, the step of matching a target intelligent fault diagnosis model adapted to the first training data in an intelligent model library according to the first training data includes:
[0014] Acquire second training data of the intelligent fault diagnosis model in the intelligent model library;
[0015] Calculating a distribution difference between the first training data and the second training data;
[0016] Recommending and ranking the intelligent fault diagnosis models in the intelligent model library according to the distribution differences;
[0017] The intelligent fault diagnosis model whose recommendation ranking is within a preset range is matched as the target intelligent fault diagnosis model.
[0018] In an embodiment of the present application, the step of calculating the distribution difference between the first training data and the second training data includes:
[0019] Calculating the MMD distance between the first training data and the second training data according to the recommendation function (1) to determine the distribution difference between the first training data and the second training data;
[0020] Formula (1);
[0021] The first training data is ,in, is the first training data in the input Sample; the second training data is ,in, is the first samples; is the norm of the matrix, is the kernel function. represents the number of samples in the first training data, represents the number of samples in the second training data, Indicates the first training data in the input samples, Indicates the first samples.
[0022] In an embodiment of the present application, the step of processing the pre-training parameters and the target training parameters to obtain the first training parameters of the target intelligent fault diagnosis model includes:
[0023] Acquire a first eigenvector of the target intelligent fault diagnosis model and a second eigenvector of the pre-screening fault diagnosis model;
[0024] Calculate the generalized Jed coefficient between the first eigenvector and the second eigenvector using formula (2);
[0025] Formula (2);
[0026] in, , represents vector product; \left | \left | {{M}_{u}} \right | \right |=\sqrt[{2}] {\sum ^{g}_{v=1} {{l}^{2}_{v}}} , represents the modulus of the first eigenvector; \left | \left | {{M}_{s}} \right | \right |=\sqrt[{2}] {\sum ^{g}_{v=1} {{s}^{2}_{v}}} , represents the modulus of the second eigenvector; represents the first eigenvector eigenvalues; represents the first eigenvector in the second eigenvector eigenvalues, and according to the generalized Jaccard coefficient, select a model corresponding to a second eigenvector with a high matching degree with the first eigenvector to perform model training to obtain a trained first training parameter.
[0027] In an embodiment of the present application, when the target intelligent fault diagnosis model deployed in multiple devices triggers the start training instruction, the start condition of the start training of the target intelligent fault diagnosis model deployed in multiple devices is:
[0028] ;
[0029] , is the accuracy of the intelligent fault diagnosis model, represents the number of true faults in the fault samples predicted by the intelligent fault diagnosis model, represents the number of false faults in the fault samples predicted by the intelligent fault diagnosis model, is a pre-set threshold;
[0030] To count the number of samples obtained during the operation of the intelligent fault diagnosis model deployed on a single device, is a pre-set threshold.
[0031] In the embodiment of the present application, the parameters of the target intelligent fault diagnosis model in the step of recalculating the third training parameters according to the second training parameters of the plurality of devices and sending them to all devices to obtain the target intelligent fault diagnosis model after the updated parameters are calculated and updated using formula (3):
[0032] Formula (3);
[0033] in, is the number of devices that have deployed intelligent fault diagnosis methods, For the The second training parameters of the device, is the recalculated third training parameter.
[0034] A second aspect of the present application provides a processor, which is configured to execute the intelligent fault diagnosis model updating method described in the above method.
[0035] A third aspect of the present application provides a machine-readable storage medium having instructions stored thereon, the instructions being used to enable a machine to execute the intelligent fault diagnosis model updating method as described in the above method.
[0036] A fourth aspect of the present application provides a fault diagnosis method, the method comprising:
[0037] Obtain equipment operating data;
[0038] Inputting the equipment operating condition data into a fault diagnosis model obtained according to the intelligent fault diagnosis model updating method as described above;
[0039] The equipment fault type is outputted through the fault diagnosis model.
[0040] A fifth aspect of the present application provides a machine-readable storage medium having instructions stored thereon, the instructions being used to enable a machine to execute the fault diagnosis method as described above.
[0041] Through the above technical scheme, firstly, the first training data input by the user is obtained; according to the first training data, the target intelligent fault diagnosis model adapted to the first training data is matched in the intelligent model library; at least one pre-screening fault diagnosis model adapted is matched in the intelligent model library; the pre-training parameters and the target training parameters are processed to obtain the first training parameters of the target intelligent fault diagnosis model, wherein the pre-training parameters are obtained by training the pre-screening fault diagnosis model, and the target training parameters are obtained by training the target intelligent fault diagnosis model; when the target intelligent fault diagnosis model deployed in multiple devices triggers the start training instruction, the target fault diagnosis model in each device is retrained, and the second training parameters of each target fault diagnosis model are obtained according to the training results; the third training parameters are recalculated according to the second training parameters of multiple devices and sent to all devices to obtain the target intelligent fault diagnosis model after the updated parameters.
