Motor fault diagnosis method and device

Abstract

A fault diagnosis method for a motor, comprising: acquiring a data set of a motor from a motor driver, uploading the data set to a cloud server or an edge device, the data set comprising a plurality of motor operating parameters collected at the same time (110); training, at the cloud server or the edge device, an auto-encoder according to the data set, downloading the trained auto-encoder to the motor driver, and loading when the motor driver runs the auto-encoder (120); identifying fault data in the data set of the motor by using the auto-encoder after the loading is completed when the auto-encoder runs, and uploading the fault data to a cloud server or an edge device (130); and training a fault diagnosis model on the cloud server or the edge device according to the fault data, and performing fault diagnosis on the fault data during the operation of the motor by using the trained fault diagnosis model (140).

Description

Motor fault diagnosis method and device Technical Field

[0001] The present invention mainly relates to the field of industrial digitalization, and in particular to a motor fault diagnosis method and device. Background Art

[0002] Predictive maintenance (PM) is used to predict when in-service equipment will require maintenance. Compared to routine or scheduled maintenance, it only performs maintenance when it is absolutely necessary, thus saving costs. Predictive maintenance plays a crucial role in motor drive systems. Deploying predictive maintenance in motor drive systems can avoid losses caused by production downtime, eliminate the need for engineers to manually collect and analyze data, and improve the accuracy of fault diagnosis.

[0003] Traditional predictive maintenance detects signals from the drive and uses signal processing algorithms to extract features, identify key system parameters, and compare these with preset thresholds to provide warnings to users before a system failure occurs. Traditional predictive maintenance algorithms are complex, requiring extensive hardware resources for analysis and calculation. Furthermore, the algorithms are embedded in the drive and difficult to upgrade, and can only diagnose a limited range of faults.

[0004] Some new predictive maintenance methods use machine learning to perform fault diagnosis. This method uses labeled data to train neural network models in the cloud or at the edge, and then uses the trained neural network models to predict and classify faults. In this type of method, because machine learning requires a large amount of hardware resources, the predictive maintenance system is deployed in the cloud or at the edge. The motor driver uploads all data to the cloud or at the edge to implement predictive maintenance. Due to the large amount of data transmitted, it requires a large amount of network bandwidth and is significantly affected by poor network conditions. In addition, a large amount of labeled data is required to train the neural network model.

[0005] Summary of the Invention

[0006] In order to solve the above technical problems, the present invention provides a motor fault diagnosis method and device to improve the intelligence and reliability of fault diagnosis, while not requiring a large amount of labeled data, increasing the speed of modeling, and thus improving the efficiency and accuracy of fault diagnosis.

[0007] To achieve the above-mentioned objectives, the present invention proposes a motor fault diagnosis method, characterized in that the method includes: obtaining a data group of the motor from a motor driver, uploading the data group to a cloud server or edge device, wherein the data group includes multiple motor operating parameters collected at the same time; training an autoencoder on the cloud server or edge device according to the data group, downloading the trained autoencoder to the motor driver, and loading the autoencoder when the motor driver operates the autoencoder; using the autoencoder to identify fault data in the motor data group after the autoencoder is loaded during operation, and uploading the fault data to the cloud server or edge device; training a fault diagnosis model on the cloud server or edge device according to the fault data, and using the trained fault diagnosis model to perform fault diagnosis on fault data during motor operation. To this end, the fault data required by the fault diagnosis model can be obtained through the autoencoder. The autoencoder is an unsupervised learning neural network model that solves the problem of insufficient fault data. During operation, the drive only needs to upload fault data instead of all operating data, which greatly reduces the consumption of network bandwidth and avoids dependence on communication conditions. In addition, the training and deployment of the autoencoder and fault diagnosis model are both in the cloud or edge, without occupying too many hardware resources of the drive, realizing the predictive maintenance function of the lightweight drive.

[0008] Optionally, training an autoencoder based on the data set on the cloud server or edge device includes: training the autoencoder under various operating conditions of the motor until the difference between input data and reconstructed data of the autoencoder is less than a threshold value under all operating conditions. Thus, the autoencoder can be adapted to various operating conditions of the motor, thereby making the neural network model of the autoencoder more accurate, and thereby making fault diagnosis more accurate.

