A method, medium and system for predicting the surface flashover voltage of a high-speed rail roof insulator
By combining simulation experiments and wind tunnel experiments with predictive neural networks, and utilizing the airflow parameters directly in front of and on the surface of the insulator on the roof of a high-speed train, the problem of predicting the flashover voltage of insulators under strong airflow conditions was solved. This achieved efficient and accurate flashover voltage prediction, providing effective guidance for the safe operation of trains.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- ELECTRIC POWER RESEARCH INSTITUTE OF STATE GRID NINGXIA ELECTRIC POWER COMPANY
- Filing Date
- 2022-11-23
- Publication Date
- 2026-06-05
AI Technical Summary
Existing technologies lack methods for predicting the surface flashover voltage of roof insulators in high-speed trains under strong airflow conditions, making it impossible to effectively assess and protect against potential threats to train safety.
By combining simulation experiments and wind tunnel experiments with a predictive neural network, the neural network is trained to predict flashover voltage using airflow parameters directly in front of the insulator on the roof of a high-speed train and surface airflow parameters. By selecting airflow parameters at feature points as input, the intrinsic relationship between airflow parameters and flashover voltage is established, thus achieving accurate prediction.
It enables accurate prediction of surface flashover voltage of roof insulators in high-speed trains under strong airflow conditions, provides design and protection references, reduces computational complexity and improves prediction accuracy.
Smart Images
Figure CN115935800B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of insulator flashover voltage technology, and in particular to a method, medium, and system for predicting surface flashover voltage of insulators on the roof of high-speed trains. Background Technology
[0002] High-speed train roof insulators are key components for high-voltage isolation and pantograph mechanical support, and their excellent insulation performance is fundamental to ensuring safe train operation. With the significant increase in train speeds, high-speed trains generate strong airflows of up to 100 m / s on their surfaces, reaching as high as 120 m / s on the roof insulator surface. This exposes the roof insulation to this strong airflow environment. Compared to traditional external insulation equipment situated in static gas, the complex, time-varying airflow field leads to new characteristics such as non-uniform distribution of air pressure / density on the insulation surface, significant differences in flow velocity / direction, and transient changes in local airflow direction / velocity. The strong airflow environment (airflow velocity up to 100 m / s) generated by the train inevitably has a significant impact on the discharge characteristics of the high-speed train roof insulation, posing a potential threat to safe train operation. Summary of the Invention
[0003] This invention provides a method, medium, and system for predicting the surface flashover voltage of high-speed train roof insulators, in order to solve the problem of the lack of existing prediction technology for the surface flashover voltage of high-speed train roof insulators under extreme environments such as strong airflow.
[0004] A first aspect provides a method for predicting the surface flashover voltage of a roof insulator on a high-speed train, wherein the high-speed train is in operation, and the method includes:
[0005] Based on the preset values of airflow parameters directly in front of the roof insulator of the high-speed train, the first simulated values of airflow parameters on the surface of the roof insulator of the high-speed train were obtained through simulation experiments.
[0006] Based on the preset values of the airflow parameters directly in front of the high-speed train roof insulator and the first simulated values of the corresponding airflow parameters on the surface of the high-speed train roof insulator, the experimental values of the surface flashover voltage of the high-speed train roof insulator under airflow conditions are obtained through wind tunnel experiments.
[0007] The prediction neural network is trained using the preset value of the airflow parameters in front of the insulator on the roof of the high-speed train, the first simulated value of the airflow parameters on the surface of the insulator on the roof of the high-speed train, and the experimental value of the flashover voltage.
[0008] Based on the measured values of the airflow parameters collected in front of the roof insulator of the high-speed train, a second simulated value of the airflow parameters on the surface of the roof insulator of the high-speed train was obtained through simulation experiments.
[0009] The measured values of the airflow parameters directly in front of the high-speed train roof insulator and the corresponding simulated values of the airflow parameters on the surface of the high-speed train roof insulator are input into the trained prediction neural network, which outputs the predicted value of the surface flashover voltage of the high-speed train roof insulator.
[0010] In a second aspect, a computer-readable storage medium is provided, wherein computer program instructions are stored on the computer-readable storage medium; when the computer program instructions are executed by a processor, they implement the method for predicting the surface flashover voltage of a high-speed train roof insulator as described in the first aspect embodiment above.
