Fault detection method and device based on dynamic weighting, medium and equipment

Abstract

The embodiment of the invention discloses a fault detection method and system based on dynamic weighting, a medium and equipment. The method comprises the steps that power signals of a driving motor under different fault types are collected; constructing corresponding base models according to the power signals of the driving motor under different fault types; monitoring performance indexes of each base model on the feature vectors of different fault types in real time; determining a self-adaptive genetic factor according to the performance index; adjusting the model weight of each base model according to the adaptive genetic factors and the performance indexes; each base model outputs a fault prediction value through the current power signal, weighted average is carried out on the fault prediction value output by each base model according to the model weight, and a subset fusion result is obtained; and carrying out overall fusion on the subset fusion result to obtain a final fault detection result. Through the multi-level fusion process, the base model can identify the fault type more accurately, and the fault detection precision is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power fault detection, and in particular to a fault detection method, device, medium and equipment based on dynamic weighting. Background Art

[0002] In modern power systems, gas-insulated switchgear disconnectors play a vital role. Their main function is to achieve electrical isolation, thereby ensuring the safe operation of the power system. With the continuous development and complexity of power systems, the operating environment of gas-insulated switchgear disconnectors has become increasingly complex, and their fault detection has also faced more and more challenges.

[0003] Traditional fault detection methods often show obvious limitations when dealing with complex and changeable fault types and cannot adapt to changing working conditions and environments. These limitations are mainly reflected in the following aspects: First, the traditional single-model fault detection method shows limited generalization ability when dealing with multiple fault types. Second, the single model has poor robustness when dealing with complex working conditions and changing environments, making it difficult to achieve stable and reliable fault detection. In addition, traditional methods lack the ability to dynamically adjust and optimize model weights and cannot adapt to changes in fault characteristics in real time, resulting in limited detection accuracy. In addition, traditional ensemble learning methods lack flexibility in weight allocation and usually adopt static weight allocation strategies, which make it difficult to adapt to dynamic changes in fault characteristics in real time. Summary of the invention

[0004] Based on this, it is necessary to propose a fault detection method, device, medium and equipment based on dynamic weighting to address the above problems.

[0005] A fault detection method based on dynamic weighting, the method comprising:

[0006] Collect the power signal of the drive motor under different fault types.

[0007] A corresponding base model is constructed according to the power signal of the drive motor under the different fault types.

[0008] The performance indicators of each of the base models on the feature vectors of the different fault types are monitored in real time.

[0009] The adaptive genetic factor is determined according to the performance index.

[0010] The model weights of the respective base models are adjusted according to the adaptive genetic factors and the performance indicators.

[0011] Each base model outputs a fault prediction value through the current power signal, and the fault prediction value output by each base model is weighted averaged according to the model weight to obtain a subset fusion result.

[0012] The subset fusion results are integrated to obtain the final fault detection result.

[0013] The step of constructing a corresponding base model according to the power signal of the drive motor under different fault types specifically includes:

[0014] The power signal of the drive motor under the different fault types is preprocessed.

[0015] The time domain, frequency domain and time-frequency domain features of the preprocessed power signal are extracted, and the time domain, frequency domain and time-frequency domain features are used as feature vectors of different fault types.

[0016] A corresponding base model is constructed according to the feature vectors of the different fault types, and the base model is trained according to the feature vectors of the different fault types.

[0017] Wherein, determining the adaptive genetic factor according to the performance index specifically includes:

[0018] An initial performance threshold is set according to the performance index of each of the base models on the feature vector.

[0019] Determine the comparison between the performance index of each base model at the current moment and the initial performance threshold.

[0020] If the performance index of each of the base models at the current moment is greater than or equal to the initial performance threshold, an adaptive genetic factor is determined according to the performance index.

[0021] Wherein, if the performance index of each of the base models at the current moment is greater than or equal to the initial performance threshold, determining the adaptive genetic factor according to the performance index specifically includes:

[0022] according to Determine the adaptive genetic factors, where β i (t) is the adaptive forgetting factor of the i-th base model at time t, γ is the adjustment parameter, Performance i (t) is the performance index of the i-th base model at time t.

