Lightning current time domain waveform correction method and device, terminal equipment and storage medium

Abstract

The invention discloses a lightning current time domain waveform correction method and device, terminal equipment and a storage medium, and belongs to the technical field of lightning current waveform correction, and the method comprises the steps: carrying out the extraction of lightning features from a lightning current time domain waveform and environment parameters through a data correction model, and extracting association characteristics describing an association relationship between the environment and lightning current characteristics, outputting corrected lightning current waveform data according to the association characteristics and the lightning characteristics, and correcting an initial lightning current model fitting based on a lightning current time domain waveform according to the corrected lightning current waveform data to obtain a lightning current model. And generating a corrected lightning current model. According to the method, the interaction between the environmental parameters and the lightning current characteristics is considered, so that the corrected data better conforms to the actual condition of the lightning current in a complex environment. According to the invention, the method can solve a problem that the correction error of the thunder and lightning model is large because the interaction between the environment parameters and the thunder and lightning current characteristics is ignored in the prior art.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of lightning current waveform correction, and in particular to a lightning current time-domain waveform correction method and device, a terminal device, and a storage medium. BACKGROUND

[0002] Accurate characterization of lightning current time-domain waveform is crucial for lightning protection design of power systems, lightning disaster assessment and other fields. However, lightning current models (such as double exponential models) usually establish standard test conditions or typical environmental parameters when describing lightning current time-domain waveform, and do not fully consider the influence of actual environmental factors on lightning current characteristics.

[0003] The traditional technology usually adjusts the parameters of the standard lightning current model through empirical formula when correcting the lightning current model, and does not consider the environmental parameters of the measurement point, thereby ignoring the influence of environmental parameters on lightning current and failing to capture the internal correlation between environmental parameters and lightning current characteristics. Due to the neglect of the interaction between environmental parameters and lightning current characteristics, the correction error of the lightning model in the complex environment by the traditional method will be large, which may cause the design scheme to fail or the evaluation result to be distorted when applying the lightning model for lightning protection design or disaster assessment, thereby increasing the failure risk of power equipment or causing resource waste. SUMMARY

[0004] The lightning current time-domain waveform correction method, device, terminal device and storage medium provided by the embodiments of the present application can include environmental parameters in the entire model correction process, and generate a correlation feature for representing the correlation between the environment and the lightning current characteristics by combining the environmental parameters and the lightning characteristics, thereby capturing the internal correlation between the environmental parameters and the lightning current characteristics, and effectively solving the problem that the correction error of the lightning model in the complex environment is large due to the neglect of the interaction between the environmental parameters and the lightning current characteristics in the prior art.

[0005] An embodiment of the present application provides a lightning current time-domain waveform correction method, comprising:

[0006] obtaining a current lightning current time-domain waveform and environmental parameters; wherein the lightning current time-domain waveform corresponds to a lightning current waveform data;

[0007] fitting the lightning current time-domain waveform to generate an initial lightning current model; wherein the lightning current model is used to represent the curve of the change of lightning current with time;

[0008] input the lightning current waveform data and the environmental parameter into a preset data correction model, so that the data correction model extracts lightning characteristics for describing lightning current characteristics according to the lightning current waveform data; generates an association feature for representing an association relationship between an environment and lightning current characteristics according to the lightning characteristics and the environmental parameter; and outputs corrected lightning current waveform data according to the association feature and the lightning characteristics.

[0009] correct an initial lightning current model according to the environmental parameter and the corrected lightning current waveform data, and generate a corrected lightning current model.

[0010] Preferably, the lightning current waveform data includes a lightning current amplitude, a wave head time and a wave tail time.

[0011] The data correction model extracts lightning characteristics for describing lightning current characteristics according to the lightning current waveform data, including:

[0012] The data correction model extracts an amplitude feature according to a lightning current amplitude in the lightning current waveform data.

[0013] The data correction model extracts a wave head time feature according to a wave head time in the lightning current waveform data.

[0014] The data correction model extracts a wave tail time feature according to a wave tail time in the lightning current waveform data.

[0015] The amplitude feature, the wave head time feature and the wave tail time feature are all taken as lightning characteristics.