[0009] Optionally, training a fault diagnosis model based on the fault data on the cloud server or edge device includes: notifying a user of the fault data, receiving fault labels added by the user to the fault data, and training the fault diagnosis model based on the fault data and the data labels. To this end, by having the user add fault labels to the fault data, the data of the fault diagnosis model is made more refined, thereby making the fault diagnosis more accurate.

[0010] Optionally, the method further includes notifying a user of the fault diagnosis result of the fault diagnosis model, receiving user feedback input by the user, and retraining the fault diagnosis model if the user feedback indicates that the fault diagnosis result is inaccurate; and retraining the autoencoder if the user feedback indicates that the fault diagnosis result does not exist. To this end, the autoencoder and fault diagnosis model are adjusted based on user feedback, further improving their accuracy, thereby improving the accuracy of the fault diagnosis.

[0011] The present invention also proposes a motor fault diagnosis method, which is executed in a motor driver. The method includes: obtaining a data group of the motor, uploading the data group to a cloud server or edge device, wherein the data group includes multiple motor operating parameters collected at the same time; downloading an autoencoder from the cloud server or edge device, and loading it when the driver operates the autoencoder, wherein the autoencoder is trained by the cloud server or edge device based on the data group; after the autoencoder is loaded during operation, the autoencoder is used to identify fault data in the data group of the motor, and upload the fault data to the cloud server or edge device, wherein the fault data is used to train a fault diagnosis model based on the fault data on the cloud server or edge device, and the trained fault diagnosis model is used to perform fault diagnosis on the fault data during motor operation.

[0012] The present invention also proposes a motor fault diagnosis device, which includes: a first data processing module, which obtains a data group of the motor from the motor driver and uploads the data group to a cloud server or edge device, wherein the data group includes multiple motor operating parameters collected at the same time; a training module, which trains an autoencoder on the cloud server or edge device according to the data group, downloads the trained autoencoder to the motor driver, and loads the autoencoder when the motor driver operates the autoencoder; a second data processing module, which uses the autoencoder to identify fault data in the motor data group after the autoencoder is loaded during operation, and uploads the fault data to the cloud server or edge device; a fault diagnosis module, which trains a fault diagnosis model on the cloud server or edge device according to the fault data, and uses the trained fault diagnosis model to perform fault diagnosis on fault data during motor operation.

[0013] Optionally, the training module trains an autoencoder based on the data group on the cloud server or edge device, including: training the autoencoder under various working conditions of the motor until the difference between the input data and the reconstructed data of the autoencoder is less than a threshold under all working conditions.

[0014] Optionally, the fault diagnosis module trains a fault diagnosis model based on the fault data on the cloud server or edge device, including: notifying the user of the fault data, receiving the fault label added by the user to the fault data, and training the fault diagnosis model based on the fault data and the data label.

[0015] Optionally, the device also includes a user feedback module, which notifies the user of the fault diagnosis result of the fault diagnosis model and receives user feedback input by the user. If the user feedback indicates that the fault diagnosis result is inaccurate, the fault diagnosis model is retrained; if the user feedback indicates that the fault diagnosis result does not exist, the autoencoder is retrained.

[0016] The present invention also proposes a motor fault diagnosis device, which is executed in a motor driver. The device includes: a data upload module, which obtains a data group of the motor and uploads the data group to a cloud server or edge device, wherein the data group includes multiple motor operating parameters collected at the same time; a download module, which downloads an autoencoder from the cloud server or edge device and loads it when the driver operates the autoencoder, wherein the autoencoder is trained by the cloud server or edge device based on the data group; a second data upload module, which uses the autoencoder to identify fault data in the motor data group after the autoencoder is loaded during operation, and uploads the fault data to the cloud server or edge device, wherein the fault data is used to train a fault diagnosis model based on the fault data on the cloud server or edge device, and use the trained fault diagnosis model to perform fault diagnosis on fault data during motor operation.

[0017] The present invention also provides a motor driver, which is configured to execute the fault diagnosis method as described above.

[0018] The present invention also provides an electronic device, comprising a processor, a memory, and instructions stored in the memory, wherein the instructions implement the above-mentioned method when executed by the processor.

[0019] The present invention also provides a computer-readable storage medium having computer instructions stored thereon. When the computer instructions are executed, the method described above is executed.

[0020] The present invention also provides a computer program product, comprising a computer program, which implements the above method when executed by a processor. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The following drawings are only intended to illustrate and explain the present invention, and do not limit the scope of the present invention.