[0042] The intelligent fault diagnosis model updating method combines the model recommendation design, the designed model parameter recommendation, and the model parameter updating method after deployment to optimize the design, deployment, and model iterative updating method of the intelligent fault diagnosis model. In general, the whole method has the following advantages:
[0043] (1) Reducing the development difficulty and technical threshold of intelligent fault diagnosis methods: The embodiments of the present invention reduce the professional knowledge requirements of developers in aspects related to intelligent fault diagnosis by intelligently recommending intelligent fault diagnosis models. At the same time, low-code technology is used to reduce the coding work of developers, thereby reducing the coding ability requirements of developers.
[0044] (2) Shortening the development cycle of the intelligent fault diagnosis method and accelerating the iteration speed of the intelligent fault diagnosis model: On the one hand, the low-code technology used in the embodiment of the present invention can reduce the programming time of developers. On the other hand, using the intelligent fault diagnosis model matching mechanism constructed by the embodiment of the present invention, developers can select multiple models similar to their own designed models for training at the same time, thereby shortening the iteration time of the model.
[0045] (3) Stabilizing the diagnostic effect after the deployment of the intelligent fault diagnosis method and reducing the difficulty of subsequent maintenance: The multi-device diagnostic algorithm model parameter update mechanism designed in the embodiment of the present invention can make full use of the data of multiple devices to improve the stability of the diagnostic effect after the deployment of the intelligent fault diagnosis method. At the same time, it can automatically update the model parameters on each device and reduce the operation of updating the model of multiple devices.
[0046] Other features and advantages of the embodiments of the present application will be described in detail in the subsequent specific implementation section. BRIEF DESCRIPTION OF THE DRAWINGS
[0047] The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present application, but do not constitute a limitation on the embodiments of the present application. In the accompanying drawings:
[0048] Figure 1 A flowchart of an intelligent fault diagnosis model updating method provided in an embodiment of the present application;
[0049] Figure 2 A flowchart of each stage of an intelligent fault diagnosis model updating method provided in an embodiment of the present application;
[0050] Figure 3 A schematic diagram of a process for updating parameters in an intelligent fault diagnosis model updating method provided in an embodiment of the present application;
[0051] Figure 4 A structural block diagram of an intelligent fault diagnosis model updating device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0052] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the specific implementation methods described herein are only used to illustrate and explain the embodiments of the present application, and are not used to limit the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.
[0053] It should be noted that if the embodiments of the present application involve directional indications (such as up, down, left, right, front, back...), such directional indications are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.
[0054] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present application, the descriptions of "first", "second", etc. are only for descriptive purposes and cannot be understood as indicating or suggesting their relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist, and is not in the flowchart of the intelligent fault diagnosis model updating method required by the present application.
[0055] At present, the intelligent fault diagnosis method combined with machine learning has the following problems:
[0056] (1) The developers of intelligent fault diagnosis algorithms not only need to be proficient in relevant programming languages, but also need to have relevant professional knowledge in the application field of fault diagnosis methods. This makes the development of intelligent fault diagnosis methods extremely difficult and the technical threshold is very high; (2) Intelligent fault diagnosis methods not only require a lot of tedious code writing work during design, but also require modification of the diagnosis algorithm during training due to poor training results. This leads to a long development cycle and slow iteration speed of intelligent fault diagnosis methods; (3) Since new data is continuously generated during the operation of application equipment, and the application equipment may be in an unknown operating state, the intelligent fault diagnosis method may have unstable diagnostic effects after deployment, that is, the diagnostic accuracy decreases; (4) In actual applications, intelligent fault diagnosis methods are often deployed on multiple devices, and the diagnostic algorithms after deployment generally need to be maintained and iterated, which makes the later maintenance of intelligent fault diagnosis methods difficult and the algorithm model iteration workload large.
[0057] Therefore, the present application aims to solve the above problems.
