Data enhancement type permanent magnet synchronous motor test method and system

Abstract

The invention relates to the technical field of motor testing, and discloses a data enhancement type permanent magnet synchronous motor testing method and system, and the method comprises the steps: obtaining a state expression coefficient of a detected motor based on a representation signal and a first machine learning model; judging whether the detected motor has an abnormal symptom or not based on the state performance coefficient; generating an actual vibration time-frequency diagram based on the vibration signal of the abnormal motor, and calling a pre-configured second machine learning model; inputting the actual vibration time-frequency diagram into a second machine learning model to evaluate the fault degree of the abnormal motor; the technical problems that in the prior art, the permanent magnet synchronous motor testing efficiency is low, the fault detection accuracy is low, the abnormal motor cannot be rapidly screened out, the fault degree of the abnormal motor cannot be determined, and the method is difficult to apply to a batch motor testing scene can be solved.

Description

Technical Field

[0001] The present invention relates to the technical field of motor testing, and more specifically, to a data-enhanced permanent magnet synchronous motor testing method and system. Background Art

[0002] With the widespread adoption of permanent magnet synchronous motors in various applications, the demand for their production quality inspection has become more and more urgent. During the production process, permanent magnet synchronous motors need to be strictly inspected to ensure that their performance meets the requirements. Traditional permanent magnet synchronous motor quality inspection methods often rely on manual visual inspection or simple testing equipment, which has problems such as low detection efficiency and low accuracy, and cannot meet the needs of modern production lines.

[0003] At present, there is a lack of efficient permanent magnet synchronous motor testing methods or systems. Although there are some intelligent permanent magnet synchronous motor testing documents, for example, the patent with publication number CN117054870A discloses a permanent magnet synchronous motor abnormality detection method and device, research and practical application of the above method and prior art have found that the above method and prior art have at least the following defects: (1) The testing efficiency of permanent magnet synchronous motors is low, and it is impossible to quickly detect whether there are abnormalities in permanent magnet synchronous motors, making it difficult to apply to mass motor production quality inspection scenarios; (2) The model training data volume is small and cannot be applied to actual detection scenarios. In addition, the detection data is single and cannot quickly identify the fault degree of abnormal permanent magnet synchronous motors based on the screening of abnormal permanent magnet synchronous motors. At the same time, the test accuracy of permanent magnet synchronous motors cannot be guaranteed. Furthermore, it is difficult to lay an important data support for the production quality inspection of permanent magnet synchronous motors. Summary of the invention

[0004] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a data-enhanced permanent magnet synchronous motor testing method and system.

[0005] To achieve the above object, the present invention provides the following technical solutions: A data-enhanced permanent magnet synchronous motor testing method, the method comprising: Measuring a characterization signal of the detected motor within a predetermined time interval, and calling a preconfigured first machine learning model; obtaining a state performance coefficient of the detected motor based on the characterization signal and the first machine learning model; the characterization signal includes a voltage value, a current value, and a resistance value; Based on the state performance coefficient, it is determined whether the detected motor has abnormal signs. If not, the corresponding detected motor is calibrated as a normal motor; if so, the corresponding detected motor is calibrated as an abnormal motor, and the vibration signal of the abnormal motor is measured within a predetermined time interval, where M is an integer greater than zero; An actual vibration time-frequency diagram is generated based on the vibration signal of the abnormal motor, and a preconfigured second machine learning model is retrieved; the actual vibration time-frequency diagram is input into the second machine learning model to evaluate the fault degree of the abnormal motor.

[0006] Furthermore, the generation logic of the preconfigured first machine learning model is as follows: Acquire historical state performance training data, and divide the historical state performance training data into a state performance training set and a state performance test set; the historical state performance training data includes a characterization signal of a predetermined time interval and its corresponding state performance coefficient; Constructing a regression network, taking the representation signal of a predetermined time interval in the state performance training set as input data of the regression network, and taking the state performance coefficient in the state performance training set as output data of the regression network, training the regression network, and obtaining an initial regression network; The initial regression network is model verified using the state performance test set, and the initial regression network that is less than or equal to the preset test error threshold is output as the preconfigured first machine learning model.