[0022] FIG1 is a flow chart of a fault diagnosis method according to an embodiment of the present invention;

[0023] FIG2 is a schematic diagram of an implementation environment of a fault diagnosis method according to an embodiment of the present invention;

[0024] FIG3 is a schematic diagram of a training process of an autoencoder according to an embodiment of the present invention;

[0025] FIG4 is a schematic diagram of an application process of an autoencoder according to an embodiment of the present invention;

[0026] FIG5 is a schematic diagram of a training process of a fault diagnosis model according to an embodiment of the present invention;

[0027] FIG6 is a schematic diagram of an application process of a fault diagnosis model according to an embodiment of the present invention;

[0028] FIG7 is a flow chart of a fault diagnosis method according to another embodiment of the present invention;

[0029] FIG8 is a schematic diagram of a fault diagnosis device according to an embodiment of the present invention;

[0030] FIG9 is a schematic diagram of a fault diagnosis device according to another embodiment of the present invention;

[0031] FIG. 10 is a schematic diagram of an electronic device according to an embodiment of the present invention.

[0032] Description of Reference Numerals

[0033] 100 Fault Diagnosis Methods

[0034] Steps 110-150

[0035] 21 motor

[0036] 22 motor drivers

[0037] 221 Data Pool

[0038] 222 Agent

[0039] 222a Autoencoder Runtime

[0040] 222b data upload unit

[0041] 23 cloud servers

[0042] 231 autoencoder training unit

[0043] 232 Autoencoder Download Unit

[0044] 233 Fault Diagnosis Model Training Unit

[0045] 234 Fault Diagnosis Model

[0046] 235 Notification Unit

[0047] 236 Feedback Receiving Unit

[0048] 300 Autoencoder training process

[0049] Steps 301-310

[0050] 400 Autoencoder Application Process

[0051] 401-405 steps

[0052] 500 Fault Diagnosis Model Training Process

[0053] Steps 501-505

[0054] Application process of 600 fault diagnosis model

[0055] Steps 601-610

[0056] 700 fault diagnosis method

[0057] Steps 710-730

[0058] 800 fault diagnosis device

[0059] 810 First Data Processing Module

[0060] 820 training modules

[0061] 830 Second Data Processing Module

[0062] 840 fault diagnosis module

[0063] 900 fault diagnosis device

[0064] 910 first data upload module

[0065] 920 Download Module

[0066] 930 Second data upload module

[0067] 1000 electronic devices

[0068] 1010 processor

[0069] 1020 memory DETAILED DESCRIPTION

[0070] In order to have a clearer understanding of the technical features, purposes and effects of the present invention, specific embodiments of the present invention are now described with reference to the accompanying drawings.

[0071] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0072] As used in this application and the claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not intended to refer to the singular but may include the plural. Generally speaking, the terms "comprises" and "include" only indicate the inclusion of the steps or elements specifically identified, and these steps and elements do not constitute an exclusive list. A method or apparatus may also include additional steps or elements.

[0073] The present invention provides a motor fault diagnosis method. FIG1 is a flow chart of a fault diagnosis method 100 according to an embodiment of the present invention. As shown in FIG1 , the method 100 includes:

[0074] Step 110 , obtaining a data set of the motor from the motor driver, and uploading the data set to a cloud server or edge device, wherein the data set includes multiple motor operating parameters collected at the same time.

[0075] The data groups of the motor are normal in most cases, and these data can be collected for preliminary judgment of the data. Figure 2 is a schematic diagram of the implementation environment of a fault diagnosis method according to an embodiment of the present invention. As shown in Figure 2, the operating parameters generated by the motor 21 during operation are sent to the data pool 221 of the motor driver 22. The data pool 221 includes data groups (data pairs) corresponding to different moments. Multiple data groups are sent to the cloud server 23 through the data upload unit 222b in the agent 222. The data group includes multiple motor operating parameters collected at the same moment. For example, a data group includes current, speed, torque, etc. at moment t. In Figure 2, the cloud server 23 is used as an example. It can be understood that the data group can also be sent to the edge device for analysis and processing. In order to simplify the explanation, the following description will be given using the cloud server as an example.