[0058] like Figure 1 As shown, the intelligent fault diagnosis model updating method of the embodiment of the present invention may include the following steps:
[0059] Step S10, obtaining first training data input by a user;
[0060] Step S20, matching a target intelligent fault diagnosis model that is compatible with the first training data in an intelligent model library according to the first training data;
[0061] Step S30, matching at least one pre-screening fault diagnosis model in the intelligent model library;
[0062] Step S40, processing the pre-training parameters and the target training parameters to obtain the first training parameters of the target intelligent fault diagnosis model, wherein the pre-training parameters are obtained by training the pre-screening fault diagnosis model, and the target training parameters are obtained by training the target intelligent fault diagnosis model;
[0063] Step S50, when the target intelligent fault diagnosis model deployed in multiple devices triggers a start training instruction, retrain the target fault diagnosis model in each device, and obtain a second training parameter of each target fault diagnosis model according to the training result;
[0064] Step S60, recalculating the third training parameters according to the second training parameters of the multiple devices and sending them to all the devices to obtain the target intelligent fault diagnosis model after the parameters are updated.
[0065] Specifically, in the embodiment of the present invention, the intelligent fault diagnosis model updating method combines the model recommendation design, the designed model parameter recommendation, and the method of updating the model parameters after deployment to optimize the design, deployment, and model iterative updating method of the intelligent fault diagnosis model. In general, the whole method has the following advantages:
[0066] (1) Reducing the development difficulty and technical threshold of intelligent fault diagnosis methods: The embodiments of the present invention reduce the professional knowledge requirements of developers in aspects related to intelligent fault diagnosis by intelligently recommending intelligent fault diagnosis models. At the same time, low-code technology is used to reduce the coding work of developers, thereby reducing the coding ability requirements of developers.
[0067] (2) Shortening the development cycle of the intelligent fault diagnosis method and accelerating the iteration speed of the intelligent fault diagnosis model: On the one hand, the low-code technology used in the embodiment of the present invention can reduce the programming time of developers. On the other hand, using the intelligent fault diagnosis model matching mechanism constructed by the embodiment of the present invention, developers can select multiple models similar to their own designed models for training at the same time, thereby shortening the iteration time of the model.
[0068] (3) Stabilizing the diagnostic effect after the deployment of the intelligent fault diagnosis method and reducing the difficulty of subsequent maintenance: The multi-device diagnostic algorithm model parameter update mechanism designed in the embodiment of the present invention can make full use of the data of multiple devices to improve the stability of the diagnostic effect after the deployment of the intelligent fault diagnosis method. At the same time, it can automatically update the model parameters on each device and reduce the operation of updating the model of multiple devices.
[0069] Specifically, the process of the embodiment of the present invention is as follows: Figure 2 As shown, the details are as follows:
[0070] Phase 1: Design Phase
[0071] At this stage, users can design the fault diagnosis model structure and its training process by dragging and dropping on the provided low-code design page. The process nodes can modify the relevant parameters of the diagnosis process by modifying the configuration parameters.
[0072] When designing, users can input the training data of the intelligent fault diagnosis process, and the intelligent fault diagnosis process recommendation mechanism will recommend appropriate process models from the process model library to facilitate users to carry out personalized customized design of the intelligent fault diagnosis process.
[0073] The specific contents of the recommendation mechanism are as follows:
[0074] The first training data input by the user is ,in, is the first training data in the input The second training data used by the intelligent fault diagnosis model in the process model library is ,in, is the first The MMD distance between the two is calculated to determine the distribution difference between the first input training data and the second training data used by the intelligent fault diagnosis model in the process model library, and the intelligent fault diagnosis process model is recommended based on this. Therefore, the recommendation function of the intelligent fault diagnosis process recommendation mechanism is:
[0075] Formula (1);
[0076] in, is the norm of the matrix, is the kernel function. In this formula, the Gaussian kernel function is used. represents the number of samples in the first training data, represents the number of samples in the second training data, Indicates the first training data in the input samples, Indicates the first samples.
[0077] The closer the value of the recommendation function is to zero, the higher the recommendation priority of the intelligent fault diagnosis process model in the process model library. Users can select the appropriate process from the recommended process model according to their needs and make personalized modifications in the low-code design interface.
[0078] Phase 2: Training Phase
[0079] When the user completes the design of the intelligent fault diagnosis model structure and its training process, the training phase can be carried out, and the training process can be started to complete the training of the intelligent fault diagnosis model. During the training process, the designed model structure may not achieve a good diagnostic effect. At this time, the designed model structure needs to be modified and adjusted before training. In order to reduce the time used to modify and adjust the model structure, the present invention proposes an intelligent fault diagnosis model matching mechanism, which recommends models with high matching degrees in the diagnosis model library to users, and trains them simultaneously with the designed models. The specific contents of the matching mechanism are as follows:
[0080] After the user designs the intelligent fault diagnosis model, the model feature vector designed by the user can be obtained based on the type, number of layers, layer parameters and other related information of the designed model. ,in, Design the eigenvalues of model-related information for users. Models in the diagnostic model library also have model eigenvectors set ,in, is the characteristic value of the relevant information of the model in the diagnostic model library. The matching function can be obtained by calculating the generalized Jedka coefficient of the two:
[0081] ;
[0082] in, , represents vector product; \left | \left | {{M}_{u}} \right | \right |=\sqrt[{2}] {\sum ^{g}_{v=1} {{l}^{2}_{v}}} , represents the modulus of the model feature vector designed by the user; \left | \left | {{M}_{s}} \right | \right |=\sqrt[{2}] {\sum ^{g}_{v=1} {{s}^{2}_{v}}} , represents the modulus of the model feature vector in the diagnostic model library; Represents the first eigenvalues; Represents the first feature values. The value range is \left [ {0, 1} \right ] The closer the value of the matching function is to 1, the higher the matching degree between the two.