[0016] Preferably, the environmental parameter includes a soil conductivity, and the association feature includes a first association feature and a second association feature.

[0017] The data correction model generates an association feature for representing an association relationship between an environment and lightning current characteristics according to the lightning characteristics and the environmental parameter, including:

[0018] The data correction model generates a first association feature for quantifying an influence degree of soil conductivity on a lightning current amplitude according to an amplitude feature and a soil conductivity.

[0019] The data correction model generates a second association feature for quantifying an influence degree of soil conductivity on a lightning current rising rate according to a wave head time feature and a soil conductivity.

[0020] Preferably, a generation process of the preset data correction model includes:

[0021] a plurality of lightning current time domain waveform samples and an environmental parameter sample corresponding to each lightning current time domain waveform sample are acquired; each lightning current time domain waveform sample corresponds to a lightning current waveform data sample.

[0022] For each lightning current time-domain waveform sample, a corresponding lightning current model sample is constructed according to the lightning current waveform data sample;

[0023] A multi-objective optimization model is constructed according to the lightning current model sample, with the objective of minimizing the time-domain root mean square error and minimizing the frequency-domain energy relative error; wherein the time-domain root mean square error is used to represent the fitting degree of the lightning current model sample to the energy distribution of the lightning current waveform in the time domain; and the frequency-domain energy relative error is used to represent the fitting degree of the lightning current model sample to the energy distribution of the lightning current waveform in the frequency domain;

[0024] The multi-objective optimization model is solved, and the corrected lightning current model sample is generated when the time-domain root mean square error is minimized and the frequency-domain energy relative error is minimized; wherein the corrected lightning current model sample includes a corrected lightning current waveform data sample;

[0025] The lightning current waveform data sample corresponding to the lightning current time-domain waveform sample and the environmental parameter sample are used as training samples;

[0026] Each training sample and the actual corrected lightning current waveform data sample of each training sample are used as input, and the lightning current waveform data sample correction prediction result of each training sample is used as output, and the data correction model to be trained is iteratively trained until the model converges, and a preset data correction model is generated.

[0027] Preferably, the data correction model includes a plurality of input layer neurons and a plurality of hidden layer neurons;

[0028] The data correction model extracts lightning characteristics for describing lightning current characteristics from lightning current waveform data, including:

[0029] Through each input layer neuron, the parameter value signal in the lightning current waveform data is transmitted to each hidden layer neuron;

[0030] For each hidden layer neuron, an input signal of the hidden layer neuron is generated according to the parameter value signal, the connection weight of the hidden layer neuron, and the preset offset of the hidden layer neuron; wherein the connection weight is used to measure the importance of the signal transmitted by the input layer neuron to the hidden layer neuron;

[0031] For each hidden layer neuron, the input signal of the hidden layer neuron is linearly transformed by an activation function to generate an output signal of the hidden layer neuron;

[0032] The output signals of each hidden layer neuron are weighted and summed to generate lightning characteristics for describing lightning current characteristics.

[0033] Preferably, the lightning current time domain waveform sample further corresponds to a preset lightning current time domain reference waveform sample;

[0034] The multi-objective optimization model is constructed based on the lightning current model sample with the goal of minimizing the time domain root mean square error and minimizing the frequency domain energy relative error, including:

[0035] The multi-objective optimization model is constructed according to the following formula:

[0036]

[0037] minε band ;

[0038]

[0039] Among them, MSE is the root mean square error in time domain, I real (t i ) is the lightning current waveform data corresponding to the lightning current time domain reference waveform sample at the i-th sampling time, I model (t i ) is the lightning current waveform data sample corresponding to the lightning current model sample at the i-th sampling time;

[0040] n is the total number of sampling time points corresponding to the lightning current model samples and the lightning current time domain reference waveform samples, ε band is the relative error of frequency domain energy, E model,band is the total energy of the lightning current waveform of the lightning current model sample, E real,band It is the total energy of the lightning current waveform of the lightning current time domain reference waveform sample.

