Urban rail train high-voltage box current detection system and method

Abstract

The invention relates to the technical field of electrical measurement, in particular to an urban rail train high-voltage box current detection system and method. The system comprises a signal acquisition module used for acquiring an original current measurement signal; the signal characteristic quantification module inputs the original current measurement signal into a current reconstruction model to generate a reconstructed current measurement signal, and calculates a signal quality index; the data association analysis module analyzes the signal quality index and the original current measurement signal, and constructs an association intensity matrix and a causal relationship graph; the numerical value correction module identifies and corrects outlier data points in the correlation intensity matrix and the causal relationship graph, and outputs a corrected current value; and the current state interpretation module calls a Gaussian process regression model to predict a probabilistic reasonable current threshold domain according to the urban rail train multi-dimensional working condition vector, compares the probabilistic reasonable current threshold domain with the corrected current value, generates a working state judgment result and outputs the working state judgment result together with the corrected current value. According to the system, the diagnosis depth and the environmental adaptability of current detection are improved, and the accuracy and the reliability of a measurement result are guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of electrical measurement, and in particular to a current detection system and method for a high-voltage box of an urban rail train. Background Art

[0002] The high-voltage system of urban rail transit trains is the energy source for core subsystems such as traction and auxiliary power supply. The high-voltage box is a key node in this system, and the high current flowing through it is a core physical quantity that reflects the system's power transmission status. Therefore, being able to accurately and real-timely obtain the operating current value of the high-voltage box is a fundamental prerequisite for electrical system status assessment and safety monitoring. However, in existing technologies, accurately and reliably measuring this current value faces many challenges, specifically in the following aspects: Traditional measurement solutions typically employ only one current sensor per high-voltage box. In the harsh operating environment of trains, characterized by severe vibration, wide temperature fluctuations, and strong electromagnetic interference, this single sensor is prone to physical degradation, such as zero-point drift, linearity degradation, and output signal distortion, or even outright failure. In these situations, the system's output value can deviate significantly from the true value.

[0003] To improve reliability, some current detection solutions use multiple sensors for redundant measurement. However, existing data processing methods are relatively rudimentary. While these methods can handle sudden, large deviations from a single sensor, they struggle to effectively identify and correct simultaneous, small, slow drifts in the same direction across multiple sensors, or more subtle anomalies at the signal waveform level. More importantly, existing methods fail to fully utilize the inherent coupling relationships and causal logic that should exist between multiple measurement signals based on physical laws to cross-validate and purify the data. As a result, the accuracy and reliability of the final measurement value output after fusion cannot be fundamentally guaranteed.

[0004] Existing technologies often rely on fixed, reasonable ranges to assess the validity of test results. However, the normal value of a train's current can vary dramatically depending on its operating conditions. Using fixed ranges to evaluate dynamically changing measurements can lead to even accurate measurements being misinterpreted as abnormal under certain operating conditions, and vice versa.

[0005] Therefore, a current detection system and method for a high-voltage box of an urban rail train are proposed. Summary of the Invention

[0006] The purpose of the present invention is to provide a current detection system and method for the high-voltage box of an urban rail train, which is used for current detection of the high-voltage box of an urban rail train. In order to solve the problems existing in the prior art, the present invention first uses a signal feature quantification module to use a current reconstruction model to perform a quality assessment on the original current measurement signal and output a signal quality index; secondly, a data association analysis module constructs a correlation strength matrix and a causal relationship diagram; then, a numerical correction module identifies and intelligently corrects outlier measurement values ​​caused by correlation failure, and outputs the corrected current value; finally, a current state judgment module calls a Gaussian process regression model according to the real-time working conditions of the train to predict a dynamic and reasonable current threshold domain, and outputs the corrected current value after state judgment. The present invention significantly improves the reliability, diagnostic depth and environmental adaptability of current detection through a layer-by-layer progressive data processing process.

[0007] To achieve the above objectives, the present invention provides a current detection system for a high-voltage box of an urban rail train, which specifically includes the following modules: Signal acquisition module: collects the current of the high-voltage box of the urban rail train in real time to obtain the original current measurement signal; Signal feature quantification module: inputs the original current measurement signal into the current reconstruction model to generate a reconstructed current measurement signal, and calculates the signal quality index of the original current measurement signal through the quality index generator; Data association analysis module: analyzing the signal quality index and the original current measurement signal, inferring and forming a correlation strength matrix and a causal relationship diagram; Numerical correction module: identifies outlier data points on the correlation strength matrix and the causal relationship diagram, identifies and corrects outlier measurement values, and outputs corrected current values; Current state judgment module: obtains the multi-dimensional working condition vector of the urban rail train in real time, calls the Gaussian process regression model, and predicts the probabilistic reasonable current threshold domain; compares the corrected current value with the probabilistic reasonable current threshold domain, generates a working state judgment result, and outputs it together with the corrected current value.

