A copper bar mechanical fault diagnosis method and system based on multi-dimensional parameter analysis

Abstract

The application discloses a copper bar mechanical fault diagnosis method and system based on multi-dimensional parameter analysis, relates to the technical field of data monitoring and fault diagnosis, and realizes real-time acquisition and recording of multi-dimensional parameters of the copper bar under the current working state under the condition of uninterrupted power supply, including contact resistance, temperature field distribution, high-frequency eddy current impedance, and load current and vibration data; based on time-frequency domain analysis, preliminary dynamic characteristics are extracted from the multi-dimensional parameters, and through multi-parameter correlation fusion analysis, correlation characteristics among the parameters are extracted, which jointly constitute a joint characteristic vector for copper bar fault diagnosis; the real-time acquired copper bar joint characteristic vector is input into a pre-trained fault diagnosis model, a hierarchical diagnosis mechanism is established based on a multi-task learning framework, and predicted copper bar mechanical fault types, fault severity and health warning information are output. The application solves the problem of resource waste caused by shutdown maintenance in the traditional detection method, and improves the efficiency and precision of copper bar fault diagnosis.

Description

Technical Field

[0001] This application relates to the field of data monitoring and fault diagnosis technology, and in particular to a method and system for diagnosing mechanical faults in copper busbars based on multidimensional parameter analysis. Background Technology

[0002] In systems such as new energy batteries, power transmission and transformation, and rail transit, copper busbars are critical high-current connection components. They are subjected to electrodynamic forces, thermal stress, and mechanical vibrations over long periods of time, making them prone to mechanical failures, including loose connections, fatigue cracks, plastic deformation, and stress corrosion cracking. These failures directly manifest as deterioration in electrical connection performance and may ultimately lead to overheating, melting, or even fire. Therefore, it is of great significance to regularly inspect the condition of copper busbars and establish a comprehensive fault detection and health management mechanism. Timely handling of early failures can prevent irreversible and serious damage, thereby extending the service life of copper busbars and their connected equipment and ensuring the reliability of the overall power system.

[0003] Traditional copper busbar maintenance methods rely on manual inspections or infrared thermography, requiring periodic shutdowns, disassembly, and checks, which consumes a lot of manpower and resources. They may also lead to problems of "over-maintenance" and "under-maintenance," and cannot detect the mechanical status in real time, making it difficult to predict sudden failures. There is an urgent need for a non-invasive online detection method that can indirectly and intelligently diagnose mechanical failures of copper busbars by monitoring their own parameters, thereby improving the efficiency of fault diagnosis. Summary of the Invention

[0004] To address the technical problems of the prior art, this application provides a method and system for diagnosing mechanical faults in copper busbars based on multidimensional parameter analysis. By monitoring and analyzing the multidimensional parameters of the copper busbar under specific excitation, it enables early identification, severity assessment, and warning of mechanical faults in the copper busbar, thereby improving the accuracy and efficiency of copper busbar fault diagnosis and ensuring the stable operation of the power system.

[0005] This application provides a method for diagnosing mechanical faults in copper busbars based on multidimensional parameter analysis, including:

[0006] Step S10: Apply a specific excitation to the copper busbar to be tested, and collect and record multi-dimensional parameters of the copper busbar in the current working state in real time under the condition of continuous power supply, including contact resistance, temperature field distribution, high-frequency eddy current impedance, load current and vibration data. Step S20: Preprocess each independent copper busbar parameter to form time series data within a set time period. Extract primary dynamic features based on time-frequency domain analysis, and extract correlation features between parameters through multi-parameter correlation fusion analysis. The two features together constitute a joint feature vector for copper busbar fault diagnosis. Step S30: Input the real-time acquired joint feature vector of the copper busbar into the pre-trained fault diagnosis model. Based on the multi-task learning framework, establish a hierarchical diagnosis mechanism, output the predicted mechanical fault type, fault severity and health warning information of the copper busbar, and complete the fault diagnosis report.

