Motor fault recognition method, device, equipment, medium and program product

Abstract

The application relates to the technical field of motor protection and artificial intelligence, and discloses a motor fault identification method, device, equipment, medium and program product, wherein the method comprises the following steps: acquiring an electrical state parameter of a motor, and the working voltage of the motor exceeding a preset value; performing time-frequency feature extraction on the electrical state parameter, and determining a first fault classification result of the motor based on time sequence evolution information of the extracted time-frequency feature; in the process of determining the first fault classification result, performing physical simulation analysis on the electrical state parameter in parallel to obtain a second fault classification result generated based on the physical simulation analysis; and determining a target fault of the motor based on cross verification of the first fault classification result and the second fault classification result. Through implementation of the technical scheme of the application, the fault diagnosis accuracy of the motor is improved.

Description

Technical Field

[0001] This invention relates to the fields of motor protection and artificial intelligence technology, specifically to methods, devices, equipment, media, and program products for fault identification of electric motors. Background Technology

[0002] Currently, most fault diagnosis devices for electric motors rely on electrical state parameters of voltage and current transformers, vibration analysis from acceleration sensors, and temperature monitoring from temperature sensors to achieve fault diagnosis and early warning. These methods can only identify a single type of fault. Although some fault diagnosis methods based on artificial intelligence exist, these methods are solely data-driven and struggle to handle extreme operating conditions or rare fault scenarios, resulting in poor fault diagnosis accuracy. Summary of the Invention

[0003] This invention provides a method, apparatus, equipment, medium, and program product for identifying faults in electric motors, in order to solve the problem of poor fault diagnosis results.

[0004] In a first aspect, the present invention provides a method for fault identification of an electric motor, comprising: Obtain the electrical status parameters of the motor; the motor's operating voltage exceeds the preset value. The electrical state parameters are extracted using time-frequency features, and the first fault classification result of the motor is determined based on the time-series evolution information of the extracted time-frequency features. In the process of determining the first fault classification result, physical simulation analysis of electrical state parameters is performed in parallel to obtain the second fault classification result generated based on the physical simulation analysis. Based on the cross-validation of the first fault classification results and the second fault classification results, the target fault of the motor is determined.

[0005] The fault identification method for electric motors provided in this invention extracts time-frequency features from electrical state parameters and determines a first fault classification result based on the time-series evolution information of the extracted time-frequency features. This method can fully capture the dynamic features and evolution patterns in the fault development process and improve the sensitivity to early fault identification. Simultaneously, a second fault classification result is obtained through parallel physical simulation analysis, which can provide auxiliary basis for fault identification from a physical perspective. The target fault is determined by cross-validating the two types of fault classification results. This method integrates the advantages of data-driven and physical models, avoids identification errors caused by a single analysis method, and improves the accuracy of fault identification.

[0006] In one optional implementation, time-frequency features are extracted from the electrical state parameters, and based on the time-series evolution information of the extracted time-frequency features, a first fault classification result for the motor is determined, including: The time-series voltage and time-series current in the electrical state parameters are sampled in a mixed segment to obtain the segmented sampling results; Perform spectral transformation on each segment of electrical signal in the segmented sampling results, and extract the multi-dimensional spectral features corresponding to each segment of electrical signal; Multi-dimensional spectral features are compressed to obtain time-frequency features; Fault identification is performed using time-frequency characteristics to obtain fault correlation characteristics; The first fault classification result is determined based on the temporal evolution characteristics of fault association features.

[0007] The fault identification method for electric motors provided in this invention, by performing mixed segmented sampling of time-series voltage and time-series current in electrical state parameters, can take into account both time continuity and local detailed features, avoiding feature loss caused by collecting only a single electrical state parameter; by performing spectral transformation on each segment of electrical signal in the segmented sampling results and extracting multi-dimensional spectral features, fault feature information in different frequency bands can be captured, enriching the information dimensions of fault identification; by performing feature compression on multi-dimensional spectral features to obtain time-frequency features, redundant and irrelevant features can be eliminated while retaining core fault identification information, reducing the amount of data computation; by performing fault identification based on time-frequency features to obtain fault association features, the corresponding association relationship between electrical signals and different fault types can be established; then, based on the time-series evolution features of fault association features, the first fault classification result is determined, which can dynamically track the changing trend of fault features over time, explore the evolution law of fault occurrence and development, and improve the sensitivity of early fault identification.

[0008] In one optional implementation, the time-series voltage and time-series current in the electrical state parameters are sampled in a mixed segment to obtain segmented sampling results, including: Obtain the sliding window for segmented sampling; Based on the sliding step size corresponding to the sliding window, the timing voltage and timing current are sampled in parallel to obtain the segmented sampling results.

[0009] The fault identification method for electric motors provided in this invention obtains a segmented sampling sliding window and performs parallel sampling of time-series voltage and time-series current based on the sliding step size corresponding to the sliding window to obtain segmented sampling results. This enables synchronous acquisition of voltage and current signals, ensuring that the two types of signals are aligned on the time axis and avoiding phase deviation caused by asynchronous sampling.

[0010] In one optional implementation, fault identification is performed on the time-frequency characteristics to obtain fault association characteristics, including: Based on a first convolutional neural network with a first preset number of layers, abnormal frequency domain features in time-frequency features are identified. Based on the correlation between abnormal frequency domain features and faults, fault-related features are extracted from the abnormal frequency domain features.

[0011] The fault identification method for electric motors provided in this embodiment of the invention can capture subtle abnormal frequency domain features by identifying abnormal frequency domain features in time-frequency features through a first convolutional neural network with a first preset number of layers. Based on the correlation between abnormal frequency domain features and faults, fault-related features are extracted from the abnormal frequency domain features, which can eliminate interference information unrelated to the fault and strengthen the correspondence between fault features and fault types.

[0012] In one optional implementation, the first fault classification result is determined based on the temporal evolution characteristics of fault association features, including: The fault-related features are input into a second convolutional neural network with a second preset number of layers, which is less than the first preset number of layers. Based on the second convolutional neural network, the temporal evolution features corresponding to the fault association features are extracted; Fault classification is performed based on temporal evolution characteristics to obtain the first fault classification result.

[0013] The fault identification method for electric motors provided in this invention involves inputting fault-related features into a second convolutional neural network with a second preset number of layers less than a first preset number of layers, and extracting temporal evolution features corresponding to the fault-related features based on the second convolutional neural network. This method can capture the dynamic change trend of fault features along the time dimension and fully capture the continuous evolution process of fault occurrence and development. Then, based on the temporal evolution features, fault classification is performed to obtain a first fault classification result, which can effectively distinguish between short-term interference signals and the actual fault evolution process, and improve the sensitivity to early fault identification.

[0014] In one optional implementation, physical analysis is performed on the electrical state parameters in parallel to obtain a second fault classification result generated based on the physical analysis, including: Abnormal parameters in the electrical state parameters are identified using an electrical circuit model, which is based on the motor's equivalent circuit parameters. Using a motor operation model, based on abnormal parameters and motor load parameters, motor abnormalities are identified. The motor operation model is cascaded with the electrical circuit model, and the motor operation model is obtained by modeling based on the motor motion equations of the motor. Using a fault identification model, the second fault classification result is determined based on the electrical circuit parameters corresponding to the motor abnormality and the motor operating parameters. The fault identification model is cascaded with the motor operating model, and the fault identification model is constructed based on the correlation between the electrical circuit parameters, the motor operating parameters, and the motor fault.

[0015] The electric motor fault identification method provided in this invention identifies abnormal parameters in electrical state parameters by utilizing an electrical circuit model obtained from modeling based on the motor's equivalent circuit parameters, thereby accurately acquiring electrical anomalies that deviate from normal operating conditions. By using a motor operation model cascaded with the electrical circuit model, motor anomalies are identified based on abnormal parameters and the motor's load parameters, achieving the fusion of electrical and mechanical operating condition information and avoiding missed fault detections caused by single-dimensional parameter identification. Furthermore, by using a fault identification model cascaded with the motor operation model, a second fault classification result is determined based on electrical circuit parameters and motor operation parameters, taking into account circuit, dynamic motion laws, and fault mapping laws, enriching the basis for fault identification, and improving the accuracy of fault classification results.