[0076] Step 120 : Train an autoencoder based on the data set on a cloud server or edge device, download the trained autoencoder to the motor driver, and load the autoencoder when the motor driver runs the autoencoder.

[0077] An autoencoder is an unsupervised neural network model that learns the implicit features of input data, called encoding, and uses these learned features to reconstruct the original input data, called decoding. For a well-trained autoencoder, the gap between the original data (original data) and the reconstructed data (reconstructed data) should be very small. However, if data with different features is fed into the trained autoencoder, the gap between the input and output data will be very large.

[0078] An autoencoder is trained on a cloud server or edge device based on a data set. Since the data set is normal in most cases, the trained autoencoder can learn the implicit features of normal data, thereby identifying normal data sets and abnormal data sets, that is, faulty data sets, from the data set. In an embodiment of the present invention, the autoencoder can be a convolutional autoencoder. As shown in Figure 2, the data upload unit 222b sends the data set to the autoencoder training unit 231 of the cloud server 23. The autoencoder training unit 231 trains the autoencoder based on the data set, including a training process and a verification process. The training process includes calculating the difference between the input data and the output data (for example, the root mean square). If the root mean square is greater than a threshold, the training parameters (for example, the number of hidden layers) are adjusted and retrained until the root mean square between the input data and the output data is less than the threshold. The verification process includes using the newly uploaded data pair to verify the autoencoder. After obtaining user confirmation, the autoencoder training is completed, and a trained autoencoder is obtained. The trained autoencoder is downloaded to the autoencoder runtime 222a of the agent 222 through the autoencoder download unit 232. The autoencoder runtime 222a includes the operating environment of the autoencoder. After the autoencoder runtime 222a loads the autoencoder at runtime, the autoencoder can run in the driver.

[0079] In some embodiments, training an autoencoder based on a data set on a cloud server or an edge device includes: training the autoencoder under various working conditions of the motor until the difference between the input data and the reconstructed data of the autoencoder is less than a threshold value under all working conditions. Specifically, the working conditions of the motor may include low-load working conditions, rated working conditions, and overload working conditions. The autoencoder is first trained under the first working condition until the difference between the input data and the reconstructed data (output data) is less than a first threshold value, and the autoencoder is continued to be trained under the second working condition. If the difference between the input data and the reconstructed data is greater than the second threshold value, the training parameters are adjusted and the autoencoder is retrained in combination with the training data under the first working condition and the training data under the second working condition until the difference between the input data and the reconstructed data is less than the second threshold value, and so on, until the difference between the input data and the reconstructed data of the autoencoder is less than the threshold value under all working conditions.

[0080] Step 130 : After the autoencoder is loaded and running, the autoencoder is used to identify fault data in the motor data set, and the fault data is uploaded to the cloud server or edge device.

[0081] After the autoencoder is loaded during runtime, the autoencoder can be run in the drive. The new data group in the data pool is sent to the autoencoder, and the autoencoder calculates the difference between the input data and the reconstructed data of the new data group. If the difference is greater than the model threshold, it indicates that the data group is faulty data, and the faulty data is uploaded to the cloud server or edge device. If the difference is less than the model threshold, it indicates that the data group is normal data, and the data group will remain in the drive. To this end, the unsupervised neural network model autoencoder can identify normal data and faulty data in the data group. As shown in Figure 2, after the autoencoder is loaded during runtime 222a, it can be run in the drive. The autoencoder can identify the new data group. If the difference between the input data and the reconstructed data calculated by the autoencoder is greater than the model threshold, it indicates that the data group is faulty data, and the faulty data is uploaded to the cloud server 23 by the data upload unit 222b.

[0082] In step 140 , a fault diagnosis model is trained on the cloud server or edge device based on the fault data, and the trained fault diagnosis model is used to perform fault diagnosis on the fault data during the operation of the motor.

[0083] Through the autoencoder running in the driver, fault data can be obtained, and this fault data is used to train a fault diagnosis model. The fault diagnosis model is used to perform fault diagnosis on fault data during motor operation, that is, new fault data, thereby realizing the predictive maintenance function of the motor. In an embodiment of the present invention, fault diagnosis can be fault classification, that is, the type of fault, such as power quality problems (voltage imbalance, harmonic distortion), variable frequency drive problems (overshoot voltage, overload), mechanical problems (imbalance, looseness, shaft wear), etc. To this end, the fault data required for the fault diagnosis model can be obtained through the autoencoder. The autoencoder is an unsupervised learning neural network model that solves the problem of insufficient fault data. During operation, the driver only needs to upload fault data instead of all operating data, which greatly reduces the consumption of network bandwidth and avoids dependence on communication conditions. In addition, the training and deployment of the autoencoder and the fault diagnosis model are both in the cloud or edge, without occupying too many hardware resources of the driver, realizing the predictive maintenance function of the lightweight driver.