[0007] Furthermore, in the process of acquiring multi-dimensional parameters of the copper busbar, two excitation methods are adopted: high-frequency eddy current excitation and DC micro-current superposition. All sensors are installed in a non-contact manner to ensure the safety of the main circuit where the copper busbar is located. The contact resistance of copper busbars is measured using the four-wire DC superposition method, including the following steps: On the current path of the copper busbar connection section to be tested, determine two voltage measurement points and two current excitation points; A pair of voltage terminals are installed on the inner side of the copper busbar surface, and a current terminal capable of carrying the excitation current is installed on the outer side of the voltage terminals along the current direction. All terminal installations do not damage the original structure of the copper busbar and do not disconnect the main circuit. A DC test current with a frequency lower than 1Hz is injected into the current terminal of the copper busbar by a DC constant current source. The current value is far more than 1% of the rated load current and has no effect on the working status of the copper busbar. Two signals are acquired simultaneously: the total voltage drop across the copper busbar voltage terminals, i.e., the voltage value generated by the combined effect of the load current and the test current, and the load current value of the copper busbar is obtained through a current transformer. Using the test current as a reference signal, the total voltage drop is digitally synchronously demodulated to separate the DC voltage signal that is in phase and frequency with the test current, and the contact resistance and resistance change rate are calculated in real time.

[0008] Temperature field distribution data of the copper busbar was acquired using a combination of infrared thermal imaging and fixed-point temperature measurement, including: The online infrared thermal imager was mounted on a stable bracket 1 meter away from the copper busbar under test to ensure coverage of the entire test area. The sampling frequency was set to automatically capture thermal images of the copper busbar surface. The highest temperature, average temperature, standard deviation of temperature distribution, and temperature matrix of the test area on the copper busbar surface are automatically extracted from each frame of the thermal image to generate isotherms of the temperature distribution on the copper busbar surface. Based on the temperature matrix, thermocouples are installed at the hottest point of the copper busbar infrared temperature measurement to simultaneously collect the fixed-point temperature and the ambient temperature, and calculate the temperature rise data and temperature gradient parameters.

[0009] The high-frequency eddy current impedance of copper busbars is acquired using the swept-frequency eddy current method, and the following detailed steps are included: The eddy current probe array is mounted above the copper busbar area to be measured using a non-metallic bracket, maintaining a constant distance, and the eddy current meter is set to the frequency sweep mode in the selected frequency band. The eddy current meter emits an excitation signal, which is transmitted to the probe coil via a coaxial cable. It simultaneously receives the complex voltage signal induced by the probe. The instrument internally compares the emitted and received signals to calculate the complex impedance. Based on the real-time acquired eddy current impedance data of the copper busbar, the differential impedance and trajectory characteristics are calculated, and the impedance amplitude and phase change at specific frequency points are recorded.

[0010] The load current of the copper busbar is obtained directly from the current transformer, and the vibration data is obtained in real time through a triaxial accelerometer installed on the copper busbar support. The raw waveform data is obtained and the effective value of the current and the effective value of the vibration acceleration are calculated simultaneously.

[0011] Furthermore, the preprocessing steps for the multidimensional parameters of the copper busbar include timestamp alignment, integrity checking, outlier detection and handling, and noise filtering. Set the length of the time window to be processed, resample the multi-dimensional parameters of the copper busbar to a uniform sampling frequency, and align the timestamps of all data through the PTP clock to form the time series of the multi-dimensional parameters of the copper busbar. The continuity of timestamps for each parameter in the sequence is checked, missing data caused by communication interruption is marked, and forward imputation, linear interpolation and regression imputation based on relevant parameters are used to adaptively impute missing data with missing time periods of less than 5 seconds according to the specific degree of data missingness. For each parameter sequence, calculate the mean of the data within the set sliding window. and standard deviation Remove more than 3 Data points; based on physical constraints, set hard thresholds to remove copper busbar temperature field distribution data below ambient temperature, delete negative resistance values, and eliminate data that does not conform to physical laws; For the collected copper busbar contact resistance data, adaptive filtering is performed based on the Kalman filter and combined with the copper busbar load current variation model. For the copper busbar vibration data, wavelet threshold denoising is used to retain the 1-5kHz frequency band of the fault feature set while suppressing high-frequency noise. Moving average filtering is applied to various temperature data to smooth data fluctuations.