[0041] Preferably, the corrected lightning current waveform data includes: a corrected lightning current amplitude, a corrected wave front time, and a corrected wave tail time; the lightning current time domain waveform also corresponds to a preset lightning current time domain reference waveform;

[0042] The modified lightning current model includes:

[0043]

[0044] Where i(t) is the instantaneous value of lightning current at time t, α is the corrected lightning current amplitude, β is the corrected wave head time, γ is the corrected wave tail time, I m is the peak current of the modified lightning current model, I pThe peak current is a time-domain reference waveform of lightning current, T1 is a head time of the time-domain reference waveform of lightning current, T2 is a tail time of the time-domain reference waveform of lightning current, sigma is soil conductivity, tau1 is a head time constant, tau2 is a tail time constant, and s is a preset parameter for affecting the speed of the size of lightning current rising over time.

[0045] On the basis of the method embodiments described above, the application correspondingly provides device embodiments.

[0046] An embodiment of the application provides a lightning current time-domain waveform correction device, which comprises a data acquisition module, a lightning current model fitting module, a data correction module and a lightning current model correction module.

[0047] The data acquisition module is configured to acquire a current lightning current time-domain waveform and environmental parameters, wherein the lightning current time-domain waveform corresponds to lightning current waveform data.

[0048] The lightning current model fitting module is configured to fit the lightning current time-domain waveform to generate an initial lightning current model, wherein the lightning current model is used to represent a curve of lightning current change over time.

[0049] The data correction module is configured to input the lightning current waveform data and the environmental parameters into a preset data correction model, so that the data correction model extracts lightning characteristics for describing lightning current characteristics according to the lightning current waveform data, generates correlation characteristics for representing the correlation between the environment and the lightning current characteristics according to the lightning characteristics and the environmental parameters, and outputs corrected lightning current waveform data according to the correlation characteristics and the lightning characteristics.

[0050] The lightning current model correction module is configured to correct the initial lightning current model according to the environmental parameters and the corrected lightning current waveform data to generate a corrected lightning current model.

[0051] On the basis of the method embodiments described above, the application correspondingly provides terminal device embodiments.

[0052] Another embodiment of the application provides a terminal device, which comprises a processor, a memory and a computer program stored in the memory and configured to be executed by the processor, and the processor implements the lightning current time-domain waveform correction method described in the above-mentioned application embodiments when executing the computer program.

[0053] On the basis of the method embodiments described above, the application correspondingly provides storage medium embodiments.

[0054] Another embodiment of the present application provides a storage medium, the computer readable storage medium comprises a stored computer program, wherein the computer readable storage medium controls the device where the computer readable storage medium is located to execute the lightning current time domain waveform correction method of the above-mentioned embodiment of the application when the computer program runs.

[0055] By implementing the present application, the following beneficial effects are achieved:

[0056] The embodiment of the present application provides a lightning current time domain waveform correction method, device, terminal equipment and storage medium. After fitting an initial lightning current model based on the current lightning current time domain waveform, the lightning current waveform data corresponding to the lightning current time domain waveform and the environmental parameters are taken as the input of the correction model, so that the lightning characteristics of the lightning current waveform data are extracted through the data correction model, the association characteristics for representing the association relationship between the environment and the lightning current characteristics are generated according to the lightning characteristics and the environmental parameters, and finally the corrected lightning current waveform data are output through the data correction model according to the association characteristics and the lightning characteristics, so as to correct the initial lightning current model according to the corrected lightning current waveform data and generate the corrected lightning current model. Compared with the prior art, the environmental parameters can be included in the whole model correction process, and by combining the environmental parameters and the lightning characteristics, the association characteristics for representing the association relationship between the environment and the lightning current characteristics are generated, and the internal association between the environmental parameters and the lightning current characteristics can be captured. Therefore, the interaction between the environmental parameters and the lightning current characteristics is considered, so that the corrected data is more consistent with the actual situation of the lightning current in the complex environment, the correction error of the lightning model in the complex environment can be effectively reduced compared with the traditional method, so that more accurate parameters can be provided when the corrected lightning current model is applied to lightning protection design or disaster assessment, the situation that the design scheme fails or the assessment result is distorted due to parameter deviation is reduced, the failure risk of the power equipment is reduced, and resource waste is avoided. BRIEF DESCRIPTION OF DRAWINGS

[0057] Figure 1 is a flowchart of a lightning current time domain waveform correction method provided by an embodiment of the present application.