[0011] Thirdly, a system for predicting the surface flashover voltage of a high-speed train roof insulator is provided, comprising: a computer-readable storage medium as described in the embodiments of the second aspect above.
[0012] Thus, this embodiment of the invention can predict the surface flashover voltage of the roof insulator of a high-speed train under extreme environments such as strong airflow, providing reference guidance for the design, protection, and risk-based operation and maintenance of external insulation under extreme environments. Attached Figure Description
[0013] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0014] Figure 1 This is a flowchart of the method for predicting the surface flashover voltage of a high-speed train roof insulator according to an embodiment of the present invention;
[0015] Figure 2 This is a side view of an insulator according to an embodiment of the present invention;
[0016] Figure 3 This is a top view of an insulator according to an embodiment of the present invention;
[0017] Figure 4 This is a schematic diagram of the predictive neural network structure according to an embodiment of the present invention. Detailed Implementation
[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0019] This invention discloses a method for predicting the surface flashover voltage of a roof insulator on a high-speed train. It should be understood that this invention is for high-speed trains in operation, with a typical speed of at least 0–100 m / s. Figure 1 As shown, the method of this embodiment of the invention includes the following steps:
[0020] Step S101: Based on the preset values of the airflow parameters directly in front of the insulator on the roof of the high-speed train, the first simulated value of the airflow parameters on the surface of the insulator on the roof of the high-speed train is obtained through simulation experiments.
[0021] The airflow directly in front of the insulator on the roof of the high-speed train refers to the initial airflow that has not yet come into contact with the insulator. The airflow parameters directly in front of the insulator on the roof of the high-speed train include: the airflow velocity and the airflow direction angle.
[0022] Existing artificial intelligence prediction technologies for insulator flashover voltage primarily select parameters from the external environment of the insulator, such as altitude, air pressure, temperature, humidity, and wind and sand. These parameters are external environmental parameters of the insulator and do not directly affect the flashover process. Using these parameters as predictions for insulator flashover voltage results in low accuracy and requires a large amount of experimental data, which cannot be utilized with software data.
[0023] During high-speed rail operation, insulators are typically exposed to a high-speed airflow environment of up to 100 m / s. Through long-term research on the flashover voltage characteristics of insulators in this airflow environment, the inventors of this invention discovered that the primary factor influencing the flashover process is not the aforementioned external environmental parameters, but rather the significant impact of this high-speed airflow field on the surface flashover voltage of the insulators on the high-speed rail roof. Specifically, the airflow distribution characteristics close to the insulator surface—that is, the airflow velocity and pressure at various points on the insulator surface (which should be understood to be different from the parameters in the external space of the insulator)—are crucial in influencing the generation and development of electrons and ions, as well as the flashover path. Because these airflow parameters affect the generation and development of electrons and ions during the flashover path, the interaction between neutral airflow molecules and electrons and ions varies under different parameters. This results in different electron ionization coefficients at various points on the insulator surface and affects the ion flow distribution on the insulator surface, ultimately causing the macroscopic flashover voltage of the insulator to change with the initial airflow velocity.
[0024] Furthermore, through long-term research on gas discharge in airflow environments, the inventors of this invention have discovered that the flow field characteristics at various points on the surface of the insulator are unevenly distributed under such high-speed airflow conditions, resulting in special regions such as the windward zone, the leeward zone, and the crosswind zone.
[0025] Specifically, when an insulator is accelerating / decelerating or operating at a certain flow field speed, its surface flashover path is very likely to occur along the middle line of each of the above-mentioned different partition surfaces.
[0026] When an insulator is in a uniform flow field, the airflow velocity on the side of the insulator is much higher than that on the windward and leeward sides. Therefore, the airflow pressure and density on the side are relatively low. Under an applied electric field, the insulator will first discharge from its side and then flashover along the surface of the insulator skirts. Thus, the flashover path may be the centerline of the side of the insulator surface. The airflow parameters along this path will become the key factors affecting the flashover process and flashover voltage of the insulator. The airflow velocity at each point along this path will affect the electron ionization coefficient, the effective electron / ion concentration, etc. The airflow direction angle will cause changes in the relative free path of electrons. The airflow pressure will affect the migration characteristics of electrons and ions. The airflow density will affect the mean free path of electrons. The average airflow velocity along the same direction partition of the insulator will affect the cumulative number of electrons and ions. These parameters will become the key parameters affecting the flashover process and flashover voltage of the insulator, and will also become the key parameters that can characterize the flashover process of the insulator. Using these parameters, the intrinsic physical relationship between the airflow parameters of the insulator along this path and the flashover voltage can be accurately established.