[0084] In some embodiments, training a fault diagnosis model based on fault data on a cloud server or edge device includes: notifying a user of the fault data, receiving fault labels added by the user to the fault data, and training the fault diagnosis model based on the fault data and the data labels. To this end, by allowing the user to add fault labels to the fault data, the data of the fault diagnosis model becomes more refined, thereby making the fault diagnosis more accurate.

[0085] In some embodiments, the method further includes step 150, in which the fault diagnosis result of the fault diagnosis model is notified to the user, user feedback is received, and if the user feedback indicates that the fault diagnosis result is inaccurate, the fault diagnosis model is retrained; if the user feedback indicates that the fault diagnosis result does not exist, the autoencoder is retrained. To this end, the autoencoder and the fault diagnosis model are adjusted based on the user feedback, further improving the accuracy of the autoencoder and the fault diagnosis model, thereby improving the accuracy of the fault diagnosis.

[0086] The following is a non-limiting and detailed description of the training process of the autoencoder, the application process of the autoencoder, the training process of the fault diagnosis model, and the application process of the fault diagnosis model.

[0087] FIG3 is a schematic diagram of a training process 300 of an autoencoder according to an embodiment of the present invention. As shown in FIG3 , the training process 300 includes:

[0088] Step 301 : the autoencoder training unit 231 of the cloud server 23 obtains a data set corresponding to a first working condition;

[0089] Step 302 , training an autoencoder using a data set corresponding to a first operating condition;

[0090] Step 303, calculating the mean square error between the input data and the reconstructed data;

[0091] Step 304, determine whether the mean square error is less than a first threshold, if yes, proceed to step 305, otherwise proceed to step 306;

[0092] Step 305 : The autoencoder training unit 231 of the cloud server 23 obtains a data set corresponding to the second working condition and uses the autoencoder to calculate the mean square error between the input data and the reconstructed data;

[0093] Step 306, adjusting the training parameters of the autoencoder;

[0094] Step 307, determine whether the mean square error is less than the second threshold, if yes, proceed to step 308, otherwise proceed to step 309;

[0095] Step 308 is used to confirm whether the training process is completed. If so, go to step 310, otherwise go to step 305;

[0096] Step 309: Adjust the training parameters based on the training data under the first working condition and the training data under the second working condition;

[0097] Step 310, end.

[0098] FIG4 is a schematic diagram of an application process 400 of an autoencoder according to an embodiment of the present invention. As shown in FIG4 , the application process 400 of the autoencoder includes:

[0099] Step 401, sending data pairs to the autoencoder;

[0100] Step 402, using the trained autoencoder to calculate the root mean square between the original data and the reconstructed data;

[0101] Step 403, determine whether the root mean square is greater than the model threshold, if yes, go to step 404, if not, return to step 401;

[0102] Step 404, sending the data group to the cloud server 23;

[0103] In step 405 , the cloud server 23 notifies the user and collects the data tags added by the user to the data group.

[0104] FIG5 is a schematic diagram of a training process 500 of a fault diagnosis model according to an embodiment of the present invention. As shown in FIG5 , the training process 500 of the fault diagnosis model includes:

[0105] Step 501: Collect fault data and tags on the cloud server;

[0106] Step 502: training a fault diagnosis model using the fault data and labels;

[0107] Step 503: determine whether the accuracy of the diagnosis result of the fault diagnosis model is greater than a threshold. If yes, proceed to step 504; otherwise, proceed to step 505.

[0108] Step 504, the fault diagnosis model training is completed;

[0109] Step 505: Adjust the training parameters to retrain the fault diagnosis model.

