Vehicle network resonance characteristic identification method and system based on DBM-SGMD

Abstract

The invention discloses a DBM-SGMD-based train network resonance characteristic identification method and system, and relates to the technical field of new energy power generation control, and the method comprises the steps: collecting target train network system data, and carrying out the preprocessing; constructing a filter bank by adopting a least square method, and obtaining optimal vehicle network resonance voltage and current data; calculating an embedded dimension by using a symplectic geometry method, and reconstructing symplectic geometry components; and classifying the symplectic geometric components through a Gaussian depth Boltzmann machine. According to the method, the vehicle network resonance phenomenon is identified through the training model, and the identification accuracy and efficiency can be improved. The deep learning algorithm can extract more complex features through multi-level data processing, thereby improving the identification capability of vehicle network resonance. Vehicle network resonance identification is not only a key for ensuring safe and stable operation of an electrified railway system, but also an important link for improving the power supply quality of a traction power supply system.

Description

Technical Field

[0001] The present invention relates to the technical field of new energy power generation control, and specifically to a method and system for identifying the resonance characteristics of a vehicle-grid based on DBM-SGMD. Background Art

[0002] In modern electrified railway systems, ensuring the safe and stable operation of electric locomotives is of crucial importance. The electrified railway system consists of a traction power supply system and electric locomotives to form a complex vehicle-grid coupling system. The stability of this system is directly related to the safety and operational reliability of the railway. Therefore, it is particularly important to quickly and accurately identify abnormal electrical phenomena in the vehicle-grid system. The stability of the vehicle-grid coupling system is affected by various factors, including the dynamic behavior of the electrical system, load changes, and the operating state of electric locomotives. In actual operation, due to the randomness and uncertainty of electrical phenomena, these factors may lead to resonance. These phenomena not only affect the safety of electrified railways but also pose a threat to the power supply quality of the traction power supply system. Resonance occurs when the harmonic frequencies present in the system match the natural frequencies of the electric locomotives, thereby triggering a resonance phenomenon. This phenomenon increases the current and voltage in the system, thus affecting the performance and lifespan of equipment.

[0003] Although these typical abnormal electrical phenomena have been relatively widely studied and understood, in actual vehicle-grid coupling systems, there is still a lack of effective methods for identifying vehicle-grid system resonance. By identifying and preventing resonance characteristic phenomena, the safe operation of the pantograph-catenary and power supply quality are ensured. Therefore, in-depth research and analysis are required.

[0004] In recent years, the rapid development of machine learning and deep learning technologies has provided new tools for the identification of abnormal electrical phenomena. These technologies can analyze large amounts of data to discover and identify abnormal patterns in the system. Traditional methods for identifying abnormal electrical phenomena often can only identify known abnormal phenomena. When the input is data of unknown abnormal electrical phenomena, it may misclassify them as one of the known abnormal phenomenon categories, resulting in misjudgment. In this case, the accuracy and reliability of traditional methods are limited.

[0005] To solve this problem, the present invention proposes to introduce advanced machine learning combined with resonance characteristic identification methods and apply them to vehicle-grid resonance identification. By training a model to identify vehicle-grid resonance phenomena, it helps to improve the accuracy and efficiency of identification. Deep learning algorithms can extract more complex features through multi-level data processing, thereby enhancing the ability to identify vehicle-grid resonance. In short, vehicle-grid resonance identification is not only the key to ensuring the safe and stable operation of electrified railway systems but also an important link in improving the power supply quality of the traction power supply system. Summary of the Invention

[0006] In view of the above problems, the present invention is proposed.

[0007] Therefore, the technical problem to be solved by the present invention is: how to accurately identify the resonance characteristics of the vehicle-network system in the electrified railway system and avoid the impact of resonance on the system safety and power supply quality.