[0012] Furthermore, time-frequency domain analysis was performed on each independent copper busbar parameter to obtain primary dynamic characteristics: For contact resistance data, the ratio of the standard deviation of the resistance change rate to the mean is calculated to obtain the fluctuation coefficient, which reflects the relative fluctuation intensity of the copper busbar contact resistance and is a key feature for identifying whether the copper busbar has loose contact. Simultaneously, the intensity ratio of the trend term of the resistance is calculated, and the trend term is separated from the time series of the copper busbar contact resistance by using the moving average method. The proportion of its energy to the total energy of the series is calculated to indicate the severity of mechanical failure of the copper busbar. For the temperature field distribution data, the maximum normalized temperature gradient is calculated. The spatial gradient amplitude is obtained from the thermal image matrix of the copper busbar. The maximum value is found and normalized with the ambient temperature to reflect whether the copper busbar has cracks or local ablation faults. Simultaneously, the spatial consistency of temperature is calculated. The thermal image is smoothed with a Gaussian filter. The standard deviation of the difference between the original thermal image and the smoothed thermal image is calculated. Dividing the result by the standard deviation of the original thermal image identifies abnormal points in the temperature field that do not conform to the normal thermal conduction law, which are used to locate the fault points of the copper busbar. Principal component analysis was performed on the real part and distribution of the complex impedance spectrum for eddy current impedance data. The ratio of the eigenvalues ​​of the first and second principal components and the angle of the first principal component vector were calculated to obtain the trajectory ellipticity and tilt angle, which were used to determine the type and severity of cracking faults in the copper busbar. Three frequency points were randomly selected from the frequency band set when the eddy current impedance data was collected, and the corresponding eddy current phase angles were obtained. The standard deviation was calculated and normalized to distinguish between copper busbar faults and normal homogeneous material changes. For vibration data, Hilbert transform is performed on the vibration signal to extract the envelope, and the power spectral density of the envelope is calculated. The maximum ratio of the envelope power spectrum to its effective value is taken as the envelope peak factor, which is a key feature for identifying bolt loosening faults in copper busbars.

[0013] Furthermore, the fault diagnosis model is based on a multi-task learning framework, which separately handles the regression task of predicting the joint feature vector of the copper busbar and the classification task of mechanical faults of the copper busbar. The model consists of two parallel branches: the upper branch is used for early warning of copper busbar health, and the lower branch is used for early mechanical fault diagnosis of copper busbars. The upper branch consists of a multi-scale processing module, a temporal analysis module, and a prediction head, which updates the network parameters of each part during the training of the fault diagnosis model. The lower branch focuses on feature extraction, extracts fault representations from the joint feature vector through convolutional layers and a multiplication-based feature enhancement module, and outputs fault diagnosis results through a mechanical fault type classifier and a fault severity determiner with fixed parameters.

[0014] This application also provides a copper busbar mechanical fault diagnosis system based on multidimensional parameter analysis, including: Multi-dimensional parameter acquisition module: used to apply specific excitation to the copper busbar under test, and to collect and record multi-dimensional parameters of the copper busbar in the current working state in real time without power interruption, including contact resistance, temperature field distribution, high-frequency eddy current impedance, as well as load current and vibration data. Joint feature vector acquisition module: used to preprocess each independent copper busbar parameter to form time series data within a set time period, extract primary dynamic features based on time-frequency domain analysis, and extract correlation features between parameters through multi-parameter correlation fusion analysis. The two types of features together constitute a joint feature vector for copper busbar fault diagnosis. Fault diagnosis result output module: It is used to input the joint feature vector of copper busbar acquired in real time into the pre-trained fault diagnosis model, establish a hierarchical diagnosis mechanism based on the multi-task learning framework, output the predicted mechanical fault type, fault severity and health warning information of copper busbar, and complete the fault diagnosis report.

[0015] This application also proposes a copper busbar mechanical fault diagnosis device based on multidimensional parameter analysis. The device includes: a memory, a processor, and programs such as a copper busbar mechanical fault diagnosis algorithm based on multidimensional parameter analysis stored in the memory and executable on the processor. The copper busbar mechanical fault diagnosis algorithm based on multidimensional parameter analysis and other programs are steps for implementing the copper busbar mechanical fault diagnosis method based on multidimensional parameter analysis as described above.

[0016] This application also provides a computer program product, which includes programs such as a copper busbar mechanical fault diagnosis algorithm based on multidimensional parameter analysis. When the copper busbar mechanical fault diagnosis algorithm based on multidimensional parameters is executed by a processor, it implements the copper busbar mechanical fault diagnosis method based on multidimensional parameter analysis as described above.

[0017] This application discloses the following technical effects: This application provides a method and system for diagnosing mechanical faults in copper busbars based on multi-dimensional parameter analysis. Through the fusion sensing of electrical, thermal, and magnetic parameters and intelligent analysis based on a deep learning model, it transforms the assessment of copper busbar mechanical conditions from "late-stage disassembly and inspection" to "early intelligent warning," improving the process of copper busbar fault prediction and health management and preventing irreversible damage. The proposed method overcomes the difficulties of traditional detection methods that require power disconnection and disassembly, enabling non-invasive fault diagnosis and monitoring of copper busbar mechanical conditions during normal equipment operation, and timely detection of potential risks. This method overcomes the limitations of single measurement by combining the synchronous acquisition and correlation analysis of multiple parameters such as resistance, temperature field, eddy current impedance, and vibration data to comprehensively and deeply reveal the mechanistic characteristics of mechanical faults, improving the accuracy and reliability of fault diagnosis. Based on data-driven deep learning models, this method achieves automatic fault diagnosis, automatically outputs the fault type and severity, predicts the timing of future fault occurrence, and provides health warnings, enabling predictive maintenance of copper busbars and improving the efficiency of copper busbar fault diagnosis and maintenance. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0019] Figure 1 This is a flowchart illustrating a method for diagnosing mechanical faults in copper busbars based on multidimensional parameter analysis, provided in an embodiment of this application.

