Railway LEU-transponder transmission system lightning stroke fault risk assessment method

Abstract

The invention discloses a railway LEU-transponder transmission system lightning stroke fault risk assessment method, and the method comprises the steps: determining a lightning current invasion path through building a lightning stroke transient electromagnetic disturbance calculation model of an electromagnetic sensitive equipment port, and calculating the voltage and current at the port when incoming waves invade. And building an electromagnetic sensitivity test experiment platform of the railway signal system based on the lightning current invasion path and the voltage and current at the port. And performing label category prediction on the to-be-predicted sample point through each support vector machine model in the target model set, and voting each prediction result. According to the invention, a complex transmission relation between lightning transient electromagnetic disturbance and equipment failure risk is established through deep learning. The fault risk of the signal system under the lightning transient electromagnetic disturbance action is predicted through transient electromagnetic effect experimental data of real equipment of the railway signal system, and powerful support is provided for improving the electromagnetic sensitivity of the signal system and guaranteeing stable operation of a railway.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of lightning protection of railway signal systems, and more particularly to a lightning strike fault risk assessment method for a railway LEU-transponder transmission system. BACKGROUND

[0002] With the large-scale construction and operation of high-speed rails around the world, lightning protection research of high-speed rail signal systems has received increasing attention and concern. The control, dispatching, communication and other functions required by modern railways largely depend on signal systems. Compared with power systems, these weak current devices usually operate at low voltage and high frequency, and have high electromagnetic sensitivity, and are easily affected by lightning-caused transient electromagnetic disturbances.

[0003] Currently, lightning protection and surge protection are commonly used to reduce or avoid direct lightning intrusion into signal systems to cause direct damage effects. However, the transient electromagnetic disturbances generated along with the protection have the characteristics of space field and road. Transient electromagnetic disturbances can cause indirect effects through electromagnetic radiation, field-line coupling and crosstalk, causing signal errors, device lockout and even system service interruption, and it is difficult to simulate and calculate the complex failure mechanism of the complete signal system under field conditions. SUMMARY

[0004] Therefore, the present application provides a lightning strike fault risk assessment method for a railway LEU-transponder transmission system, which establishes a complex transfer relationship between lightning transient electromagnetic disturbances and device failure risks through deep learning, and predicts the fault risk of signal systems under the action of lightning transient electromagnetic disturbances.

[0005] To achieve the above-mentioned purpose, the present scheme is as follows:

[0006] A lightning strike fault risk assessment method for a railway LEU-transponder transmission system, comprising:

[0007] A lightning transient electromagnetic disturbance calculation model of an electromagnetic sensitive device port is established, the lightning current intrusion path is determined, and the voltage and current at the port when the incoming wave intrudes are calculated;

[0008] An electromagnetic sensitivity test experiment platform of the railway signal system is built based on the lightning current intrusion path and the voltage and current at the port, and a data sample set is obtained;

[0009] The data sample set is preprocessed and randomly divided into several training sets and test sets;

[0010] Support vector machine models are trained and evaluated based on each training set and test set, and a target model set is obtained, wherein the target model set includes at least one trained support vector machine model;

[0011] The support vector machine models in the target model set are used to predict the label categories of the to-be-predicted sample points, and the prediction results are voted to obtain a fault prediction result.

[0012] Preferably, the process of establishing the lightning transient electromagnetic disturbance calculation model of the electromagnetic sensitive device port, determining the lightning current intrusion path, and calculating the voltage and current at the port when the incoming wave intrudes, comprises:

[0013] Based on the transient electromagnetic response of the multi-conductor transmission line in the signal system under external electromagnetic field excitation, a lightning transient electromagnetic disturbance calculation model of the port connection line and cable of the electromagnetic sensitive device is established;

[0014] Based on the lightning transient electromagnetic disturbance calculation model, the voltage and current at the port when the incoming wave intrudes are calculated using the transmission line theory.

[0015] Preferably, the electromagnetic sensitivity test experimental platform comprises: a pulse injection module, a measurement and data acquisition module, and a to-be-tested signal system.

[0016] The pulse injection module outputs pulses to the to-be-tested signal system through a lightning surge generator.

