Electric locomotive converter grounding fault diagnosis method and device based on MLP

Abstract

The invention discloses an electric locomotive converter grounding fault diagnosis method and device based on MLP. Relates to the technical field of electric locomotive operation. The method mainly comprises the following steps: acquiring historical operation data of an electric locomotive converter, wherein the historical operation data comprises operation state data and operation condition data corresponding to the operation state data; constructing an electric locomotive converter grounding fault diagnosis model based on a multi-layer perceptron model, and training the electric locomotive converter grounding fault diagnosis model by taking the operation state characteristic data as the input of the model and taking the corresponding operation condition data as the output of the model; and inputting to-be-diagnosed operation state characteristic data into the trained electric locomotive converter grounding fault diagnosis model, and obtaining a diagnosis result output by the electric locomotive converter grounding fault diagnosis model. According to the method, historical data are collected to train the MLP model, a small number of features are screened and applied to online fault diagnosis, and therefore the grounding fault occurrence point can be located more accurately.

Description

Technical Field

[0001] The present invention relates to the technical field of electric locomotive operation and maintenance, and in particular to an MLP-based electric locomotive converter grounding fault diagnosis method and device. Background Art

[0002] The converter of an electric locomotive is one of the core components that ensures efficient and stable operation. However, ground faults, as a common type of electrical fault, seriously impact converter reliability and safety. Traditional ground fault diagnosis methods rely primarily on empirical rules and simple threshold judgments, and are capable of locating faults within a relatively large area, such as the four-quadrant side, the positive end of the intermediate DC circuit, the negative end of the intermediate DC circuit, and the inverter side. However, with the advancement of power electronics technology, the requirements for fault diagnosis accuracy are increasing, and traditional diagnostic methods are unable to meet practical needs. Accurately locating the fault point is crucial for reducing repair time and manpower and material costs, especially in complex operating conditions.

[0003] The currently commonly used fault diagnosis methods for electric locomotive converters mainly include:

[0004] Methods based on the rate of change of current or voltage. This type of method primarily relies on monitoring the rate of change of current and voltage at key nodes in the converter. By setting specific thresholds, when the detected changes exceed these thresholds, the system triggers an alarm, indicating that a ground fault has occurred. However, this method can only determine that the fault roughly occurs in a large area such as the four-quadrant side, the positive end of the intermediate DC circuit, the negative end of the intermediate DC circuit, and the inverter side, and cannot accurately locate specific circuit components. Furthermore, this method can only determine that the fault roughly occurs in a large area such as the four-quadrant side, the positive end of the intermediate DC circuit, the negative end of the intermediate DC circuit, and the inverter side, and cannot accurately locate specific circuit components. At the same time, this method has poor adaptability. For different types of electric locomotives or different operating conditions, the thresholds need to be readjusted. Moreover, because it relies on fixed threshold judgments, it does not respond promptly to rapidly changing faults.

[0005] Spectral analysis. Spectral analysis performs frequency-domain analysis on signals in the converter circuit to identify abnormal frequency components and thus determine whether a ground fault has occurred. However, this method requires high-performance data acquisition and processing equipment for spectrum analysis, increasing system cost and complexity. Furthermore, spectrum analysis is ineffective for non-periodic signal variations, potentially leading to missed detections or misjudgments. Furthermore, this method's high demand for computing resources to process large amounts of data in real time has limited its widespread adoption in practical applications.

[0006] Fault detection and isolation based on model predictive control (MPC). MPC is an advanced control strategy that predicts potential fault modes by establishing a mathematical model of the converter system and updating the model parameters based on real-time data. However, MPC involves complex mathematical modeling and algorithm design, making it difficult to implement, especially in a dynamic environment where the adjustment of model parameters is very difficult, and the debugging cycle is long. At the same time, MPC involves complex mathematical modeling and algorithm design, making it difficult to implement, especially in a dynamic environment where the adjustment of model parameters is very difficult. In addition, due to the changeable working environment of electric locomotives, it is difficult for the model to fully and accurately reflect the actual situation, which may lead to inaccurate prediction results. Summary of the Invention

[0007] In view of the shortcomings of the existing technology, the present invention provides a method and device for diagnosing ground faults in electric locomotive converters based on MLP. This method collects historical data to train an MLP model, screens a small number of features, and applies them to online fault diagnosis to achieve more accurate positioning of the ground fault.