[0023] The step of adjusting the model weights of each base model according to the adaptive genetic factor and the performance index specifically includes:

[0024] according to Adjust the model weight of the base model, where w i (t+1) is the model weight of the i-th base model at time t+1, w i (t) is the model weight of the i-th base model at time t, N is a positive integer, Performance i (t) is the performance index of the i-th base model at time t, Performance j (t) is the performance index of the j-th base model at time t, w i (t) is the model weight of the i-th base model at time t.

[0025] The step of determining the comparison between the performance index of each base model at the current moment and the initial performance threshold value further specifically includes:

[0026] If the performance index of each base model at the current moment is less than the initial performance threshold, the weight of the base model below the initial performance threshold is reduced according to the performance index.

[0027] Each of the base models outputs a fault prediction value through the current power signal, and the fault prediction value output by each base model is weighted averaged according to the model weight to obtain a subset fusion result, which specifically includes:

[0028] according to Perform weighted averaging on the fault prediction values ​​output by each base model to obtain a subset fusion result, where Y k is the subset fusion result, S k is the base model, y i is the fault prediction value of the ith base model, w i is the model weight of the ith base model.

[0029] The overall fusion of the subset fusion results to obtain the final fault detection result specifically includes:

[0030] according to The subset fusion results are integrated to obtain the final fault detection result, where Y is the final fault detection result, Y k is the subset fusion result, w k is the model weight of the base model.

[0031] Wherein, after the subset fusion results are integrated to obtain the final fault detection result, the following steps are specifically included:

[0032] Acquire feature data after overall fusion of each base model.

[0033] A fault classification model is constructed according to the feature data and the feature vectors of different fault types.

[0034] The fault classification model determines the fault type through the current feature data.

[0035] A fault detection system based on dynamic weighting, the system comprising:

[0036] The power signal acquisition module is used to collect the power signal of the drive motor under different fault types.

[0037] The base model determination module is used to construct a corresponding base model according to the power signal of the drive motor under the different fault types.

[0038] The performance indicator monitoring module is used to monitor the performance indicators of each base model on the characteristic vectors of different fault types in real time.

[0039] The adaptive genetic factor determination module is used to determine the adaptive genetic factor according to the performance index.

[0040] The model weight adjustment module is used to adjust the model weight of each base model according to the adaptive genetic factor and the performance index.

[0041] The subset fusion result acquisition module is used for each base model to output a fault prediction value through the current power signal, and to perform weighted averaging on the fault prediction values ​​output by each base model according to the model weight to obtain the subset fusion result.

[0042] The final fault detection result acquisition module performs overall fusion on the subset fusion results to obtain the final fault detection result.

[0043] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor executes the steps of the method described above.

[0044] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the above method.

[0045] The embodiments of the present invention have the following beneficial effects:

[0046] The present invention selects multiple base models so that each base model can be good at different types of fault identification and detection, fully utilizes the advantages of different base models, and makes up for the shortcomings of a single model, thereby improving the accuracy of the overall prediction. When a base model performs poorly under specific circumstances, other base models can provide supplementary information to make the overall model more robust. Further, an adaptive genetic factor is introduced to dynamically adjust the weights of each base model according to the performance indicators of the base model under different fault types, so that the overall model can adapt to the real-time changing fault characteristics, and more weights are allocated to the base model that performs better under the current fault type and real-time data, thereby improving the overall accuracy and robustness of fault detection. Further, the fault prediction value of each base model is weighted averaged according to the model weight to obtain a subset fusion result, reduce the deviation of the single model prediction result, and make the subset fusion result more stable and reliable. Finally, the multiple subset fusion results are comprehensively evaluated through the overall fusion process, so as to obtain a more comprehensive and accurate final fault detection result. The present invention enables each base model to more accurately identify the fault type and improve the accuracy of fault detection through a multi-level fusion process. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0048] in:

[0049] Figure 1 A schematic diagram of a flow chart of an embodiment of a fault detection method based on dynamic weighting provided by the present invention;

[0050] Figure 2 A schematic diagram of a flow chart of another embodiment of a fault detection method based on dynamic weighting provided by the present invention;

[0051] Figure 3 A structural schematic diagram of an embodiment of a fault detection system based on dynamic weighting provided by the present invention;

[0052] Figure 4 A schematic structural diagram of an embodiment of the device provided by the present invention;

[0053] Figure 5 A schematic structural diagram of an embodiment of the medium provided by the present invention. DETAILED DESCRIPTION

[0054] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0055] like Figure 1 As shown, Figure 1 A schematic diagram of a flow chart of an embodiment of a fault detection method based on dynamic weighting provided by the present invention. A fault detection method based on dynamic weighting, the method comprising:

[0056] S101: Collecting power signals of the drive motor under different fault types.