[0027] When a high-speed train accelerates, a low-pressure turbulence zone will be generated on the leeward side of the insulator. This zone will first trigger flashover of the insulator. The flashover process is similar to that on the crosswind side, so the airflow parameters are selected in the same way.
[0028] During the process of a high-speed train stopping and decelerating, flashover may occur in the windward area of the insulator, and the process is similar to flashover on the crosswind and leeward sides.
[0029] Based on the above analysis, the central surface of each region is selected as the main flashover path, and characteristic points of each insulator skirt are selected along this path. The selection of these characteristic points should represent the main characteristics of the insulator flashover process. The flashover process is closely related to the airflow parameters at each characteristic point of the skirt along this path. Due to the different structural distributions of each skirt, the airflow parameter distributions on the upper and lower surfaces of each skirt differ. Specifically, characteristic points on the upper and lower surfaces of each skirt are selected, namely the midpoint and root point of the skirt; since the upper and lower surfaces of the skirt converge at the skirt edge, a point on the skirt edge is also selected. Figure 2 and 3 As shown, each umbrella skirt has 12 feature points. It should be understood that the midpoint of the umbrella skirt is the point exactly midway between the root point and the edge point of the umbrella skirt.
[0030] Specifically, the edges of the sheds are the main locations where gap short circuits occur between adjacent sheds, primarily affecting the transition of electron flow between adjacent sheds during insulator flashover. The electric field strength is relatively uniform near the midpoint of the sheds, mainly affecting the flow of electrons on the shed surface during flashover. The electric field strength is greatest at the root of the sheds, primarily affecting the generation and development path of electron flow during flashover.
[0031] By selecting these feature points, the flashover path of insulators under airflow conditions can be described, and the airflow parameters at these feature points can be used to characterize the physical process of insulator flashover. Furthermore, the intrinsic relationship between these airflow parameters and insulator flashover voltage can be established, thereby achieving the goal of predicting insulator flashover voltage.
[0032] Therefore, based on the above analysis, prior to this step, the embodiment of the present invention pre-determines the feature points on the surface of the insulator. The specific process is as follows:
[0033] (1) Based on the airflow direction angle directly in front of the insulator on the roof of the high-speed train, the surface of each skirt of the insulator is divided into four fan-shaped regions of equal area.
[0034] Specifically, such as Figure 3 As shown, the fan-shaped area facing the airflow directly in front of the insulator on the roof of the high-speed train is the windward side, the fan-shaped area opposite the windward side is the leeward side, and the remaining two fan-shaped areas are the crosswind sides.
[0035] (2) Determine the feature points of each sector region.
[0036] The feature points of each sector include: the edge point of the umbrella skirt on the center line of each sector, the root point of the umbrella skirt on the center line of the upper and lower surfaces of each sector, and the middle point of the umbrella skirt on the center line of the upper and lower surfaces of each sector.
[0037] It should be understood that the center line is the midline that divides a sector into two symmetrical parts.
[0038] Furthermore, based on the above analysis, the specific airflow parameters of the high-speed train roof insulator surface are determined to include: airflow velocity at the characteristic point, airflow direction angle at the characteristic point, airflow pressure at the characteristic point, airflow density at the characteristic point, average airflow velocity along the same direction fan-shaped area of the insulator, and average airflow pressure along the same direction fan-shaped area of the insulator.
[0039] Taking the windward side as an example, the average airflow velocity along the same direction of the insulator's fan-shaped area refers to the average airflow velocity at all characteristic points on the windward side of the insulator's skirts from top to bottom; the average airflow pressure along the same direction of the insulator's fan-shaped area refers to the average airflow pressure at all characteristic points on the windward side of the insulator's skirts from top to bottom.
[0040] The flow field was calculated using simulation software to obtain the aforementioned airflow parameters at the locations of characteristic points on the surface of each insulator skirt.
[0041] Specifically, the simulation experiments in this step can be conducted using existing simulation software, such as ANSYS and Comsol. The flow field on the insulator surface is simulated using the fluid-structure interaction module of Comsol software. By setting various values in the laminar flow section, the airflow direction and pressure are made to meet the conditions of a strong airflow environment. Then, through the solid component, relevant parameters of the insulator, such as the insulator's moving direction and speed, are set to ensure the airflow velocity meets the conditions of a strong airflow environment. This yields the relevant airflow parameters of the high-speed train's airflow environment during operation.