[0016] In one optional implementation, the target fault of the motor is determined based on cross-validation of the first fault classification result and the second fault classification result, including: Obtain the first abnormal parameter, the first motor abnormality, and the first fault classification corresponding to the first fault classification result; The first abnormal parameter is compared with the abnormal parameters identified by the electrical circuit model to determine the first deviation; The first motor anomaly is compared with the motor anomaly identified by the motor operation model to determine the second deviation; The first fault classification is compared with the second fault classification result output by the fault identification model to determine the third deviation; Based on the weighted results of the first deviation, the second deviation, and the third deviation, the target deviation and the target fault corresponding to the target deviation are determined.

[0017] The fault identification method for electric motors provided in this invention compares the acquired first abnormal parameter with the abnormal parameter identified by the electrical circuit model to determine the first deviation. This allows for cross-verification of data-driven and circuit model identification results at the electrical state parameter level, quantifying the difference between the two in extracting abnormal electrical quantities. Comparing the acquired first motor abnormality with the motor abnormality output by the motor operation model to determine the second deviation allows for cross-verification of data-driven and physical model analysis results at the electromechanical operation state level, quantifying the difference between the two in identifying motor operation abnormalities. Comparing the acquired first fault classification with the second fault classification result output by the fault identification model to determine the third deviation quantifies the discrepancy between the two methods in fault classification. Then, based on the weighted calculation results of the first, second, and third deviations, the target deviation and corresponding target fault are determined. This quantifies the error weights of each identification stage, integrates multi-level verification deviations to obtain the true fault, and improves the accuracy of electric motor fault location.

[0018] In one optional implementation, the target fault corresponding to the target deviation includes: Obtain the deviation threshold corresponding to the running parameters; If the target deviation exceeds the deviation threshold, the first fault classification result is calibrated, and the target fault of the motor is determined based on the second fault classification result. If the target deviation does not exceed the deviation threshold, the target fault of the motor is determined based on the first fault classification result.

[0019] The fault identification method for electric motors provided in the embodiments of the invention provides a quantitative benchmark for judging the difference in multi-model identification deviations by obtaining a deviation threshold corresponding to the operating parameters. If the target deviation exceeds the deviation threshold, the first fault classification result is calibrated, and the target fault of the electric motor is determined based on the second fault classification result. This method can correct the data-driven results based on physical model analysis, thereby reducing the risk of fault misjudgment. If the target deviation does not exceed the deviation threshold, the target fault of the electric motor is determined based on the first fault classification result, which can improve the efficiency of fault identification while ensuring accurate identification.

[0020] In an optional implementation, the method further includes: If the target deviation is greater than the first threshold but less than the deviation threshold, then the generation and display of the first warning message are triggered; the first threshold is less than the deviation threshold. If the target deviation is greater than the second threshold but less than the deviation threshold, the fault monitoring cycle is shortened, and the second early warning information is sent to the remote terminal; the second threshold is greater than the first threshold. If the target deviation is greater than the third threshold but less than the deviation threshold, the protection mechanism is triggered; the third threshold is greater than the second threshold.

[0021] The fault identification method for electric motors provided in the embodiments of the invention sets a first threshold, a second threshold, and a third threshold that increase in stages, and all three thresholds are less than the deviation threshold. This allows for the matching of differentiated response measures based on the target deviation between data-driven results and physical model analysis, thereby achieving refined hierarchical control of fault severity.

[0022] Secondly, the present invention provides a fault identification device for an electric motor, comprising: The signal acquisition module is used to acquire the electrical status parameters of the motor, and the operating voltage of the motor exceeds the preset value; The time-frequency feature extraction module is used to extract time-frequency features from electrical state parameters and determine the first fault classification result of the motor based on the time-series evolution information of the extracted time-frequency features. The physical simulation analysis module is used to perform physical simulation analysis on electrical state parameters in parallel during the process of determining the first fault classification result, and to obtain the second fault classification result generated based on the physical simulation analysis. The target fault identification module is used to determine the target fault of the motor based on the cross-validation of the first fault classification result and the second fault classification result.

[0023] Thirdly, the present invention provides an electronic device, comprising: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the motor fault identification method of the first aspect or any corresponding embodiment described above.

[0024] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the fault identification method for an electric motor according to the first aspect or any corresponding embodiment thereof.

[0025] Fifthly, the present invention provides a computer program product, including computer instructions for causing a computer to execute the fault identification method for an electric motor according to the first aspect or any corresponding embodiment described above. Attached Figure Description

[0026] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0027] Figure 1 This is a schematic diagram of an application scenario according to an embodiment of the present invention; Figure 2 This is a schematic flowchart of a first method for identifying faults in an electric motor according to an embodiment of the present invention. Figure 3 This is a schematic diagram of a second process for a fault identification method for an electric motor according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the third process of the fault identification method for an electric motor according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the fourth process of the fault identification method for an electric motor according to an embodiment of the present invention; Figure 6 This is a structural block diagram of a fault identification device for an electric motor according to an embodiment of the present invention; Figure 7 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] It is understood that before using the technical solutions disclosed in the various embodiments of the present invention, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in the present invention and their authorization should be obtained in accordance with relevant laws and regulations through appropriate means.

[0030] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0031] As an optional application scenario of this invention, such as Figure 1 As shown, the fault identification system for the electric motor may include at least one electronic device and at least one server. Figure 1 The system is illustrated in the example, which includes a computer 101, a mobile terminal 102, and a server 103, and the electronic devices such as the computer 101 and the mobile terminal 102 are connected to the server 103 via a network 110.

[0032] Specifically, the electronic devices can be desktop computers, laptops, or similar devices. These devices allow users to view real-time operating status parameters of the motor, frequency domain and time-series feature extraction results, fault classification results, cross-validation results, and final fault warning information, enabling operation and maintenance monitoring and fault trend viewing.

[0033] Server 103 can be a standalone physical server, a server cluster, a distributed system, or a cloud server that provides cloud services, used to execute all the method steps of the present invention; network 110 can be a wired network or a wireless network, examples of which include, but are not limited to, the Internet, corporate intranets, local area networks, wide area networks, mobile communication networks, and combinations thereof.

[0034] Currently, most fault diagnosis devices for electric motors rely on electrical state parameters of voltage and current transformers, vibration analysis from acceleration sensors, and temperature monitoring from temperature sensors to achieve fault diagnosis and early warning. These methods can only identify a single type of fault. Although some fault diagnosis methods based on artificial intelligence exist, these methods are solely data-driven and struggle to handle extreme operating conditions or rare fault scenarios, resulting in poor fault diagnosis accuracy.

[0035] Based on this, the technical solution of the present invention performs time-series feature analysis and multi-level electromechanical coupling physical simulation analysis on electrical state parameters in parallel using a two-layer convolutional network to obtain two types of fault results respectively, and performs weighted cross-validation on the two types of results to realize bidirectional constraint verification of data-driven network model and physical model, thereby improving the accuracy of motor fault diagnosis.

[0036] According to an embodiment of the present invention, a method for fault identification of an electric motor is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0037] This embodiment provides a method for identifying faults in electric motors, which can be used in the aforementioned electronic devices, such as laptops and desktop computers. Figure 2 This is a flowchart of a fault identification method for an electric motor according to an embodiment of the present invention, such as... Figure 2 As shown, the process includes the following steps: Step S201: Obtain the electrical status parameters of the motor.

[0038] The operating voltage of the motor exceeds the preset value.

[0039] The preset value is a pre-defined minimum rated voltage, such as 2.5kV. Accordingly, the motor can be an industrial drive motor with a rated voltage level of 3kV, 6kV, or 10kV.

[0040] Electrical condition parameters refer to electrical monitoring data that characterize the operating condition of equipment collected throughout the entire operation of the motor. For example, electrical condition parameters may include three-phase voltage, three-phase current, zero-sequence leakage current, active power, reactive power, etc.

[0041] Specifically, electrical status parameters can be acquired in real time using field sensing and data acquisition devices, and the acquired data can be transmitted to a server via a data acquisition gateway. For example, electrical status parameters can be acquired by using voltage transformers, current transformers, and zero-sequence current transformers installed on the motor power supply line, sampling continuous three-phase voltage, three-phase current time-series data, and zero-sequence leakage current data at a sampling rate of 128 points per second. All acquired data is uploaded to electronic equipment via industrial Ethernet. Correspondingly, the electronic equipment can acquire all electrical status parameters in real time.