[0110] FIG6 is a schematic diagram of an application process 600 of a fault diagnosis model according to an embodiment of the present invention. As shown in FIG6 , the application process 600 of the fault diagnosis model includes:

[0111] Step 601, sending the data set to the autoencoder of the driver;

[0112] Step 602, the autoencoder calculates the difference between the input data and the reconstructed data;

[0113] Step 603, determine whether the difference is greater than the model threshold, if yes, go to step 604, otherwise return to step 601;

[0114] Step 604, sending the fault data to the cloud server;

[0115] Step 605: perform fault diagnosis on the fault data using the fault diagnosis model;

[0116] Step 606: Notify the user of the fault diagnosis result and collect user feedback;

[0117] Step 607: Determine user feedback. If the user feedback indicates that the fault does not exist, proceed to step 608. If the user feedback indicates that the fault exists and the fault diagnosis is accurate, proceed to step 609. If the user feedback indicates that the fault exists but the fault diagnosis is inaccurate, proceed to step 610.

[0118] Step 608, retraining the autoencoder;

[0119] Step 609, end;

[0120] Step 610: retrain the fault diagnosis model.

[0121] An embodiment of the present invention provides a motor fault diagnosis method. Fault data required by a fault diagnosis model can be obtained through an autoencoder. The autoencoder is a neural network model for unsupervised learning, which solves the problem of insufficient fault data. During operation, the driver only needs to upload fault data instead of all operating data, which greatly reduces the consumption of network bandwidth and avoids dependence on communication conditions. In addition, the training and deployment of the autoencoder and the fault diagnosis model are both in the cloud or at the edge, without occupying too many hardware resources of the driver, thereby realizing a lightweight predictive maintenance function for the driver.

[0122] The present invention further provides a motor fault diagnosis method, which is executed in a motor driver. FIG7 is a flow chart of a fault diagnosis method 700 according to another embodiment of the present invention. As shown in FIG7 , the method 700 includes:

[0123] Step 710 , obtaining a data set of the motor and uploading the data set to a cloud server or edge device, wherein the data set includes multiple motor operating parameters collected at the same time;

[0124] Step 720 , downloading an autoencoder from a cloud server or edge device and loading it when the driver runs the autoencoder, the autoencoder being trained by the cloud server or edge device based on the data set;

[0125] Step 730: After the autoencoder is loaded during operation, the autoencoder is used to identify the fault data in the motor data group, and the fault data is uploaded to the cloud server or edge device. The fault data is used to train a fault diagnosis model based on the fault data on the cloud server or edge device, and the trained fault diagnosis model is used to perform fault diagnosis on the fault data during motor operation.

[0126] The present invention further provides a motor fault diagnosis device. FIG8 is a schematic diagram of a fault diagnosis device according to an embodiment of the present invention. As shown in FIG8 , the device 800 includes:

[0127] A first data processing module 810 acquires a data set of the motor from the motor driver and uploads the data set to a cloud server or edge device. The data set includes multiple motor operating parameters collected at the same time.

[0128] A training module 820 trains an autoencoder based on the data set on a cloud server or edge device, downloads the trained autoencoder to the motor driver, and loads the autoencoder when the motor driver runs the autoencoder;

[0129] The second data processing module 830 uses the autoencoder to identify fault data in the motor data set after the autoencoder is loaded and runs, and uploads the fault data to the cloud server or edge device;

[0130] The fault diagnosis module 840 trains a fault diagnosis model based on the fault data on the cloud server or edge device, and uses the trained fault diagnosis model to perform fault diagnosis on the fault data during motor operation.

[0131] In some embodiments, the training module 820 trains an autoencoder based on the data group on a cloud server or edge device, including: training the autoencoder under various working conditions of the motor until the difference between the input data and the reconstructed data of the autoencoder is less than a threshold under all working conditions.

[0132] In some embodiments, the fault diagnosis module 840 trains a fault diagnosis model based on fault data on a cloud server or edge device, including: notifying the user of the fault data, receiving fault labels added by the user to the fault data, and training the fault diagnosis model based on the fault data and data labels.

[0133] In some embodiments, the device also includes a user feedback module 850, which notifies the user of the fault diagnosis result of the fault diagnosis model and receives user feedback input by the user. If the user feedback fault diagnosis result is inaccurate, the fault diagnosis model is retrained; if the user feedback fault diagnosis result does not exist, the autoencoder is retrained.

[0134] The present invention further provides a motor fault diagnosis device, which is implemented in a motor driver. FIG9 is a schematic diagram of a fault diagnosis device 900 according to another embodiment of the present invention. The device 900 includes:

[0135] A first data uploading module 910 acquires a data set of the motor and uploads the data set to a cloud server or edge device. The data set includes multiple motor operating parameters collected at the same time.