[0008] Preferably, the process of the signal acquisition module acquiring the original current measurement signal includes: using a sensor arranged inside the high-voltage box to convert the current into an analog voltage measurement signal; using a multi-channel synchronous analog-to-digital conversion interface to synchronously sample and digitize the analog voltage measurement signals of all channels to generate the original current measurement signal.

[0009] Preferably, the signal feature quantization module includes a current reconstruction model and a quality index generator; wherein, the current reconstruction model includes an encoder and a decoder; the original current measurement signal is input into the current reconstruction model, processed by the encoder, and compressed into a current waveform eigenvector; the current waveform eigenvector is input into the decoder, decoded and generates a reconstructed current measurement signal; the current reconstruction model is trained offline by healthy current data.

[0010] Preferably, the specific processing process of the quality index generator includes: the quality index generator calculates the point-by-point error between the original current measurement signal and the reconstructed current measurement signal, integrates and generates a reconstructed error energy value; inputs the reconstructed error energy value into the error-index response curve for normalization, and calculates and obtains a signal quality index; the signal quality index is inversely proportional to the error energy value.

[0011] Preferably, the specific operations of the data association analysis module include: screening according to the signal quality index of each channel, and reorganizing the retained original current measurement signals into a synchronous current measurement matrix; solving the normalized mutual information between any two original current measurement signals in the synchronous current measurement matrix within a sliding time window to construct a correlation strength matrix; performing Granger causality test on signal pairs with a correlation degree higher than a threshold in the correlation strength matrix to infer a causal relationship diagram.

[0012] Preferably, the specific operations of the numerical correction module include: comparing the correlation strength matrix and the causal relationship diagram with a preset benchmark correlation paradigm item by item; when the normalized mutual information value and causal relationship of the original current measurement signal significantly deviate from the benchmark correlation paradigm, determining that the original current measurement signal is an outlier measurement value; when the outlier measurement value is identified, calling a pre-trained multidimensional current sequence coupling model; the multidimensional current sequence coupling model receives the original current measurement signals of multiple healthy channels as input, analyzes nonlinearity and timing dependence, calculates and outputs the optimal estimated current waveform; uses the current value of the optimal estimated current waveform to replace the outlier measurement value to complete the correction, and outputs the corrected current value.

[0013] Preferably, the multi-dimensional operating condition vector includes: real-time train speed, real-time train acceleration, slope information of the line on which the train is located, traction or braking command level output by the driver controller, high-voltage system bus voltage, and external ambient temperature.

[0014] Preferably, the specific operations of the current state judgment module include: obtaining a multidimensional operating condition vector in real time through the train network; inputting the multidimensional operating condition vector into a Gaussian process regression model to output a predicted current expectation value and a predicted current uncertainty; combining the predicted current expectation value and the predicted current uncertainty with a preset confidence level to calculate and construct a probabilistic reasonable current threshold domain; comparing the input value of the corrected current value with the probabilistic reasonable current threshold domain to obtain the working state judgment result, and outputting the working state judgment result and the corrected current value.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. By performing a quality assessment on the original current measurement signal, the present invention can identify signal distortion caused by sensor failure or complex electromagnetic interference online and in real time, thus solving the problem that traditional single-sensor systems cannot verify their credibility.

[0016] 2. This invention introduces mutual information analysis and causality testing. By comparing real-time multi-channel correlation patterns with benchmark paradigms, it can identify hidden faults where individual channel values ​​are normal but their group correlation behavior is abnormal, thereby improving the system's ability to detect connectivity faults and early performance drift.

[0017] 3. By employing a Gaussian process regression model, this invention dynamically generates probabilistic and reasonable current ranges with confidence intervals based on the train's real-time, complex, and ever-changing operating conditions. This intelligent and flexible discrimination mechanism effectively captures true anomalies while reducing false alarms caused by drastic changes in operating conditions, thereby enhancing the robustness and usability of the detection system.