[0042] In some optional implementations, environmental parameters of the motor can also be acquired simultaneously to assist in the subsequent determination of target faults.

[0043] Step S202: Extract time-frequency features from electrical state parameters, and determine the first fault classification result of the motor based on the time-series evolution information of the extracted time-frequency features.

[0044] Time-frequency characteristics refer to the comprehensive features that characterize the information of an electrical signal in the time and frequency domains. For example, time-frequency characteristics include the time-domain variation pattern and frequency-domain harmonic distortion information.

[0045] Temporal evolution information is a sequence of time-frequency characteristics arranged in chronological order, reflecting the development process of faults from minor to severe over time.

[0046] The first fault classification result is a determination of the current fault type of the motor based on data-driven analysis. For example, 0 to 7 can be used to identify 8 different types of fault results.

[0047] The fault results can be categorized as follows: 0—normal operation, 1—inter-turn short circuit, 2—grounding fault, 3—rotor bar breakage, 4—bearing fault, 5—overload, 6—unbalanced operation, and 7—stator winding fault.

[0048] Among them, data-driven analysis is a method that uses historical operating data or collected measured operating data as a basis to mine fault features based on multi-layer convolutional networks, and makes fault judgments based on the mapping relationship between fault features and fault types.

[0049] Step S203: In the process of determining the first fault classification result, physical simulation analysis is performed on the electrical state parameters in parallel to obtain the second fault classification result generated based on the physical simulation analysis.

[0050] The second fault classification result is the determination result of the current fault type of the motor obtained based on physical simulation analysis.

[0051] Physical simulation analysis is a method that uses mathematical models constructed based on the electromagnetic and dynamic physical laws of electric motors to numerically simulate and solve electrical state parameters, quantifying the degree of deterioration under various operating conditions of the motor and thus identifying abnormal motor states. For example, physical simulation analysis can be achieved using three types of mathematical models: electrical circuit models, motor kinematics models, and electromechanical coupling models.

[0052] Specifically, while determining the first fault classification result based on the time-series evolution information, the constructed mathematical model is used to perform parallel physical simulation analysis on the electrical state parameters to obtain the second fault classification result obtained from the physical simulation analysis.

[0053] Step S204: Based on the cross-validation of the first fault classification result and the second fault classification result, determine the target fault of the motor.

[0054] Cross-validation involves calculating the deviation between the first and second fault classification results, and then using a weighted fusion method to calculate a weighted average deviation as the final deviation value. The weighted fusion dynamically adjusts the weights based on the historical validation accuracy of the physical simulation analysis; the higher the accuracy, the greater the proportion of the deviation result in the final deviation value. The target fault is the fault type selected based on the cross-validation results. Specifically, the first and second fault classification results are cross-validated to determine the validity of the first fault classification result. If the first fault classification result is valid, the target fault currently existing in the motor is determined based on it; if the first fault classification result is invalid, the target fault currently existing in the motor is determined based on the second fault classification result.

[0055] The fault identification method for electric motors provided in this invention obtains operating state parameters and extracts time-frequency features from the electrical state parameters. Based on the time-series evolution information of the extracted time-frequency features, a first fault classification result is determined, which can fully capture the dynamic features and evolution patterns in the fault development process and improve the sensitivity to early fault identification. At the same time, the parallel physical simulation analysis obtains a second fault classification result, which can provide auxiliary basis for fault identification from a physical level. The target fault is determined by cross-validating the two types of fault classification results. This method integrates the advantages of data-driven and physical models, avoids the identification error caused by a single analysis method, and improves the accuracy of fault identification.

[0056] This embodiment provides a method for identifying faults in electric motors, which can be used in the aforementioned electronic devices, such as laptops and desktop computers. Figure 3 This is a flowchart of a fault identification method for an electric motor according to an embodiment of the present invention, such as... Figure 3 As shown, the process includes the following steps: Step S301: Obtain the motor's operating status parameters. The motor's operating voltage exceeds a preset value. For details, please refer to [link to relevant documentation]. Figure 2 Step S201 of the illustrated embodiment will not be described again here.

[0057] Step S302: Extract time-frequency features from electrical state parameters, and determine the first fault classification result of the motor based on the time-series evolution information of the extracted time-frequency features.

[0058] Specifically, step S302 includes: Step S3021: Perform mixed segmented sampling of the time-series voltage and time-series current in the electrical state parameters to obtain the segmented sampling results.

[0059] Time-series voltage and time-series current refer to the sequences of instantaneous three-phase voltage and current values ​​of a motor, continuously and synchronously acquired according to a unified timestamp. They are dual-channel one-dimensional time-domain signal waveforms that change synchronously with time. For example, the instantaneous three-phase current values ​​are continuously recorded at 128 sampling points per second.

[0060] Segmented sampling results in a set of paired voltage and current segments obtained after mixed segmented sampling.

[0061] Specifically, segmented sampling parameters (such as segmented sampling step size, sliding window, etc.) are set, and the time-series voltage and time-series current in the electrical state parameters are simultaneously sampled in mixed segments according to these parameters, resulting in a set of paired voltage segments and current segments. Each voltage segment and current segment contains the same fixed number of sampling points, and the voltage segments and current segments correspond strictly to each other on the time axis.

[0062] In some optional implementations, step S3021 above includes: Step a1: Obtain the sliding window for segmented sampling.

[0063] A sliding window is a time interval with a certain window length defined on a long-term continuous signal. This interval moves gradually along the time axis according to certain rules, and each time it moves, a new segment of the signal is extracted.

[0064] The window length refers to the total number of sampling points contained in a single window, which determines the data size of a single electrical signal segment.

[0065] Specifically, the sliding window can be pre-configured or adaptively determined based on the time window corresponding to the timing voltage and timing current; no specific limitation is made here.

[0066] Step a2: Based on the sliding step size corresponding to the sliding window, the timing voltage and timing current are sampled in parallel to obtain the segmented sampling results.

[0067] The sliding step size refers to the number of sampling points that a single window moves in each translation, and it is used to control the overlap rate between two adjacent signal segments. When the sliding step size is less than the window length, there is data overlap between adjacent windows; when the sliding step size is equal to the window length, there is no overlapping data between adjacent windows.

[0068] The overlap rate is used to measure the proportion of shared sampling points between the two sliding windows of the quantization vector. A higher overlap rate means that there is more overlapping sampling data between the two signal segments, which can preserve the timing details of the slow evolution of the fault; a lower overlap rate means less data redundancy and higher computational efficiency. The specific calculation formula is as follows:

[0069] Specifically, using the same sliding window start position and window length, time-series voltage and current within the same time interval are sampled in parallel to obtain paired voltage and current segments as a set of segmented sampling results. For example, setting the window length to 1024 sampling points and the sliding step size to 512 sampling points, the overlap rate is calculated to be 50%. The seven channels of time-series data (time-series voltage and current) are synchronously truncated. After each sliding window truncates, a set of paired voltage and current segments is generated, and the signals from each channel are integrated to form a 7×1024 mixed data matrix. The synchronous truncating operation is repeated by shifting the sliding window by 512 sampling points each time. There is 50% overlap in sampling data between adjacent sliding windows. After completely traversing all time-series data, all segments are summarized to obtain the segmented sampling results.

[0070] The fault identification method for electric motors provided in this invention obtains a segmented sampling sliding window and performs parallel sampling of time-series voltage and time-series current based on the sliding step size corresponding to the sliding window to obtain segmented sampling results. This enables synchronous acquisition of voltage and current signals, ensuring that the two types of signals are aligned on the time axis and avoiding phase deviation caused by asynchronous sampling.

[0071] Step S3022: Perform spectral transformation on each segment of electrical signal in the segmented sampling results to extract the multi-dimensional spectral features corresponding to each segment of electrical signal.

[0072] Multidimensional spectral features are a set of feature parameters obtained by extracting different types of frequency domain feature parameters from the spectral transformation results and then splicing and fusing them. For example, the amplitude spectrum, phase spectrum, and power spectral density data of a signal can be extracted separately, and the three types of frequency domain data can be spliced ​​together to obtain multidimensional spectral features.

[0073] Specifically, Fourier transform, wavelet transform, and other spectrum transformation algorithms are performed on each segment of electrical signal in the segmented sampling results to calculate the amplitude spectrum, phase spectrum, power spectral density, and other information of the corresponding signal, and thus obtain the spectrum transformation result.