[0083] Users can select several recommended models with high matching degree for training at the same time according to their needs. After the training is completed, the system will record the relevant parameters of each model. Users can select appropriate model parameters based on the training results for use in subsequent model deployment.
[0084] Phase 3: Deployment
[0085] After the training phase is over, the user selects the trained intelligent fault diagnosis model and modifies the node configuration on the pre-built unified deployment process metamodel, including the configuration for reading the data to be diagnosed, the configuration for processing the data to be diagnosed, and the parameters of the intelligent fault diagnosis model. After the modification is completed, the deployment process can be started to perform device fault diagnosis.
[0086] During the operation of the deployment process, data that can be used for model training will be continuously obtained. At the same time, since the training data input by the user cannot contain all the information of the device operation and the device operation process will also generate data that is completely different from the past, the previously trained model will make incorrect judgments. At the same time, in practical applications, intelligent fault diagnosis methods are often deployed on multiple devices. Therefore, in order to make full use of the data obtained during the operation of the deployment process on multiple devices and to ensure the accuracy of the intelligent fault diagnosis model, the present invention proposes a diagnostic algorithm model parameter update mechanism applicable to multiple devices, as follows:
[0087] Count the number of samples obtained during the deployment process on a single device , when the number of samples obtained Greater than or equal to a preset threshold , the system will start the training process and retrain the intelligent fault diagnosis model. The training data is the data reconstructed from the previously input training data and the newly obtained sample data. Therefore, the retraining start function associated with the data obtained during the deployment process is:
[0088] ;
[0089] At the same time, calculate the accuracy of the intelligent fault diagnosis model during the deployment process on a single device ,in, represents the number of true faults in the fault samples predicted by the intelligent fault diagnosis model, It indicates the number of false faults in the fault samples predicted by the intelligent fault diagnosis model. Less than a preset threshold , the system will start the training process and retrain the intelligent fault diagnosis model. The training data is the data reconstructed from the previously input training data and the sample data obtained during this period. Therefore, the retraining start function associated with the accuracy of the model during the deployment process is:
[0090] ;
[0091] The actual retraining start function can be obtained by ORing the two:
[0092] ;
[0093] when When the value of is 1, the system starts the training process and retrains the intelligent fault diagnosis model on a single device.
[0094] like Figure 3As shown, when multiple devices are trained, each device retrains the model to obtain the model parameters. At the same time, each device will upload its own model parameters to the server. After the server collects the model parameters of each device, it calculates and obtains the new model parameters. , and sends it to each device. Each device updates the parameters of its own model to the model parameters sent by the server. At this point, the diagnostic algorithm model parameters of multiple devices have been updated. The formula for calculating the new model parameters by the server is as follows:
[0095] ;
[0096] in, The number of devices on which intelligent fault diagnosis methods are deployed.
[0097] Figure 4 This is a structural block diagram of an intelligent fault diagnosis model updating device provided in an embodiment of the present application. Figure 4 As shown, the embodiment of the present application provides an intelligent fault diagnosis model updating device, which may include:
[0098] The memory 210 is configured to store instructions; and the processor 220 is configured to call instructions from the memory 210 and implement the above-mentioned intelligent fault diagnosis model updating method when executing the instructions.
[0099] An embodiment of the present application also provides a machine-readable storage medium, on which instructions are stored, and the instructions are used to enable a machine to execute the above-mentioned intelligent fault diagnosis model updating method.
[0100] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.
[0101] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0102] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0103] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0104] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0105] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.
[0106] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.
[0107] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.
[0108] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.