[0007] Furthermore, the generation logic of the state performance coefficient in the historical state performance training data is as follows: In a given time interval, actual measurement data of the test motor under given conditions are obtained, as well as expected measurement data of the test motor under given conditions; the actual measurement data include actual speed measurement values, actual temperature measurement values, and actual torque measurement values, and the expected measurement data include expected speed mean values, expected temperature mean values, and expected torque mean values; Calculate a speed evaluation coefficient, a temperature evaluation coefficient, and a torque evaluation coefficient based on actual measurement data and expected measurement data; The speed evaluation coefficient, the temperature evaluation coefficient and the torque evaluation coefficient are input into a mathematical calculation model of a pre-built state performance coefficient to obtain the state performance coefficient.

[0008] Furthermore, the calculation logic of the speed evaluation coefficient is as follows: Extract the actual measured value of the rotation speed at each moment in the actual measurement data; The speed evaluation coefficient is calculated based on the actual speed measurement value and the expected speed mean value; the calculation formula is: ; Where: is the speed evaluation coefficient, is the measured value of the speed at the rth moment, is the expected mean speed, The duration of a given time interval; The calculation logic of the temperature evaluation coefficient is as follows: Extract the actual temperature value at each moment in the actual measurement data; The temperature evaluation coefficient is calculated based on the measured temperature value and the expected average temperature value; the calculation formula is: ; Where: is the temperature evaluation coefficient, is the measured temperature value at the rth moment, is the expected mean temperature; The calculation logic of the torque evaluation coefficient is as follows: Extract the torque measured value at each moment in the actual measurement data; The torque evaluation coefficient is calculated based on the measured torque value and the expected torque mean value; the calculation formula is: ; is the torque evaluation coefficient, is the measured torque value at the rth moment, is the expected mean torque.

[0009] Furthermore, the step of determining whether the detected motor has any abnormal sign includes: comparing the state performance coefficient with a preset state performance coefficient interval; If the state performance coefficient is within the preset state performance coefficient range, it is determined that the detected motor has no abnormal signs; If the state performance coefficient is not within the preset state performance coefficient range, it is determined that the detected motor has abnormal signs.

[0010] Furthermore, the generation logic of the preconfigured second machine learning model is as follows: Acquire fault detection expansion training data, and divide the historical fault degree training data into a fault degree training set and a fault degree test set; the fault detection expansion training data includes a Q-amplitude actual vibration time-frequency diagram and annotated labels corresponding to the fault degree, and the annotated labels of the fault degree include a "mild" label, a "moderate" label, and a "severe" label, and Q is an integer greater than zero; Construct a classifier, use the actual vibration time-frequency diagram in the fault degree training set as the input data of the classifier, and use the labeled labels of the fault degree in the fault degree training set as the output data of the classifier, train the classifier, and obtain an initial classifier; The initial classifier is model verified using the fault degree test set, and the output of the initial classifier that is greater than or equal to the preset test accuracy threshold is used as the preconfigured second machine learning model.

[0011] Furthermore, the labeling process of the fault degree label is as follows: Extracting actual vibration time-frequency diagram and extracting pre-stored template vibration time-frequency diagram; Set the step to 1, and perform cross-correlation calculation between the template vibration time-frequency diagram and the actual vibration time-frequency diagram through a sliding window method to obtain the difference degree of each cross window; Filter cross windows whose difference is greater than or equal to a preset difference threshold as difference windows, and count the number of difference windows and the difference of each difference window; According to the number and degree of difference of the difference windows, the fault degree assessment coefficient of the abnormal motor corresponding to each actual vibration time-frequency diagram is calculated; the specific calculation formula is: ; Where: is the fault severity assessment coefficient, is the number of difference windows, is the difference of the ith difference window, is the maximum difference among G difference windows, is the minimum difference among G difference windows; The fault degree is marked on the corresponding actual vibration time-frequency diagram according to the fault degree assessment coefficient.

[0012] Furthermore, the marking of the fault degree of the corresponding actual vibration time-frequency diagram includes: Setting fault degree assessment coefficient thresholds Td1 and Td2, Td1>Td2, and comparing the fault degree assessment coefficient with the fault degree assessment coefficient threshold; If the fault degree assessment coefficient is greater than or equal to the fault degree assessment coefficient threshold value Td1, the actual vibration time-frequency graph corresponding to the fault degree assessment coefficient is marked with a "severe" label; If the fault degree assessment coefficient is less than the fault degree assessment coefficient threshold value Td1 and greater than the fault degree assessment coefficient threshold value Td2, the actual vibration time-frequency graph corresponding to the fault degree assessment coefficient is marked with a "moderate" label; If the fault degree assessment coefficient is less than or equal to the fault degree assessment coefficient threshold value Td2, the actual vibration time-frequency diagram corresponding to the fault degree assessment coefficient is marked with a "mild" label.