[0074] Step S3023: Perform feature compression on the multi-dimensional spectral features to obtain time-frequency features.

[0075] Specifically, the multi-dimensional spectral features are standardized to eliminate dimensional differences, and then a dimensionality reduction mapping operation is performed to compress the high-dimensional multi-dimensional spectral features into a fixed-length one-dimensional feature vector to obtain the time-frequency features.

[0076] Step S3024: Fault identification is performed on the time-frequency features to obtain fault-related features.

[0077] Fault correlation features are obtained by filtering all abnormal features in time-frequency features and obtaining a feature subset that is strongly correlated with the fault type.

[0078] Specifically, time-frequency features can be used to mine the transient time-domain features and steady-state frequency-domain features of the motor, and identify the core features that are strongly correlated with the fault type, fault degree and fault location. These core features are the fault association features.

[0079] Furthermore, by using a pre-trained recognition model, abnormal information related to various fault types of motors can be detected and located from time-frequency features, thus obtaining fault association features.

[0080] The identification model can be trained based on a neural network or deep learning; there are no specific limitations here, as long as it can achieve fault identification.

[0081] In some optional implementations, step S3024 above includes: Step b1: Based on the first convolutional neural network with a first preset number of layers, identify abnormal frequency domain features in the time-frequency features.

[0082] Step b2: Based on the correlation between abnormal frequency domain features and faults, extract fault-related features from the abnormal frequency domain features.

[0083] The first convolutional neural network is a one-dimensional convolutional neural network with a first preset number of layers. For example, the first preset number of layers can be 3 layers. Here, a structure of three layers of 1×1 convolutional and pooling layers stacked alternately can be used as the first convolutional neural network.

[0084] Abnormal frequency domain features refer to frequency domain features that deviate from the normal value range or exhibit abnormal changes such as distortion and shift in time-frequency characteristics.

[0085] Specifically, a first convolutional neural network with a pre-configured first preset number of network layers is used as the input of the first convolutional neural network. Through each layer of convolutional kernels, local feature extraction, weight parameter learning and feature superposition mapping are completed layer by layer. The network automatically captures latent abnormal information such as frequency shift, harmonic abrupt change, frequency band energy anomaly, and local frequency domain distortion in the time-frequency features, so as to accurately distinguish between normal frequency domain features and deviation features, and output and identify abnormal frequency domain features in the time-frequency features.

[0086] Correlation is a quantitative indicator used to measure the degree of matching between a certain abnormal frequency domain feature and the type of motor fault. The higher the correlation value, the more likely the abnormal feature is to be caused by a certain specific fault.

[0087] Specifically, the linear correlation and significance weight between each abnormal feature and different fault types, fault locations, and fault damage degrees can be quantitatively calculated. By setting a correlation threshold, weakly correlated, interfering, and redundant abnormal features are filtered out, while core abnormal features that highly match the actual faults of the motor and are representative are retained. These core abnormal features are the fault correlation features.

[0088] In an optional implementation, each type of abnormal frequency domain feature can be paired with corresponding fault samples and normal samples; the correlation between abnormal frequency domain features and faults can be calculated by jointly using Pearson correlation coefficient and grey relational degree, and the Pearson correlation coefficient and grey relational degree can be weighted and fused to obtain the comprehensive correlation degree; a correlation degree threshold is set, features with comprehensive correlation degree higher than the correlation degree threshold are retained, and weakly correlated interference features are removed to obtain the final fault correlation features.

[0089] Among them, the grey relational degree is used to calculate the curve similarity between the abnormal feature sequence and the standard fault feature sequence. The closer the value is to 1, the stronger the correlation. The Pearson correlation coefficient is used to quantify the linear correlation between the feature amplitude change and the degree of fault damage.

[0090] The fault identification method for electric motors provided in this embodiment of the invention can capture subtle abnormal frequency domain features by identifying abnormal frequency domain features in time-frequency features through a first convolutional neural network with a first preset number of layers. Based on the correlation between abnormal frequency domain features and faults, fault-related features are extracted from the abnormal frequency domain features, which can eliminate interference information unrelated to the fault and strengthen the correspondence between fault features and fault types.

[0091] Step S3025: Determine the first fault classification result based on the temporal evolution characteristics of the fault association features.

[0092] Temporal evolution features are time series composed of fault-related features spliced ​​together from continuous time windows. The features extracted through convolution operations are used to characterize the development pattern of faults.

[0093] The time series is obtained by splicing together the fault association features output by multiple consecutive sliding windows in chronological order of acquisition time.

[0094] Specifically, fault-related features from multiple time windows are concatenated into a time-series sequence according to the chronological order of data collection. Dynamic evolution characteristics of these fault-related features on this time-series sequence are then extracted, such as feature amplitude drift, fluctuation trends, growth rates, periodic changes, and steady-state shifts. These dynamic evolution characteristics are used to characterize the dynamic evolution of the fault.

[0095] The temporal evolution features are input into a pre-trained fault classification model. The standard temporal evolution templates and feature change patterns corresponding to each type of fault are matched. The confidence probability of each fault category is calculated, and the fault type corresponding to the highest confidence is selected to obtain the first fault classification result.

[0096] The fault classification model adopts a temporal feature classification structure, which can be trained based on a temporal convolutional network or a long short-term memory network. There is no specific limitation here, as long as it can capture information such as the long-term trend of features, abrupt inflection points, and temporal fluctuation patterns to adapt to the evolution scenarios such as gradual fault change and sudden anomalies.

[0097] In some optional implementations, step S3025 above includes: Step c1: Input the fault-related features into a second convolutional neural network with a second preset number of layers, where the second preset number of layers is less than the first preset number of layers.

[0098] Step c2: Based on the second convolutional neural network, extract the temporal evolution features corresponding to the fault association features.

[0099] Step c3: Based on the temporal evolution characteristics, perform fault classification to obtain the first fault classification result.

[0100] The second convolutional neural network is a one-dimensional convolutional neural network with a second preset number of layers, where the second preset number of layers is less than the first preset number of layers. For example, if the first preset number of layers is 3, the second preset number of layers can be 2. In a specific example, the backbone of the second convolutional neural network can adopt a structure of two layers of 1×1 convolutional and pooling layers stacked alternately, and incorporate multi-head self-attention to mine long sequence dependencies.

[0101] Specifically, the time-series sequence obtained by concatenating fault-related features in chronological order is input into the second convolutional neural network with a second preset number of layers; the time-series evolution features are obtained by feature mining; the time-series evolution features are then passed through a flattening layer and a fully connected layer to complete feature mapping; finally, the matching probability of various faults is calculated by normalization through an activation layer to obtain the first fault classification result.

[0102] The fault identification method for electric motors provided in this invention involves inputting fault-related features into a second convolutional neural network with a second preset number of layers less than a first preset number of layers, and extracting temporal evolution features corresponding to the fault-related features based on the second convolutional neural network. This method can capture the dynamic change trend of fault features along the time dimension and fully capture the continuous evolution process of fault occurrence and development. Then, based on the temporal evolution features, fault classification is performed to obtain a first fault classification result, which can effectively distinguish between short-term interference signals and the actual fault evolution process, and improve the sensitivity to early fault identification.

[0103] Step S303: During the process of determining the first fault classification result, a physical simulation analysis is performed on the electrical state parameters in parallel to obtain the second fault classification result generated based on the physical simulation analysis. For details, please refer to [link to relevant documentation]. Figure 2 Step S203 of the illustrated embodiment will not be described again here.

[0104] Step S304: Based on the cross-validation of the first fault classification result and the second fault classification result, determine the target fault of the motor. For details, please refer to [link to relevant documentation]. Figure 2 Step S204 of the illustrated embodiment will not be described again here.

[0105] The fault identification method for electric motors provided in this invention, by performing mixed segmented sampling of time-series voltage and time-series current in electrical state parameters, can take into account both time continuity and local detailed features, avoiding feature loss caused by collecting only a single electrical state parameter; by performing spectral transformation on each segment of electrical signal in the segmented sampling results and extracting multi-dimensional spectral features, fault feature information in different frequency bands can be captured, enriching the information dimensions of fault identification; by performing feature compression on multi-dimensional spectral features to obtain time-frequency features, redundant and irrelevant features can be eliminated while retaining core fault identification information, reducing the amount of data computation; by performing fault identification based on time-frequency features to obtain fault association features, the corresponding association relationship between electrical signals and different fault types can be established; then, based on the time-series evolution features of fault association features, the first fault classification result is determined, which can dynamically track the changing trend of fault features over time, explore the evolution law of fault occurrence and development, and improve the sensitivity of early fault identification.