[0136] A download module 920 downloads an autoencoder from a cloud server or edge device and loads it when the driver runs the autoencoder, where the autoencoder is trained by the cloud server or edge device based on the data set;

[0137] The second data uploading module 930 uses the autoencoder to identify the fault data in the motor data group after the autoencoder is loaded during operation, and uploads the fault data to the cloud server or edge device. The fault data is used to train a fault diagnosis model based on the fault data on the cloud server or edge device, and use the trained fault diagnosis model to perform fault diagnosis on the fault data during motor operation.

[0138] The present invention also provides a motor driver, which is configured to execute the fault diagnosis method 700 described above.

[0139] The present invention also provides an electronic device 1000. FIG10 is a schematic diagram of an electronic device 1000 according to an embodiment of the present invention. As shown in FIG10 , the electronic device 1000 includes a processor 1010 and a memory 1020. The memory 1020 stores instructions, wherein the instructions, when executed by the processor 1010, implement the method 100 described above.

[0140] The present invention further provides a computer-readable storage medium having computer instructions stored thereon. When the computer instructions are executed, the methods 100 and 700 described above are executed.

[0141] The present invention further provides a computer program product, comprising a computer program, which implements the methods 100 and 700 described above when the computer program is executed by a processor.

[0142] Some aspects of the methods and apparatus of the present invention may be performed entirely by hardware, entirely by software (including firmware, resident software, microcode, etc.), or by a combination of hardware and software. The above hardware or software may be referred to as "data blocks," "modules," "engines," "units," "components," or "systems." The processor may be one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DAPDs), programmable logic devices (PLCs), field programmable gate arrays (FPGAs), processors, controllers, microcontrollers, microprocessors, or combinations thereof. In addition, various aspects of the present invention may be embodied as computer products in one or more computer-readable media, the product including computer-readable program code. For example, computer-readable media may include, but are not limited to, magnetic storage devices (e.g., hard disks, floppy disks, tapes, etc.), optical disks (e.g., compact disks (CDs), digital versatile disks (DVDs), etc.), smart cards, and flash memory devices (e.g., cards, sticks, key drives, etc.).

[0143] Flowcharts are used herein to illustrate the operations performed by the methods according to the embodiments of the present application. It should be understood that the preceding operations are not necessarily performed in exact order. Instead, the various steps may be performed in reverse order or simultaneously. Furthermore, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0144] It should be understood that although this specification is described according to various embodiments, not every embodiment contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

[0145] The above description is only an illustrative embodiment of the present invention and is not intended to limit the scope of the present invention. Any equivalent changes, modifications and combinations made by those skilled in the art without departing from the concept and principle of the present invention shall fall within the scope of protection of the present invention.

Claims

1. A motor fault diagnosis method (100), characterized in that: The method (100) comprises: Acquire a data set of the motor from a motor driver, and upload the data set to a cloud server or an edge device, wherein the data set includes a plurality of motor operating parameters collected at the same time (110); Training an autoencoder based on the data set on the cloud server or edge device, downloading the trained autoencoder to the motor driver, and loading the autoencoder when the motor driver performs runtime operation (120); After the autoencoder is loaded during operation, the autoencoder is used to identify fault data in the data set of the motor, and the fault data is uploaded to the cloud server or edge device (130); A fault diagnosis model is trained on the cloud server or edge device based on the fault data, and the trained fault diagnosis model is used to perform fault diagnosis on the fault data during motor operation (140).

2. The fault diagnosis method (100) according to claim 1, characterized in that: Training an autoencoder according to the data group on the cloud server or edge device includes: training the autoencoder under various working conditions of the motor until the difference between the input data and the reconstructed data of the autoencoder is less than a threshold under all working conditions.

3. The fault diagnosis method (100) according to claim 1 or 2, characterized in that: Training a fault diagnosis model based on the fault data on the cloud server or edge device includes: notifying the user of the fault data, receiving a fault label added by the user to the fault data, and training the fault diagnosis model based on the fault data and the data label.