[0018] 4. Through three progressive processing steps, namely signal quality assessment, multi-channel consistency verification, and adaptive working condition identification, the present invention ultimately refines the original, uncertain measurement signal into a highly reliable measurement result that has undergone internal quality verification, external cross-validation, and scenario rationality verification, providing high-quality data support for the safe operation and intelligent operation and maintenance of trains. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A schematic structural diagram of a current detection system for a high-voltage box of an urban rail train provided by an embodiment of the present invention; Figure 2 A flow chart of a method for detecting current in a high-voltage box of an urban rail train provided by an embodiment of the present invention; Figure 3 This is an internal data flow diagram of a current detection system and method for a high-voltage box of an urban rail train provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0020] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 making creative efforts are within the scope of protection of the present invention.

[0021] See also Figures 1 to 3 The present invention provides a system and method for detecting current of a high-voltage box of an urban rail train. The technical solution is as follows: A system and method for detecting current in high-voltage box of urban rail train, referring to Figure 1 , the specific modules of the system proposed by the present invention include: Signal acquisition module: collects the current of the high-voltage box of the urban rail train in real time to obtain the original current measurement signal; Signal feature quantification module: inputs the original current measurement signal into the current reconstruction model to generate a reconstructed current measurement signal, and calculates the signal quality index of the original current measurement signal through the quality index generator; Data association analysis module: analyzing the signal quality index and the original current measurement signal, inferring and forming a correlation strength matrix and a causal relationship diagram; Numerical correction module: identifies outlier data points on the correlation strength matrix and the causal relationship diagram, identifies and corrects outlier measurement values, and outputs corrected current values; Current state judgment module: obtains the multi-dimensional working condition vector of the urban rail train in real time, calls the Gaussian process regression model, and predicts the probabilistic reasonable current threshold domain; compares the corrected current value with the probabilistic reasonable current threshold domain, generates a working state judgment result, and outputs it together with the corrected current value.

[0022] Example 1 This embodiment provides a specific application of a high-voltage box current detection system for urban rail trains. Its typical application scenario is an urban rail train containing six high-voltage boxes, which introduces a high-voltage box current detection system for regular detection. Figure 2 、 Figure 3 This embodiment performs detection according to the process of a method for detecting current of a high-voltage box of an urban rail train provided by the present invention, and displays the data flow.

[0023] Furthermore, the current of the high-voltage box of the urban rail train is collected in real time to obtain the original current measurement signal. The specific process is as follows: The signal acquisition module's functions are implemented by current sensors distributed within six high-voltage boxes and the data synchronization acquisition interface of the onboard data processing controller. Each high-voltage box contains a current sensor that linearly converts the measured main circuit high current into a continuously varying analog voltage measurement signal. The onboard data processing controller, through its multi-channel, high-precision analog-to-digital conversion interface, rigorously and synchronously samples and digitizes these six analog voltage measurement signals at a sampling rate of 20,000 times per second. The resulting raw digital signal, sampled at 20,000 Hz, passes through a digital low-pass filter. The filter's cutoff frequency is set slightly below the Nyquist frequency of the target post-decimation sampling rate. The low-pass filtered signal then passes through a decimator. This decimator operates at a 4:1 decimation ratio, retaining one point for every four samples, reducing the signal's sampling rate from 20,000 Hz to 5,000 Hz. This results in a smoother, cleaner, pre-processed raw current measurement signal with a sampling rate of 5,000 Hz. By reducing the sampling rate from 20,000 Hz to 5,000 Hz, the computational complexity and resource consumption of the subsequent system are greatly reduced, and the signal quality and signal-to-noise ratio are significantly improved, ensuring the effectiveness and stability of subsequent analysis.

[0024] The signal acquisition process enables the current data between different high-voltage boxes to be aligned in time with high accuracy. After digital conversion, six parallel raw current measurement signal data streams are formed that can accurately characterize the instantaneous changes in current.

[0025] Furthermore, the original current measurement signal is input into a current reconstruction model to generate a reconstructed current measurement signal, and a quality index generator calculates a signal quality index of the original current measurement signal. The specific process is as follows: The raw current measurement signal from each channel is fed into a signal feature quantization module. This module incorporates a built-in current reconstruction model, consisting of an encoder and a decoder. This model is trained offline using a large amount of healthy current data. This data covers all normal operating conditions of the train, including stationary trains, traction at different acceleration levels, braking at different power levels, and auxiliary loads operating under various seasons and passenger density conditions.