[0057] For example, first collect the power signal of the drive motor under different fault types. Use high-precision current sensors, high-precision voltage sensors and data acquisition equipment. Specifically, the current sensor is used to measure the current change of the drive motor and should have the characteristics of high precision and fast response. The specific requirements are as follows:

[0058] Measurement range: The full operating current range of the drive motor should be covered to ensure that all key current changes can be captured.

[0059] Accuracy: The measurement accuracy of the sensor should be within ±0.1% to ensure the accuracy of the data.

[0060] Response time: The response time of the sensor should be fast enough to capture instantaneous changes in current, usually within milliseconds.

[0061] High-precision voltage sensors are used to measure the voltage changes of drive motors and also need to have the characteristics of high precision and fast response. The specific requirements are as follows:

[0062] Measuring range: Should cover the entire operating voltage range of the drive motor.

[0063] Accuracy: The measurement accuracy of the sensor should be within ±0.1%.

[0064] Response time: The response time of the sensor should be fast enough to capture instantaneous changes in voltage, usually within milliseconds.

[0065] The data acquisition device is responsible for receiving the output signal of the sensor and performing digital processing. The specific requirements are as follows:

[0066] Sampling rate: The sampling rate of the data acquisition device should be high enough to capture all key information during the operation of the drive motor. It is recommended to set the sampling rate above 10kHz.

[0067] Resolution: The resolution of the acquisition device should be 16 bits or higher to ensure that data details are not lost.

[0068] Storage capacity: The device should have sufficient large-capacity storage to ensure the integrity of long-term data collection.

[0069] S102: Constructing a corresponding base model according to the power signal of the drive motor under different fault types.

[0070] Exemplarily, when collecting data, the sampling frequency and sampling time window are set. To ensure that the details of the data are not lost, the sampling frequency is set above 10kHz. The sampling time window should cover the entire process of driving the motor from starting to stopping to ensure that complete data of the motor operation is collected. In the startup phase, the current and voltage changes when the motor starts are recorded, so as to record the power signal changes of the motor in the startup phase and capture possible startup faults; in the stable operation phase, the power signal of the motor in the stable operation phase is recorded to analyze the characteristics under normal working conditions; in the stop phase, the power signal changes when the motor stops are recorded to capture possible stop faults.

[0071] In order to ensure the quality of data acquisition, safeguards need to be taken, including equipment calibration, sensor installation and data verification. Use standard signal sources to calibrate high-precision current sensors, high-precision voltage sensors and data acquisition equipment to ensure that their output is accurate and reliable. Among them, the installation position and method of high-precision current sensors and high-precision voltage sensors have an important impact on the accuracy of data acquisition. High-precision current sensors and high-precision voltage sensors should be installed at the position closest to the drive motor to minimize interference and loss during signal transmission. And they need to be firmly fixed to avoid data errors caused by vibration or displacement during the acquisition process. Data verification is a preliminary inspection and verification of the collected data to ensure the integrity and consistency of the data. During the data acquisition process, the changes in data (power signals) are monitored in real time to detect and handle abnormal situations in a timely manner. After the acquisition is completed, the data (power signal) is fully checked to remove obvious noise and interference signals to ensure the quality of the data. Through the above equipment configuration, sampling methods and quality assurance measures, high-quality drive motor power signal data can be obtained.

[0072] Furthermore, the power signal is preprocessed, and the time domain, frequency domain and time-frequency domain features of the preprocessed power signal are extracted, and the time domain, frequency domain and time-frequency domain features are used as feature vectors of different fault types.

[0073] Furthermore, corresponding base models are constructed according to the feature vectors of different fault types, and the base models are trained according to the feature vectors of different fault types. By selecting a variety of different types of machine learning models as the model type of the base model, combining the feature vectors of different fault types to construct the corresponding base model, and training and evaluating these base models, the diversity and complementarity of the model can be ensured, thereby improving the accuracy and robustness of fault detection.