[0017] The measurement and data acquisition module measures and records the data of the to-be-tested signal system through a four-channel oscilloscope.

[0018] Preferably, the process of measuring and recording the data of the to-be-tested signal system through the four-channel oscilloscope comprises:

[0019] The first channel, the second channel, and the third channel are used to collect the electromagnetic disturbance effect on the signal cable between the LEU and the ground transponder in the to-be-tested signal system.

[0020] The fourth channel is used to measure the input current in the pulse injection module.

[0021] Preferably, each sample data in the data sample set comprises core wire-to-ground voltage, core wire current, two-core wire differential voltage, and current of the injection coupling device.

[0022] Preferably, the process of preprocessing the data sample set comprises:

[0023] Feature extraction is performed on each sample data in the sample data set.

[0024] The feature values of all sample data and the corresponding label categories are combined respectively, and the sample data set is divided into a feature set and a label vector set.

[0025] Preferably, the process of randomly dividing the training set and the test set comprises:

[0026] The feature set and the label vector set are hierarchically and randomly divided to obtain several training sets and test sets which are different, wherein the proportion of each label category in the training set and the test set is consistent with the proportion in the sample data set.

[0027] Preferably, the process of training and evaluating the support vector machine model based on each training set and test set respectively comprises:

[0028] According to the original features in each training set, a disturbance signal is added respectively to obtain each updated training set.

[0029] A support vector machine model is trained based on different training sets, and the support vector machine model is a linear kernel function, a polynomial kernel function or a Gaussian radial basis kernel function.

[0030] The parameters of the support vector machine model are optimized by using a grid search method.

[0031] The trained support vector machine model is verified by the corresponding test set of the training set, the accuracy of each support vector machine model is calculated, the support vector machine model with an accuracy higher than a preset value is screened, and a target model set is obtained.

[0032] According to the specific embodiments provided by the present application, the following technical effects are disclosed:

[0033] The railway LEU-responder transmission system lightning fault risk assessment method provided by the present application establishes a lightning transient electromagnetic disturbance calculation model of the port of the electromagnetic sensitive device, determines the lightning current invasion path and calculates the voltage and current at the port when the incoming wave invades. Based on the lightning current invasion path and the voltage and current at the port, an electromagnetic sensitivity test experiment platform of the railway signal system is built to obtain a data sample set. Each support vector machine model in the target model set is used to predict the label category of the to-be-predicted sample point, and each prediction result is voted. The present application establishes a complex transfer relationship between lightning transient electromagnetic disturbance and device failure risk through deep learning. Through the transient electromagnetic effect experiment data of the real device of the railway signal system, the failure risk of the signal system under the action of lightning transient electromagnetic disturbance is predicted. And it provides strong support for improving the electromagnetic sensitivity of the signal system and ensuring the stable operation of the railway. BRIEF DESCRIPTION OF DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only embodiments of the present application, and those skilled in the art can obtain other drawings according to the provided drawings without creative labor.

[0035] Figure 1A flow chart of a lightning stroke fault risk assessment method of a railway LEU-transponder transmission system is provided for the embodiment of the present application.

[0036] Figure 2 A lightning current invasion schematic diagram at lightning stroke is provided for the embodiment of the present application.

[0037] Figures 3a-3b A configuration diagram and an equivalent circuit diagram of a lightning stroke transient electromagnetic disturbance calculation model are respectively provided for the embodiment of the present application.

[0038] Figure 4 A voltage waveform of a transmission line terminal of an electromagnetic sensitive device is provided for the embodiment of the present application.

[0039] Figure 5 A schematic diagram of an electromagnetic sensitivity test experiment platform is provided for the embodiment of the present application.

[0040] Figure 6 A working principle diagram of a transponder signal transmission system is provided for the embodiment of the present application.

[0041] Figure 7 A model training flow chart is provided for the embodiment of the present application.

[0042] Figure 8 A sample feature set confusion matrix schematic diagram is provided for the embodiment of the present application.