[0026] The original current measurement signal first flows through the encoder, which consists of five one-dimensional convolutional layers and three maximum pooling layers interleaved. As the original current measurement signal flows through the encoder, its high-dimensional time-domain features are extracted and compressed layer by layer, compressing the complex waveform information into a low-dimensional current waveform eigenvector. This vector can be understood as a condensed representation of the most essential and core information of the original waveform. This current waveform eigenvector is then input into the decoder, which consists of five one-dimensional deconvolutional layers and three upsampling layers. Through progressive decoding and dimensionality increase, the decoder reconstructs the current waveform eigenvector into a reconstructed current measurement signal with the same dimensionality as the original input.

[0027] The current reconstruction model enables the system to not only evaluate signal quality, but also extract the eigenvectors that characterize the core dynamics of the current, providing high-quality, highly concentrated input for subsequent deeper diagnosis and state classification of the current, greatly improving efficiency and accuracy.

[0028] At the same time, the quality index generator begins operating. It synchronously receives the original and reconstructed current measurement signals and calculates the root mean square error (RMS) between the original and reconstructed current measurement signals point by point, obtaining a reconstructed error energy value. This reconstructed error energy value is then normalized using a pre-defined, monotonically decreasing error-exponent response curve. This function, an inverse S-shaped curve, smoothly and nonlinearly maps the reconstructed error energy value to a signal quality index between 0 and 1. The error-exponent response curve is a dynamically adaptive curve whose shape adjusts based on the RMS value of the original current measurement signal within the current time window. The system first calculates the RMS value of the original current measurement signal within the current time window as a quantitative indicator of the current energy level of the original current measurement signal. A response curve is then selected from a pre-defined function family based on this RMS value. This implementation ensures that the signal quality index evaluation criteria can smoothly and automatically adapt to continuous changes in signal load, thereby providing accurate and reliable signal quality assessment under all operating conditions. The signal quality index is in strict inverse proportion to the reconstruction error energy value. The higher the signal quality index is, the more reliable the original signal quality is.

[0029] The signal quality index generation process is concretized into calculating the reconstruction error and normalizing it using the response curve, providing a quantitative evaluation method so that the signal quality of different sensors and at different times has a unified and horizontally comparable evaluation scale.

[0030] Furthermore, the signal quality index and the original current measurement signal are analyzed to infer and construct a correlation strength matrix and a causal relationship diagram. The specific process is as follows: The data association analysis module receives all raw current measurement signals and their corresponding signal quality indices. It first screens them based on the signal quality index, retaining only channels with an index above 0.9 for subsequent collaborative analysis to ensure the reliability of the analysis foundation. The filtered signals are reorganized into a six-column synchronized current measurement matrix. The data association analysis module continuously performs in-depth analysis of this matrix within a sliding time window of 1 second. It calculates the normalized mutual information between any two high-voltage box current signals and constructs a real-time correlation strength matrix. The correlation strength matrix quantifies the strength of the nonlinear coupling between all channels.

[0031] For strongly correlated channel pairs with mutual information values ​​greater than 0.6, the data association analysis module initiates a parallel dynamic relationship analysis process, performing Granger causality tests on these channel pairs. Based on all test results, the data association analysis module generates and updates a directed real-time causal relationship graph. The time delay calculator parallelly calculates the cross-correlation functions of these strongly correlated signal pairs and identifies the time delay corresponding to the peak point to accurately quantify the signal transmission delay τ between them. The data association analysis module integrates and mutually supports the causal direction and signal transmission delay τ inferred from the real-time causal relationship graph.

[0032] This real-time dynamic relationship description, which includes direction and delay information, will be used for deeper comparison with the benchmark correlation paradigm, thereby identifying more hidden system failures that not only have abnormal correlation strength but also have changes in transmission delay or causal logic.

[0033] By using mutual information and Granger causality tests to construct correlations, the system can go beyond simple linear correlation analysis and explore the complex nonlinear dependencies and "leader-follower" conduction patterns between the currents of each high-voltage box, thereby enabling earlier and more accurate detection of system-level hidden faults and facilitating precise current detection.