[0013] A data-enhanced permanent magnet synchronous motor test system, comprising: A data collection module, used to measure a characterization signal of the detected motor within a predetermined time interval, and to retrieve a preconfigured first machine learning model; based on the characterization signal and the first machine learning model, to obtain a state performance coefficient of the detected motor; the characterization signal includes a voltage value, a current value, and a resistance value; The abnormality judgment module is used to judge whether the detected motor has abnormal signs based on the state performance coefficient. If not, the corresponding detected motor is calibrated as a normal motor; if so, the corresponding detected motor is calibrated as an abnormal motor, and the vibration signal of the abnormal motor is measured within a predetermined time interval, where M is an integer greater than zero; A fault testing module is used to generate an actual vibration time-frequency diagram based on the vibration signal of the abnormal motor and call a preconfigured second machine learning model; the actual vibration time-frequency diagram is input into the second machine learning model to evaluate the fault degree of the abnormal motor.

[0014] Compared with the prior art, the present invention has the following beneficial effects: The present application discloses a data-enhanced permanent magnet synchronous motor testing method and system, which obtains a state performance coefficient of a detected motor based on a characterization signal and a first machine learning model; determines whether the detected motor has abnormal signs based on the state performance coefficient; generates an actual vibration time-frequency diagram based on the vibration signal of the abnormal motor, and calls a preconfigured second machine learning model; inputs the actual vibration time-frequency diagram into the second machine learning model to evaluate the fault degree of the abnormal motor; based on the above process, the present invention quickly detects whether the permanent magnet synchronous motor has an abnormality; further, on the basis of screening out abnormal permanent magnet synchronous motors, the present invention is conducive to quickly detecting the fault degree of abnormal permanent magnet synchronous motors; at the same time, through operations such as data expansion and model training, the present invention can ensure the test accuracy of the permanent magnet synchronous motor, which is conducive to laying an important data support for the production quality inspection of permanent magnet synchronous motors. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A flow chart of a data-enhanced permanent magnet synchronous motor testing method provided by the present invention; Figure 2 A schematic diagram of the module structure of a data-enhanced permanent magnet synchronous motor test system provided by the present invention. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] Example 1: Please refer to Figure 1 As shown, this embodiment discloses a data-enhanced permanent magnet synchronous motor testing method, the method comprising: S101: measuring a characterization signal of the detected motor within a predetermined time interval, and calling a preconfigured first machine learning model; obtaining a state performance coefficient of the detected motor based on the characterization signal and the first machine learning model; the characterization signal includes a voltage value, a current value, and a resistance value; It should be understood that: before measuring the characteristic signal of the detected motor within a predetermined time interval, the detected motor (i.e., the detected permanent magnet synchronous motor) needs to be powered on to cause it to run. During the running process, the characteristic signal of the detected motor within the predetermined time interval is measured by various sensors; the various sensors include but are not limited to voltage sensors, current sensors, and resistance sensors; the duration of the predetermined time interval is set in advance by the technicians based on specific experiments or experience; In implementation, the generation logic of the preconfigured first machine learning model is as follows: Acquire historical state performance training data, and divide the historical state performance training data into a state performance training set and a state performance test set; the historical state performance training data includes a characterization signal of a predetermined time interval and its corresponding state performance coefficient; It should be understood that: the characterization signal of a predetermined time interval in the historical state performance training data is measured by various sensors when the test motor is powered on and running; The generation logic of the state performance coefficient in the historical state performance training data is as follows: In a given time interval, actual measurement data of the test motor under given conditions are obtained, as well as expected measurement data of the test motor under given conditions; the actual measurement data include actual speed measurement values, actual temperature measurement values, and actual torque measurement values, and the expected measurement data include expected speed mean values, expected temperature mean values, and expected torque mean values; It should be noted that the actual measurement data under the given conditions are measured by technicians after installing various types of sensor equipment on the test motor in advance; the various types of sensor equipment include but are not limited to speed sensors, temperature sensors and torque sensors, etc.; wherein, the given conditions refer to measurement