[0043] Figure 9 A signal cable two-core line differential voltage maximum value experimental result diagram is provided for the embodiment of the present application. DETAILED DESCRIPTION

[0044] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0045] First, combined with Figure 1 A railway LEU-transponder transmission system lightning stroke fault risk assessment method is introduced for the embodiment of the present application, as shown in Figure 1 The method comprises the following steps:

[0046] Step S01, a lightning stroke transient electromagnetic disturbance calculation model of an electromagnetic sensitive device port is established, the lightning current invasion path is determined, and the voltage and current at the port when the incoming wave invades are calculated.

[0047] Specifically, the propagation path of the signal system that can suffer from lightning transient electromagnetic disturbance is analyzed, the engineering status, topology structure, and terminal characteristics of the railway signal system are investigated, and the system parameters are reasonably selected in combination with the engineering actual situation. For example, a lightning hazard event occurred in a station, the signal equipment in the station was damaged by lightning overvoltage, and multiple trains were delayed in the section. The schematic diagram of the intrusion path of lightning overvoltage inferred according to the on-site investigation is shown in Figure 2 The signal room of the EMU depot where the lightning hazard event occurred is not provided with a comprehensive grounding system, resulting in a high resistance value of the grounding resistance of the buildings such as the overhead contact system support that are prone to lightning strikes. When the overhead contact system support and other high buildings are struck by lightning, no good grounding network is provided for discharging lightning current, and the ground potential near the overhead contact system support is instantaneously greatly raised. There is a large potential difference between the inside and outside of the signal cable near the overhead contact system support struck by lightning. Because the shielding effect of the balise signal cable far from the end of the signal room is poor, and the shielding layer of the signal cable is single-ended grounded at this location, the grounding point is located at the signal room, the insulation of the balise signal cable is easily broken down, and the lightning current intrudes along the steel belt, shielding layer, core wire, etc. to the wayside signal box and the signal room, causing the failure and damage of the wayside equipment and the equipment in the machine room, and affecting the safe, efficient, and stable operation of the subsequent trains in the section. The transformer of the power board in the LEU cabinet in the signal room is damaged in the on-site investigation.

[0048] First, a lightning transient electromagnetic disturbance calculation model of the port of an electromagnetic sensitive device is established based on the multi-conductor transmission line theory. The electrical equipment in the signal system is connected through a multi-conductor transmission line network to form a small independent electrical system. The transient electromagnetic disturbance source caused by lightning can be coupled with the multi-conductor transmission line in the signal system, causing the abnormal operation or damage of various electrical equipment. The transient electromagnetic response of the multi-conductor transmission line in the signal system under external electromagnetic field excitation is analyzed, and a lightning transient electromagnetic disturbance calculation model of the port connection line and cable of the electromagnetic sensitive device is established for calculation. The configuration of the lightning transient electromagnetic disturbance calculation model (multi-conductor transmission line model of the signal system) is shown in Figure 3a , and the equivalent circuit is shown in Figure 3b .

[0049] Then, the voltage and current responses at the port when the wave intrudes are calculated using the transmission line theory. In the frequency domain, the multi-conductor transmission line equation is decomposed into the sum of the responses of the transmission line equivalent distribution voltage and the load end voltage excitation, the voltage and current responses at the port are calculated, and the time-domain form of the voltage and current waveforms of the transmission line terminal are obtained through inverse Fourier transform.

[0050] The impedance matrix Z(s) and the admittance matrix Y(s) are fitted into a rational function expression and decomposed into the following partial fraction form:

[0051]

[0052] where c k , d k are the residues and poles of the partial fraction expression of impedance Z(s), N Z is the number of terms of the partial fraction of impedance Z(s) function, s is a complex frequency variable, s = σ + jω, where σ is the real part of the damping factor, ω is the imaginary part of the angular frequency, j is the imaginary unit, s is used to describe the frequency response and transient behavior of the system. L is a constant term, representing the static or DC impedance component of the system.

[0053]

[0054] where a k , b k are the residues and poles of the partial fraction expression of admittance Y(s), N Y is the number of terms of the partial fraction of admittance Y(s) function, C is the admittance.