[0106] This embodiment provides a method for identifying faults in electric motors, which can be used in the aforementioned electronic devices, such as laptops and desktop computers. Figure 4 This is a flowchart of a fault identification method for an electric motor according to an embodiment of the present invention, such as... Figure 4 As shown, the process includes the following steps: Step S401: Obtain the motor's operating status parameters. For details, please refer to [link to relevant documentation]. Figure 2 Step S201 of the illustrated embodiment will not be described again here.

[0107] Step S402 involves extracting time-frequency features from the electrical state parameters and determining the first fault classification result of the motor based on the time-series evolution information of the extracted time-frequency features. For details, please refer to [link to relevant documentation]. Figure 3 Step S302 of the illustrated embodiment will not be described again here.

[0108] Step S403: In the process of determining the first fault classification result, physical simulation analysis is performed on the electrical state parameters in parallel to obtain the second fault classification result generated based on the physical simulation analysis.

[0109] Specifically, step S403 includes: Step S4031: Using the electrical circuit model, identify abnormal parameters in the electrical state parameters. The electrical circuit model is obtained by modeling the equivalent circuit parameters of the motor.

[0110] An electrical circuit model is a mathematical model established based on the equivalent circuit parameters of an electric motor, used to describe the electromagnetic relationships of the motor. For example, the voltage equation, flux linkage equation, and electromagnetic torque equation in a two-phase synchronous rotating coordinate system can be used to characterize the electromagnetic relationships of the motor. The specific calculation steps are as follows: The three-phase voltage, three-phase current, and zero-sequence leakage current in the electrical state parameters are transformed using the Parker transformation:

[0111]

[0112] Where, transformation matrix Include , , Elements, synchronous electric angular velocity ， , Represents the grid frequency; for motors without a neutral connection, the zero-sequence component... ; Indicates three-phase voltage; Indicates three-phase current; This represents the transformed three-phase voltage; This represents the transformed three-phase current.

[0113] Based on the transformed coordinates, the stator voltage equation is constructed as follows:

[0114] in, and stator , Shaft voltage; and stator , shaft current; This represents the equivalent resistance per phase of the stator; and stator , Axial magnetic flux.

[0115] Based on the transformed coordinates, setting the rotor voltage to 0, the rotor voltage equation is constructed as follows:

[0116] in, This represents the equivalent resistance of each phase of the rotor; and Indicates rotor , shaft current; Indicates the rotor's electrical angular velocity; and Indicates rotor , Axial magnetic flux.

[0117] Based on the transformed coordinates, construct the flux linkage equation:

[0118] in, ; ; Indicates the total inductance of the stator; Indicates the rotor's total inductance; Indicates stator leakage inductance; Indicates rotor leakage inductance; This represents the magnetizing inductance.

[0119] Based on the above, the stator , shaft flux linkage and stator , Based on shaft current, construct the electromagnetic torque equation based on magnetic flux linkage and current:

[0120] in, Indicates electromagnetic torque; This indicates the number of pole pairs of the motor.

[0121] Based on the equivalent resistance per phase of the rotor and the synchronous electrical angular velocity obtained above, a torque formula based on the equivalent circuit is constructed:

[0122]

[0123]

[0124] in, This represents the rotor resistance referred to the stator side; Indicates stator leakage reactance; This represents the converted rotor leakage reactance; This represents the effective value of the stator phase voltage; Indicates the critical slip ratio; This indicates the maximum torque.

[0125] Using the voltage equation, flux linkage equation, and electromagnetic torque equation above, we can calculate the values ​​of key electrical state parameters such as stator current, rotor current, stator and rotor flux linkage, and electromagnetic torque under the current operating conditions, which can be used for subsequent abnormal parameter identification and motor operating status analysis.

[0126] Step S4032: Using the motor operation model, motor anomalies are identified based on abnormal parameters and motor load parameters.

[0127] A motor operation model is a mathematical model based on the motor's motion equations, used to characterize the electromechanical coupling operation characteristics of the motor. For example, the rotor motion equations can be used to construct a motor operation model.

[0128] Load parameters refer to the characteristic parameters of the load driven by the motor. For example, the corresponding load parameters can be determined by using a fan load torque model.

[0129] Motor anomalies are identified by using abnormal parameters and load parameters in the motor motion equations to derive abnormal quantities at the operational state level, thereby identifying the corresponding motor anomalies. For example, abnormal parameters output from the electrical circuit model and the fan load torque model are input together into the rotor motion equations for solving, yielding abnormal quantities at the operational state level, and thus identifying the corresponding motor anomalies. The specific calculation steps are as follows: Based on the load parameters, a wind turbine load torque model is constructed to obtain the load torque:

[0130]

[0131]

[0132]

[0133]

[0134]

[0135]

[0136]

[0137] in, Indicates mechanical angular velocity; Indicates the load ratio factor; Indicates load torque; The rated power of the fan is indicated on the nameplate. Indicates load power; Indicates the rated speed of the fan; Indicates standard air density; Indicates the current air density; Indicates the rated torque; Indicates the rated power of the motor; This indicates the rated speed of the motor.

[0138] Based on the electromagnetic torque and load torque, substituting them into the rotor motion equation, the rotor electric angular velocity is obtained:

[0139]

[0140]

[0141] in, This represents the total moment of inertia of the motor and the fan; Indicates mechanical angular velocity; Indicates the coefficient of viscous friction; This represents the torque lost due to dry friction and wind resistance.

[0142] Step S4033: Using the fault identification model, based on the electrical circuit parameters corresponding to the motor abnormality and the motor operating parameters, determine the second fault classification result.

[0143] The fault identification model is cascaded with the motor operation model and is constructed based on the correlation between electrical circuit parameters, motor operation parameters and motor faults.

[0144] Specifically, the electrical circuit parameters and motor operating parameters corresponding to the motor abnormality are summarized, the actual values ​​calculated from each group of abnormal parameters are compared with the benchmark values, and feature matching is performed in combination with the evolution patterns of various faults to determine the second fault classification result.

[0145] The baseline value is the theoretical parameter value calculated under normal, fault-free operating conditions.

[0146] For example, a fault identification model can be constructed using an electromechanical coupling model; the specific calculation steps are as follows: Eliminating the flux linkage term from the stator voltage equation and flux linkage equation yields the stator current equation:

[0147]

[0148]

[0149] in, This represents the leakage flux coefficient.

[0150] The stator current equation, flux linkage equation, and rotor motion equation are converted into standard state-space form to obtain the electromechanical coupling model:

[0151]

[0152]

[0153]

[0154]

[0155]

[0156]

[0157]

[0158]

[0159]

[0160]

[0161] in, Represents state variables; Indicates input variables; Indicates the rotor time constant; Represents the stator transient time constant; This represents the rotor coupling coefficient.

[0162] Then, the values ​​of the state variables were obtained by solving the electromechanical coupling model using the fourth-order Runge-Kutta method:

[0163]

[0164]

[0165]

[0166]

[0167] in, This is the simulated step size.

[0168] Next, the baseline values ​​of the state variables are calculated using the following formula:

[0169] in, Indicates the resistivity of copper; Indicates the average coil length; Indicates the cross-sectional area of ​​the conductor; This indicates the number of turns per phase of the stator winding.

[0170] Copper resistivity at 75°C It needs to be corrected according to the actual temperature. The temperature correction formula is as follows:

[0171] in, , This indicates a reference temperature, usually taken as 75℃.

[0172]

[0173]

[0174] in, Indicates the conversion factor; Indicates the number of rotor slots; Indicates the stator winding coefficient; This represents the equivalent resistance of the rotor bars themselves. This represents the equivalent resistance of the end rings at both ends of the rotor.

[0175]

[0176] in, Indicates the permeability of free space; Indicates the stator inner diameter; Indicates the effective length of the iron core; Indicates the air gap length; This represents the Carter coefficient.

[0177]

[0178]

[0179] in, Indicates stator slot leakage inductance; Indicates end leakage; Indicates the specific leakage permeability; Indicates the length of the end protrusion; Indicates the end coefficient; This indicates the number of slots per pole per phase of the stator.