[0074] Specifically, in the model type selection stage of the base model, multiple machine learning models can be combined to take advantage of the advantages of each model and make up for the shortcomings of a single model. The model types of the selected base models include:

[0075] Decision tree: A decision tree is a classification and regression model based on a tree structure. It is easy to understand and explain and is suitable for processing data with obvious classification boundaries.

[0076] Support vector machine: Support vector machine is a classification and regression model based on the maximum margin principle. It can effectively process data in high-dimensional space and performs well with a small number of samples. It has the advantages of strong high-dimensional data processing capability, strong robustness, and the ability to handle nonlinear problems.

[0077] Random Forest: Random Forest is an integrated model based on multiple decision trees. It improves the accuracy and stability of predictions by voting or averaging the prediction results of multiple trees. It has the advantages of high accuracy, strong resistance to overfitting, and strong ability to process high-dimensional data.

[0078] Neural network: Neural network is a deep learning model based on simulating the neuron structure of the human brain, with powerful feature extraction and learning capabilities. It is characterized by its ability to handle complex nonlinear problems and is suitable for large-scale data. It has the advantages of strong feature extraction capabilities, strong nonlinear problem processing capabilities, and is suitable for big data.

[0079] Furthermore, after the base model is constructed, each base model is trained according to the feature vectors of different fault types. Specifically, when training the base model, the feature vectors of different fault types are divided into training sets, validation sets, and test sets to ensure that the training, validation, and testing processes of the base model are not affected by data leakage. During the training process, the base model is trained and validated multiple times through cross-validation to select the optimal model parameters to avoid overfitting and underfitting. The hyperparameters of the model are tuned through grid search, random search, and other methods to ensure the optimal performance of the base model.

[0080] S103: Monitor the performance indicators of each base model on the feature vectors of different fault types in real time.

[0081] Exemplarily, the performance indicators of each base model on the feature vectors of different fault types are determined through the historical and real-time performance (accuracy, etc.) of each base model on the training set and the validation set, and the optimal performance of the model in fault diagnosis is ensured through this performance indicator.

[0082] S104: Determine adaptive genetic factors according to the performance indicators.

[0083] Exemplarily, an initial performance threshold is set according to the performance index of each base model on the feature vector; the comparison between the performance index of each base model at the current moment and the initial performance threshold is determined; if the performance index of each base model at the current moment is greater than or equal to the initial performance threshold, an adaptive genetic factor is determined according to the performance index. Specifically, the adaptive genetic factor is determined according to the formula shown below:

[0084]

[0085] Among them, βi(t) is the adaptive forgetting factor of the i-th base model at time t, γ is the adjustment parameter, and Performancei(t) is the performance indicator of the i-th base model at time t.

[0086] S105: Adjust the model weights of each base model according to the adaptive genetic factors and the performance indicators.

[0087] Exemplarily, the model weights of each base model are dynamically adjusted according to the following formula:

[0088]

[0089] Among them, w i (t+1) is the model weight of the i-th base model at time t+1, w i (t) is the model weight of the i-th base model at time t, N is a positive integer, Performance i (t) is the performance index of the i-th base model at time t, Performance j (t) is the performance index of the jth base model at time t, and DPT is the initial performance threshold.

[0090] At the same time, if the performance index of each base model at the current moment is less than the initial performance threshold, the base model below the initial performance threshold is weighted according to the performance index, and the model weight of the base model after weight reduction is determined according to the following formula:

[0091] w i (t+1)=w i (t)·δPerformance i (t) <DPT;

[0092] Among them, w i (t+1) is the model weight of the i-th base model at time t+1, w i (t) is the model weight of the ith base model at time t, DPT is the initial performance threshold, and δ is the reduction coefficient, which ranges from 0<δ<1 and is used to reduce the weight of the poorly performing model.

[0093] S106: Each base model outputs a fault prediction value through the current power signal, and the fault prediction value output by each base model is weighted averaged according to the model weight to obtain a subset fusion result.

[0094] Exemplarily, each base model outputs a fault prediction value through the current power signal, and the fault prediction result of each base model is weighted averaged according to the model weight to obtain a subset fusion result, which is determined according to the following formula:

[0095]

[0096] Among them, Y k is the subset fusion result, S k is the base model, y i is the fault prediction value of the ith base model, w i is the model weight of the ith base model.