[0008] The technical means adopted in the present invention are as follows:

[0009] A method for diagnosing ground faults in an electric locomotive converter based on MLP includes the following steps:

[0010] Acquiring historical operating data of the electric locomotive converter, wherein the historical operating data of the electric locomotive converter includes operating status data and operating condition data corresponding to the operating status data;

[0011] Performing time domain and frequency domain analysis on the operating status data, and calculating operating status characteristic data based on the analysis results;

[0012] Building an electric locomotive converter grounding fault diagnosis model based on a multi-layer perceptron model, using the operating state characteristic data as the model input and the corresponding operating condition data as the model output to train the electric locomotive converter grounding fault diagnosis model;

[0013] Acquiring the operating status data to be diagnosed in real time, and calculating the operating status characteristic data to be diagnosed based on the operating status data to be diagnosed;

[0014] The operating state characteristic data to be diagnosed is input into the trained electric locomotive converter grounding fault diagnosis model to obtain a diagnosis result output by the electric locomotive converter grounding fault diagnosis model.

[0015] Furthermore, the operating status data is analyzed in the time domain and the frequency domain respectively, and the operating status characteristic data is filtered based on the analysis results, including:

[0016] Calculating a characteristic value for each state quantity in the operating state data,

[0017] The permutation importance method is used to sort the eigenvalues, and the operating status characteristic data is determined according to the sorting results.

[0018] Furthermore, the operating status characteristic data includes: frequency variance of the ground voltage, frequency root mean square value of the ground voltage, phase difference between the ground voltage and the current at the same end of the four quadrants, and phase difference between the ground voltage and the U-phase current.

[0019] Furthermore, the electric locomotive converter grounding fault diagnosis model is trained by using the operating status characteristic data as the input of the model and the corresponding operating condition data as the output of the model, including converting the operating condition data into a vector form through one-hot encoding.

[0020] Furthermore, the operating status data includes the converter grounding voltage, the four-quadrant like-name terminal current and the U-phase current, and the operating condition data includes the normal working state and the four-quadrant like-name terminal grounding state, the four-quadrant opposite-name terminal grounding state, the intermediate circuit positive terminal grounding state, the intermediate circuit negative terminal grounding state, the inverter U-phase grounding state, the inverter V-phase grounding state and the inverter W-phase grounding state.

[0021] The present invention also discloses an MLP-based electric locomotive converter grounding fault diagnosis device for executing the above-mentioned fault diagnosis method, comprising:

[0022] A historical data acquisition unit, which is used to acquire historical operating data of the electric locomotive converter, wherein the historical operating data of the electric locomotive converter includes operating status data and operating condition data corresponding to the operating status data;

[0023] a historical data analysis unit, configured to perform time domain and frequency domain analysis on the operating status data, and calculate operating status characteristic data based on the analysis results;

[0024] a model building unit, which is used to build an electric locomotive converter grounding fault diagnosis model based on a multi-layer perceptron model, and train the electric locomotive converter grounding fault diagnosis model by taking the operating state characteristic data as the input of the model and taking the corresponding operating condition data as the output of the model;

[0025] A real-time data acquisition unit, which is used to acquire the operating status data to be diagnosed in real time and calculate the operating status characteristic data to be diagnosed based on the operating status data to be diagnosed;

[0026] The fault diagnosis unit is used to input the operating state characteristic data to be diagnosed into the trained electric locomotive converter grounding fault diagnosis model to obtain the diagnosis result output by the electric locomotive converter grounding fault diagnosis model.

[0027] The present invention also discloses an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the above-mentioned MLP-based electric locomotive converter grounding fault diagnosis method through the computer program.

[0028] Compared with the prior art, the present invention has the following advantages:

[0029] By introducing the MLP model, this invention not only improves the accuracy of ground fault diagnosis but also reduces the false alarm rate. Furthermore, thanks to the use of a machine learning algorithm, the system possesses self-learning capabilities, continuously improving diagnostic accuracy over time. This will significantly enhance the safety and reliability of electric locomotive operation while effectively reducing maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0031] Figure 1 This is a flow chart of a method for diagnosing grounding faults in an electric locomotive converter based on MLP according to the present invention.

[0032] Figure 2 Schematic diagram of the main circuit of the traction converter in the embodiment.

[0033] Figure 3 Flowchart of the fault diagnosis solution based on MLP in the embodiment.

[0034] Figure 4 This is a specific execution flow of fault diagnosis using the method of the present invention in an embodiment.

[0035] Figure 5 This is a ranking diagram of the importance of alternative features in the embodiment. DETAILED DESCRIPTION

[0036] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described 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 should fall within the scope of protection of the present invention.