conditions without changes, and the given conditions include but are not limited to measurement environment conditions (such as temperature, humidity, etc.) and operating conditions (such as constant load, etc.). An exemplary explanation is that if the measurement environment conditions change (such as humidity changes), the measured values ​​will not be a true reflection of the test motor. Therefore, in order to eliminate the interference of other conditions and factors and ensure measurement accuracy, the limiting conditions remain unchanged; It should also be noted that: the expected measurement data is pre-stored in the system database; wherein the expected average speed is obtained by the technicians' prior experimental records of the normal permanent magnet synchronous motor; the expected average speed indicates the value that the normal permanent magnet synchronous motor should reach within a given time interval and under given conditions; similarly, the expected average temperature and expected average torque are also the same; Calculate a speed evaluation coefficient, a temperature evaluation coefficient, and a torque evaluation coefficient based on actual measurement data and expected measurement data; The calculation logic of the speed evaluation coefficient is as follows: Extract the actual measured value of the rotation speed at each moment in the actual measurement data; The speed evaluation coefficient is calculated based on the actual speed measurement value and the expected speed mean value; the calculation formula is: ; Where: is the speed evaluation coefficient, is the measured value of the speed at the rth moment, is the expected mean speed, The duration of a given time interval; The calculation logic of the temperature evaluation coefficient is as follows: Extract the actual temperature value at each moment in the actual measurement data; The temperature evaluation coefficient is calculated based on the measured temperature value and the expected average temperature value; the calculation formula is: ; Where: is the temperature evaluation coefficient, is the measured temperature value at the rth moment, is the expected mean temperature; The calculation logic of the torque evaluation coefficient is as follows: Extract the torque measured value at each moment in the actual measurement data; The torque evaluation coefficient is calculated based on the measured torque value and the expected torque mean value; the calculation formula is: ; is the torque evaluation coefficient, is the measured torque value at the rth moment, is the expected mean torque; The speed evaluation coefficient, the temperature evaluation coefficient and the torque evaluation coefficient are input into the mathematical calculation model of the pre-built state performance coefficient to obtain the state performance coefficient; the formula of the mathematical calculation model is: ; Where: , and is a weight factor greater than zero, , is the state performance coefficient; Constructing a regression network, taking the representation signal of a predetermined time interval in the state performance training set as input data of the regression network, and taking the state performance coefficient in the state performance training set as output data of the regression network, training the regression network, and obtaining an initial regression network; Using the state performance test set to perform model verification on the initial regression network, outputting an initial regression network that is less than or equal to a preset test error threshold as a preconfigured first machine learning model; It should be noted that the regression network is a specific one of algorithm models such as linear regression, decision tree regression, support vector machine regression, random forest regression, polynomial regression or neural network.

[0018] S102: judging whether the detected motor has abnormal signs based on the state performance coefficient, if not, calibrating the corresponding detected motor as a normal motor; if so, calibrating the corresponding detected motor as an abnormal motor, and measuring the vibration signal of the abnormal motor within a predetermined time interval, where M is an integer greater than zero; In implementation, the step of determining whether the detected motor has abnormal signs includes: comparing the state performance coefficient with a preset state performance coefficient interval; If the state performance coefficient is within the preset state performance coefficient range, it is determined that the detected motor has no abnormal signs; If the state performance coefficient is not within the preset state performance coefficient range, it is determined that the detected motor has abnormal signs; It can be understood that: when the state performance coefficient is not within the preset state performance coefficient range, that is, when the detected motor has abnormal signs, it means that the detected motor has abnormal performance due to a fault; It should be understood that: the vibration signal of the abnormal motor is collected by a vibration sensor; It should be understood that: based on the characterization signal of the detected motor, the state performance coefficient of the detected motor is obtained, and whether the detected motor is normal is judged according to the state performance coefficient. Compared with the prior art, the processing process of the present invention is convenient and simple, and can quickly detect whether the detected motor has abnormal signs. Furthermore, normal motors and abnormal motors can be quickly screened out based on abnormal signs. It is suitable for batch permanent magnet synchronous motor quality inspection scenarios, and greatly improves the abnormality detection efficiency of permanent magnet synchronous motors.