[0097] S107: Perform overall fusion on the subset fusion results to obtain a final fault detection result.

[0098] Exemplarily, the subset fusion results are integrated to obtain a final fault detection result, which is determined according to the following formula:

[0099]

[0100] Among them, Y is the final fault detection result, Y k is the subset fusion result, w k is the model weight of the base model.

[0101] It can be seen from the above description that the present invention selects multiple base models so that each base model can be good at different types of fault identification and detection, fully utilizes the advantages of different base models, and makes up for the shortcomings of a single model, thereby improving the accuracy of the overall prediction. When a base model performs poorly under certain circumstances, other base models can provide supplementary information to make the overall model more robust. Further, an adaptive genetic factor is introduced to dynamically adjust the weights of each base model according to the performance indicators of the base model under different fault types, so that the overall model can adapt to the real-time changing fault characteristics, and more weights are allocated to the base model that performs better under the current fault type and real-time data, thereby improving the overall accuracy and robustness of fault detection. Further, the fault prediction value of each base model is weighted averaged according to the model weight to obtain the subset fusion result, reduce the deviation of the single model prediction result, and make the subset fusion result more stable and reliable. Finally, the multiple subset fusion results are comprehensively evaluated through the overall fusion process to obtain a more comprehensive and accurate final fault detection result. Through the multi-level fusion process, the present invention enables each base model to more accurately identify the fault type and improve the accuracy of fault detection.

[0102] like Figure 2 As shown, Figure 2 A flow chart of another embodiment of a fault detection method based on dynamic weighting provided by the present invention. A fault detection method based on dynamic weighting, the method comprising:

[0103] S201: Collecting power signals of the drive motor under different fault types.

[0104] It should be noted that step S201 Figure 1 This has been discussed in detail in the implementation scenario shown and will not be repeated here.

[0105] S202: Preprocessing the power signal of the drive motor under different fault types.

[0106] For example, the power signal is normalized to ensure that the power signal is within a uniform scale range for subsequent analysis. Through normalization, the influence of different dimensions on feature extraction and fault detection can be eliminated. Specifically, the Z-score normalization method is used, as shown in the following formula:

[0107]

[0108] Wherein, P′ is the normalized power signal, P is the power signal, μ is the mean of the power signal, and σ is the standard deviation of the signal.

[0109] S203: extracting the time domain, frequency domain and time-frequency domain features of the preprocessed power signal, and using the time domain, frequency domain and time-frequency domain features as feature vectors of different fault types.

[0110] Exemplarily, by extracting the time domain, frequency domain and time-frequency domain features of the preprocessed power signal, the characteristics of the signal are fully reflected, providing an important basis for subsequent fault classification and detection.

[0111] Specifically, the time domain features mainly reflect the basic statistical characteristics of the signal and help capture the changing laws of the power signal. The time domain features include mean, standard deviation and peak value.

[0112] The mean value reflects the average level of the power signal and is determined according to the following formula:

[0113]

[0114] Wherein, Mean is the mean value of the power signal after preprocessing, N is the number of sample points of the power signal after preprocessing, and P′(i) is the preprocessed power signal of the i-th sample point.

[0115] The standard deviation reflects the degree of dispersion of the signal and is determined according to the following formula:

[0116]

[0117] Wherein, StdDev is the standard deviation of the preprocessed power signal, Mean is the mean of the preprocessed power signal, N is the number of sample points of the preprocessed power signal, and P′(i) is the preprocessed power signal of the i-th sample point.

[0118] The peak value reflects the maximum amplitude of the signal and is determined according to the following formula:

[0119] Peak = max(P′);

[0120] Wherein, Peak is the peak value of the preprocessed power signal, and P′(i) is the preprocessed power signal of the i-th sample point.

[0121] Frequency domain features are extracted through Fourier transform, which can reveal the frequency composition of the signal and help identify periodic fault characteristics. Frequency domain features include spectrum energy and main frequency components. Spectral energy reflects the energy distribution of the signal in the frequency domain and is determined according to the following formula:

[0122]

[0123] Wherein, E is the spectrum energy, F(P′)(k) is the kth Fourier transform coefficient of the preprocessed power signal, and N is the power spectrum density.