[0042] Step S102: Based on the preset values of the airflow parameters directly in front of the high-speed train roof insulator and the first simulated values of the corresponding airflow parameters on the surface of the high-speed train roof insulator, the experimental values of the surface flashover voltage of the high-speed train roof insulator under airflow conditions are obtained through wind tunnel experiments.
[0043] A high-speed rail model was placed in a wind tunnel, and tests were conducted under preset values of airflow parameters directly in front of the insulator on the roof of the high-speed rail and the first simulated values of airflow parameters on the surface of the insulator. The experimental values of the surface flashover voltage of the insulator on the roof of the high-speed rail were collected. Wind tunnel experiments are a common technique in fluid mechanics and will not be elaborated upon here.
[0044] Step S103: Train the prediction neural network using the preset values of the airflow parameters directly in front of the high-speed train roof insulator, the first simulated values of the corresponding airflow parameters on the surface of the high-speed train roof insulator, and the experimental values of the corresponding flashover voltage.
[0045] Because the physical relationship between airflow parameters and insulator flashover voltage is extremely complex, being a multidimensional and nonlinear relationship, it is difficult to directly establish a physical model of the relationship between airflow parameters and flashover voltage along this path. Therefore, the powerful nonlinear approximation capability of neural networks can be used to better approximate this multidimensional nonlinear relationship between airflow parameter conditions and insulator surface flashover voltage, so as to establish the correlation between insulator surface airflow parameters and flashover voltage, and then use these key airflow parameters to achieve accurate prediction of insulator flashover voltage.
[0046] Specifically, the predictive neural network in this embodiment of the invention can be a backpropagation (BP) neural network. For example... Figure 4As shown, the prediction neural network comprises an input layer, hidden layers, and an output layer connected sequentially. Specifically, the input layer consists of eight first neurons, each receiving an input for a specific airflow parameter. The hidden layer comprises five second neurons, and the output layer comprises one third neuron. The activation function of the hidden layer is the sigmoid function. The specific form of the sigmoid function is:
[0047] The predictive neural network uses preset values of airflow parameters directly in front of the insulator on the roof of the high-speed train, and the first simulated values of the corresponding airflow parameters on the surface of the insulator as its input vector. The output vector is the surface flashover voltage of the insulator under these airflow parameters. The input vector can be denoted as... v1, θ1, p1, and ρ1 represent the airflow velocity, direction angle, pressure, and density at various characteristic points of the insulator, respectively. Let v1 and v2 represent the average airflow velocity and pressure along the surface of each sector in the same direction of the insulator, respectively. Let v2 and θ2 represent the airflow velocity and direction angle directly in front of the insulator on the roof of the high-speed train, respectively. The output vector can be denoted as Y = [U]. T U represents the surface flashover voltage of the insulator.
[0048] Since the input to the predictive neural network should be a dimensionless vector, and in order to further reflect the magnitude of the change in the output vector caused by any change in the input vector, preferably, each type of airflow parameter is normalized to the interval [0,1].
[0049] Specifically, the normalization process includes:
[0050] For each type of airflow parameter, if the airflow parameter is positively correlated with the surface flashover voltage of the insulator on the roof of the high-speed train, then... Normalize these airflow parameters; otherwise (i.e., negatively correlated), use... Normalize these airflow parameters.
[0051] Among them, f n ′(x) represents the normalized airflow parameter for this type of flow, f n (x) represents the airflow parameters of this type before normalization, f max f represents the preset maximum value of this type of airflow parameter. min This indicates the preset minimum value for this type of airflow parameter. max and f min This can be preset empirically, for example, for the airflow velocity at a feature point, f max For 100m / s, f min The speed is 0 m / s.
[0052] The correlation can be determined by summarizing single-factor experiments in wind tunnel tests.
[0053] The preset values of airflow parameters directly in front of the insulator on the roof of the high-speed train, the first simulated values of airflow parameters on the surface of the insulator, and the experimental values of the corresponding flashover voltage are used as a set of training samples. The number of training samples can be selected according to needs; for example, in this embodiment of the invention, the number of training samples is 100 sets. The training process adopts a supervised learning method, adjusting the number of neurons in the hidden layer to ensure that the error does not exceed a preset value, and finally obtaining an optimal predictive neural network.