[0019] S103: Generate an actual vibration time-frequency diagram based on the vibration signal of the abnormal motor, and retrieve a preconfigured second machine learning model; input the actual vibration time-frequency diagram into the second machine learning model to evaluate the fault degree of the abnormal motor; It should be noted that: the actual vibration time-frequency diagram is obtained by processing the vibration signal of the abnormal motor using the time-frequency analysis technology; wherein the time-frequency analysis technology includes but is not limited to Short-Time Fourier Transform (STFT) or Wavelet Transform; Specifically, the generation logic of the preconfigured second machine learning model is as follows: Acquire fault detection expansion training data, and divide the historical fault degree training data into a fault degree training set and a fault degree test set; the fault detection expansion training data includes a Q-amplitude actual vibration time-frequency diagram and annotated labels corresponding to the fault degree, and the annotated labels of the fault degree include a "mild" label, a "moderate" label, and a "severe" label, and Q is an integer greater than zero; In a specific embodiment, the acquisition logic of the fault detection expansion training data includes: acquiring historical fault detection training data, wherein the historical fault detection training data includes U-amplitude actual vibration time-frequency diagrams and their corresponding fault degree annotation labels, U<Q; extracting U-amplitude actual vibration time-frequency diagrams in the historical fault detection training data, adding random noise, such as Gaussian noise, salt and pepper noise, etc., to each actual vibration time-frequency diagram to obtain new K-amplitude actual vibration time-frequency diagrams and their corresponding fault degree annotation labels, K is an integer greater than zero, and merging the new K-amplitude actual vibration time-frequency diagrams and their corresponding fault degree annotation labels into the historical fault detection training data to obtain fault detection expansion training data; In another specific implementation, the logic for acquiring the fault detection expansion training data further includes: Collecting an original data set containing multiple actual vibration time-frequency graphs and their corresponding fault degree annotation labels; Build a generative network based on the CNN convolutional neural network architecture, add a random noise vector to the actual vibration time-frequency diagram in the original data set, and use it as the output of the generative network to output a synthetic actual vibration time-frequency diagram; Build a discriminant network based on the CNN convolutional neural network architecture, synthesize the actual vibration time-frequency diagram output by the generation network as the input of the discriminant network to output the true probability value; The generative network and the discriminative network are combined into a generative adversarial network, and the generative network and the discriminative network are trained alternately until the mean square error reaches a preset error threshold, and the trained generative adversarial network is output; It should be understood that: Generative Adversarial Network (GAN) is a deep learning model. GAN consists of two main parts: Generator and Discriminator, which compete with each other to achieve learning through adversarial learning; Generator: The task of the generator is to generate realistic data samples, such as images, text, etc. from random noise; it receives random noise as input and tries to generate samples similar to real data; the goal of the generator is to deceive the discriminator as much as possible so that it cannot distinguish between generated samples and real samples; Discriminator: The task of the discriminator is to distinguish between fake samples generated by the generator and real data samples; it receives samples as input and outputs a value indicating the probability that the sample comes from real data; the goal of the discriminator is to distinguish between real and fake samples as accurately as possible; The core idea of ​​the Generative Adversarial Network (GAN) is to let the generator and the discriminator compete with each other, which leads to a dynamic game process; the generator constantly tries to generate more realistic samples to deceive the discriminator, while the discriminator constantly improves its discrimination ability to distinguish between true and false samples; through this competition process, the generator and the discriminator gradually learn the distribution characteristics of real data, and finally the generator can generate very realistic samples; Generate new actual vibration time-frequency graphs using the trained generative adversarial network, label the new actual vibration time-frequency graphs, and merge the labeled new actual vibration time-frequency graphs into the original data set to obtain fault detection expansion training data; It can be understood that: by expanding the samples of the training data and enhancing the diversity of the training data, it