[0020] Figure 2 This is a schematic diagram of the structure of the fault diagnosis model provided in the embodiments of this application.

[0021] Figure 3 This is a schematic diagram of a copper busbar mechanical fault diagnosis system based on multidimensional parameter analysis, provided in an embodiment of this application. Detailed Implementation

[0022] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application.

[0023] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description of this application will be provided in conjunction with the accompanying drawings. The described embodiments should not be considered as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] In the following description, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only.

[0025] Example 1: This application provides a method for diagnosing mechanical faults in copper busbars based on multidimensional parameter analysis, such as... Figure 1 As shown, the method includes: Step S10: Apply a specific excitation to the copper busbar to be tested, and collect and record multi-dimensional parameters of the copper busbar in the current working state in real time under the condition of continuous power supply, including contact resistance, temperature field distribution, high-frequency eddy current impedance, load current and vibration data.

[0026] In the process of acquiring multi-dimensional parameters of the copper busbar in this embodiment, two excitation methods are adopted: high-frequency eddy current excitation and DC micro-current superposition. All sensors are installed in a non-contact manner to ensure the safety of the main circuit where the copper busbar is located. The contact resistance of copper busbars is measured using the four-wire DC superposition method, including the following steps: On the current path of the copper busbar connection section to be tested, determine two voltage measurement points and two current excitation points; A pair of voltage terminals are installed on the inner side of the copper busbar surface, and a current terminal capable of carrying the excitation current is installed on the outer side of the voltage terminals along the current direction. All terminal installations do not damage the original structure of the copper busbar and do not disconnect the main circuit. A DC test current with a frequency lower than 1Hz is injected into the current terminal of the copper busbar by a DC constant current source. The current value is far more than 1% of the rated load current and has no effect on the working status of the copper busbar. Two signals are acquired simultaneously: the total voltage drop across the copper busbar voltage terminals, i.e., the voltage value generated by the combined effect of the load current and the test current, and the load current value of the copper busbar is obtained through a current transformer. Using the test current as a reference signal, the total voltage drop is digitally synchronously demodulated to separate the DC voltage signal that is in phase and frequency with the test current, and the contact resistance and resistance change rate are calculated in real time.

[0027] Temperature field distribution data of the copper busbar was acquired using a combination of infrared thermal imaging and fixed-point temperature measurement, including: The online infrared thermal imager was mounted on a stable bracket 1 meter away from the copper busbar under test to ensure coverage of the entire test area. The sampling frequency was set to automatically capture thermal images of the copper busbar surface. The highest temperature, average temperature, standard deviation of temperature distribution, and temperature matrix of the test area on the copper busbar surface are automatically extracted from each frame of the thermal image to generate isotherms of the temperature distribution on the copper busbar surface. Based on the temperature matrix, thermocouples are installed at the hottest point of the copper busbar infrared temperature measurement to simultaneously collect the fixed-point temperature and the ambient temperature, and calculate the temperature rise data and temperature gradient parameters.

[0028] The high-frequency eddy current impedance of copper busbars is acquired using the swept-frequency eddy current method, and the following detailed steps are included: The eddy current probe array is mounted above the copper busbar area to be measured using a non-metallic bracket, maintaining a constant distance, and the eddy current meter is set to the frequency sweep mode in the selected frequency band. The eddy current meter emits an excitation signal, which is transmitted to the probe coil via a coaxial cable. It simultaneously receives the complex voltage signal induced by the probe. The instrument internally compares the emitted and received signals to calculate the complex impedance. Based on the real-time acquired eddy current impedance data of the copper busbar, the differential impedance and trajectory characteristics are calculated, and the impedance amplitude and phase change at specific frequency points are recorded.

[0029] The load current of the copper busbar is obtained directly from the current transformer, and the vibration data is obtained in real time through a triaxial accelerometer installed on the copper busbar support. The raw waveform data is obtained and the effective value of the current and the effective value of the vibration acceleration are calculated simultaneously.