[0055] The cable complex frequency domain multi-conductor transmission line equation can be written as follows:

[0056]

[0057] where U(x, s), I(x, s) are the voltage and current column vectors, Z(s), Y(s) are the unit length impedance and admittance matrix of the cable composed of multi-conductor transmission line, x is the length coordinate along the cable axis.

[0058] Considering the need to involve the process simulation of surge protector action and breakdown flashover in the calculation of cable overvoltage, this paper chooses to solve the cable conductor transmission line calculation model in time domain. Based on the Laplace inverse transform, the time domain telegraph equation with frequency variable parameters can be obtained as follows:

[0059]

[0060] Where u and i represent the components of U and I corresponding to each segment respectively.

[0061] For example, as shown in Figure 4 , when 5kV lightning is injected into the overhead contact system support, there is about 350V potential on the through ground wire and parallel conductor. If the average voltage of lightning current is calculated as 30kV, there is about 2.1kV potential in the parallel conductor.

[0062] Step S02, based on the lightning current intrusion path and the voltage and current at the port, an electromagnetic sensitivity test platform for railway signal system is built.

[0063] Specifically, in order to conveniently obtain the effect data of the railway signal system under the transient electromagnetic disturbance, evaluate the failure risk of the high-speed railway signal system under the lightning transient electromagnetic interference, and according to the lightning current invasion path of the railway signal system and the voltage and current at the port, an electromagnetic sensitivity test experimental platform for the railway signal system is built. For example, the balise transmission system in the railway signal system can be taken as the experimental object, the 1.2 / 50 μs double exponential wave generated by the lightning surge generator is coupled to the signal cable between the ground balise and the ground electronic unit (LEU) through the coupling device, so as to simulate the process of the lightning transient electromagnetic disturbance invading the balise transmission system. The state data of the experimental object under the electromagnetic interference injected by the experimental platform and the corresponding label are recorded as the data sample for subsequent model evaluation, and the data sample set is obtained. The platform field schematic diagram is shown in Figure 5 .

[0064] The electromagnetic sensitivity test experimental platform mainly includes three parts: a pulse injection module, a measurement and data acquisition module, and a signal system to be tested.

[0065] (1) Pulse injection module:

[0066] ① Laboratory electromagnetic interference effect test pulse source (lightning surge generator 1.2 / 50 μs);

[0067] ② Electromagnetic coupling device (inductive coupler);

[0068] ③ A number of matching resistors and pulse injection cables.

[0069] (2) Measurement and data acquisition module:

[0070] ① Voltage probe;

[0071] ② Current sensor;

[0072] ③ Four-channel oscilloscope;

[0073] ④ A number of BNC lines.

[0074] (3) Signal system to be tested:

[0075] The railway signal system is composed of a centralized dispatch control system (CTC), a train control system (CTCS-3 level), a station interlocking system (CBI), a signal centralized monitoring system (CSM), a power supply system, etc. Among them, the train control system is composed of ground and on-board equipment, mainly including a radio block center (RBC), a temporary speed limit server (TSRS), a train control center (TCC), a track circuit, a balise and a ground electronic unit (LEU), a GSM-R wireless communication system, on-board ATP equipment, etc.

[0076] The following will be introduced as an example of the components of the transponder signal transmission system. The working principle of the transponder signal transmission system is shown in Figure 6 The system comprises:

[0077] ① Ground electronic unit (LEU) and its power module;

[0078] ② Ground (active) transponder;

[0079] ③ Signal cable;

[0080] ④ Vehicle-mounted antenna;

[0081] ⑤ Vehicle-mounted transponder signal transmission unit (BTM host) and its power module;

[0082] ⑥ PC (installing the supporting software of the BTM host, used for monitoring the signal transmission of the system);

[0083] ⑦ Optical fiber (used for connecting the BTM host and the PC software end)