Claims
1. A method for updating an intelligent fault diagnosis model, characterized in that: The method comprises: Acquire first training data input by a user; Matching a target intelligent fault diagnosis model adapted to the first training data in an intelligent model library according to the first training data; Personalize and modify the target intelligent fault diagnosis model; Matching at least one pre-screened fault diagnosis model in the intelligent model library; Processing the pre-training parameters and the target training parameters to obtain the first training parameters of the target intelligent fault diagnosis model, wherein the pre-training parameters are obtained by training the pre-screening fault diagnosis model, and the target training parameters are obtained by training the target intelligent fault diagnosis model; When the target intelligent fault diagnosis model deployed in multiple devices triggers a start training instruction, retrain the target intelligent fault diagnosis model in each device, and obtain a second training parameter of each target intelligent fault diagnosis model according to the training result; The third training parameters are recalculated based on the second training parameters of the multiple devices and sent to all devices to obtain the target intelligent fault diagnosis model after the updated parameters. The step of matching a target intelligent fault diagnosis model adapted to the first training data in an intelligent model library according to the first training data includes: Acquire second training data of the intelligent fault diagnosis model in the intelligent model library; Calculating a distribution difference between the first training data and the second training data; Recommending and ranking the intelligent fault diagnosis models in the intelligent model library according to the distribution differences; Matching the intelligent fault diagnosis model whose recommended ranking is within a preset range as the target intelligent fault diagnosis model, The step of processing the pre-training parameters and the target training parameters to obtain the first training parameters of the target intelligent fault diagnosis model includes: Acquire a first eigenvector of the target intelligent fault diagnosis model and a second eigenvector of the pre-screening fault diagnosis model; Calculate the generalized Jaccard coefficient between the first eigenvector and the second eigenvector using formula (1); Formula (1) in, , represents vector product; \left | \left | {{M}_{u}} \right | \right |=\sqrt[{2}] {\sum ^{g}_{v=1} {{l}^{2}_{v}}} , represents the modulus of the first eigenvector; \left | \left | {{M}_{s}} \right | \right |=\sqrt[{2}] {\sum ^{g}_{v=1} {{s}^{2}_{v}}} , represents the modulus of the second eigenvector; represents the vth eigenvalue in the first eigenvector; represents the vth eigenvalue in the second eigenvector, and according to the generalized Jaccard coefficient, selects a model corresponding to the second eigenvector with a high matching degree with the first eigenvector for model training to obtain a trained first training parameter.
2. The intelligent fault diagnosis model updating method according to claim 1, characterized in that: The step of calculating the distribution difference between the first training data and the second training data includes: Calculating the MMD distance between the first training data and the second training data according to formula (2) to determine the distribution difference between the first training data and the second training data; Formula (2) The first training data is ,in, is the first training data in the input Sample; the second training data is ,in, is the first samples; is the norm of the matrix, is the kernel function, represents the number of samples in the first training data, represents the number of samples in the second training data, Indicates the first training data in the input samples, Indicates the first samples.
3. The intelligent fault diagnosis model updating method according to claim 1, characterized in that: The starting condition for starting the training of the target intelligent fault diagnosis model deployed in the multiple devices when the target intelligent fault diagnosis model deployed in the multiple devices triggers the start training instruction is: ； , is the accuracy of the intelligent fault diagnosis model, represents the number of true faults in the fault samples predicted by the intelligent fault diagnosis model, represents the number of false faults in the fault samples predicted by the intelligent fault diagnosis model, is a pre-set threshold; To count the number of samples obtained during the operation of the intelligent fault diagnosis model deployed on a single device, is a pre-set threshold.
4. The intelligent fault diagnosis model updating method according to claim 1, characterized in that: In the step of recalculating the third training parameters according to the second training parameters of the plurality of devices and sending the third training parameters to all the devices to obtain the target intelligent fault diagnosis model after the updated parameters, the parameters of the target intelligent fault diagnosis model are calculated and updated using formula (3): Formula (3) in, is the number of devices that have deployed intelligent fault diagnosis methods, For the The second training parameters of the device, is the recalculated third training parameter.
5. A processor, characterized in that: The method is configured to execute the intelligent fault diagnosis model updating method according to any one of claims 1 to 4.
6. A machine-readable storage medium, characterized in that: The machine-readable storage medium stores instructions for enabling a machine to execute the intelligent fault diagnosis model updating method according to any one of claims 1 to 4.
7. A fault diagnosis method, characterized in that: The method comprises: Obtain equipment operating data; Inputting the equipment operating condition data into an intelligent fault diagnosis model obtained by the intelligent fault diagnosis model updating method according to any one of claims 1 to 4; The equipment fault type is output through the intelligent fault diagnosis model.
8. A machine-readable storage medium, characterized in that: The machine-readable storage medium stores instructions, which are used to enable the machine to execute the fault diagnosis method according to claim 7.