[0030] Step S20: Preprocess each independent copper busbar parameter to form time series data within a set time period. Extract primary dynamic features based on time-frequency domain analysis, and extract correlation features between parameters through multi-parameter correlation fusion analysis. The two types of features together constitute a joint feature vector for copper busbar fault diagnosis.

[0031] In this embodiment, the steps for preprocessing the multidimensional parameters of the copper busbar include timestamp alignment, integrity check, outlier detection and processing, and noise filtering. Set the length of the time window to be processed, resample the multi-dimensional parameters of the copper busbar to a uniform sampling frequency, and align the timestamps of all data through the PTP clock to form the time series of the multi-dimensional parameters of the copper busbar. The continuity of timestamps for each parameter in the sequence is checked, missing data caused by communication interruption is marked, and forward imputation, linear interpolation and regression imputation based on relevant parameters are used to adaptively impute missing data with missing time periods of less than 5 seconds according to the specific degree of data missingness. For each parameter sequence, calculate the mean of the data within the set sliding window. and standard deviation Remove more than 3 Data points; based on physical constraints, set hard thresholds to remove copper busbar temperature field distribution data below ambient temperature, delete negative resistance values, and eliminate data that does not conform to physical laws; For the collected copper busbar contact resistance data, adaptive filtering is performed based on the Kalman filter and combined with the copper busbar load current variation model; for the copper busbar vibration data, wavelet threshold denoising is used to retain the 1-5kHz frequency band of the fault feature set while suppressing high-frequency noise, and moving average filtering is applied to various temperature data to smooth data fluctuations. Time-frequency domain analysis was performed on each independent copper busbar parameter to obtain primary dynamic characteristics: For contact resistance data, the ratio of the standard deviation of the resistance change rate to the mean is calculated to obtain the fluctuation coefficient, which reflects the relative fluctuation intensity of the copper busbar contact resistance and is a key feature for identifying whether the copper busbar has loose contact. Simultaneously, the intensity ratio of the trend term of the resistance is calculated, and the trend term is separated from the time series of the copper busbar contact resistance by using the moving average method. The proportion of its energy to the total energy of the series is calculated to indicate the severity of mechanical failure of the copper busbar. For the temperature field distribution data, the maximum normalized temperature gradient is calculated. The spatial gradient amplitude is obtained from the thermal image matrix of the copper busbar. The maximum value is found and normalized with the ambient temperature to reflect whether the copper busbar has cracks or local ablation faults. Simultaneously, the spatial consistency of temperature is calculated. The thermal image is smoothed with a Gaussian filter. The standard deviation of the difference between the original thermal image and the smoothed thermal image is calculated. Dividing the result by the standard deviation of the original thermal image identifies abnormal points in the temperature field that do not conform to the normal thermal conduction law, which are used to locate the fault points of the copper busbar. Principal component analysis was performed on the real part and distribution of the complex impedance spectrum for eddy current impedance data. The ratio of the eigenvalues ​​of the first and second principal components and the angle of the first principal component vector were calculated to obtain the trajectory ellipticity and tilt angle, which were used to determine the type and severity of cracking faults in the copper busbar. Three frequency points were randomly selected from the frequency band set when the eddy current impedance data was collected, and the corresponding eddy current phase angles were obtained. The standard deviation was calculated and normalized to distinguish between copper busbar faults and normal homogeneous material changes. For vibration data, Hilbert transform is performed on the vibration signal to extract the envelope, the power spectral density of the envelope is calculated, and the maximum ratio of the envelope power spectrum to its effective value is extracted as the envelope peak factor, which is a key feature for identifying bolt loosening faults in copper busbars. In the multi-parameter correlation and fusion analysis of copper busbars, the correlation coefficient between the resistance change rate sequence and the temperature change rate is calculated, a linear regression model of resistance and temperature is established and the curve is plotted to obtain the hysteresis area of ​​the curve; simultaneously, the cross-correlation function between the eddy current impedance phase change rate and the effective value of vibration acceleration is calculated, the maximum cross-correlation value is extracted, and the covariance between the eddy current impedance amplitude and the vibration signal in a specific frequency band is calculated; finally, the trend direction of all copper busbar multidimensional parameters within the time window is calculated, and the trend consistency ratio is statistically analyzed through linear fitting slope quantification to predict the future characteristic performance of copper busbar parameters and achieve fault early warning; The primary dynamic features and associated features of the copper busbar are concatenated according to the channel dimension and converted into multi-dimensional time series data, ensuring time alignment and consistent time window size, thus forming a joint feature vector for diagnosing mechanical faults in the copper busbar.