[0180]

[0181]

[0182] in, Indicates the bearing friction coefficient; Indicates the bearing friction power; This indicates the bearing friction torque; The value is the rotational speed.

[0183] The evolution patterns of various faults are shown in Table 1: Table 1

[0184] F1 Stator Turn-to-Turn Short Circuit: Insulation breakdown occurs between some turns of the same phase of the stator winding, forming a short-circuit loop. Assume the short-circuit turn ratio is:

[0185] in, Indicates the number of short-circuited turns after the fault; Indicates the total number of turns per phase when the circuit is healthy.

[0186] Early days: Slightly reduced; Slight decrease Basically unchanged or slightly decreased; negative sequence component appears in the current, and the current in a certain phase increases slightly; local temperature rises, and the overall temperature is slightly higher (+5~10℃); slight torque pulsation ( ).

[0187] Mid-term: Continuously decreasing (effective number of turns decreases); A significant decrease (5%~15%). The current begins to decrease, the negative sequence current increases (3%~10%), and the phase current imbalance increases; the hot spot temperature rises significantly, and the periodic fluctuations of accelerated insulation aging intensify. ),Appear Frequency pulsation.

[0188] Post-production: The resistance may drop suddenly (it will recover if the short-circuit ring burns out, and continue to drop if it expands) or local overheating may cause the resistance to rise in reverse. Significant drop (decline of more than 15%) Due to the severe imbalance of leakage flux saturation nonlinear change current ( This could lead to localized overheating and burnout of the protection mechanism, potentially developing into a ground fault or phase-to-phase short circuit with strong pulsation, increased vibration and noise, and possible loss of synchronism or stalling.

[0189] F2 rotor bar breakage: Initial stage: 1-2 broken branches; Slightly increased ( ); speed appears The frequency oscillates slightly.

[0190] Mid-game: Breaks expand, Increase ( Significant torque pulsation and the appearance of sidebands in the stator current. .

[0191] Later stages: Multiple broken strips or even broken loops. Significant increase; longer start-up time, decreased maximum torque, and ultimately, inability to start under load.

[0192] The abnormal threshold ranges are shown in Table 2: Table 2

[0193]

[0194]

[0195]

[0196]

[0197]

[0198]

[0199] in, This represents the maximum value among the three-phase currents; This represents the minimum value of the three-phase current; This represents the average value of the three-phase current; Indicates the fluctuation range of rotational speed; Indicates the rated electric angular velocity; This represents the change in stator resistance; Indicates the stator resistance value; The reference value for the stator resistance; This represents the change in stator inductance; Indicates the total inductance of the stator; The reference value for stator inductance; This represents the effective value of the vibration velocity; Represents the instantaneous sampled value of the vibration signal; Indicates the number of sampling points.

[0200] Step S404: Based on the cross-validation of the first fault classification result and the second fault classification result, determine the target fault of the motor. For details, please refer to [link to relevant documentation]. Figure 2 Step S204 of the illustrated embodiment will not be described again here.

[0201] The electric motor fault identification method provided in this invention identifies abnormal parameters in electrical state parameters by utilizing an electrical circuit model obtained from modeling based on the motor's equivalent circuit parameters, thereby accurately acquiring electrical anomalies that deviate from normal operating conditions. By using a motor operation model cascaded with the electrical circuit model, motor anomalies are identified based on abnormal parameters and the motor's load parameters, achieving the fusion of electrical and mechanical operating condition information and avoiding missed fault detections caused by single-dimensional parameter identification. Furthermore, by using a fault identification model cascaded with the motor operation model, a second fault classification result is determined based on electrical circuit parameters and motor operation parameters, taking into account circuit, dynamic motion laws, and fault mapping laws, enriching the basis for fault identification, and improving the accuracy of fault classification results.

[0202] This embodiment provides a method for identifying faults in electric motors, which can be used in the aforementioned electronic devices, such as laptops and desktop computers. Figure 5 This is a flowchart of a fault identification method for an electric motor according to an embodiment of the present invention, such as... Figure 5 As shown, the process includes the following steps: Step S501: Obtain the motor's operating status parameters. For details, please refer to [link to relevant documentation]. Figure 2 Step S201 of the illustrated embodiment will not be described again here.

[0203] Step S502 involves extracting time-frequency features from the electrical state parameters and determining the first fault classification result of the motor based on the time-series evolution information of the extracted time-frequency features. For details, please refer to [link to relevant documentation]. Figure 3 Step S302 of the illustrated embodiment will not be described again here.

[0204] Step S503: During the process of determining the first fault classification result, a physical simulation analysis is performed on the electrical state parameters in parallel to obtain the second fault classification result generated based on the physical simulation analysis. For details, please refer to [link to relevant documentation]. Figure 4 Step S403 of the illustrated embodiment will not be described again here.

[0205] Step S504: Based on the cross-validation of the first fault classification result and the second fault classification result, determine the target fault of the motor.

[0206] Specifically, step S504 includes: Step S5041: Obtain the first abnormal parameter, the first motor abnormality, and the first fault classification corresponding to the first fault classification result.

[0207] The first abnormal parameter is a quantified value of an electrical characteristic that deviates from the normal range and is detected during the data-driven analysis process. Examples include quantifiable indicators such as characteristic timing increases, frequency offsets, and harmonic amplitudes.

[0208] The normal range is the range of standard values ​​for various electrical characteristics obtained from a large number of samples under the condition that the motor is running stably without faults.

[0209] The first motor anomaly is a motor operation anomaly identified by data-driven analysis combined with the analysis of the deviation and variation pattern of the first anomaly parameter. Examples include three-phase current imbalance, continuous increase in rotor sidefrequency, and periodic fluctuation of load current.

[0210] The first fault classification is the final fault determination result output by data-driven analysis.

[0211] Step S5042: Compare the first abnormal parameter with the abnormal parameter identified by the electrical circuit model to determine the first deviation.

[0212] Abnormal parameters identified by the electrical circuit model refer to parameter values ​​that deviate from the rated values ​​or normal ranges after substituting electrical state parameters into the electrical circuit model.

[0213] Specifically, the first abnormal parameter identified by data-driven analysis and the abnormal parameter identified by the electrical circuit model are compared under a unified dimension or a unified parameter space. The difference is calculated using methods such as parameter mapping comparison and normalized residual method to obtain the first deviation.

[0214] Step S5043: Compare the first motor abnormality with the motor abnormality identified by the motor operation model to determine the second deviation.

[0215] Motor anomalies identified by the motor operation model are abnormal phenomena at the motor operation state level that are determined by substituting the output of the electrical circuit model into the motor operation model.

[0216] Specifically, the first motor anomaly identified by data-driven analysis and the motor anomaly identified by the motor operation model are compared under a unified dimension or a unified parameter space. The difference is calculated using methods such as state coding mapping and entity unified mapping to obtain the second deviation.

[0217] Step S5044: Compare the first fault classification with the second fault classification result output by the fault identification model to determine the third deviation.

[0218] Specifically, the difference between the first fault classification obtained from data-driven analysis and the second fault classification result output by the fault identification model is calculated using methods such as the confusion matrix weighting method and the probability distribution distance method to obtain the third deviation.

[0219] Step S5045: Based on the weighted result of the first deviation, the second deviation, and the third deviation, determine the target deviation and the target fault corresponding to the target deviation.

[0220] Target fault refers to the type of motor fault that has the smallest target deviation and has been determined through cross-validation of data-driven analysis and physical simulation analysis.

[0221] Specifically, the first, second, and third deviations are multiplied by their respective weighting coefficients based on their importance, and then weighted and summed to obtain the target deviation. For example, if the first, second, and third deviations are 0.2, 0.12, and 0.15 respectively; the preset weights are 0.2, 0.3, and 0.5 respectively; and the target deviation is 0.20 × 0.2 + 0.12 × 0.3 + 0.15 × 0.5 = 0.151.

[0222] The weighting coefficients can be dynamically adjusted based on historical verification accuracy. The higher the weighting coefficient of an indicator with high importance, the stronger its impact on target deviation.

[0223] In some optional implementations, step S5045 above includes: Step d1: Obtain the deviation threshold corresponding to the running parameters.

[0224] Step d2: If the target deviation exceeds the deviation threshold, the first fault classification result is calibrated, and the target fault of the motor is determined based on the second fault classification result.