[0084] The above electromagnetic sensitivity test experimental platform is used to carry out electromagnetic sensitivity test experiments of the railway transponder transmission system. For example: the output amplitude of the lightning surge generator of the pulse injection module is from 0.2 kV to 2.3 kV, and a set of test experiments is carried out every 0.1 kV. The waveform data and the corresponding running state of the signal system to be tested under each set of experiments are measured and collected. The measurement and data collection module uses a four-channel oscilloscope to measure and collect experimental data. Among them, the first channel, the second channel and the third channel are used to collect the electromagnetic interference effect on the signal cable between LEU and the ground transponder in the signal system to be tested; the fourth channel is used to measure the current input into the inductive coupler in the pulse injection module. The current is a standard double exponential waveform, which is used as a trigger channel in the experiment. Through the experiment, a total of 62 sample data can be obtained. There are three system running states in the experimental process, including the normal running state of the system and the fault states of two fault levels, which are "transponder transmission system appears signal error code, can be self-recovered" and "transponder transmission system appears signal error code, needs to restart the system to recover". The above three states are represented as labels "0", "1" and "2" respectively.

[0085] Step S03, pre-process the data sample set and randomly divide it into several training sets and test sets.

[0086] Specifically, the sample data set under the three system states classified in the experiment is subjected to data pre-processing. Among them, the four signals of each sample data include the core wire to ground voltage, the core wire current, the differential voltage of the two core wires and the current of the injection coupling device.

[0087] First, the sample data is preprocessed. Feature extraction can be performed on each sample data in the sample data set. For example: feature extraction is performed on the four signals of the sample data respectively, and 11 feature values are obtained for each sample data. The feature values of all sample data and the corresponding label categories are combined to obtain the final data sample set.

[0088] Then, the data sample set is randomly divided into several groups, and several groups of different training sets and test sets are obtained for support vector machine (SVM) model training and comprehensive evaluation. The proportion of each label category in the training set and the test set is consistent with the proportion in the sample data set.

[0089] The data is read from the sample data set and divided into a feature set X and a label vector set y. In order to reduce the influence of accidental factors on the model training effect when dividing the training set and the test set, and to make full use of the sample data, the data set composed of the feature set X and the label vector set y is divided by a stratified random division method, for example: 100 times of division can obtain 100 groups of different training sets (X_train, y_train) and test sets (X_test, y_test). The proportion of each label category in the training set and the test set obtained by each division is consistent with that in the original sample data set.

[0090] Step S04, training and evaluation of the support vector machine model based on each training set and test set.

[0091] Specifically, the training set obtained by each division is used to train the support vector machine model (SVM), and a trained support vector machine model is obtained. The support vector machine model can be a linear kernel function, a polynomial kernel function, and a Gaussian radial basis kernel function. Compared with the training effect of the three functions, the accuracy of the SVM model using the Gaussian radial basis kernel function is higher than that of the linear kernel function and the polynomial kernel function. Therefore, the SVM model using the Gaussian radial basis kernel function can be used. During the training of each support vector machine model, the gamma parameter of the Gaussian radial basis kernel function is optimized by grid search, and each trained support vector machine model is verified under the corresponding test set to calculate the accuracy, precision, recall, and F1 score of each support vector machine model. The support vector machine model with an accuracy higher than a preset value is selected to obtain a target model set, and the target model set includes at least one trained support vector machine model. The training process is as follows Figure 7As shown, after the data set is divided, the original characteristics of the signal cable transmission signal in the data set are obtained. The standard deviation, maximum value, minimum value, kurtosis and root mean square of the core-to-ground voltage (V) in the data set, the maximum value, minimum value and root mean square of the core current (A), the maximum value of the differential voltage (signal voltage) (V) and the information entropy, etc. 12 characteristic values can be extracted.

[0092] A random disturbance signal of ±20V is added to the data in the data set, and the features are selected and sorted by the feature disturbance method. The feature with the largest feature importance in the model training, "signal cable differential voltage maximum value", is obtained, and the system state label prediction is performed under this feature value. For example, the support vector machine model with an accuracy of greater than 85% can be saved and applied to subsequent prediction models. The average evaluation index of the SVM model in the target model set is shown in Table 1.

[0093] Table 1

[0094] Evaluation metrics Value / % Mean accuracy 93.79 Mean precision 95.28 Mean recall 93.54 Mean F1 score 93.68

[0095] Step S05, label category prediction of the to-be-predicted sample point is performed by each support vector machine model in the target model set, and voting is performed on each prediction result.