[0032] Step S30: Input the real-time acquired joint feature vector of the copper busbar into the pre-trained fault diagnosis model. Based on the multi-task learning framework, establish a hierarchical diagnosis mechanism, output the predicted mechanical fault type, fault severity and health warning information of the copper busbar, and complete the fault diagnosis report.

[0033] In this embodiment, the structure of the fault diagnosis model is as follows: Figure 2 As shown, the model adopts a multi-task learning framework to handle the regression task of predicting the joint feature vector of the copper busbar and the classification task of the mechanical fault of the copper busbar. The model consists of two parallel branches: the upper branch is used for early warning of copper busbar health, and the lower branch is used for early mechanical fault diagnosis of copper busbars. The upper branch consists of a multi-scale processing module, a temporal analysis module, and a prediction head, which updates the network parameters of each part during the training of the fault diagnosis model. The lower branch focuses on feature extraction, extracts fault representations from the joint feature vector through convolutional layers and a multiplication-based feature enhancement module, and outputs fault diagnosis results through a mechanical fault type classifier with fixed parameters and a fault severity determiner. The model uses the joint feature vector of the copper busbar as the common input feature of the upper and lower branches. The input feature is first fed into the feature extraction stage, and then high-level feature representation is extracted through multiple modules. Among them, the multiplication-based feature enhancement module divides the output of the convolutional layer into two parts. One part is activated by the activation function, and the other part is not activated. The two parts of features are multiplied element by element, and then passed through depthwise separable convolution and residual connection. The multiplication feature processing method calculates attention weights through the feature itself and recalibrates the feature to reduce computational complexity. Secondly, the high-level feature representation of the joint feature vector of the copper busbar is input to the multi-scale processing module in the upper branch. After average pooling operation, the multi-dimensional parameter time series data of the copper busbar is divided into multiple scales to analyze the manifestation of mechanical faults of the copper busbar in different time windows of different sizes, so as to achieve comprehensive and accurate capture of fault features. Next, the time series analysis module applies the KAN network to perform in-depth analysis of the multi-scale high-level features of the copper busbar. Through a hierarchical fusion strategy from fine-grained to coarse-grained, it aggregates multi-scale features. The KAN network is based on the Kolmogorov representation theorem and adopts a hierarchical structure of interconnection nodes with different levels of complexity. It uses a combination of simple functions to represent complex functions and models the time series relationship of the multi-dimensional parameters of the copper busbar. Finally, the joint feature vector of the copper busbar is output by the prediction head for a specific time period in the future. After multi-parameter correlation analysis, the health index of the copper busbar is calculated, a threshold is set to make a mild or severe warning, and corresponding maintenance decisions are made based on the warning. In parallel, the lower branch inputs the high-level feature representation into the mechanical fault type classifier and the fault severity determiner to achieve hierarchical fault diagnosis. Their network parameters are fixed in advance through training. The mechanical fault type classifier is trained using the historical joint feature vector of the copper busbar and the corresponding fault type label as the dataset. The fault types include loose connection, crack defect, local ablation, plastic deformation and compound fault. The fault severity determiner is trained using the historical joint feature vector of the copper busbar and the severity label of each fault as the dataset. The severity levels include mild, moderate and severe.

[0034] Example 2: The copper busbar mechanical fault diagnosis system based on multidimensional parameter analysis provided in this embodiment of the invention can execute the copper busbar mechanical fault diagnosis method based on multidimensional parameter analysis provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method, such as... Figure 3 As shown, it includes the following modules: Multi-dimensional parameter acquisition module: used to apply specific excitation to the copper busbar under test, and to collect and record multi-dimensional parameters of the copper busbar in the current working state in real time without power interruption, including contact resistance, temperature field distribution, high-frequency eddy current impedance, as well as load current and vibration data. Joint feature vector acquisition module: used to preprocess each independent copper busbar parameter to form time series data within a set time period, extract primary dynamic features based on time-frequency domain analysis, and extract correlation features between parameters through multi-parameter correlation fusion analysis. The two types of features together constitute a joint feature vector for copper busbar fault diagnosis. Fault diagnosis result output module: It is used to input the joint feature vector of copper busbar acquired in real time into the pre-trained fault diagnosis model, establish a hierarchical diagnosis mechanism based on the multi-task learning framework, output the predicted mechanical fault type, fault severity and health warning information of copper busbar, and complete the fault diagnosis report.

[0035] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.