[0058] Figure 2 is a flowchart of a prediction correction function value based on a neural network provided by an embodiment of the present application.

[0059] Figure 3 is a neural network model structure diagram provided by an embodiment of the present application.

[0060] Figure 4 is a structure diagram of a lightning current time domain waveform correction device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0061] The technical solutions in the embodiments of the present application will be clearly and completely described 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 of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative effort belong to the scope of protection of the present application.

[0062] As shown in the formula (1), in order to solve the problem that the correction error of the lightning model in a complex environment is large due to the neglect of the interaction between the environmental parameters and the lightning current characteristics, the design scheme is invalid or the evaluation result is distorted due to the parameter deviation when the lightning model is applied to lightning protection design or disaster evaluation, and the fault risk of the power equipment is increased or the resource is wasted, an embodiment of the present application provides a correction method for a lightning current time-domain waveform, comprising the following steps: Figure 1 Step s1: acquiring a current lightning current time-domain waveform and environmental parameters; wherein the lightning current time-domain waveform corresponds to lightning current waveform data;

[0063] Illustratively, the present application can collect the lightning current waveform generated in natural lightning strike or artificial lightning test by lightning observation equipment (such as a high-speed oscilloscope, a Rogowski coil, etc.), the waveform records the original data of the change of the lightning current with time (i.e. the lightning current time-domain waveform), then the key parameters can be extracted from the lightning current time-domain waveform as the lightning current waveform data. And the environmental data related to the lightning current waveform is also collected synchronously, such as the environmental parameters directly affecting the propagation and loss of lightning energy in the soil.

[0064] Step s2: fitting the lightning current time-domain waveform to generate an initial lightning current model; wherein the lightning current model is used to represent the curve of the change of the lightning current with time;

[0065] Illustratively, the present application can preliminarily fit the measured lightning current time-domain waveform by using the classic lightning current waveform model (such as the Heidler model) to obtain the initial model parameters, so as to generate a curve preliminarily reflecting the change trend of the lightning current, thereby constructing the initial lightning current model. It can be understood that in this step, a benchmark model based on the measured data can be established as the starting point for subsequent optimization, and the initial lightning current model may have fitting error due to the neglect of the environmental parameters, and then needs to be further corrected.

[0066]

[0067] ​Step s3: Inputting the lightning current waveform data and environmental parameters into a preset data correction model, so that the data correction model extracts lightning features for describing lightning current characteristics based on the lightning current waveform data; generating correlation features for characterizing the correlation relationship between the environment and the lightning current characteristics based on the lightning features and the environmental parameters; and outputting corrected lightning current waveform data based on the correlation features and the lightning features;

[0068] In schematic form, based on the trained data correction model, lightning characteristics and related features can be extracted according to the currently measured lightning current waveform data and environmental parameters, so as to adjust the original waveform data according to the above characteristics and generate corrected lightning current waveform data to assist in the subsequent correction of the initial lightning current model.

[0069] Step s4: modifying the initial lightning current model according to the environmental parameters and the modified lightning current waveform data to generate a modified lightning current model;

[0070] Indicatively, the corrected waveform data obtained in step s3 can be substituted into the initial lightning current model, and the parameters of the lightning current waveform data in the model can be adjusted so that the corrected model can be closer to the measured data, realizing the model correction process combined with environmental parameters, and significantly improving the fitting accuracy of the model.

[0071] Furthermore, accurate lightning current models can optimize the parameter design of power system lightning protection equipment (such as lightning arresters and grounding systems), reducing the risks of tripping and equipment damage caused by lightning strikes.

[0072] Moreover, the present invention can automatically learn the lightning current characteristics under different environments through the data correction model. It can be applied to multi-source data such as natural lightning strikes and artificial lightning, as well as complex scenes such as high altitudes, coastal areas, and mountainous areas, avoiding the limitations of traditional models that rely on a single data source.

[0073] For step s1, in a preferred embodiment, the lightning current waveform data includes: lightning current amplitude, wave front time and wave tail time; the environmental parameters include: soil conductivity.

[0074] Schematically, the lightning current amplitude refers to the maximum current value in the waveform (unit: kA), reflecting the intensity of the lightning discharge.