is beneficial to improve the generalization ability of the subsequent second machine learning model, prevent the subsequent second machine learning model from overfitting, and at the same time, it is beneficial to improve the accuracy of the subsequent fault degree test; The labeling process of the fault degree label is as follows: Extracting actual vibration time-frequency diagram and extracting pre-stored template vibration time-frequency diagram; It should be noted that: the template vibration time-frequency diagram is pre-stored in the system database, and the template vibration time-frequency diagram is a normal vibration time-frequency waveform of a normal permanent magnet synchronous motor in a normal (ie, fault-free) state; Set the step to 1, and perform cross-correlation calculation between the template vibration time-frequency diagram and the actual vibration time-frequency diagram through a sliding window method to obtain the difference degree of each cross window; It should be understood that: cross-correlation calculation is a method of image processing, and its basic idea is to use one image as a template and another image as a matching object, place the matching object on the template, and slide the matching object in the template. When a part of the template image and the matching image overlap, the overlapping part is used as a cross window; by calculating the difference of each cross window, the difference degree of the cross window can be obtained; wherein, the calculation formula of the difference is the Euclidean distance algorithm; Filter cross windows whose difference is greater than or equal to a preset difference threshold as difference windows, and count the number of difference windows and the difference of each difference window; According to the number and degree of difference of the difference windows, the fault degree assessment coefficient of the abnormal motor corresponding to each actual vibration time-frequency diagram is calculated; the specific calculation formula is: ; Where: is the fault severity assessment coefficient, is the number of difference windows, is the difference of the ith difference window, is the maximum difference among G difference windows, is the minimum difference among G difference windows; Mark the fault degree of the corresponding actual vibration time-frequency diagram according to the fault degree assessment coefficient; Specifically, the marking of the fault degree of the corresponding actual vibration time-frequency diagram includes: Setting fault degree assessment coefficient thresholds Td1 and Td2, Td1>Td2, and comparing the fault degree assessment coefficient with the fault degree assessment coefficient threshold; If the fault degree assessment coefficient is greater than or equal to the fault degree assessment coefficient threshold value Td1, the actual vibration time-frequency graph corresponding to the fault degree assessment coefficient is marked with a "severe" label; If the fault degree assessment coefficient is less than the fault degree assessment coefficient threshold value Td1 and greater than the fault degree assessment coefficient threshold value Td2, the actual vibration time-frequency graph corresponding to the fault degree assessment coefficient is marked with a "moderate" label; If the fault degree assessment coefficient is less than or equal to the fault degree assessment coefficient threshold value Td2, the actual vibration time-frequency graph corresponding to the fault degree assessment coefficient is marked with a "mild" label; It can be understood that: when the corresponding actual vibration time-frequency graph is marked with a "severe" label, it means that the corresponding abnormal motor has a large and serious fault; similarly, when the corresponding actual vibration time-frequency graph is marked with a "moderate" label or a "mild" label, it means that the corresponding abnormal motor has a general degree of fault or a minor fault; Construct a classifier, use the actual vibration time-frequency diagram in the fault degree training set as the input data of the classifier, and use the labeled labels of the fault degree in the fault degree training set as the output data of the classifier, train the classifier, and obtain an initial classifier; Using the fault degree test set to perform model verification on the initial classifier, outputting an initial classifier that is greater than or equal to a preset test accuracy threshold as a preconfigured second machine learning model; It should be noted that: the classifier is a specific one of the algorithm models such as decision tree classification, support vector machine classification, random forest classification, Bayesian classification or neural network; It can be understood that: on the basis of screening out abnormal motors, by performing model analysis on the vibration signals of the abnormal motors, the present invention can quickly and accurately screen out the fault degree of the abnormal motors, thereby laying an important data support for the production quality inspection of permanent magnet synchronous motors, and further, it is beneficial to improve the efficiency of batch testing of abnormal motors.