[0008] To solve the above technical problem, the present invention provides the following technical solution: a method for identifying the resonance characteristics of the vehicle-network based on DBM-SGMD, which includes the following steps

[0009] Collect the data of the vehicle-network system of the target train and perform preprocessing;

[0010] Construct a filter bank by using the least squares method to obtain the optimal vehicle-network resonance voltage and current data;

[0011] Use the symplectic geometry method to calculate the embedding dimension and reconstruct the symplectic geometry component;

[0012] Classify the symplectic geometry components through the Gaussian deep Boltzmann machine.

[0013] As a preferred solution of the method for identifying the resonance characteristics of the vehicle-network based on DBM-SGMD according to the present invention, wherein: the data of the vehicle-network system of the target train includes obtaining the time-series data of voltage and current;

[0014] The preprocessing is to divide the voltage and current data recorded by the vehicle-network system of the target train within the same acquisition period into time windows of the same time length according to time, and the expression is:

[0015] [δ] = [δ 1 , δ 2 ,..., δ lw×fs l w×fs

[0016] wherein, lw represents the window length, f is the sampling frequency, and δ is the sampling data sample.

[0017] As a preferred solution of the method for identifying the resonance characteristics of the vehicle-network based on DBM-SGMD according to the present invention, wherein: the step of constructing a filter bank by using the least squares method to obtain the optimal vehicle-network resonance voltage and current data includes

[0018] Introduce an adaptive triplet half-band filter, and construct a filter bank by using the least squares method to reduce the error caused by the passband and stopband to the data and obtain the optimal vehicle-network resonance voltage and current data. The expression is:

[0019]

[0020] wherein, S o (s), S1 (s), S 2 (s) represents the kernel obtained by the half-band filter, 1 / 2(1 + S o (s)), 1 / 2(1 + S 1 (s)), 1 / 2(1 + S 2 (s)) is the half-band filter.

[0021] As a preferred solution of a method for identifying the resonance characteristics of a vehicle-grid based on DBM - SGMD according to the present invention, wherein: calculating the embedding dimension using the symplectic geometry method, and reconstructing the symplectic geometry components includes,

[0022] Taking the optimal vehicle-grid resonance voltage and current data as the embedding dimension of the symplectic geometry method, reconstructing the symplectic geometry components using the normalized mean square error, and establishing a symplectic geometry model of the pantograph-catenary system. The expression is:

[0023]

[0024] Among them, p represents the embedding dimension of the reconstruction space, r is the time delay, and the number of points in the p-dimensional reconstructed attractor is represented by g = n - (p - 1).

[0025] As a preferred solution of a method for identifying the resonance characteristics of a vehicle-grid based on DBM - SGMD according to the present invention, wherein: classifying the symplectic geometry components through a Gaussian deep Boltzmann machine includes,

[0026] Classifying the vehicle-grid resonance components reconstructed by symplectic geometry through the pipeline of a Gaussian deep Boltzmann machine, taking the distribution on the binary vector [0, 1] as the feature, and modeling and training the real data as a probability. The expression is:

[0027]

[0028] Among them, W ij is used to describe the correlation between the i-th node in the visible layer and the j-th node in the hidden layer, α i represents the bias of the i-th node in the visible layer, b j is the bias of the j-th node in the hidden layer. The visible layer is composed of N k binary random variables, denoted as vector k, and the hidden layer is composed of N y binary random variables, denoted as vector y. The variable ki corresponds to the data of the i-th node in the visible layer, and the variable y i corresponds to the data of the j-th node in the hidden layer;

[0029] After determining E(k, y), the multi-dimensional probability distribution between the two layers is obtained. The expression is:

[0030]

[0031] Z θ = ∑ k,y e -Eθ(k,y)

[0032] Among them, Z θ is the partition function.

[0033] As a preferred solution of a method for identifying the resonance characteristics of a vehicle-grid based on DBM-SGMD according to the present invention, wherein: the Gaussian deep Boltzmann machine includes,

[0034] The Gaussian deep Boltzmann machine is composed of a visible layer and a hidden layer. The upper layer is the hidden layer, which has N y hidden units; the lower layer represents the visible layer, which has N k visible units.