[0075] The wave front time refers to the time required for the lightning current to rise from 10% to 90% of the peak value (unit: μs), which reflects the speed of the current rise.

[0076] The tail time refers to the time required for the lightning current to drop from the peak value to 50% of the peak value (unit: μs), reflecting the characteristics of the current attenuation.

[0077] In the embodiments of the present application, the soil conductivity is taken as an environmental parameter, the value of which is related to the soil type, humidity, salt content, etc., and can directly affect the propagation and loss of lightning energy in the soil.

[0078] Illustratively, the smaller the soil conductivity, the smaller the lightning peak current and the longer the time to reach the peak. That is, the soil conductivity can affect the amplitude of the lightning current and the time for the wave front to reach the peak, so the present embodiments can consider the influence of the soil conductivity, accurately correct the lightning current waveform data, and thus further realize accurate correction of the lightning current model.

[0079] In a preferred embodiment, after collecting the current lightning time-domain waveform, the present application can perform preprocessing on the lightning time-domain waveform before fitting the lightning time-domain waveform, specifically:

[0080] By eliminating high-frequency noise in the lightning time-domain waveform through wavelet transform and performing smoothing processing on the lightning time-domain waveform, important features in the lightning time-domain waveform can be retained, and the quality and analyzability of the lightning time-domain waveform can be significantly improved.

[0081] For step s2, in a preferred embodiment, based on the preprocessed lightning time-domain waveform, the preprocessed lightning time-domain waveform can be preliminarily fitted by the Heidler model to generate an initial lightning current model.

[0082] Illustratively, the parameters of the Heidler model can be initialized according to the lightning current amplitude, wave front time, and wave tail time in the lightning time-domain waveform to generate a curve that can reflect the trend of the change of the lightning current over time.

[0083] It can be understood that the Heidler model fitting the lightning time-domain waveform will have a large fitting error; for example, when the lightning current amplitude input is constant, the wave tail time is fixed, the wave front time is increased, and the wave tail time will actually also increase, and the lightning current amplitude will decrease. When the lightning current amplitude input is constant, the wave front time is fixed, the wave tail time is increased, and the wave front time will actually slightly increase; if the input amplitude is changed, the wave front time and the wave tail time of the waveform will also change.

[0084] Therefore, only by fitting the naturally observed lightning current and the artificially induced lightning test lightning current through the Heidler model, a large fitting error will occur, and a nonlinear correction function needs to be introduced to optimize the Heidler model, so as to optimize the Heidler model based on the value of the correction function, obtain a corrected Heidler model, and improve the accuracy of lightning data measurement.

[0085] For step s3, in one preferred embodiment, the data correction model of the present application is a BP neural network (BackPropagation Neural Network), which is a multi-layer feedforward neural network based on error backpropagation algorithm. In the present application, the BP neural network is the core network component of the data correction model, which is used to extract the features of the lightning waveform data and generate the corrected data.

[0086] In one preferred embodiment, the generation process of the preset data correction model comprises:

[0087] Obtaining a plurality of lightning current time-domain waveform samples and a plurality of environmental parameter samples corresponding to the plurality of lightning current time-domain waveform samples; wherein each lightning current time-domain waveform sample corresponds to a lightning current waveform data sample;

[0088] For each lightning current time-domain waveform sample, constructing a corresponding lightning current model sample according to the lightning current waveform data sample;

[0089] According to the lightning current model sample, constructing a multi-objective optimization model with the objective of minimizing the time-domain root mean square error and minimizing the frequency-domain energy relative error; wherein the time-domain root mean square error is used to represent the fitting degree of the lightning current model sample to the lightning current waveform energy distribution in the time domain; and the frequency-domain energy relative error is used to represent the fitting degree of the lightning current model sample to the lightning current waveform energy distribution in the frequency domain;

[0090] Solving the multi-objective optimization model, and generating a corrected lightning current model sample when the time-domain root mean square error is minimized and the frequency-domain energy relative error is minimized; wherein the corrected lightning current model sample comprises a corrected lightning current waveform data sample;

[0091] Taking the lightning current waveform data sample corresponding to the lightning current time-domain waveform sample and the environmental parameter sample as the training sample;

[0092] Taking each training sample and the actual corrected lightning current waveform data sample of each training sample as input, and taking the lightning current waveform data sample correction prediction result of each training sample as output, iteratively training the data correction model to be trained until the model converges, and generating a preset data correction model.