[0037] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0038] like Figure 1-4 As shown, the embodiment of the present application discloses a method for diagnosing ground faults of electric locomotive converters based on MLP, comprising the following steps:

[0039] S1. Acquire historical operating data of an electric locomotive converter, the historical operating data of the electric locomotive converter including operating status data and operating condition data corresponding to the operating status data. In an embodiment of the present application, the operating status data includes the converter ground voltage, the four-quadrant same-name terminal current, and the U-phase current. The operating condition data includes normal operating status and the four-quadrant same-name terminal grounding state, the four-quadrant opposite-name terminal grounding state, the intermediate circuit positive terminal grounding state, the intermediate circuit negative terminal grounding state, the inverter U-phase grounding state, the inverter V-phase grounding state, and the inverter W-phase grounding state.

[0040] Specifically, if Figure 2 The figure shows the main circuit diagram of the traction converter of an electric locomotive. The present invention locates the location of the ground fault to the grounding of the same-name terminal of the four-quadrant, the grounding of the opposite-name terminal of the four-quadrant, the grounding of the positive terminal of the intermediate circuit, the grounding of the negative terminal of the intermediate circuit, the grounding of the U phase of the inverter, the grounding of the V phase of the inverter, and the grounding of the W phase of the inverter. The ground voltage U is collected when the main circuit of the traction converter is in normal working state and when a ground fault occurs at each point (hereinafter referred to as each fault condition). G 、The four-quadrant same-name terminal current i A and U phase current i UAs the data source, the historical operation data of the electric locomotive converter is constructed.

[0041] Furthermore, to train and test the neural network model, each fault condition is defined as a fault label. To eliminate the misleading tendency of increasing numerical values ​​in the fault labels and ensure mutual exclusivity between categories, the category labels are converted to vector form using one-hot encoding, as shown in Table 1. This simplifies the loss function calculation and effectively handles multi-class classification problems. After removing non-numeric data, the historical operating data set is divided into training, test, and validation sets.

[0042] Table 1 Fault label definition and one-hot encoding

[0043]

[0044] S2. Perform time domain and frequency domain analysis on the operating status data, and calculate operating status characteristic data based on the analysis results.

[0045] Specifically, after obtaining the category labels of the historical operation data, the operation status data must be analyzed and processed. This application preferably divides the data within 1 second after the fault occurs into 50 time windows at every 0.02 seconds, and calculates the characteristic values ​​under each fault condition in each time window, including the maximum value, mean, root mean square value, skewness value, pulse index, kurtosis value in the time domain, center of gravity frequency, frequency root mean square, average frequency, frequency variance, U in the frequency domain, and the eigenvalues ​​of ... G with i A Phase difference, U G with i U phase difference.

[0046] Time-domain features provide information about a signal's temporal behavior and characteristics, such as its overall level and the sharpness of its distribution. Frequency-domain features provide information about a signal's frequency components, such as its primary frequency components and the degree of its distribution. To more comprehensively analyze signal characteristics, feature engineering analyzes both time-domain and frequency-domain features to generate candidate features.

[0047] Because there are significant numerical differences between features, they are normalized. Considering that there are negative numbers in the feature values, the maximum normalization method is used to return the feature values ​​to [-1,1]. In order to further complete the screening of the constructed alternative features, the "Permutation Importance" method is used to complete the feature importance ranking. The permutation importance method is to randomly disrupt the numerical order of each alternative feature in the created multi-class classification ECOC model. If the model accuracy drops significantly, it means that the feature is very important to the model prediction. The importance ranking of the alternative features is obtained as follows Figure 5As shown, it can effectively identify the "contribution" of each feature to the model prediction, and select the top four feature quantities as the input of MLP. Here, U is selected G Frequency variance, U G with i A Phase difference, U G Frequency RMS value, U G with i U The phase difference is taken as input.

[0048] S3. Construct an electric locomotive converter grounding fault diagnosis model based on a multi-layer perceptron model, and train the electric locomotive converter grounding fault diagnosis model by taking the operating status characteristic data as the input of the model and the corresponding operating condition data as the output of the model.

[0049] The electric locomotive converter grounding fault diagnosis model of the present application preferably adopts a multilayer perceptron (MLP) type of neural network, which is a feedforward artificial neural network model. It can capture the complex nonlinear relationship in the input data through multi-layer nonlinear transformation, avoid the neural network model from believing that there is a progressive relationship between fault labels, and convert the category label into a vector form through one-hot encoding to effectively handle multi-class classification problems.