[0036] Example 3: This application provides a copper busbar mechanical fault diagnosis device based on multidimensional parameter analysis. The copper busbar mechanical fault diagnosis device based on multidimensional parameter analysis includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute a copper busbar mechanical fault diagnosis method based on multidimensional parameter analysis in the above embodiment.

[0037] In embodiment four, this application provides a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via a communication system, or installed from a storage system. When the computer program is executed by a processing system, it performs the functions defined in the method of embodiment one of this application.

[0038] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application. In some cases, the actions or steps described in this application can be performed in a different order than that shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A method for diagnosing mechanical faults in copper busbars based on multidimensional parameter analysis, characterized in that, The method includes: Step S10: Apply a specific excitation to the copper busbar to be tested, and collect and record multi-dimensional parameters of the copper busbar in the current working state in real time under the condition of continuous power supply, including contact resistance, temperature field distribution, high-frequency eddy current impedance, load current and vibration data. Step S20: Preprocess each independent copper busbar parameter to form time series data within a set time period. Extract primary dynamic features based on time-frequency domain analysis, and extract correlation features between parameters through multi-parameter correlation fusion analysis. The two features together constitute a joint feature vector for copper busbar fault diagnosis. Step S30: Input the real-time acquired joint feature vector of the copper busbar into the pre-trained fault diagnosis model. Based on the multi-task learning framework, establish a hierarchical diagnosis mechanism, output the predicted mechanical fault type, fault severity and health warning information of the copper busbar, and complete the fault diagnosis report.

2. The method for diagnosing mechanical faults in copper busbars based on multidimensional parameter analysis as described in claim 1, characterized in that, In step S10, two excitation methods, high-frequency eddy current excitation and DC microcurrent superposition, are used to obtain the contact resistance and eddy current complex impedance spectrum data of the copper busbar. The temperature field distribution data of the copper busbar is collected based on the combination of infrared thermography and fixed-point temperature measurement. The load current and vibration data are directly obtained through the corresponding sensors. All sensors are installed in a non-contact manner to ensure the safety of the main circuit where the copper busbar is located.

3. The method for diagnosing mechanical faults in copper busbars based on multidimensional parameter analysis as described in claim 1, characterized in that, In step S20, time-frequency domain analysis is performed on each copper busbar parameter to obtain primary dynamic characteristics: For contact resistance data, the ratio of the standard deviation of the resistance change rate to the mean is calculated to obtain the fluctuation coefficient; the intensity ratio of the trend term of the resistance is calculated simultaneously, and the trend term is separated from the time series of copper busbar contact resistance by the moving average method. The proportion of its energy to the total energy of the series is calculated to indicate the severity of mechanical failure of the copper busbar. For the temperature field distribution data, the maximum normalized temperature gradient is calculated. The spatial gradient amplitude is obtained from the thermal image matrix of the copper busbar. The maximum value is found and normalized with the ambient temperature to reflect whether the copper busbar has cracks or local ablation faults. Simultaneously, the spatial consistency of temperature is calculated. The thermal image is smoothed with a Gaussian filter. The standard deviation of the difference between the original thermal image and the smoothed thermal image is calculated. Dividing the result by the standard deviation of the original thermal image identifies abnormal points in the temperature field that do not conform to the normal thermal conduction law, which are used to locate the fault points of the copper busbar. Principal component analysis was performed on the real part and distribution of the complex impedance spectrum for eddy current impedance data. The ratio of the eigenvalues ​​of the first and second principal components and the angle of the first principal component vector were calculated to obtain the trajectory ellipticity and tilt angle, which were used to determine the type and severity of cracking faults in the copper busbar. Three frequency points were randomly selected from the frequency band set when the eddy current impedance data was collected, and the corresponding eddy current phase angles were obtained. The standard deviation was calculated and normalized to distinguish between copper busbar faults and normal homogeneous material changes. For vibration data, Hilbert transform is performed on the vibration signal to extract the envelope, the power spectral density of the envelope is calculated, and the maximum ratio of the envelope power spectrum to its effective value is extracted as the envelope peak factor, which is a key feature for identifying bolt loosening faults in copper busbars.

4. The method for diagnosing mechanical faults in copper busbars based on multidimensional parameter analysis as described in claim 1, characterized in that, In step S20, during the multi-parameter correlation fusion analysis of the copper busbar, the following correlation features are obtained: Calculate the correlation coefficient between the resistance change rate sequence and the temperature change rate, establish a linear regression model of resistance and temperature and plot the curve to obtain the hysteresis area of ​​the curve. Simultaneously calculate the cross-correlation function between the phase change rate of eddy current impedance and the effective value of vibration acceleration, extract the maximum cross-correlation value, and calculate the covariance between the amplitude of eddy current impedance and the vibration signal in a specific frequency band. The trend direction of all copper busbar multidimensional parameters is calculated within a time window. The slope is quantified by linear fitting, and the proportion of trend consistency is statistically analyzed to predict the future characteristic performance of copper busbar parameters.