[0096] Specifically, in the prediction process, first, the models in the target model set are called in turn to predict the label category of each to-be-predicted sample point, and the prediction results of each sample point under all models are voted respectively, so as to determine the prediction result of the final integrated model. Taking the sample feature set in the experimental data as an example for verification, 62 groups of data are collected, and the model prediction label is compared with the actual label, and the confusion matrix of the prediction model on the sample feature set can be obtained as shown in Figure 8 As shown in Figure 8 For label 0, 16 prediction models predict correctly 15, with a correct rate of 93.75%, and for label 1, the prediction accuracy reaches 92.85%, and the prediction accuracy of label 2 can reach 88.88%. Therefore, it can be seen that the prediction accuracy of the label is high.

[0097] In addition, in order to support the support vector machine model under the condition of low feature dimension, the model can still maintain the highest prediction accuracy as possible. On the basis of the original support vector machine model, the feature perturbation method can be used to measure the contribution of each feature in the sample data set to the model classification and prediction. By perturbing the specific value of a feature, the original correlation between the feature and the label category is disturbed, and the change of the model performance before and after the perturbation is observed, so as to judge the contribution of the feature value to the classification and prediction of the support vector machine model. For example: the feature with the largest classification and prediction contribution in the aforementioned sample data set under the aforementioned model is the "maximum differential voltage of the two core wires of the signal cable". Therefore, the maximum differential voltage of the two core wires of the signal cable under different state labels in the responder transmission system electromagnetic pulse interference injection experiment is shown in FIG. 8, and in addition, different colors can be used to represent different types of labels, for example: red represents label 2, yellow represents label 1, and green represents label 0. Figure 9

[0098] Next, the trained support vector model is used to predict the probability of the feature "maximum differential voltage of the two core wires of the signal cable" in the sample data set range [0, 250] near each label category. The threshold point of the feature between adjacent labels is obtained. The process is as follows:

[0099] First, the trained multiple support vector models are used for prediction, and 1000 equally spaced data points between 0 and 250 are generated for the feature input of the model and the label prediction. Then, the decision value of each data point under each support vector model is calculated, and the decision value is converted into output probability by Platt scaling. For multi-classification problems, the probability of each label category is calculated and normalized, so as to obtain the probability of the sample data under different label categories of each support vector model. Finally, the average probability of the sample data under different label categories of all support vector models is calculated. In addition, the threshold value between adjacent categories can be determined by the intersection point of the probability curve between adjacent categories. The threshold value of the "maximum differential voltage of the two core wires of the signal cable" feature between label 0 (i.e. normal working state of the system) and label 1 (i.e. system error code fault and self-recovery operation) is 60.31V; the threshold value of the "maximum differential voltage of the two core wires of the signal cable" feature between label 1 (i.e. system error code fault and self-recovery) and label 2 (i.e. system error code fault and restart of the device for recovery) is 168.92V.

[0100] The embodiment of the present application uses deep neural network to approximate the probability result, introduces feature coding and support vector machine model, realizes the overall model from excitation to port and then to the probability of lightning electromagnetic sensitive problem, and comprehensively evaluates the lightning transient electromagnetic disturbance effect and fault risk of the railway signal system.

[0101] ​The railway LEU-responder transmission system lightning failure risk assessment method provided by the embodiment of the present application determines the lightning current invasion path and calculates the voltage and current at the port when the incoming wave invades by establishing a lightning transient electromagnetic disturbance calculation model of the electromagnetic sensitive device port. An electromagnetic sensitivity test experiment platform of the railway signal system is built based on the lightning current invasion path and the voltage and current at the port, and a data sample set is obtained. The label category of a to-be-predicted sample point is predicted by each support vector machine model in the target model set, and the prediction results are voted. The embodiment of the present application establishes the complex transmission relationship between lightning transient electromagnetic disturbance and device failure risk through deep learning. The failure risk of the signal system under the action of lightning transient electromagnetic disturbance is predicted through the transient electromagnetic effect experiment data of the real-type device of the railway signal system. And it provides strong support for improving the electromagnetic sensitivity of the signal system and ensuring the stable operation of the railway.