[0054] Therefore, the embodiments of the present invention select the flow field parameters at feature points on the insulator surface that are closely related to the flashover process, as well as the airflow parameters in front of the insulator on the roof of the high-speed train that affect the flow field parameters at the feature points. This can better describe the flashover process of the insulator. These airflow parameters are closely related to the flashover path and flashover voltage of the insulator. Using these airflow parameters as input parameters for artificial intelligence prediction to train the prediction neural network is more accurate, and can reduce experimental data, make extensive use of software simulation data, and improve the speed, effectiveness and accuracy of insulator flashover voltage prediction.
[0055] Step S104: Based on the measured values of the airflow parameters directly in front of the insulator on the roof of the high-speed train, a second simulated value of the airflow parameters on the surface of the insulator on the roof of the high-speed train is obtained through a simulation experiment.
[0056] This simulation experiment is the same as the simulation experiment in step S101, and will not be described again here.
[0057] Step S105: Input the measured values of the airflow parameters directly in front of the high-speed train roof insulator and the corresponding second simulated values of the airflow parameters on the surface of the high-speed train roof insulator into the trained prediction neural network, and output the predicted value of the surface flashover voltage of the high-speed train roof insulator.
[0058] It should be understood that the measured values of the collected airflow parameters, as well as the second simulated values of the obtained airflow parameters, also need to be normalized. The normalization method is the same as the normalization method in the training steps described above, and will not be repeated here.
[0059] Therefore, this embodiment of the invention selects some characteristic points on the surface of the insulator that can characterize the flashover process of the insulator by using the surface flow field distribution characteristics and airflow parameter values of each point on the surface of the insulator under high-speed airflow environment, and uses flow field simulation software to obtain the flow field parameter values at these characteristic points. Using the obtained macroscopic flashover voltage results, combined with artificial intelligence algorithms, the flashover voltage of the insulator in high-speed airflow environment is predicted.
[0060] Furthermore, predictive neural networks can also analyze the influence of certain airflow parameters on the surface flashover voltage of insulators. Specifically, each airflow parameter is assigned a variation range, and these parameters are then freely combined. For example, to obtain the change in surface flashover voltage of insulators caused by changes in airflow velocity and pressure at a feature point using this predictive neural network, the airflow velocity at the feature point is allowed to vary within the range of [100m / s, 150m / s], and the airflow pressure within the range of [7.73e+04Pa, 1.08e+05Pa]. The remaining airflow parameters are kept at a stable value. The airflow velocity and pressure at the feature point are freely combined within this variation range to obtain the input matrix of airflow velocity and pressure at the feature point. This input matrix is then fed into the predictive neural network for calculation, and the output result is the change in surface flashover voltage of the insulator when both airflow velocity and airflow pressure at the feature point change simultaneously. The method for analyzing the influence of other airflow parameter changes on the surface flashover voltage of insulators on the roof of high-speed trains is similar.
[0061] This invention also discloses a computer-readable storage medium storing computer program instructions; when executed by a processor, the computer program instructions implement the method for predicting the surface flashover voltage of a high-speed train roof insulator as described in the above embodiments.
[0062] This invention also discloses a surface flashover voltage prediction system for high-speed train roof insulators, comprising: a computer-readable storage medium as described in the above embodiments.
[0063] In summary, this embodiment of the invention, based on the flow field characteristics of the high-speed rail roof insulator, rationally selects the airflow parameter types that are strongly correlated with the predicted flashover voltage of the high-speed rail roof insulator, simulates and obtains the airflow parameters of each feature point on the insulator surface, and then uses the insulator flow field characteristics to predict the flashover voltage of the insulator surface under airflow conditions. This reduces computational difficulty and complexity while eliminating the influence of multiple weakly correlated or irrelevant variables, thus improving prediction accuracy. It also reduces the uncertainty of the artificial intelligence algorithm, allowing for the verification of network performance using multiple sets of experimental data to obtain the optimal artificial neural network model for the flashover voltage of the high-speed rail roof insulator surface. Furthermore, by using a neural network to better approximate the multidimensional nonlinear relationship between environmental conditions and the flashover voltage of the insulator surface, the prediction results become more accurate.