5. The method for diagnosing mechanical faults in copper busbars based on multidimensional parameter analysis as described in claim 1, characterized in that, In step S30, the fault diagnosis model is based on a multi-task learning framework, which handles the regression task of predicting the joint feature vector of the copper busbar and the classification task of the mechanical fault of the copper busbar respectively. The model consists of two parallel branches: the upper branch is used for early warning of copper busbar health, and the lower branch is used for early mechanical fault diagnosis of copper busbars. The upper branch consists of a multi-scale processing module, a temporal analysis module, and a prediction head, which updates the network parameters of each part during the training of the fault diagnosis model. The lower branch focuses on feature extraction, extracts fault representations from the joint feature vector through convolutional layers and a multiplication-based feature enhancement module, and outputs fault diagnosis results through a mechanical fault type classifier and a fault severity determiner with fixed parameters.

6. The method for diagnosing mechanical faults in copper busbars based on multidimensional parameter analysis as described in claim 5, characterized in that, The multi-scale processing module divides the multi-dimensional parameter time-series data of the copper busbar into multiple scales through average pooling operations, analyzes the manifestation of mechanical faults of the copper busbar in different time windows of different sizes, and achieves comprehensive and accurate capture of fault characteristics. The time series analysis module applies the KAN network to perform in-depth analysis of the multi-scale high-level features of the copper busbar. Through a hierarchical fusion strategy from fine-grained to coarse-grained, it aggregates multi-scale features. The KAN network is based on the Kolmogorov representation theorem and adopts a hierarchical structure of interconnection nodes with different levels of complexity. It uses a combination of simple functions to represent complex functions and models the time series relationship of the multi-dimensional parameters of the copper busbar.

7. The method for diagnosing mechanical faults in copper busbars based on multidimensional parameter analysis as described in claim 5, characterized in that, The network parameters of the fault type classifier and the fault severity determiner are fixed in advance through training. The mechanical fault type classifier is trained using the historical joint feature vector of the copper busbar and the corresponding fault type label as the dataset. The fault types include loose connection, crack defect, local ablation, plastic deformation and compound fault. The fault severity determiner is trained using the historical joint feature vector of the copper busbar and the severity label of each fault as the dataset. The severity levels include mild, moderate and severe.

8. A copper busbar mechanical fault diagnosis system based on multidimensional parameter analysis, characterized in that, The system is used to implement the copper busbar mechanical fault diagnosis method based on multidimensional parameter analysis as described in any one of claims 1-7, the system comprising: Multi-dimensional parameter acquisition module: used to apply specific excitation to the copper busbar under test, and to collect and record multi-dimensional parameters of the copper busbar in the current working state in real time without power interruption, including contact resistance, temperature field distribution, high-frequency eddy current impedance, as well as load current and vibration data. Joint feature vector acquisition module: used to preprocess each independent copper busbar parameter to form time series data within a set time period, extract primary dynamic features based on time-frequency domain analysis, and extract correlation features between parameters through multi-parameter correlation fusion analysis. The two types of features together constitute a joint feature vector for copper busbar fault diagnosis. Fault diagnosis result output module: It is used to input the joint feature vector of copper busbar acquired in real time into the pre-trained fault diagnosis model, establish a hierarchical diagnosis mechanism based on the multi-task learning framework, output the predicted mechanical fault type, fault severity and health warning information of copper busbar, and complete the fault diagnosis report.

9. A copper busbar mechanical fault diagnosis device based on multidimensional parameter analysis, characterized in that, The aforementioned agricultural product traceability device based on digital signature includes: The system includes a memory, a processor, and a copper busbar mechanical fault diagnosis program based on multidimensional parameter analysis, which is stored in the memory and can run on the processor. When the processor executes the copper busbar mechanical fault diagnosis program based on multidimensional parameter analysis, it implements a copper busbar mechanical fault diagnosis method based on multidimensional parameter analysis as described in any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes a copper busbar mechanical fault diagnosis program based on multidimensional parameter analysis. When the copper busbar mechanical fault diagnosis program based on multidimensional parameter analysis is executed by the processor, it implements a copper busbar mechanical fault diagnosis method based on multidimensional parameter analysis as described in any one of claims 1 to 7.

Images