[0035] As a preferred solution of a method for identifying the resonance characteristics of a vehicle-grid based on DBM-SGMD according to the present invention, wherein: the classification of the symplectic geometric components by the Gaussian deep Boltzmann machine further includes,

[0036] Collect vehicle-grid resonance data and input it into the Gaussian deep Boltzmann machine model. Select the oscillation frequency of the vehicle-grid resonance, set the number of layers of the Gaussian deep Boltzmann machine and the number of neurons in each layer. Use an unsupervised learning algorithm to train the Gaussian deep Boltzmann machine model, optimize the weights of the Gaussian deep Boltzmann machine model, perform supervised fine-tuning using labeled data, evaluate the performance of the model through methods such as cross-validation, check the accuracy and recall rate indicators, deploy the trained Gaussian deep Boltzmann machine model to actual applications, and perform frequency screening.

[0037] Another object of the present invention is to provide a vehicle-grid resonance characteristic identification system, which can automatically extract resonance characteristics and classify and identify abnormal signals through multi-level data processing and model training, and solves the problem of inaccurate identification of existing methods when dealing with dynamic and complex electrical phenomena.

[0038] To solve the above technical problems, the present invention provides the following technical solution: a vehicle-grid resonance characteristic identification system, including: a data acquisition and preprocessing module, a filter design and data optimization module, a symplectic geometry calculation and feature reconstruction module, and a classification and model training module;

[0039] The data acquisition and preprocessing module collects the required voltage and current data from the vehicle-grid system of the target train, performs preprocessing, records the data within the same acquisition period, and divides it into windows with the same time length to form standardized time series data;

[0040] The filter design and data optimization module processes the preprocessed time series data through an adaptive triplet half-band filter, constructs a filter bank using the least squares method to reduce the errors in the passband and stopband, and optimizes the obtained vehicle-grid resonance voltage and current data;

[0041] The symplectic geometry calculation and feature reconstruction module calculates the embedding dimension of the vehicle-grid resonance through the symplectic geometry method and reconstructs the symplectic geometry components using the normalized mean square error;

[0042] The classification and model training module uses DBM to perform binary distribution modeling and training on the symplectic geometry components, optimizes the weights of the model through unsupervised learning algorithms, and performs fine-tuning through supervised learning algorithms. Finally, an optimized classification model is formed. Through the trained model, the oscillation frequency of the vehicle-grid resonance characteristics is identified and applied to the resonance frequency screening of the actual vehicle-grid system.

[0043] A computer device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps of a vehicle-grid resonance feature identification method based on DBM-SGMD as described above are implemented.

[0044] A computer-readable storage medium stores a computer program thereon, and when the computer program is executed by a processor, the steps of a vehicle-grid resonance feature identification method based on DBM-SGMD as described above are implemented.

[0045] Advantages of the present invention: Through the acquisition and preprocessing of the voltage and current data of the vehicle-grid system, the present invention constructs a filter bank to optimize the data, calculates the embedding dimension using the symplectic geometry method and reconstructs the symplectic geometry components, and then performs classification and model training through the Gaussian deep Boltzmann machine (DBM), realizing the accurate identification and screening of the vehicle-grid resonance characteristics. The present invention not only improves the accuracy and reliability of identification through deep learning technology, effectively solves the problem of easy misjudgment of traditional methods when dealing with unknown resonance characteristics, but also reduces noise interference, enhances the ability to capture non-linear dynamic characteristics, and ensures the safe operation of the electrified railway system and the stability of the power supply quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts. Among them:

[0047] Figure 1 It is the overall flowchart of a vehicle-grid resonance feature identification method based on DBM-SGMD provided by the first embodiment of the present invention;

[0048] Figure 2 The structural diagram of DBM - SGMD in a method for identifying the resonance characteristics of vehicle - grid based on DBM - SGMD provided by the first embodiment of the present invention;

[0049] Figure 3 The decomposition flowchart of the symplectic geometric algorithm in a method for identifying the resonance characteristics of vehicle - grid based on DBM - SGMD provided by the first embodiment of the present invention;