[0093] The lightning current waveform data sample prediction correction result is: the data correction result predicted by the data correction model to be trained according to the lightning current waveform data sample and the environmental parameter sample in the training sample after correcting the lightning current waveform data sample.

[0094] Illustratively, in the embodiments of the present application, the generation process of the data correction model combines multi-objective optimization and neural network training, that is, in the training process of the neural network, the training samples of the model are obtained through multi-objective model optimization, so that the corrected lightning current waveform data samples are used as the actual results that should be generated during model training.

[0095] Specifically, first is sample acquisition and model construction, that is, a plurality of lightning current time-domain waveform samples (including amplitude, wave head time, and wave tail time) and corresponding environmental parameters (such as soil conductivity) are acquired.

[0096] For each waveform sample, an initial lightning current model sample is generated by using the Heidler model fitting. In order to correct each initial lightning current model sample and obtain an accurate lightning current model sample after correction (which can be used as a subsequent model training sample), the present application first constructs a multi-objective optimization model according to the lightning current model sample.

[0097] In a preferred embodiment, the lightning current time-domain waveform sample also corresponds to a preset lightning current time-domain reference waveform sample.

[0098] Therefore, when constructing the multi-objective optimization model with the objectives of minimizing the time-domain root mean square error and minimizing the frequency domain energy relative error, the specific process is as follows:

[0099] The multi-objective optimization model is constructed according to the following formula:

[0100]

[0101] minε band ;

[0102]

[0103] Wherein, MSE is the time-domain root mean square error, I real (t i ) is the lightning current waveform data corresponding to the i-th sampling time of the lightning current time-domain reference waveform sample, I model (t i ) is the lightning current waveform data sample corresponding to the i-th sampling time of the lightning current model sample.

[0104] n is the total number of sampling time points corresponding to the lightning current model sample and the lightning current time-domain reference waveform sample, ε band is the frequency domain energy relative error, E model,band is the total energy of the lightning current waveform of the lightning current model sample, E real,band is the total energy of the lightning current waveform of the lightning current time-domain reference waveform sample.

[0105] It can be understood that the multi-objective optimization model established by the application is to minimize the time domain root mean square error and the frequency energy relative error, and then the multi-objective optimization model is solved by an optimization algorithm, so that the corrected function value can be obtained, so that the neural network model can be trained based on the lightning current parameters (peak current, wave head time, wave tail time), environmental parameters (soil conductivity) and the corrected function value (amplitude correction function value, wave head time correction function value and wave tail time correction function value), so that the neural network model can mine the complex nonlinear relationship between the lightning current parameters and the correction parameter function, so as to establish the mapping relationship between the input and the output, and the corrected function value can be quickly predicted according to the measured lightning current parameters and environmental parameters in subsequent application, the calculation speed and the accuracy of the model are improved, and the output corrected function value is realized. The initial lightning current model can be corrected.

[0106] In a preferred embodiment, as shown in the flowchart, Figure 2 The specific processing steps are as follows:

[0107] Step 1: Collect the lightning current time domain waveform of natural observation and artificial lightning, eliminate high frequency noise by wavelet transform, smooth the waveform, remove interference to improve data quality, and extract feature parameters from the preprocessed waveform, including lightning current amplitude, wave head time and wave tail time;

[0108] Step 2: Three waveform correction functions are proposed for the Heidler model, the influence of soil conductivity on lightning is considered, and the frequency energy error between the measured waveform and the waveform of the Heidler model is calculated. The Heidler model is a mathematical model for simulating lightning current changes, and the waveform is the sum of two exponential functions.

[0109] Step 3: A multi-objective optimization model is established and solved, and the values of the three correction functions under different lightning current parameters are obtained; wherein the three correction functions under different lightning current parameters can be: lightning current amplitude correction function, wave head time correction function and wave tail time correction function; and the values of different correction functions can be: corrected lightning current amplitude, corrected wave head time and corrected wave tail time.