[0225] Step d3: If the target deviation does not exceed the deviation threshold, then the target fault of the motor is determined based on the first fault classification result.

[0226] The deviation threshold is a pre-set critical value that represents the maximum allowable difference between data-driven analysis and physical simulation analysis, used to determine whether the target deviation is within an acceptable range. For example, it can be dynamically adjusted based on real-time parameters such as the motor's power factor, load rate, and temperature.

[0227] Specifically, when collecting the current operating parameters of the motor, the deviation threshold corresponding to the current operating condition is retrieved; the target deviation is compared with the deviation threshold corresponding to the current operating condition. If the target deviation exceeds the deviation threshold, the process of obtaining the first fault classification result is iteratively optimized, and the first fault classification result is determined to be unreliable and not used as a reference for fault judgment. The target fault of the motor is directly determined based on the second fault classification result; if the target deviation does not exceed the deviation threshold, the first fault classification result is determined to be reliable, and the first fault classification result is directly output as the target fault of the motor.

[0228] The fault identification method for electric motors provided in this embodiment of the invention provides a quantitative benchmark for judging the difference in multi-model identification deviations by obtaining a deviation threshold corresponding to the operating parameters. If the target deviation exceeds the deviation threshold, the first fault classification result is calibrated, and the target fault of the electric motor is determined based on the second fault classification result. This method can correct the data-driven results based on physical model analysis, reducing the risk of fault misjudgment. If the target deviation does not exceed the deviation threshold, the target fault of the electric motor is determined based on the first fault classification result, which can improve the efficiency of fault identification while ensuring accurate identification.

[0229] The electric motor fault identification method provided in this invention compares the acquired first abnormal parameter with the abnormal parameter identified by the electrical circuit model to determine the first deviation. This allows for cross-verification of data-driven and circuit model identification results at the electrical state parameter level, quantifying the difference between the two in extracting abnormal electrical quantities. Comparing the acquired first motor abnormality with the motor abnormality output by the motor operation model to determine the second deviation allows for cross-verification of data-driven and physical model analysis results at the electromechanical operation state level, quantifying the difference between the two in identifying motor operation abnormalities. Comparing the acquired first fault classification with the second fault classification result output by the fault identification model to determine the third deviation quantifies the discrepancy between the two methods in fault classification. Then, based on the weighted calculation results of the first, second, and third deviations, the target deviation and corresponding target fault are determined. This quantifies the error weights of each identification stage, integrates multi-level verification deviations to obtain the true fault, and improves the accuracy of electric motor fault location.

[0230] In some optional implementations, the above method further includes: Step e1: If the target deviation is greater than the first threshold and less than the deviation threshold, then the generation and display of the first warning information are triggered; the first threshold is less than the deviation threshold.

[0231] Step e2: If the target deviation is greater than the second threshold but less than the deviation threshold, the fault monitoring cycle is shortened and the second warning information is sent to the remote terminal; the second threshold is greater than the first threshold.

[0232] Step e3: If the target deviation is greater than the third threshold but less than the deviation threshold, the protection mechanism is triggered; the third threshold is greater than the second threshold.

[0233] The first warning message is a mild alert triggered when the operating status parameters deviate slightly from the normal range. For example, the first warning message can be processed by displaying a warning message.

[0234] Shortening the fault monitoring cycle involves reducing the original time interval for fault identification and analysis, and increasing the frequency and density of sampling operating status parameters to continuously track abnormal change trends.

[0235] The second warning message is a prompt message triggered when the operating status parameters continuously deviate from the normal range or when the trend changes significantly.

[0236] Among these, the processing of the second warning information can be achieved by sending warning information to a remote terminal.

[0237] The protection mechanism is a mandatory safety operation triggered when the target deviation does not exceed the deviation threshold, but a serious fault or threat to equipment safety is detected. For example, the protection mechanism can be implemented through tripping, alarms, or remote emergency notifications.

[0238] The fault identification method for electric motors provided in this invention sets progressively increasing first, second, and third thresholds, all of which are less than the deviation threshold. This allows for the matching of differentiated response measures based on the target deviation between data-driven results and physical model analysis, thereby achieving refined hierarchical control of fault severity.

[0239] This embodiment also provides a fault identification device for an electric motor, which is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0240] This embodiment provides a fault identification device for an electric motor, such as... Figure 6 As shown, it includes: The signal acquisition module 601 is used to acquire the electrical status parameters of the motor.

[0241] The time-frequency feature extraction module 602 is used to extract time-frequency features from electrical state parameters and determine the first fault classification result of the motor based on the time-series evolution information of the extracted time-frequency features.

[0242] The physical simulation analysis module 603 is used to perform physical simulation analysis on electrical state parameters in parallel during the process of determining the first fault classification result, so as to obtain the second fault classification result generated based on the physical simulation analysis.

[0243] The target fault identification module 604 is used to determine the target fault of the motor based on the cross-validation of the first fault classification result and the second fault classification result.

[0244] In some optional implementations, the time-frequency feature extraction module 602 includes: The segmented sampling unit is used to perform mixed segmented sampling of time-series voltage and time-series current in electrical state parameters to obtain segmented sampling results.

[0245] The spectrum feature extraction unit is used to perform spectrum transformation on each segment of electrical signal in the segmented sampling results and extract the multi-dimensional spectrum features corresponding to each segment of electrical signal.

[0246] The feature compression unit is used to compress multi-dimensional spectral features to obtain time-frequency features.

[0247] The fault identification unit is used to identify faults based on time-frequency characteristics and obtain fault-related characteristics.

[0248] The first result determination unit is used to determine the first fault classification result based on the temporal evolution characteristics of fault association features.

[0249] In some alternative implementations, the segmented sampling unit includes: The window acquisition sub-unit is used to acquire the sliding window for segmented sampling.

[0250] The parallel sampling subunit is used to sample the timing voltage and timing current in parallel based on the sliding step size corresponding to the sliding window, so as to obtain the segmented sampling results.

[0251] In some optional implementations, the fault identification unit includes: The first convolutional subunit is used to identify abnormal frequency domain features in time-frequency features based on the first convolutional neural network with a first preset number of layers.

[0252] The correlation feature extraction subunit is used to extract fault correlation features from the abnormal frequency domain features based on the correlation between the abnormal frequency domain features and the fault.

[0253] In some optional implementations, the fault result determination unit includes: The second convolutional subunit is used to input fault-related features into a second convolutional neural network with a second preset number of layers, which is less than the first preset number of layers.

[0254] The temporal feature extraction subunit is used to extract the temporal evolution features corresponding to the fault association features based on the second convolutional neural network.

[0255] The fault classification subunit is used to classify faults based on temporal evolution characteristics and obtain the first fault classification result.

[0256] In some alternative implementations, the physical simulation analysis module 603 includes: The electrical state parameter identification unit is used to identify abnormal parameters in the electrical state parameters using an electrical circuit model. The electrical circuit model is obtained by modeling the equivalent circuit parameters of the motor.

[0257] The motor anomaly identification unit is used to identify motor anomalies by utilizing the motor operation model and based on the abnormal parameters and the motor load parameters. The motor operation model is cascaded with the electrical circuit model, and the motor operation model is obtained by modeling the motor motion equations of the motor.

[0258] The second result determination unit is used to determine the second fault classification result based on the electrical circuit parameters and motor operating parameters corresponding to the motor abnormality using the fault identification model. The fault identification model is cascaded with the motor operating model, and the fault identification model is constructed based on the correlation between the electrical circuit parameters, motor operating parameters and motor faults.

[0259] In some alternative implementations, the target fault identification module 604 includes: The parameter acquisition unit is used to acquire the first abnormal parameter, the first motor abnormality, and the first fault classification corresponding to the first fault classification result.

[0260] The first determining unit is used to compare the first abnormal parameter with the abnormal parameter identified by the electrical circuit model to determine the first deviation.

[0261] The second determining unit is used to compare the first motor abnormality with the motor abnormality identified by the motor operation model to determine the second deviation.

[0262] The third determining unit is used to compare the first fault classification with the second fault classification result output by the fault identification model to determine the third deviation.

[0263] The fourth determining unit is used to determine the target deviation and the target fault corresponding to the target deviation based on the weighted result of the first deviation, the second deviation and the third deviation.