[0050] Furthermore, during the model training process, regularization techniques are used to optimize hyperparameters, and early stopping is used to monitor model performance in real time to prevent overfitting. Specifically: Regularization is a key strategy used to limit model complexity and avoid overfitting of training data. By imposing constraints on model parameters, regularization not only helps to reduce the risk of overfitting, but also improves the generalization ability of the model. In this application, an implicit random regularization constraint is imposed on the weight parameters of the fully connected layer by introducing the Dropout mechanism during neural network training. This mechanism randomly "turns off" some neurons with a certain probability p (usually set to 0.5) in each training iteration, thereby forcing the network to learn more robust and generalized feature representations. Its mathematical expression is as follows:

[0051] Assume that the output of the neurons in the lth layer is:

[0052] a (l) =f(W (l) a (l-1) +b (l) )

[0053] After introducing Dropout, the output becomes:

[0054] where ξ i ~Bernoulli(p)

[0055] where ξ is a Bernoulli random vector of the same dimension as a(l), which simulates the random inactivation of neurons. This mechanism is equivalent to applying a structured regularization to the weight matrix W, preventing it from over-adapting to noise or local patterns in the training data, thereby improving the generalization performance of the model.

[0056] In this invention, the neural network uses the cross-entropy loss function (Cross-Entropy Loss), which is suitable for multi-classification tasks. The multi-class cross-entropy loss function is defined as follows:

[0057] Let the true label be the one-hot encoded vector y∈{0,1} C , the probability distribution of the model output is Then the cross entropy loss is:

[0058]

[0059] in:

[0060] C: Total number of categories.

[0061] y c : Whether the true label belongs to the cth category.

[0062] The probability of class c predicted by the model.

[0063] This is the default loss function used by MATLAB and is applicable to classificationLayer, so no additional definition is required.

[0064] In this invention, although L1 or L2 regularization terms are not explicitly added to the loss function, indirect control of model complexity is achieved through the following methods:

[0065] (1) Use the Adam optimizer to update parameters and take advantage of its adaptive learning rate feature to prevent some parameters from being updated too much.

[0066] (2) Set the maximum number of training rounds (epochs) to 200 to prevent the model from overfitting the training data.

[0067] (3) Introduce a validation set monitoring mechanism to stop training when the validation set performance no longer improves, thus achieving early stopping. In MATLAB, this mechanism is implemented by setting the net.trainParam.max_fail parameter. Setting this parameter to 10 means that if the error on the validation set does not decrease for 10 consecutive epochs, training will automatically stop.

[0068] (4) A Dropout layer (with a dropout rate of 0.5) is introduced into the neural network to randomly mask the hidden layer neurons and enhance the generalization ability of the model.

[0069] (5) The original data is divided into training set and test set in a ratio of 7:3 to ensure that the model has sufficient training samples while retaining reliable verification data.

[0070] This approach ensures that the model can not only achieve good performance on the training set, but also maintain stable prediction performance on unseen data, thereby enhancing the generalization ability of the model.

[0071] Furthermore, this application uses the MBD model to use s-function to compile the trained neural network model into C language, burn the header file and source file into the chip to complete the model deployment. Through this feasible model deployment solution, MLP can realize real-time diagnosis function in the converter control system.

[0072] S4. Acquire the operating status data to be diagnosed in real time, and calculate the operating status characteristic data to be diagnosed based on the operating status data to be diagnosed.

[0073] S5. Input the operating state characteristic data to be diagnosed into the trained electric locomotive converter grounding fault diagnosis model to obtain a diagnosis result output by the electric locomotive converter grounding fault diagnosis model.

[0074] This application uses historical data to construct and filter features through importance sorting to build an MLP model suitable for converter ground fault diagnosis. This model achieves accurate fault location from the macroscopic area to the specific port level, that is, expanding the "ground fault occurrence point diagnosis to the four-quadrant side, the intermediate DC circuit positive end, the intermediate DC circuit negative end, and the inverter side" to diagnosing the ground fault occurrence point to the four-quadrant same-name end, the four-quadrant opposite-name end, the intermediate DC circuit positive end, the intermediate DC circuit negative end, the inverter side U phase, the inverter side V phase, and the inverter side W phase.

[0075] The present invention also discloses an MLP-based electric locomotive converter grounding fault diagnosis device for executing the above-mentioned fault diagnosis method, comprising:

[0076] A historical data acquisition unit, which is used to acquire historical operating data of the electric locomotive converter, wherein the historical operating data of the electric locomotive converter includes operating status data and operating condition data corresponding to the operating status data;

[0077] a historical data analysis unit, configured to perform time domain and frequency domain analysis on the operating status data, and calculate operating status characteristic data based on the analysis results;

[0078] a model building unit, which is used to build an electric locomotive converter grounding fault diagnosis model based on a multi-layer perceptron model, and train the electric locomotive converter grounding fault diagnosis model by taking the operating state characteristic data as the input of the model and taking the corresponding operating condition data as the output of the model;

[0079] A real-time data acquisition unit, which is used to acquire the operating status data to be diagnosed in real time and calculate the operating status characteristic data to be diagnosed based on the operating status data to be diagnosed;

[0080] The fault diagnosis unit is used to input the operating state characteristic data to be diagnosed into the trained electric locomotive converter grounding fault diagnosis model to obtain the diagnosis result output by the electric locomotive converter grounding fault diagnosis model.