[0034] Furthermore, outlier data points are identified on the correlation strength matrix and the causal relationship diagram, and outlier measurement values ​​are identified and corrected, and the corrected current value is output. The specific process is as follows: The numerical correction module then compares the correlation strength matrix and real-time causal relationship graph with a locally stored, preset benchmark correlation paradigm for the current train operating mode. When the system maintains stable operation under specific operating conditions, if the signal quality index of all channels remains high and no outliers are identified, the numerical correction module activates the paradigm learner. The paradigm learner collects the correlation strength matrix and causal relationship graph during the stable operation period and calculates real-time statistical characteristics. If these real-time statistical characteristics exhibit small but stable systematic deviations from the existing benchmark correlation paradigm, the learner calculates correction coefficients. After obtaining authorization, the numerical correction module uses these correction coefficients to fine-tune and update the stored benchmark correlation paradigm. Dynamic fine-tuning of the benchmark correlation paradigm avoids the long-term accumulation of misjudgments caused by a fixed benchmark, thereby enhancing the reliability of the detection system. The adaptive update mechanism eliminates the need for the preset benchmark paradigm to be perfect at the initial stage, reducing the stringent requirements for initial model training data and enhancing the model's generalization capabilities across diverse application scenarios.

[0035] During the comparison process, if the numerical correction module finds that any of the mutual information and causal relationships of a channel deviates significantly from the baseline, it will determine that channel as an outlier measurement. At this point, the numerical correction module invokes a pre-trained multidimensional current sequence coupling model. This model takes the raw current measurement signals of multiple healthy channels as input and, by analyzing the complex nonlinear and temporal dependencies between them, calculates and outputs the optimal estimated current waveform for the outlier channel. The numerical correction module uses the current value of this optimal estimated waveform to replace the original outlier measurement, completing the correction and outputting the final corrected current value.

[0036] By proposing an outlier correction method based on a multidimensional current sequence coupling model, the difficult problem of how to deal with bad data after identification is solved; using the information of healthy channels to repair data ensures that even in the event of transient failures in some sensors, the system can still continuously output high-quality data streams, thereby enhancing the fault tolerance and data integrity of the entire detection system.

[0037] Furthermore, a multi-dimensional operating condition vector of the urban rail train is obtained in real time, and a Gaussian process regression model is called to predict a probabilistic reasonable current threshold domain; the corrected current value is compared with the probabilistic reasonable current threshold domain, and a working state discrimination result is generated and output together with the corrected current value. The specific process is as follows: The current state judgment module is the final decision-making and output unit of this system. The current state judgment module obtains various parameters that constitute the multi-dimensional working condition vector in real time through the train's high-speed data network. These parameters include: the train's real-time speed of 65 kilometers per hour, the train's real-time acceleration of 1m / s 2, the slope information of the line on which the train is located is 0.5% uphill, the traction command level 3 output by the driver controller, the high-voltage system bus voltage is 760 volts and the outdoor ambient temperature is 28 degrees Celsius.

[0038] Enumerating the specific physical quantities that constitute the multidimensional operating condition vector provides a solid physical foundation for the subsequent adaptive operating condition judgment; it enables the model to fully consider the key external factors that affect current changes, ensuring the accuracy and reliability of dynamic threshold judgment.

[0039] The current status judgment module inputs the real-time multidimensional operating condition vector into a Gaussian process regression model. The model predicts the expected current value of 60A for each high-voltage box under the current specific operating conditions, and also provides a prediction uncertainty with a standard deviation of 4A. Based on this, the system constructs a probabilistic reasonable current threshold range with a 99.7% confidence level, namely a dynamic confidence interval ranging from 48A to 72A. The current status judgment module compares all values ​​in the received corrected current value vector with the probabilistic reasonable current threshold range. If all values ​​fall within this interval, the module generates a "normal" operating status judgment result. Finally, the current status judgment module integrates the six-dimensional corrected current value vector and its accompanying "normal" status label into a standard data package and publishes it to the train's main monitoring system for real-time display and long-term health record recording.

[0040] By streamlining the specific operational processes for current state interpretation, alarm decisions are no longer simply threshold comparisons. This makes the system more robust when dealing with fuzzy states, effectively suppressing false alarms caused by slight signal fluctuations and improving the accuracy of current measurement.

[0041] The present invention improves the reliability, accuracy and depth of current measurement results by constructing an innovative data processing process that progresses step by step from single-channel confidence assessment and multi-channel coupling analysis to adaptive interpretation of working conditions. While enhancing the system's fault tolerance, it provides a technical means for achieving predictive maintenance and lean operation and maintenance of rail transit vehicle high-voltage systems, and provides technical support for ensuring the safe and efficient operation of trains.