4. The fault diagnosis method (100) according to any one of claims 1 to 3, characterized in that: The method (100) further includes: notifying a user of a fault diagnosis result of the fault diagnosis model, receiving user feedback input by the user, and retraining the fault diagnosis model if the user feedback indicates that the fault diagnosis result is inaccurate, and retraining the autoencoder (150) if the user feedback indicates that the fault diagnosis result does not exist.

5. A motor fault diagnosis method (700), the fault diagnosis method being executed in a motor driver, characterized in that: The method (700) comprises: Acquiring a data set of the motor, and uploading the data set to a cloud server or an edge device, wherein the data set includes a plurality of motor operating parameters collected at the same time (710); Downloading an autoencoder from the cloud server or edge device and loading it when the driver runs the autoencoder, the autoencoder being trained by the cloud server or edge device based on the data set (720); After the autoencoder is loaded during operation, the autoencoder is used to identify fault data in the data group of the motor, and the fault data is uploaded to the cloud server or edge device. The fault data is used to train a fault diagnosis model based on the fault data on the cloud server or edge device, and the trained fault diagnosis model is used to perform fault diagnosis on the fault data during motor operation (730).

6. A motor fault diagnosis device (800), characterized in that: The device (800) comprises: a first data processing module (810) for acquiring a data set of the motor from a motor driver and uploading the data set to a cloud server or an edge device, the data set including a plurality of motor operating parameters collected at the same time; a training module (820) for training an autoencoder based on the data set on the cloud server or edge device, downloading the trained autoencoder to the motor driver, and loading the autoencoder when the motor driver is operating the autoencoder; a second data processing module (830), which uses the autoencoder to identify fault data in the data set of the motor after the autoencoder is loaded and runs, and uploads the fault data to the cloud server or edge device; A fault diagnosis module (840) trains a fault diagnosis model on the cloud server or edge device based on the fault data, and uses the trained fault diagnosis model to perform fault diagnosis on the fault data during motor operation.

7. The fault diagnosis device (800) according to claim 6, characterized in that: The training module (820) trains an autoencoder based on the data group on the cloud server or edge device, including: training the autoencoder under various working conditions of the motor until the difference between the input data and the reconstructed data of the autoencoder is less than a threshold under all working conditions.

8. The fault diagnosis device (800) according to claim 6 or 7, characterized in that: The fault diagnosis module (840) trains a fault diagnosis model based on the fault data on the cloud server or edge device, including: notifying the user of the fault data, receiving a fault label added by the user to the fault data, and training the fault diagnosis model based on the fault data and the data label.

9. The fault diagnosis device (800) according to any one of claims 6 to 8, characterized in that: The device (800) further includes a user feedback module (850), which notifies the user of the fault diagnosis result of the fault diagnosis model and receives user feedback input by the user. If the user feedback indicates that the fault diagnosis result is inaccurate, the fault diagnosis model is retrained; if the user feedback indicates that the fault diagnosis result does not exist, the autoencoder is retrained.

10. A motor fault diagnosis device (900), wherein the fault diagnosis device (900) is executed in a motor driver, characterized in that: The device (900) comprises: a first data uploading module (910) for acquiring a data set of the motor and uploading the data set to a cloud server or an edge device, wherein the data set includes a plurality of motor operating parameters collected at the same time; a download module (920) for downloading an autoencoder from the cloud server or edge device and loading the autoencoder when the driver runs the autoencoder, wherein the autoencoder is trained by the cloud server or edge device based on the data set; A second data uploading module (930) uses the autoencoder to identify fault data in the data group of the motor after the autoencoder is loaded during operation, and uploads the fault data to the cloud server or edge device. The fault data is used to train a fault diagnosis model based on the fault data on the cloud server or edge device, and the trained fault diagnosis model is used to perform fault diagnosis on the fault data during motor operation.

11. A motor driver, characterized in that: The motor driver is configured to execute the fault diagnosis method according to claim 5 .

12. An electronic device (1000), comprising a processor (1010), a memory (1020), and instructions stored in the memory (1020), wherein the instructions, when executed by the processor (1010), implement the method (100, 700) according to any one of claims 1 to 5.

13. A computer-readable storage medium having computer instructions stored thereon, wherein the computer instructions, when executed, perform the method (100, 700) according to any one of claims 1 to 5.

14. A computer program product, characterized in that The invention comprises a computer program for implementing the method (100, 700) according to any one of claims 1 to 5 when the computer program is executed by a processor.

Images