[0081] The present invention also discloses an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the above-mentioned MLP-based electric locomotive converter grounding fault diagnosis method through the computer program.

[0082] The serial numbers of the above embodiments of the present invention are for description only and do not represent the advantages or disadvantages of the embodiments.

[0083] In the above embodiments of the present invention, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0084] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only exemplary. For example, the division of the units can be a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0085] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0086] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0087] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk, etc. Various media that can store program codes.

[0088] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for diagnosing ground faults in electric locomotive converters based on MLP, characterized in that: The following steps are involved: Acquiring historical operating data of the electric locomotive converter, wherein the historical operating data of the electric locomotive converter includes operating status data and operating condition data corresponding to the operating status data; Performing time domain and frequency domain analysis on the operating status data, and calculating operating status characteristic data based on the analysis results; Building an electric locomotive converter grounding fault diagnosis model based on a multi-layer perceptron model, using the operating state characteristic data as the model input and the corresponding operating condition data as the model output to train the electric locomotive converter grounding fault diagnosis model; Acquiring the operating status data to be diagnosed in real time, and calculating the operating status characteristic data to be diagnosed based on the operating status data to be diagnosed; The operating state characteristic data to be diagnosed is input into the trained electric locomotive converter grounding fault diagnosis model to obtain a diagnosis result output by the electric locomotive converter grounding fault diagnosis model.

2. The MLP-based electric locomotive converter ground fault diagnosis method according to claim 1, characterized in that: Performing time domain and frequency domain analysis on the operating status data, and filtering the operating status characteristic data based on the analysis results, including: Calculating a characteristic value for each state quantity in the operating state data, The permutation importance method is used to sort the eigenvalues, and the operating status characteristic data is determined according to the sorting results.

3. The MLP-based electric locomotive converter ground fault diagnosis method according to claim 2, characterized in that: The operating status characteristic data include: frequency variance of the ground voltage, frequency root mean square value of the ground voltage, phase difference between the ground voltage and the current of the same name terminal in the four quadrants, and phase difference between the ground voltage and the U-phase current.

4. The method for diagnosing ground fault of an electric locomotive converter based on MLP according to claim 1, characterized in that: The electric locomotive converter grounding fault diagnosis model is trained by using the operating state characteristic data as the input of the model and the corresponding operating condition data as the output of the model, including converting the operating condition data into a vector form through one-hot encoding.

5. The method for diagnosing ground fault of an electric locomotive converter based on MLP according to claim 1, characterized in that: The operating status data includes the converter grounding voltage, the four-quadrant like-name terminal current and the U-phase current, and the operating condition data includes the normal operating state and the four-quadrant like-name terminal grounding state, the four-quadrant opposite-name terminal grounding state, the intermediate circuit positive terminal grounding state, the intermediate circuit negative terminal grounding state, the inverter U-phase grounding state, the inverter V-phase grounding state and the inverter W-phase grounding state.

6. An MLP-based electric locomotive converter ground fault diagnosis device, used to execute the fault diagnosis method according to claim 1, characterized in that: include: A historical data acquisition unit, which is used to acquire historical operating data of the electric locomotive converter, wherein the historical operating data of the electric locomotive converter includes operating status data and operating condition data corresponding to the operating status data; a historical data analysis unit, configured to perform time domain and frequency domain analysis on the operating status data, and calculate operating status characteristic data based on the analysis results; a model building unit, which is used to build an electric locomotive converter grounding fault diagnosis model based on a multi-layer perceptron model, and train the electric locomotive converter grounding fault diagnosis model by taking the operating state characteristic data as the input of the model and taking the corresponding operating condition data as the output of the model; A real-time data acquisition unit, which is used to acquire the operating status data to be diagnosed in real time and calculate the operating status characteristic data to be diagnosed based on the operating status data to be diagnosed; The fault diagnosis unit is used to input the operating state characteristic data to be diagnosed into the trained electric locomotive converter grounding fault diagnosis model to obtain the diagnosis result output by the electric locomotive converter grounding fault diagnosis model.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: The processor executes the MLP-based electric locomotive converter grounding fault diagnosis method according to any one of claims 1 to 5 by running the computer program.

Images