[0050] Figure 4 The model diagram of the Gaussian Boltzmann machine in a method for identifying the resonance characteristics of vehicle - grid based on DBM - SGMD provided by the first embodiment of the present invention;

[0051] Figure 5 The comparison of execution costs of a method for identifying the resonance characteristics of vehicle - grid based on DBM - SGMD provided by the third embodiment of the present invention under different numbers of tasks;

[0052] Figure 6 The modal decomposition diagram of a method for identifying the resonance characteristics of vehicle - grid based on DBM - SGMD provided by the third embodiment of the present invention;

[0053] Figure 7 The oscillation frequency identification diagram of a method for identifying the resonance characteristics of vehicle - grid based on DBM - SGMD provided by the third embodiment of the present invention. Detailed implementation manners

[0054] To make the above - mentioned objects, features, and advantages of the present invention more obvious and understandable, the following detailed description of the specific implementation manners of the present invention will be given in conjunction with the accompanying drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0055] Embodiment 1, referring to Figures 1 to 4 An embodiment of the present invention provides a method for identifying the resonance characteristics of vehicle - grid based on DBM - SGMD, including:

[0056] First, obtain the voltage and current time - series data recorded by the target train vehicle - grid system within the same acquisition period. Through pre - processing, the obtained data is segmented into time windows with a time length of 8 seconds, and a one - dimensional window with a time step is retained for each of the above window lengths.

[0057] Introduce an adaptive triplet half - band filter, and construct a filter bank using the least - squares method to reduce the errors caused by the passband and stopband to the data, and obtain the optimal vehicle - grid resonance voltage and current data

[0058] The embedding dimension of the nonlinear vehicle-grid resonance voltage and current time series is estimated by the symplectic geometry method (SGMD), and the symplectic geometry components are reconstructed using the normalized mean square error for the classification of vehicle-grid resonance signals.

[0059] Finally, the symplectic geometry components are classified through the pipeline of the Gaussian deep Boltzmann machine (DBM), characterized by the distribution on the binary vector [0, 1], allowing the real data to be modeled and trained as probabilities to identify the oscillation characteristics of vehicle-grid resonance.

[0060] Preprocess the voltage and current time series data recorded by the target train vehicle-grid system during the same acquisition period.

[0061] [δ] = [δ 1 , δ 2 ,..., δ lw×fs lw×fs (1)

[0062] where: l w represents the window length in seconds, f s is the sampling frequency in Hz, and δ is the sampling data sample.

[0063] Refer to Figure 2 , the model structure diagram of DBM-SGMD, and the adaptive triplet half-band filter structure for reducing passband and stopband errors constructed based on the least squares method is designed as follows:

[0064]

[0065] where S o (s), S 1 (s), S 2 (s) represent the kernels obtained by the half-band filter, and 1 / 2(1 + S o (s)), 1 / 2(1 + S 1 (s)), 1 / 2(1 + S 2 (s)) are the half-band filters.

[0066] where T 1 (s) = S -1 T 0 (-s), U 1 (s) = SU 0 (-s).

[0067] Use the symplectic geometry method (SGMD) to estimate the embedding dimension of the nonlinear vehicle-grid resonance voltage and current time series, refer to Figure 3 ​Reconstruct the symplectic geometric components using the normalized mean square error to establish a symplectic geometric model of the pantograph-catenary system, which usually involves describing the dynamic behavior and equations of the pantograph-catenary system. Linearize the pantograph-catenary system to transform the nonlinear system into a linear system. Select an appropriate symplectic operator for decomposition to ensure that the system maintains its symplectic property in numerical solutions. Apply the symplectic geometric algorithm to decompose the Hamiltonian of the system to reveal the characteristics of pantograph-catenary resonance. Analyze the frequency and mode of pantograph-catenary resonance through the decomposition results to obtain the characteristics of voltage and current time series data. Use these sub-components to reconstruct the signal to evaluate its effectiveness and the ability to retain the basic characteristics of the signal. The symplectic geometric model composed of the original data is as follows:

[0068]

[0069] Where p represents the embedding dimension of the reconstruction space. The time delay is represented by r, and the number of points in the p-dimensional reconstructed attractor is represented by g = n - (p - 1).