[0020] Example 2: Please refer to Figure 2 As shown, based on the same inventive concept, this embodiment discloses a data-enhanced permanent magnet synchronous motor test system. The details not detailed in this embodiment refer to the specific description in the above embodiment 1. The system includes: The data collection module 210 is used to measure the characterization signal of the detected motor within a predetermined time interval and retrieve a preconfigured first machine learning model; based on the characterization signal and the first machine learning model, obtain the state performance coefficient of the detected motor; the characterization signal includes a voltage value, a current value, and a resistance value; The abnormality judgment module 220 is used to judge whether the detected motor has abnormal signs based on the state performance coefficient. If not, the corresponding detected motor is calibrated as a normal motor; if so, the corresponding detected motor is calibrated as an abnormal motor, and the vibration signal of the abnormal motor is measured within a predetermined time interval, where M is an integer greater than zero; The fault testing module 230 is used to generate an actual vibration time-frequency diagram based on the vibration signal of the abnormal motor, and call a preconfigured second machine learning model; the actual vibration time-frequency diagram is input into the second machine learning model to evaluate the fault degree of the abnormal motor.

[0021] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters, weights and thresholds in the formula are set by technicians in this field according to actual conditions.

[0022] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website, computer, server or data center to another website, computer, server or data center via a wired network or a wireless network. The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media sets. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD) or a semiconductor medium. The semiconductor medium may be a solid-state hard disk.

[0023] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present invention can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0024] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0025] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the units is only one, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0026] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0027] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0028] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0029] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A data-enhanced permanent magnet synchronous motor testing method, characterized in that: The method comprises: Measuring a characterization signal of the detected motor within a predetermined time interval, and calling a preconfigured first machine learning model; obtaining a state performance coefficient of the detected motor based on the characterization signal and the first machine learning model; the characterization signal includes a voltage value, a current value, and a resistance value; Based on the state performance coefficient, it is determined whether the detected motor has abnormal signs. If not, the corresponding detected motor is calibrated as a normal motor; if so, the corresponding detected motor is calibrated as an abnormal motor, and the vibration signal of the abnormal motor is measured within a predetermined time interval, where M is an integer greater than zero; An actual vibration time-frequency diagram is generated based on the vibration signal of the abnormal motor, and a preconfigured second machine learning model is retrieved; the actual vibration time-frequency diagram is input into the second machine learning model to evaluate the fault degree of the abnormal motor.

2. The data-enhanced permanent magnet synchronous motor testing method according to claim 1, characterized in that: The generation logic of the preconfigured first machine learning model is as follows: Acquire historical state performance training data, and divide the historical state performance training data into a state performance training set and a state performance test set; the historical state performance training data includes a characterization signal of a predetermined time interval and its corresponding state performance coefficient; Constructing a regression network, taking the representation signal of a predetermined time interval in the state performance training set as input data of the regression network, and taking the state performance coefficient in the state performance training set as output data of the regression network, training the regression network, and obtaining an initial regression network; The initial regression network is model verified using the state performance test set, and the initial regression network that is less than or equal to the preset test error threshold is output as the preconfigured first machine learning model.

3. The data-enhanced permanent magnet synchronous motor testing method according to claim 2, characterized in that: The generation logic of the state performance coefficient in the historical state performance training data is as follows: In a given time interval, actual measurement data of the test motor under given conditions are obtained, as well as expected measurement data of the test motor under given conditions; the actual measurement data include actual speed measurement values, actual temperature measurement values, and actual torque measurement values, and the expected measurement data include expected speed mean values, expected temperature mean values, and expected torque mean values; Calculate a speed evaluation coefficient, a temperature evaluation coefficient, and a torque evaluation coefficient based on actual measurement data and expected measurement data; The speed evaluation coefficient, the temperature evaluation coefficient and the torque evaluation coefficient are input into a mathematical calculation model of a pre-built state performance coefficient to obtain the state performance coefficient.

4. The data-enhanced permanent magnet synchronous motor testing method according to claim 3 is characterized in that: The calculation logic of the speed evaluation coefficient is as follows: Extract the actual measured value of the rotation speed at each moment in the actual measurement data; The speed evaluation coefficient is calculated based on the actual speed measurement value and the expected speed mean value; the calculation formula is: ; Where: is the speed evaluation coefficient, is the measured value of the speed at the rth moment, is the expected mean speed, The duration of a given time interval; The calculation logic of the temperature evaluation coefficient is as follows: Extract the actual temperature value at each moment in the actual measurement data; The temperature evaluation coefficient is calculated based on the measured temperature value and the expected average temperature value; the calculation formula is: ; Where: is the temperature evaluation coefficient, is the measured temperature value at the rth moment, is the expected mean temperature; The calculation logic of the torque evaluation coefficient is as follows: Extract the torque measured value at each moment in the actual measurement data; The torque evaluation coefficient is calculated based on the measured torque value and the expected torque mean value; the calculation formula is: ; is the torque evaluation coefficient, is the measured torque value at the rth moment, is the expected mean torque.