[0124] The main frequency component reflects the frequency component with the strongest energy in the signal and is determined according to the following formula:

[0125] f d = argmax f |F(P′)(f)|;

[0126] Among them, f d is the main frequency component, and F(P′)(f) is the Fourier transform amplitude of the preprocessed power signal at frequency f.

[0127] The time-frequency domain features are extracted through time-frequency transform, which can capture the changes of the signal in time and frequency, and help identify instantaneous fault characteristics. The time-frequency domain features include wavelet coefficients and instantaneous frequency. The wavelet coefficients are obtained through wavelet transform, which reflects the frequency characteristics of the power signal at different scales and time positions, and are determined according to the following formula:

[0128]

[0129] Among them, W ψ (a, b) are wavelet coefficients, ψ * is the complex conjugate of the mother wavelet function, a is the scale factor, b is the time shift factor, and P′(t) is the preprocessed power signal at time t.

[0130] The instantaneous frequency reflects the frequency component of the power signal at a specific point in time and is determined according to the following formula:

[0131]

[0132] Among them, f i (t) is the instantaneous frequency, and φ(t) is the phase of the preprocessed power signal.

[0133] S204: construct corresponding base models according to the feature vectors of different fault types, and train the base models according to the feature vectors of different fault types.

[0134] S205: Monitor the performance indicators of each base model on the feature vectors of different fault types in real time.

[0135] S206: Setting an initial performance threshold according to the performance indicators of each base model on the feature vector.

[0136] S207: Determine the comparison between the performance index of each base model at the current moment and the initial performance threshold.

[0137] S2071: If the performance index of each base model at the current moment is greater than or equal to the initial performance threshold, the adaptive genetic factor is determined according to the performance index.

[0138] S208: Adjusting the model weights of each base model according to the adaptive genetic factors and the performance index.

[0139] S2072: If the performance index of each base model at the current moment is less than the initial performance threshold, the weight of the base model below the initial performance threshold is reduced according to the performance index.

[0140] Exemplarily, the model weight of the base model is determined according to the following formula:

[0141]

[0142] Among them, w i (t+1) is the model weight of the i-th base model at time t+1, w i (t) is the model weight of the i-th base model at time t, N is a positive integer, Performance i (t) is the performance index of the i-th base model at time t, Performance j (t) is the performance index of the jth base model at time t, DPT is the initial performance threshold, δ is the reduction coefficient, which takes the value between 0<δ<1 and is used to reduce the weight of the poorly performing model.

[0143] S209: Each base model outputs a fault prediction value through the current power signal, and the fault prediction value output by each base model is weighted averaged according to the model weight to obtain a subset fusion result.

[0144] S210: Perform overall fusion on the subset fusion results to obtain a final fault detection result.

[0145] It should be noted that steps S209-S210 are Figure 1 This has been discussed in detail in the implementation scenario shown and will not be repeated here.

[0146] S211: Obtain feature data after overall fusion of each base model.

[0147] S212: Construct a fault classification model based on the feature data and feature vectors of different fault types.

[0148] Exemplarily, a machine learning model suitable for fault classification is selected as the model type of the fault classification model, such as decision tree, support vector machine, random forest, neural network, etc., and a fault classification model is constructed by combining feature data and feature vectors of different fault types. The feature data and feature vectors of different fault types are divided into a training set and a test set, and the training set is used to train the fault classification model, and the model parameters are optimized to ensure its best performance on the training set.

[0149] S213: The fault classification model determines the fault type through the current feature data.

[0150] like Figure 3 As shown, Figure 3 A structural diagram of an embodiment of a fault detection system based on dynamic weighting provided by the present invention. A fault detection system based on dynamic weighting 10, the system comprises:

[0151] The power signal acquisition module 11 is used to acquire the power signal of the drive motor under different fault types.

[0152] In the base model building module 12, the user builds a corresponding base model according to the power signal of the drive motor under different fault types.

[0153] The performance indicator monitoring module 13 is used to monitor the performance indicators of each base model on the feature vectors of different fault types in real time.

[0154] The adaptive genetic factor determination module 14 is used to determine the adaptive genetic factor according to the performance index.

[0155] The model weight adjustment module 15 is used to adjust the model weights of each base model according to the adaptive genetic factors and performance indicators.