[0070] Classify the symplectic geometric reconstructed pantograph-catenary resonance components through the pipeline of the Gaussian deep Boltzmann machine (DBM), characterized by the distribution on the binary vector [0, 1], allowing real data to be modeled and trained as probabilities. The DBM consists of a visible layer and a hidden layer. Refer to Figure 4 In the Gaussian Boltzmann machine model, the upper layer is the hidden layer with Ny hidden units; the lower layer represents the visible layer with Nk visible units. Any two visible units or any two hidden units in the model do not affect each other, and thus are "restricted". The model of the Gaussian deep Boltzmann machine is as follows:

[0071]

[0072] In the formula: W ij is used to describe the correlation between the i-th node in the visible layer and the j-th node in the hidden layer, α i represents the bias of the i-th node in the visible layer, and b j is the bias of the j-th node in the hidden layer. The visible layer is composed of N k binary random variables, denoted as vector k; the hidden layer is composed of N y binary random variables, denoted as vector y; the variable ki corresponds to the data of the i-th node in the visible layer, and the variable y i corresponds to the data of the j-th node in the hidden layer. After determining E(k, y), the multi-dimensional probability distribution between the two layers is immediately obtained, as follows:

[0073]

[0074] In the formula: Z θ = ∑ k,y e -Eθ(k,y)is the partition function, or is also known as the normalization factor.

[0075] Classify the pipeline of the symplectic geometric feature components into the Gaussian deep Boltzmann machine. Screen the characteristic parameters of the vehicle network resonance.

[0076] Example 2, an embodiment of the present invention, provides a system for a vehicle network resonance feature identification method based on DBM-SGMD, including: a data acquisition and preprocessing module, a filter design and data optimization module, a symplectic geometry calculation and feature reconstruction module, and a classification and model training module;

[0077] The data acquisition and preprocessing module collects the required voltage and current data from the target train vehicle network system, performs preprocessing, records the data within the same acquisition period, and divides it into windows with the same time length to form standardized time series data;

[0078] The filter design and data optimization module processes the preprocessed time series data through an adaptive triplet half-band filter, constructs a filter bank using the least squares method to reduce the errors in the passband and stopband, and optimizes the obtained vehicle network resonance voltage and current data;

[0079] The symplectic geometry calculation and feature reconstruction module calculates the embedding dimension of the vehicle network resonance through the symplectic geometry method, and reconstructs the symplectic geometry components using the normalized mean square error;

[0080] The classification and model training module uses DBM to perform binary distribution modeling and training on the symplectic geometry components, optimizes the weights of the model through an unsupervised learning algorithm, and performs fine-tuning through a supervised learning algorithm. Finally, an optimized classification model is formed. Through the trained model, the oscillation frequency of the vehicle network resonance feature is identified and applied to the resonance frequency screening of the actual vehicle network system.

[0081] If the described function 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, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. And the aforementioned storage medium includes: USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs and other various media that can store program codes.

[0082] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or used in conjunction with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0083] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection part (electronic device) having one or more wirings, a portable computer disk cartridge (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or otherwise processing as appropriate, and then stored in a computer memory.

[0084] It should be understood that various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well-known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0085] Example 3, referring to Figures 5 to 7 is another embodiment of the present invention. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments. In this embodiment, experiments are respectively carried out on the existing traditional method and the method of this embodiment.

[0086] By analyzing the actual measurement data of vehicle-grid resonance in a certain place in Yunnan by using the method proposed in this method, it can be verified that the algorithm proposed in this scheme can effectively separate the fundamental frequency component and the vehicle-grid resonance component, and effectively obtain their frequency and amplitude characteristics, providing more data support for the next step of research. Figure 5It is a measured waveform diagram containing the vehicle-network resonance component. First, through the preprocessing proposed by the present invention, the measured data is denoised to remove interferences such as noise during data collection, and a relatively smooth optimal vehicle-network resonance data waveform diagram is obtained, as Figure 6 shown.