[0264] In some optional implementations, the fourth determining unit includes: The threshold acquisition subunit is used to acquire the deviation threshold corresponding to the running parameters.

[0265] The first threshold determination subunit is used to calibrate the first fault classification result and determine the target fault of the motor based on the second fault classification result if the target deviation exceeds the deviation threshold.

[0266] The second threshold determination subunit is used to determine the target fault of the motor based on the first fault classification result if the target deviation does not exceed the deviation threshold.

[0267] In some optional implementations, the fourth determining unit further includes: The first early warning subunit is used to trigger the generation and display of the first early warning information if the target deviation is greater than a first threshold and less than the deviation threshold; the first threshold is less than the deviation threshold.

[0268] The second early warning subunit is used to shorten the fault monitoring cycle and trigger the transmission of the second early warning information to the remote terminal if the target deviation is greater than the second threshold but less than the deviation threshold; the second threshold is greater than the first threshold.

[0269] The third early warning subunit is used to trigger the protection mechanism if the target deviation is greater than the third threshold but less than the deviation threshold; the third threshold is greater than the second threshold.

[0270] The motor fault identification device provided in this embodiment of the invention can execute the motor fault identification method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects for executing the method. Further functional descriptions of the various modules and units described above are the same as in the corresponding embodiments described above, and will not be repeated here.

[0271] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0272] The following is a detailed reference. Figure 7 This diagram illustrates a suitable structural schematic for implementing an electronic device according to embodiments of the present invention. The electronic device may include a processor (e.g., a central processing unit, graphics processor, etc.) 701, which can perform various appropriate actions and processes based on a program stored in a read-only memory (ROM) 702 or a program loaded from memory 708 into random access memory (RAM) 703. The RAM 703 also stores various programs and data required for the operation of the electronic device. The processor 701, ROM 702, and RAM 703 are interconnected via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0273] Typically, the following devices can be connected to I / O interface 705: input devices 706 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 707 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 708 including, for example, magnetic tapes, hard disks, etc.; and communication devices 709. Communication device 709 allows electronic devices to exchange data via wireless or wired communication with other devices. Although Figure 7 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0274] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 709, or installed from a memory 708, or installed from a ROM 702. When the computer program is executed by the processor 701, it performs the functions defined in the motor fault identification method of the embodiments of the present invention.

[0275] Figure 7 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0276] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the motor fault identification method shown in the above embodiments is implemented.

[0277] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0278] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for fault identification of an electric motor, characterized in that, The method includes: Obtain the electrical status parameters of the motor; the operating voltage of the motor exceeds a preset value; The electrical state parameters are subjected to time-frequency feature extraction, and the first fault classification result of the motor is determined based on the time-series evolution information of the extracted time-frequency features. During the process of determining the first fault classification result, physical simulation analysis is performed on the electrical state parameters in parallel to obtain the second fault classification result generated based on the physical simulation analysis. Based on the cross-validation of the first fault classification result and the second fault classification result, the target fault of the motor is determined.

2. The fault identification method for an electric motor according to claim 1, characterized in that, The step of extracting time-frequency features from the electrical state parameters and determining the first fault classification result of the motor based on the time-series evolution information of the extracted time-frequency features includes: The timing voltage and timing current in the electrical state parameters are sampled in a mixed segment to obtain the segmented sampling results; Perform spectral transformation on each segment of electrical signal in the segmented sampling results to extract the multi-dimensional spectral features corresponding to each segment of electrical signal; The time-frequency features are obtained by performing feature compression on the multi-dimensional spectral features. Fault identification is performed on the time-frequency features to obtain fault association features; Based on the temporal evolution characteristics of the fault association features, the first fault classification result is determined.

3. The fault identification method for an electric motor according to claim 2, characterized in that, The step of performing mixed segmented sampling of the time-series voltage and time-series current in the electrical state parameters to obtain segmented sampling results includes: Obtain the sliding window for segmented sampling; Based on the sliding step size corresponding to the sliding window, the timing voltage and the timing current are sampled in parallel to obtain the segmented sampling results.

4. The fault identification method for an electric motor according to claim 2, characterized in that, The process of fault identification based on the time-frequency features to obtain fault association features includes: Based on a first convolutional neural network with a first preset number of layers, abnormal frequency domain features in the time-frequency features are identified. Based on the correlation between the abnormal frequency domain features and the fault, the fault-related features are extracted from the abnormal frequency domain features.

5. The fault identification method for an electric motor according to claim 4, characterized in that, The determination of the first fault classification result based on the temporal evolution features of the fault association features includes: The fault-related features are input into a second convolutional neural network with a second preset number of layers, where the second preset number of layers is less than the first preset number of layers. Based on the second convolutional neural network, extract the temporal evolution features corresponding to the fault association features; Based on the aforementioned temporal evolution characteristics, fault classification is performed to obtain the first fault classification result.

6. The fault identification method for an electric motor according to claim 1, characterized in that, The parallel physical analysis of the electrical state parameters yields a second fault classification result based on the physical analysis, including: Abnormal parameters in the electrical state parameters are identified using an electrical circuit model, which is obtained by modeling the equivalent circuit parameters of the motor. Using a motor operation model, based on the abnormal parameters and the load parameters of the motor, motor abnormalities are identified. The motor operation model is cascaded with the electrical circuit model, and the motor operation model is obtained by modeling based on the motor motion equations of the motor. Using the fault identification model, the second fault classification result is determined based on the electrical circuit parameters and motor operating parameters corresponding to the motor abnormality; the fault identification model is cascaded with the motor operating model, and the fault identification model is constructed based on the correlation between the electrical circuit parameters, the motor operating parameters, and the motor fault.

7. The fault identification method for an electric motor according to claim 6, characterized in that, The step of determining the target fault of the motor based on the cross-validation of the first fault classification result and the second fault classification result includes: Obtain the first abnormal parameter, the first motor abnormality, and the first fault classification corresponding to the first fault classification result; The first abnormal parameter is compared with the abnormal parameter identified by the electrical circuit model to determine the first deviation; The first motor anomaly is compared with the motor anomaly identified by the motor operation model to determine the second deviation; The first fault classification is compared with the second fault classification result output by the fault identification model to determine the third deviation; Based on the weighted result of the first deviation, the second deviation, and the third deviation, the target deviation and the target fault corresponding to the target deviation are determined.

8. The fault identification method for an electric motor according to claim 7, characterized in that, Determining the target fault corresponding to the target deviation includes: Obtain the deviation threshold corresponding to the operating parameters; If the target deviation exceeds the deviation threshold, the first fault classification result is calibrated, and the target fault of the motor is determined based on the second fault classification result. If the target deviation does not exceed the deviation threshold, then the target fault of the motor is determined based on the first fault classification result.

9. The fault identification method for an electric motor according to claim 8, characterized in that, Also includes: If the target deviation is greater than the first threshold and less than the deviation threshold, then the generation and display of the first warning information are triggered; The first threshold is less than the deviation threshold; If the target deviation is greater than the second threshold but less than the deviation threshold, the fault monitoring cycle is shortened, and a second early warning message is sent to the remote terminal; the second threshold is greater than the first threshold. If the target deviation is greater than the third threshold but less than the deviation threshold, a protection mechanism is triggered; the third threshold is greater than the second threshold.

10. A fault identification device for an electric motor, characterized in that, The device includes: The signal acquisition module is used to acquire the electrical status parameters of the motor, wherein the operating voltage of the motor exceeds a preset value; The time-frequency feature extraction module is used to extract time-frequency features from the electrical state parameters and determine the first fault classification result of the motor based on the time-series evolution information of the extracted time-frequency features; The physical simulation analysis module is used to perform physical simulation analysis on the electrical state parameters in parallel during the process of determining the first fault classification result, so as to obtain the second fault classification result generated based on the physical simulation analysis. The target fault identification module is used to determine the target fault of the motor based on the cross-validation of the first fault classification result and the second fault classification result.

11. An electronic device, characterized in that, include: A memory and a processor are interconnected, the memory storing computer instructions, and the processor executing the computer instructions to perform the fault identification method for an electric motor as described in any one of claims 1 to 9.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing a computer to execute the fault identification method for an electric motor according to any one of claims 1 to 9.

13. A computer program product, characterized in that, It includes computer instructions for causing a computer to execute the fault identification method for an electric motor according to any one of claims 1 to 9.

Images