[0110] Step 4: Establish a neural network, train the neural network based on the corrected lightning current amplitude, corrected wave head time and corrected wave tail time in step 3, the feature parameters extracted in step 2, and the corresponding soil conductivity. After the neural network is trained, the corrected function value can be directly predicted by the neural network based on the measured lightning current feature parameters and soil conductivity.

[0111] Therefore, the embodiment of the present application establishes a multi-objective optimization model, solves and obtains the correction function value under different lightning current parameters, and then can predict the correction function value based on the neural network, using the measured lightning current characteristic parameters and soil conductivity to enhance the adaptability and prediction ability of the model, so that the correction result can be quickly and accurately given.

[0112] In a preferred embodiment, the lightning current time domain reference waveform sample and the model fitted lightning current time domain waveform can be respectively subjected to discrete Fourier transform to obtain a reference waveform discrete frequency spectrum sequence and a fitted waveform discrete frequency spectrum sequence; the reference waveform discrete frequency spectrum sequence and the fitted waveform discrete frequency spectrum sequence are respectively subjected to energy spectrum density calculation and summation to obtain a reference waveform total energy and a fitted waveform total energy; and finally, the relative error between the measured waveform total energy and the fitted waveform total energy is calculated to obtain the frequency energy relative error ε band .

[0113] Specifically, the reference lightning current time domain waveform and the model fitted lightning current time domain waveform are respectively subjected to discrete Fourier transform, and the calculation formula is as follows: In the formula, F real (k) is the discrete frequency spectrum sequence of the reference lightning current time domain waveform, F model (k) is the discrete frequency spectrum sequence of the model fitted lightning current time domain waveform, k is the index of the frequency component, n is the total number of sampling time points corresponding to the lightning current model sample and the lightning current time domain reference waveform sample, I real (t i ) is the lightning current waveform data corresponding to the i-th sampling time of the lightning current time domain reference waveform sample, and I model (t i ) is the lightning current waveform data sample corresponding to the i-th sampling time of the lightning current model sample.

[0114] According to the above discrete frequency spectrum sequence, the energy spectrum density is calculated, and the calculation formula is as follows:

[0115]

[0116] In the formula, E real (k) is the value of the energy spectrum density of the reference lightning current waveform at the k-th frequency component, F real (k) is the value of the discrete frequency spectrum sequence of the reference lightning current time domain waveform at the k-th frequency component, E model (k) is the value of the energy spectrum density of the model fitted lightning current waveform at the k-th frequency component, F model (k) is the value of the discrete frequency spectrum sequence of the model fitted lightning current time domain waveform at the k-th frequency component.

[0117] By summing the above energy density discrete points, the total energy of the lightning current waveform of the lightning current model sample is obtained as E model,band The total energy of the lightning current waveform of the lightning current time domain reference waveform sample (i.e. the reference lightning current waveform) is E real,band , and thus calculate the relative error of frequency domain energy ε band .

[0118] Therefore, when optimizing the multi-objective model, the present invention can achieve dual-objective optimization, ensuring that the multi-objective model is consistent with the actual lightning data in both the time domain (waveform shape) and the frequency domain (energy distribution), so that the correction samples generated by the multi-objective optimization meet the accuracy requirements of the time domain and frequency domain at the same time, providing more accurate training data for the neural network, so that the neural network can more accurately capture the nonlinear relationship between the lightning current parameters and the correction function by learning these samples, and ultimately output a more reliable correction result.

[0119] In a preferred embodiment, before training the data correction model, characteristic parameters may be standardized, wherein the characteristic parameters may be lightning current waveform data samples and environmental parameter samples corresponding to lightning current time domain waveform samples.

[0120] Specifically, because the units of each feature parameter are different, the order of magnitude of each feature parameter may also be different. Small-scale feature parameters are often ignored due to the presence of large-scale feature parameters. Therefore, in order to eliminate the influence of unit and scale differences between features, each dimension can be treated equally, thereby standardizing the feature parameters. The standardization formula is as follows:

[0121]

[0122] Where, T'1 is the normalized wave head time, T'2 is the normalized wave tail time, and I' p For standardization

[0123] The lightning amplitude after the normalization, σ is the normalized soil conductivity, μ is the mean value of the characteristic parameter, S is the variance of the characteristic parameter, T1 is the wave head time of the lightning current time domain reference waveform, T2 is the wave tail time of the lightning current time domain reference waveform, I p is the peak current of the lightning current time domain reference waveform, and σ is the soil conductivity.