[0087] Then, a symplectic geometry calculation and feature reconstruction module is introduced. The optimal vehicle-network resonance data is reconstructed into symplectic geometry components through normalized mean square error, and a symplectic geometry model of the pantograph-catenary system is established. Through symplectic geometry similarity transformation, the corresponding symplectic geometry components are extracted, such as the fundamental frequency symplectic geometry component, the vehicle-network resonance symplectic geometry component, and the symplectic geometry components of other frequencies.

[0088] Finally, the above-mentioned feature components such as the fundamental frequency symplectic geometry component, the vehicle-network resonance symplectic geometry component, and the symplectic geometry components of other frequencies are classified through the pipeline of a Gaussian deep Boltzmann machine. Binary distribution modeling and training are performed on the symplectic geometry components. The weights of the model are optimized through unsupervised learning algorithms, and fine-tuning is performed through supervised learning algorithms. Finally, an optimized classification model is formed. Through the trained model, the oscillation frequency of the vehicle-network resonance feature is identified, and the separated fundamental frequency and resonance waveforms are as Figure 7 shown. It can be seen from the figure that the oscillation frequency of the vehicle-network resonance is relatively high, which is significantly different from the fundamental wave. The detailed data is shown in Table 1.

[0089] Table 1 Data Display Table

[0090] Vehicle network resonance mode Frequency (Hz) Amplitude (kV) Fundamental frequency component 50 36.49 Resonance component 2100 4.47

[0091] Table 1 records the results of the vehicle-network resonance characteristics identified by the method proposed by the present invention. It includes the frequency and amplitude of the fundamental wave, as well as the frequency and amplitude characteristics of the vehicle-network resonance characteristics. From the overall data, the algorithm proposed by the present invention can well distinguish the fundamental frequency component and the resonance component, and can provide relevant data support, providing an effective research basis for further studying the suppression scheme of vehicle-network resonance.

[0092] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A vehicle-grid resonance feature identification method based on DBM-SGMD, characterized in that: include: Collect target train network system data and perform pre-processing; The least square method is used to construct a filter bank to obtain the optimal vehicle-grid resonant voltage and current data; Use symplectic geometry methods to calculate the embedding dimension and reconstruct the symplectic geometry components; The symplectic geometric components are classified by Gaussian Deep Boltzmann Machine.

2. The vehicle-grid resonance feature identification method based on DBM-SGMD as claimed in claim 1, characterized in that: The target train network system data includes obtaining time series data of voltage and current; The preprocessing is to divide the voltage and current data recorded by the target train network system in the same acquisition period into time windows of the same time length according to time division, and the expression is: [δ]=[δ1,δ2,...,δ lw×fs ] lw×fs Among them, lw represents the window length, f is the sampling frequency, and δ is the sampled data sample.

3. The vehicle-grid resonance feature identification method based on DBM-SGMD as claimed in claim 2, characterized in that: The least square method is used to construct a filter bank to obtain the optimal vehicle network resonant voltage and current data, including: An adaptive triplet half-band filter is introduced, and the filter bank is constructed using the least squares method to reduce the error caused by the passband and stopband to the data and obtain the optimal vehicle-grid resonant voltage and current data. The expression is: Among them, S o (s), S1(s), S2(s) represent the kernels obtained by half-band filters, 1 / 2(1+S o (s)), 1 / 2(1+S1(s)), and 1 / 2(1+S2(s)) are half-band filters.

4. The vehicle-grid resonance feature identification method based on DBM-SGMD as claimed in claim 3, characterized in that: The method of using symplectic geometry to calculate the embedding dimension and reconstructing the symplectic geometry components includes: The optimal vehicle-grid resonant voltage and current data are used as the embedding dimension of the symplectic geometry method, and the symplectic geometry components are reconstructed using the normalized mean square error to establish the symplectic geometry model of the pantograph-catenary system. The expression is: Where p represents the embedding dimension of the reconstructed space, r is the time delay, and the number of points in the p-dimensional reconstructed attractor is represented by g=n-(p-1).