[0156] The subset fusion result acquisition module 16 is used for each base model to output a fault prediction value through the current power signal, and to perform weighted averaging on the fault prediction values ​​output by each base model according to the model weight to obtain the subset fusion result.

[0157] The final fault detection result acquisition module 17 is used to perform overall fusion of the subset fusion results to obtain the final fault detection result.

[0158] Exemplarily, in the power signal acquisition module 11, the power signal of the drive motor under different fault types is collected. In the base model construction module 12, the power signal of the drive motor under different fault types is preprocessed; the time domain, frequency domain and time-frequency domain features of the preprocessed power signal are extracted, and the time domain, frequency domain and time-frequency domain features are used as feature vectors of different fault types; the corresponding base model is constructed according to the feature vectors of different fault types, and the base model is trained according to the feature vectors of different fault types. In the performance indicator monitoring module 13, the performance indicators of each base model on the feature vectors of the different fault types are monitored in real time. In the adaptive genetic factor determination module 14, the initial performance threshold is set according to the performance indicators of each base model on the feature vector; the comparison between the performance indicators of each base model at the current moment and the initial performance threshold is determined; if the performance indicators of each base model at the current moment are greater than or equal to the initial performance threshold, the adaptive genetic factor is determined according to the performance indicator. In the model weight adjustment module 15, the model weights of each base model are adjusted according to the adaptive genetic factor and the performance indicator. Furthermore, in the subset fusion result acquisition module 16, each base model outputs a fault prediction value through the current power signal, and the fault prediction value output by each base model is weighted averaged according to the model weight to obtain the subset fusion result. Finally, in the final fault detection result acquisition module 17, the subset fusion results are integrated to obtain the final fault detection result.

[0159] like Figure 4 As shown, Figure 4 The device 20 includes a memory 21 and a processor 22. The memory 21 stores a computer program, and the processor 22 executes the computer program when working to implement the following. Figure 1 and Figure 2 The method shown.

[0160] The specific technical details of a dynamic weighted fault detection method implemented when the device 20 executes a computer program have been discussed in detail in the aforementioned method steps, so they will not be repeated here.

[0161] like Figure 5 As shown, Figure 5 The structure diagram of an embodiment of the medium provided by the present invention is shown in FIG. The medium 30 stores at least one computer program 31, and the computer program 31 is executed by the processor 22 to implement the following Figure 1 and Figure 2 In one embodiment, the medium 30 may be a storage chip, a hard disk, a mobile hard disk, a USB flash drive, an optical disk, or other readable and writable storage tools, or a server, etc.

[0162] The above describes specific embodiments of the present specification, and other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily have to be performed in the specific order or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0163] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer-readable storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0164] The apparatus, device, non-volatile computer-readable storage medium and method provided in the embodiments of this specification correspond to each other, and therefore, the apparatus, device, and non-volatile computer storage medium also have similar beneficial technical effects as the corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the corresponding apparatus, device, and non-volatile computer storage medium will not be repeated here.

[0165] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0166] For the convenience of description, the above device is described by being divided into various units according to their functions and described separately. Of course, when implementing this specification, the functions of each unit can be implemented in one or more software and / or hardware. It should be understood by those skilled in the art that this specification embodiment can be provided as a method, system, or computer program product. Therefore, this specification embodiment can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, this specification embodiment can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0167] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0168] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0169] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0170] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0171] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0172] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0173] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0174] This specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0175] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0176] The above disclosure is only the preferred embodiment of the present invention, which certainly cannot be used to limit the scope of the present invention. Therefore, equivalent changes made according to the claims of the present invention are still within the scope of the present invention.

Claims

1. A fault detection method based on dynamic weighting, characterized in that: The method comprises: Collect the power signal of the drive motor under different fault types; Constructing a corresponding base model according to the power signal of the driving motor under the different fault types; Real-time monitoring of the performance indicators of each of the base models on the feature vectors of the different fault types; determining adaptive genetic factors according to the performance indicators; According to the adaptive genetic factors and the performance indicators, adjusting the model weights of the base models; Each of the base models outputs a fault prediction value through the current power signal, and the fault prediction values ​​output by each of the base models are weighted averaged according to the model weights to obtain a subset fusion result; The subset fusion results are integrated to obtain the final fault detection result.