[0064] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. A method for predicting the surface flashover voltage of a roof insulator in a high-speed train, characterized in that, The high-speed train is in operation, and the method includes: Based on the preset values of airflow parameters directly in front of the roof insulator of the high-speed train, the first simulated values of airflow parameters on the surface of the roof insulator of the high-speed train were obtained through simulation experiments. Based on the preset values of the airflow parameters directly in front of the high-speed train roof insulator and the first simulated values of the corresponding airflow parameters on the surface of the high-speed train roof insulator, the experimental values of the surface flashover voltage of the high-speed train roof insulator under airflow conditions are obtained through wind tunnel experiments. The prediction neural network is trained using the preset value of the airflow parameters in front of the insulator on the roof of the high-speed train, the first simulated value of the airflow parameters on the surface of the insulator on the roof of the high-speed train, and the experimental value of the flashover voltage. Based on the measured values of the airflow parameters collected in front of the roof insulator of the high-speed train, a second simulated value of the airflow parameters on the surface of the roof insulator of the high-speed train was obtained through simulation experiments. The measured values of the airflow parameters directly in front of the high-speed train roof insulator and the corresponding simulated values of the airflow parameters on the surface of the high-speed train roof insulator are input into the trained prediction neural network, which outputs the predicted value of the surface flashover voltage of the high-speed train roof insulator.
2. The method for predicting the surface flashover voltage of a high-speed train roof insulator according to claim 1, characterized in that, The airflow parameters directly in front of the high-speed train roof insulator include: the airflow velocity directly in front of the high-speed train roof insulator, and the airflow direction angle directly in front of the high-speed train roof insulator.
3. The method for predicting the surface flashover voltage of a high-speed train roof insulator according to claim 1, characterized in that, Before the step of obtaining the first simulated value of the airflow parameters on the surface of the high-speed train roof insulator through simulation experiments, the method further includes: Based on the airflow direction angle directly in front of the high-speed train roof insulator, the surface of each skirt of the insulator is divided into four fan-shaped areas of equal area. The fan-shaped area directly facing the airflow in front of the high-speed train roof insulator is the windward side, the fan-shaped area opposite to the windward side is the leeward side, and the remaining two fan-shaped areas are the crosswind side. The feature points of each of the fan-shaped regions are determined, wherein the feature points of each of the fan-shaped regions include: the edge point of the skirt on the center line of each fan-shaped region, the root point of the skirt on the center line of the upper and lower surfaces of each fan-shaped region, and the middle point of the skirt on the center line of the upper and lower surfaces of each fan-shaped region.
4. The method for predicting the surface flashover voltage of a high-speed train roof insulator according to claim 3, characterized in that, The airflow parameters on the surface of the high-speed train roof insulator include: airflow velocity at the feature point, airflow direction angle at the feature point, airflow pressure at the feature point, airflow density at the feature point, average airflow velocity along the fan-shaped area of the insulator in the same direction, and average airflow pressure along the fan-shaped area of the insulator in the same direction.
5. The method for predicting the surface flashover voltage of a high-speed train roof insulator according to claim 4, characterized in that, The predictive neural network comprises an input layer, a hidden layer, and an output layer connected in sequence.
6. The method for predicting the surface flashover voltage of a high-speed train roof insulator according to claim 5, characterized in that: The input layer comprises eight first neurons, the hidden layer comprises five second neurons, and the output layer comprises one third neuron.
7. The method for predicting the surface flashover voltage of a high-speed train roof insulator according to claim 1, characterized in that: The airflow parameters for each category have been normalized.
8. The method for predicting the surface flashover voltage of a high-speed train roof insulator according to claim 7, characterized in that, The normalization process includes: For each type of airflow parameter, if the airflow parameter of that type is positively correlated with the surface flashover voltage of the insulator on the roof of the high-speed train, then by... Normalize the airflow parameters of this type; otherwise, through... The airflow parameters described herein are then normalized. Among them, f n ′(x) represents the normalized airflow parameter of this type, f n (x) represents the airflow parameters of this type before normalization, f max f represents the preset maximum value of the airflow parameter of this type. min This indicates the preset minimum value of the airflow parameters described in this category.
9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer program instructions; when the computer program instructions are executed by a processor, they implement the method for predicting the surface flashover voltage of a high-speed train roof insulator as described in any one of claims 1 to 8.
10. A system for predicting the surface flashover voltage of a high-speed train roof insulator, characterized in that, include: The computer-readable storage medium as described in claim 9.