[0042] Example 2 In this embodiment, a system and method for detecting current of a high-voltage box of an urban rail train is applied to the current detection of the high-voltage box of a B urban rail train, focusing on the quality assessment of the original current signal and the multi-channel data fusion analysis process.

[0043] First, the system's signal acquisition module starts operating. Current sensors installed within each high-voltage box convert the real-time current into a corresponding analog voltage measurement signal. A multi-channel, synchronized analog-to-digital conversion interface samples and digitizes the analog voltage measurement signals from all channels at a preset high sampling frequency, generating multiple, parallel raw current measurement signals that represent the continuously changing current waveforms and transmits them to subsequent processing units.

[0044] The signal acquisition process is concretized as multi-channel synchronous analog-to-digital conversion, so that the original current measurement signals of all high-voltage boxes are aligned on the time basis; this time synchronization is the prerequisite for subsequent multi-channel mutual information analysis, causal relationship testing and correlation collaborative diagnosis, and provides a solid data foundation to ensure the accuracy of the results.

[0045] Next, the signal feature quantization module receives these raw current measurement signals. The core of this module is a current reconstruction model trained offline, which logically consists of an encoder and a decoder. When a raw current measurement signal is input, the encoder first processes it, extracting its deep temporal features and compressing it into a low-dimensional eigenvector of the current waveform. This eigenvector is then fed into the decoder, which decodes it and attempts to restore the original waveform to generate a reconstructed current measurement signal.

[0046] The current reconstruction model enables the system to not only evaluate signal quality but also extract the eigenvectors that characterize the core dynamics of the current; this provides high-quality, highly concentrated input for subsequent fault diagnosis and state classification, improving the accuracy of the results.

[0047] At the same time, the quality index generator begins operating. It receives the original and reconstructed current measurement signals, calculates the point-by-point error between them, and integrates it to obtain a scalarized reconstruction error energy value. This energy value is then normalized using a preset, monotonically decreasing error-exponential response curve. Ultimately, a signal quality index (SQI) is calculated, which is inversely proportional to the error energy value. This index intuitively quantifies the degree of deviation of the original signal from the healthy paradigm, i.e., its inherent credibility.

[0048] The signal quality index generation process is concretized into calculating the reconstruction error and normalizing it using the response curve, providing a quantitative evaluation method so that the signal quality of different sensors and at different times has a unified and horizontally comparable evaluation scale.

[0049] The data correlation analysis module then begins a collaborative analysis of the data from all channels. It first screens each channel based on its signal quality index, retaining only those channels with an index exceeding a preset reliability threshold and reorganizing them into a synchronized current measurement matrix. Within a continuously rolling sliding time window, the module performs in-depth computations on the data in the matrix: first, it calculates the normalized mutual information between any two channels to construct a real-time correlation strength matrix; second, it performs Granger causality tests on strongly correlated channel pairs to infer potential leader-follower relationships, ultimately forming a causal relationship graph. Together, these two results fully characterize the complex coupling state between the currents in the high-voltage boxes of the entire train at the current moment.

[0050] By using mutual information and Granger causality tests to construct correlations, the system can go beyond simple linear correlation analysis and explore the complex nonlinear dependencies and "leader-follower" conduction patterns between the currents of each high-voltage box, thereby enabling earlier and more accurate detection of system-level hidden faults and facilitating precise current detection.

[0051] Finally, the numerical correction module performs data optimization and correction. It compares the real-time correlation strength matrix and causal relationship diagram against a locally stored benchmark correlation paradigm representing the train's healthy operating status. When the mutual information value or causal relationship of a channel deviates significantly from the benchmark paradigm, the raw current measurement signal of that channel is identified as an outlier. Once an outlier measurement is identified, the system does not simply remove it. Instead, it calls upon a pre-trained multidimensional current sequence coupling model. This model takes the raw current measurement signals of multiple other healthy channels as input and calculates an optimal estimated current waveform for the outlier channel by analyzing the complex nonlinear and temporal dependencies between them. Finally, the module uses the current instantaneous value of this optimal estimated waveform to replace the original outlier measurement value, completing the correction and outputting a corrected current value that is consistent and reliable with internal data.

[0052] Example 3 In this embodiment, a system and method for detecting current of a high-voltage box of an urban rail train are applied to the current detection of the high-voltage box of a C urban rail train, focusing on the intelligent correction of electrical outliers and the adaptive state judgment process based on working conditions.