[0102] Finally, it should be noted that in this document, the terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed or inherent to such a process, method, article or device. Without more limitations, the element defined by the statement "comprising a" does not exclude the presence of additional identical elements in the process, method, article or device including the element.

[0103] The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the difference from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0104] The above description of the disclosed embodiments enables a person skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for assessing lightning fault risk in a railway LEU-balise transmission system, characterized in that: include: Establish a calculation model for lightning transient electromagnetic disturbance at the ports of electromagnetically sensitive equipment, determine the lightning current intrusion path, and calculate the voltage and current at the port when the wave intrudes; Based on the lightning current intrusion path and the voltage and current at the port, an electromagnetic susceptibility test platform for the railway signal system was established to obtain a data sample set; Preprocess the data sample set and randomly divide it into several training sets and test sets; Training and evaluating the support vector machine model based on each training set and test set respectively to obtain a target model set, wherein the target model set includes at least one trained support vector machine model; The label category of the sample points to be predicted is predicted by each support vector machine model in the target model set, and each prediction result is voted to obtain the fault prediction result.

2. The method for assessing lightning failure risk of a railway LEU-balise transmission system according to claim 1, characterized in that: The process of establishing a lightning transient electromagnetic disturbance calculation model for the port of an electromagnetically sensitive device, determining the lightning current intrusion path, and calculating the voltage and current at the port when the incoming wave intrudes includes: Based on the transient electromagnetic response of multi-conductor transmission lines in signal systems under external electromagnetic field excitation, a calculation model for lightning transient electromagnetic disturbance of port connection lines and cables of electromagnetic sensitive equipment is established; Based on the calculation model of lightning transient electromagnetic disturbance, the voltage and current at the port when the wave invades are calculated using transmission line theory.

3. The method for assessing lightning failure risk of a railway LEU-balise transmission system according to claim 1, characterized in that: The electromagnetic susceptibility test experimental platform includes: a pulse injection module, a measurement and data acquisition module, and a signal system to be tested; The pulse injection module outputs pulses to the signal system to be tested through the lightning surge generator; The measurement and data acquisition module measures and records the data of the signal system to be measured through a four-channel oscilloscope.

4. The method for assessing lightning failure risk of a railway LEU-balise transmission system according to claim 3, characterized in that: The process of measuring and recording the data of the signal system under test using a four-channel oscilloscope includes: The first channel, the second channel and the third channel are used to collect the electromagnetic disturbance effects on the signal cable between the LEU and the ground transponder in the signal system to be tested; The fourth channel is used to measure the current input to the pulse injection module.

5. The method for assessing lightning failure risk of a railway LEU-balise transmission system according to claim 1, characterized in that: Each sample data in the data sample set includes a core wire-to-ground voltage, a core wire current, a two-core wire differential voltage, and a current injected into a coupling device.

6. The method for assessing lightning failure risk of a railway LEU-balise transmission system according to claim 1, characterized in that: The process of preprocessing the data sample set includes: Extract features from each sample data in the sample data set; The feature values ​​and corresponding label categories of all sample data are combined respectively, and the sample data set is divided into a feature set and a label vector set.

7. The method for assessing lightning failure risk of a railway LEU-balise transmission system according to claim 6, characterized in that: The process of randomly stratifying and partitioning to obtain several training and test sets includes: The feature set and label vector set are randomly divided into layers to obtain several different training sets and test sets, where the proportion of each label category in the training set and test set is consistent with the proportion in the sample data set.

8. The method for assessing lightning failure risk of a railway LEU-balise transmission system according to claim 7, characterized in that: The process of training and evaluating the support vector machine model based on each training set and test set includes: According to the original features in each training set, the disturbance signal is added to obtain the updated training sets; Based on different training sets, a support vector machine model is trained respectively, wherein the support vector machine model is a linear kernel function, a polynomial kernel function or a Gaussian radial basis kernel function; The grid search method is used to optimize the parameters of the support vector machine model; The trained support vector machine model is verified through the test set corresponding to the training set, the accuracy of each support vector machine model is calculated, and the support vector machine model with an accuracy higher than the preset value is screened to obtain the target model set.

Images