2. A fault detection method based on dynamic weighting according to claim 1, characterized in that: The step of constructing a corresponding base model according to the power signal of the driving motor under the different fault types specifically includes: Preprocessing the power signal of the driving motor under the different fault types; Extracting the time domain, frequency domain and time-frequency domain features of the preprocessed power signal, and using the time domain, frequency domain and time-frequency domain features as feature vectors of different fault types; A corresponding base model is constructed according to the feature vectors of the different fault types, and the base model is trained according to the feature vectors of the different fault types.

3. A fault detection method based on dynamic weighting according to claim 2, characterized in that: Determining the adaptive genetic factor according to the performance index specifically includes: Setting an initial performance threshold according to the performance indicators of each of the base models on the feature vector; Determine a comparison between the performance index of each base model at the current moment and the initial performance threshold; If the performance index of each of the base models at the current moment is greater than or equal to the initial performance threshold, an adaptive genetic factor is determined according to the performance index.

4. A fault detection method based on dynamic weighting according to claim 3, characterized in that: If the performance index of each base model at the current moment is greater than or equal to the initial performance threshold, determining the adaptive genetic factor according to the performance index specifically includes: according to Determine the adaptive genetic factors, where β i (t) is the adaptive forgetting factor of the i-th base model at time t, γ is the adjustment parameter, Performance i (t) is the performance index of the i-th base model at time t.

5. A fault detection method based on dynamic weighting according to claim 4, characterized in that: The step of adjusting the model weights of each base model according to the adaptive genetic factor and the performance index specifically includes: according to β i (t) adjusting the model weight of the base model, where w i (t+1) is the model weight of the i-th base model at time t+1, w i (t) is the model weight of the i-th base model at time t, N is a positive integer, Performance i (t) is the performance index of the i-th base model at time t, Performance j (t) is the performance index of the jth base model at time t.

6. A fault detection method based on dynamic weighting according to claim 3, characterized in that: The determining of the comparison between the performance index of each base model at the current moment and the initial performance threshold value further specifically includes: If the performance index of each base model at the current moment is less than the initial performance threshold, the weight of the base model below the initial performance threshold is reduced according to the performance index.

7. A fault detection method based on dynamic weighting according to claim 6, characterized in that: Each of the base models outputs a fault prediction value through the current power signal, and weighted averages the fault prediction values ​​output by each base model according to the model weight to obtain a subset fusion result, specifically including: according to Perform weighted averaging on the fault prediction values ​​output by each base model to obtain a subset fusion result, where Y k is the subset fusion result, S k is the base model, y i is the fault prediction value of the ith base model, w i is the model weight of the ith base model.

8. A fault detection method based on dynamic weighting according to claim 7, characterized in that: The overall fusion of the subset fusion results to obtain the final fault detection result specifically includes: according to The subset fusion results are integrated to obtain the final fault detection result, where Y is the final fault detection result, Y k is the subset fusion result, w k is the model weight of the base model.

9. A fault detection method based on dynamic weighting according to claim 8, characterized in that: After the subset fusion results are integrated to obtain the final fault detection result, the method further specifically includes: Acquire feature data after overall fusion of each of the base models; Building a fault classification model according to the feature data and the feature vectors of different fault types; The fault classification model determines the fault type through the current feature data.

10. A fault detection system based on dynamic weighting, characterized in that: The system comprises: A power signal acquisition module is used to collect the power signal of the drive motor under different fault types; A base model determination module, used to construct a corresponding base model according to the power signal of the drive motor under the different fault types; A performance indicator monitoring module, used for real-time monitoring of the performance indicators of each base model on the characteristic vectors of different fault types; An adaptive genetic factor determination module, used to determine the adaptive genetic factor according to the performance index; A model weight adjustment module, used for adjusting the model weight of each base model according to the adaptive genetic factor and the performance index; A subset fusion result acquisition module, used for each base model to output a fault prediction value through the current power signal, and to perform weighted averaging on the fault prediction values ​​output by each base model according to the model weight to obtain a subset fusion result; The final fault detection result acquisition module performs overall fusion on the subset fusion results to obtain the final fault detection result.

11. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 9.

12. A computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method according to any one of claims 1 to 9.

Images