[0053] First, the system performs a current calibration. The numerical correction module receives the correlation strength matrix and causal relationship diagram from the previous module. The numerical correction module internally stores a preset baseline correlation paradigm. This baseline correlation paradigm, derived through offline learning of a large amount of data under normal operating conditions, encompasses the ideal correlation strength and causal relationships between different sensor channels.

[0054] The module compares the real-time correlation strength matrix and causal relationship diagram against the baseline correlation paradigm item by item. If the normalized mutual information value and causal relationship between a channel's signal and other channels deviate significantly from the baseline paradigm, the system determines that the current measurement value of that channel is an outlier. Once an outlier measurement value is identified, the system does not directly eliminate it. Instead, it uses a pre-trained multidimensional current sequence coupling model for intelligent correction. This model receives as input the raw current measurement signal sequence of all other channels determined to be healthy. By analyzing the nonlinearities and temporal dependencies between these healthy signals, it calculates and outputs an optimal estimated current waveform for the outlier channel in real time. The system then uses the current value of this optimal estimated current waveform to replace the original outlier measurement value, thereby completing the data correction. Finally, the module outputs the verified corrected current value.

[0055] The outlier correction method solves the problem of how to deal with bad data after it is identified; it uses information from healthy channels to repair data, so that even if some sensors suffer transient failures, the system can still output high-quality data streams uninterruptedly, enhancing the fault tolerance and data integrity of the entire detection system.

[0056] Next, the system performs current status determination. The current status determination module is responsible for making the final determination of the rationality of the corrected current value. This module acquires a multi-dimensional operating condition vector in real time via the train network. In this embodiment, this vector includes: the train's real-time speed, real-time acceleration, the slope of the line the train is on, the traction or braking command level output by the driver's controller, the high-voltage system bus voltage, and the ambient temperature outside the vehicle.

[0057] Clarifying the specific physical quantities that constitute the multi-dimensional operating condition vector enables the model to fully consider the key external factors that affect current changes, providing a solid physical foundation for subsequent adaptive operating condition judgment and ensuring the accuracy and reliability of dynamic threshold judgment.

[0058] This multidimensional operating condition vector is input into a Gaussian process regression model. Based on the current operating conditions, the model generates a real-time output of the expected current value μ and the predicted current uncertainty σ. These two values ​​describe the most likely current value of the high-voltage box under the current operating conditions and its reasonable fluctuation range. Subsequently, based on a preset confidence level, the system calculates and constructs a dynamic probabilistic reasonable current threshold range within the range [μ-3σ, μ+3σ].

[0059] Finally, the current status determination module compares the corrected current value from the previous module with this dynamic threshold range. If the value falls within the threshold range, the operating status is determined to be normal; if it significantly exceeds the threshold range, the operating status is determined to be abnormal. Finally, the system outputs this determination result along with the corrected current value for display and recording by the train monitoring system.

[0060] The specific operational process of current state interpretation means that alarm judgment is no longer a simple comparison; this makes the system more robust when dealing with fuzzy states, can effectively suppress false alarms caused by small signal fluctuations, and improve the accuracy of current measurement.

[0061] Through the process described in this embodiment, the system can not only intelligently repair the data generated by the faulty sensor, but also dynamically and intelligently judge the rationality of the current in combination with the complex operating conditions of the train, greatly improving the accuracy and reliability of detection, and effectively avoiding the false alarms and missed alarms caused by traditional fixed threshold alarm methods.

[0062] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A current detection system for a high-voltage box of an urban rail train, characterized in that: include: Signal acquisition module: used to collect the current of the high-voltage box of the urban rail train in real time and obtain the original current measurement signal; A signal feature quantification module is configured to input the original current measurement signal into a current reconstruction model to generate a reconstructed current measurement signal, and calculate a signal quality index of the original current measurement signal through a quality index generator; Data association analysis module: used to analyze the signal quality index and the original current measurement signal, infer and construct a correlation strength matrix and a causal relationship diagram; Numerical correction module: used to identify outlier data points on the correlation strength matrix and the causal relationship diagram, identify and correct outlier measurement values, and output corrected current values; Current state judgment module: used to obtain the multi-dimensional working condition vector of the urban rail train in real time, call the Gaussian process regression model, and predict the probabilistic reasonable current threshold domain; compare the corrected current value with the probabilistic reasonable current threshold domain, generate a working state judgment result, and output it together with the corrected current value.