5. The vehicle-grid resonance feature identification method based on DBM-SGMD as claimed in claim 4, characterized in that: The classifying of the symplectic geometric components by a Gaussian deep Boltzmann machine includes: The vehicle-grid resonant components reconstructed by symplectic geometry are classified through the pipeline of Gaussian deep Boltzmann machine, characterized by the distribution on the binary vector [0,1], and the real data is modeled and trained as probability, expressed as: Among them, W ij It is used to describe the correlation between the i-th node in the visible layer and the j-th node in the hidden layer, α i represents the bias of the i-th node in the visible layer, b j is the bias of the jth node in the hidden layer, and the visible layer consists of N k binary random variables, denoted as vector k, and the hidden layer consists of N y binary random variables, denoted as vector y, variable ki corresponds to the data of the i-th node in the visible layer, and variable y i Then it corresponds to the data of the jth node of the hidden layer; After determining E(k,y), the multidimensional probability distribution between the two layers is obtained, and the expression is: With θ =∑ k,y yes -Eθ(k,y) Among them, Z θ is the partition function.

6. The vehicle-grid resonance feature identification method based on DBM-SGMD as claimed in claim 5, characterized in that: The Gaussian deep Boltzmann machine includes: The Gaussian Deep Boltzmann Machine consists of a visible layer and a hidden layer. The upper layer is a hidden layer with N y hidden units; the lower layer represents the visible layer, with N k Visible units.

7. The vehicle-grid resonance feature identification method based on DBM-SGMD as claimed in claim 6, characterized in that: The classifying of the symplectic geometric components by a Gaussian deep Boltzmann machine also includes: Collect vehicle-grid resonance data and input it into the Gaussian deep Boltzmann machine model, select the oscillation frequency of vehicle-grid resonance, set the number of layers of the Gaussian deep Boltzmann machine and the number of neurons in each layer, use an unsupervised learning algorithm to train the Gaussian deep Boltzmann machine model, optimize the weights of the Gaussian deep Boltzmann machine model, use labeled data for supervised fine-tuning, evaluate model performance through cross-validation and other methods, check accuracy and recall rate indicators, deploy the trained Gaussian deep Boltzmann machine model to actual applications, and perform frequency screening.

8. A system using the vehicle-grid resonance feature identification method based on DBM-SGMD as claimed in any one of claims 1 to 7, characterized in that: It includes data acquisition and preprocessing module, filter design and data optimization module, symplectic geometry calculation and feature reconstruction module, and classification and model training module; The data acquisition and preprocessing module collects the required voltage and current data from the target train network system, performs preprocessing, records the data in the same acquisition period, and divides it into windows with the same time length to form standardized time series data; The filter design and data optimization module processes the preprocessed time series data through an adaptive triplet half-band filter, uses the least squares method to construct a filter bank to reduce the errors of the passband and stopband, and optimizes the obtained vehicle network resonant voltage and current data; The symplectic geometry calculation and feature reconstruction module calculates the embedding dimension of the vehicle-grid resonance by using the symplectic geometry method, and reconstructs the symplectic geometry components by using the normalized mean square error; The classification and model training module uses DBM to perform binary distribution modeling and training on symplectic geometric components, optimizes the model weights through an unsupervised learning algorithm, and performs fine-tuning through a supervised learning algorithm to ultimately form an optimized classification model. The trained model is used to identify the oscillation frequency of the vehicle-grid resonance characteristic and is applied to the resonance frequency screening of the actual vehicle-grid system.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of a vehicle-grid resonance feature identification method based on DBM-SGMD according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a vehicle network resonance characteristic identification method based on DBM-SGMD as described in any one of claims 1 to 7 are implemented.

Images