5. The data-enhanced permanent magnet synchronous motor testing method according to claim 4, characterized in that: The step of judging whether the detected motor has abnormal signs includes: comparing the state performance coefficient with a preset state performance coefficient interval; If the state performance coefficient is within the preset state performance coefficient range, it is determined that the detected motor has no abnormal signs; If the state performance coefficient is not within the preset state performance coefficient range, it is determined that the detected motor has abnormal signs.

6. The data-enhanced permanent magnet synchronous motor testing method according to claim 5, characterized in that: The generation logic of the preconfigured second machine learning model is as follows: Acquire fault detection expansion training data, and divide the historical fault degree training data into a fault degree training set and a fault degree test set; the fault detection expansion training data includes Q-amplitude actual vibration time-frequency diagrams and their corresponding fault degree annotation labels, the fault degree annotation labels include a "mild" label, a "moderate" label, and a "severe" label, and Q is an integer greater than zero; Construct a classifier, use the actual vibration time-frequency diagram in the fault degree training set as the input data of the classifier, and use the labeled labels of the fault degree in the fault degree training set as the output data of the classifier, train the classifier, and obtain an initial classifier; The initial classifier is model verified using the fault degree test set, and the output of the initial classifier that is greater than or equal to the preset test accuracy threshold is used as the preconfigured second machine learning model.

7. The data-enhanced permanent magnet synchronous motor testing method according to claim 6, characterized in that: The labeling process of the fault degree label is as follows: Extracting actual vibration time-frequency diagram and extracting pre-stored template vibration time-frequency diagram; Set the step to 1, and perform cross-correlation calculation between the template vibration time-frequency diagram and the actual vibration time-frequency diagram through a sliding window method to obtain the difference degree of each cross window; Filter cross windows whose difference is greater than or equal to a preset difference threshold as difference windows, and count the number of difference windows and the difference of each difference window; According to the number and degree of difference of the difference windows, the fault degree assessment coefficient of the abnormal motor corresponding to each actual vibration time-frequency diagram is calculated; The specific calculation formula is: ; Where: is the fault severity assessment coefficient, is the number of difference windows, is the difference of the ith difference window, is the maximum difference among G difference windows, is the minimum difference among G difference windows; The fault degree is marked on the corresponding actual vibration time-frequency diagram according to the fault degree assessment coefficient.

8. The data-enhanced permanent magnet synchronous motor testing method according to claim 7, characterized in that: The marking of the fault degree of the corresponding actual vibration time-frequency diagram includes: Setting fault degree assessment coefficient thresholds Td1 and Td2, Td1>Td2, and comparing the fault degree assessment coefficient with the fault degree assessment coefficient threshold; If the fault degree assessment coefficient is greater than or equal to the fault degree assessment coefficient threshold value Td1, the actual vibration time-frequency graph corresponding to the fault degree assessment coefficient is marked with a "severe" label; If the fault degree assessment coefficient is less than the fault degree assessment coefficient threshold value Td1 and greater than the fault degree assessment coefficient threshold value Td2, the actual vibration time-frequency graph corresponding to the fault degree assessment coefficient is marked with a "moderate" label; If the fault degree assessment coefficient is less than or equal to the fault degree assessment coefficient threshold value Td2, the actual vibration time-frequency diagram corresponding to the fault degree assessment coefficient is marked with a "mild" label.

9. A data-enhanced permanent magnet synchronous motor test system, characterized in that: include: A data collection module, used to measure a characterization signal of the detected motor within a predetermined time interval and retrieve a preconfigured first machine learning model; Based on the characterization signal and the first machine learning model, a state performance coefficient of the detected motor is obtained; the characterization signal includes a voltage value, a current value and a resistance value; The abnormality judgment module is used to judge whether the detected motor has abnormal signs based on the state performance coefficient. If not, the corresponding detected motor is calibrated as a normal motor; if so, the corresponding detected motor is calibrated as an abnormal motor, and the vibration signal of the abnormal motor is measured within a predetermined time interval, where M is an integer greater than zero; A fault testing module is used to generate an actual vibration time-frequency diagram based on the vibration signal of the abnormal motor and call a preconfigured second machine learning model; the actual vibration time-frequency diagram is input into the second machine learning model to evaluate the fault degree of the abnormal motor.

Images