2. The urban rail train high-voltage box current detection system according to claim 1, characterized in that: The process of the signal acquisition module acquiring the original current measurement signal includes: The current is converted into an analog voltage measurement signal by using a sensor arranged inside the high-voltage box; the analog voltage measurement signals of all channels are synchronously sampled and digitized using a multi-channel synchronous analog-to-digital conversion interface to generate the original current measurement signal.

3. The urban rail train high-voltage box current detection system according to claim 1, characterized in that: The signal feature quantization module includes a current reconstruction model and a quality index generator; wherein, the current reconstruction model includes an encoder and a decoder; the original current measurement signal is input into the current reconstruction model and processed by the encoder to be compressed into a current waveform eigenvector; the current waveform eigenvector is input into the decoder, decoded and generates a reconstructed current measurement signal; the current reconstruction model is trained offline using healthy current data.

4. The urban rail train high-voltage box current detection system according to claim 1, characterized in that: The specific processing process of the quality index generator includes: The quality index generator calculates the point-by-point error between the original current measurement signal and the reconstructed current measurement signal, integrates the error, and generates a reconstructed error energy value; inputs the reconstructed error energy value into an error-index response curve for normalization, and calculates a signal quality index; the signal quality index is inversely proportional to the error energy value.

5. The urban rail train high-voltage box current detection system according to claim 1, characterized in that: The specific operations of the data association analysis module include: The signal quality index of each channel is screened and the retained original current measurement signals are reorganized into a synchronous current measurement matrix. The normalized mutual information between any two original current measurement signals in the synchronous current measurement matrix is ​​calculated within a sliding time window to construct a correlation strength matrix. A Granger causality test is performed on signal pairs with a correlation degree higher than a threshold in the correlation strength matrix to infer a causal relationship graph.

6. The urban rail train high-voltage box current detection system according to claim 1, characterized in that: The specific operations of the numerical correction module include: The correlation strength matrix and the causal relationship diagram are compared item by item with a preset benchmark correlation paradigm; when the normalized mutual information value and the causal relationship of the original current measurement signal significantly deviate from the benchmark correlation paradigm, the original current measurement signal is determined to be an outlier measurement value; when the outlier measurement value is identified, a pre-trained multidimensional current sequence coupling model is called; the multidimensional current sequence coupling model receives the original current measurement signals of multiple healthy channels as input, analyzes nonlinearity and timing dependence, calculates and outputs an optimal estimated current waveform; the current value of the optimal estimated current waveform is used to replace the outlier measurement value to complete the correction, and the corrected current value is output.

7. The urban rail train high-voltage box current detection system according to claim 1, characterized in that: The multi-dimensional operating condition vector includes: real-time train speed, real-time train acceleration, gradient information of the line on which the train is located, traction or braking command level output by the driver controller, high-voltage system bus voltage, and external ambient temperature.

8. The urban rail train high-voltage box current detection system according to claim 1, characterized in that: The specific operations of the current state judgment module include: A multi-dimensional operating condition vector is acquired in real time through the train network; the multi-dimensional operating condition vector is input into a Gaussian process regression model to output a predicted current expectation value and a predicted current uncertainty; the predicted current expectation value and the predicted current uncertainty are combined with a preset confidence level to calculate and construct a probabilistic reasonable current threshold domain; the numerical value of the input corrected current value is compared with the probabilistic reasonable current threshold domain to obtain the working state judgment result, and the working state judgment result and the numerical value of the corrected current value are output.

9. A method for detecting current of a high-voltage box of an urban rail train, characterized in that: include: Real-time collection of current from the high-voltage box of urban rail trains to obtain the original current measurement signal; Inputting the original current measurement signal into a current reconstruction model to generate a reconstructed current measurement signal, and calculating a signal quality index of the original current measurement signal through a quality index generator; Analyzing the signal quality index and the original current measurement signal to infer and construct a correlation strength matrix and a cause-effect relationship diagram; Identifying outlier data points on the correlation strength matrix and the causal relationship diagram, identifying and correcting outlier measurement values, and outputting corrected current values; A multi-dimensional operating condition vector of the urban rail train is obtained in real time, a Gaussian process regression model is called to predict a probabilistic reasonable current threshold domain; the corrected current value is compared with the probabilistic reasonable current threshold domain, a working state discrimination result is generated and output together with the corrected current value.

Images