[0124] In a preferred embodiment, Figure 3 As shown, the data correction model includes an input layer, a hidden layer, and an output layer. There can be several nodes between layers, and the connection status of the nodes between layers is reflected by weights.

[0125] Each node can be regarded as a neuron, each neuron receives input signals from other neurons, each signal is transmitted through a connection with a weight, the neuron adds up the signals to obtain a total input value, then compares the total input value with the threshold value of the neuron (simulates the threshold potential), processes the final output through an activation function, and the output is used as the input of the next neuron.

[0126] Specifically, in the embodiments of the present application, the input layer has 4 nodes corresponding to 4 input parameters, the output layer has 3 nodes corresponding to 3 corrected function values, and the number of nodes in the hidden layer can be adjusted according to actual conditions.

[0127] In a preferred embodiment, the data correction model comprises: an input layer composed of a plurality of input layer neurons, and a hidden layer composed of a plurality of hidden layer neurons; when lightning characteristics used to describe lightning current characteristics are extracted from lightning current waveform data, specifically comprising:

[0128] Through each input layer neuron, the parameter value signal in the lightning current waveform data is transmitted to each hidden layer neuron;

[0129] For each hidden layer neuron, an input signal of the hidden layer neuron is generated according to the parameter value signal, a connection weight of the hidden layer neuron, and a preset offset of the hidden layer neuron; wherein the connection weight is used to measure the importance of the signal transmitted by the input layer neuron to the hidden layer neuron;

[0130] For each hidden layer neuron, the input signal of the hidden layer neuron is linearly transformed by an activation function to generate an output signal of the hidden layer neuron;

[0131] The output signals of the hidden layer neurons are weighted and summed to generate lightning characteristics used to describe lightning current characteristics.

[0132] Illustratively, through the nonlinear transformation of the activation function, the data correction model can automatically extract the complex nonlinear relationship between the lightning current parameters. For example: the influence of soil conductivity on lightning current amplitude is not linear, low conductivity soil may cause rapid amplitude decay, while high conductivity soil has less influence. The relationship between the wave head time and the soil conductivity may present different trends in different regions, then the data correction model of the present application can adaptively learn these regional characteristics through different neurons and different weights, so as to obtain the nonlinear relationship between the lightning current parameters.

[0133] In a preferred embodiment, the data correction model of the present application can realize forward propagation, so that information enters the network from the input layer, and is calculated through each layer in turn to obtain the final output layer result. Then:

[0134] From input layer to hidden layer:

[0135]

[0136] where, a h represents the input of the hth hidden layer neuron, v ih represents the weight of the ith input to the hth hidden layer neuron, θ h represents the bias of the input to the hth hidden layer neuron, d represents the number of input variables, and in the embodiment of the present application, d = 4.

[0137] From hidden layer to output layer:

[0138]

[0139] where, β j represents the input of the jth output neuron, w hj represents the weight of the hth hidden layer to the jth output neuron, b h represents the output of the hth hidden layer, θ j represents the bias of the hidden layer neuron to the jth output, and q represents the number of hidden layer neurons.

[0140] Further, the present application can also adjust the network parameters by calculating the error between the output layer and the expected value, so as to make the error smaller.

[0141] Through continuous iteration and weight update, the error will eventually reach the minimum, and the network is completed. At this time, the weights and biases from the hidden layer to the output layer, and the weights and biases from the input layer to the hidden layer are known. The input is substituted into the forward propagation process to predict the corrected lightning current waveform data.

[0142] In a preferred embodiment, the number of layers of the hidden layer of the neural network of the data correction model is set to 12-16 layers, which can better cope with complex nonlinear problems, ensuring the fitting accuracy and faster calculation speed. The training samples can also be divided into a training set, a validation set and a test set, the data correction model is trained, and its performance is evaluated according to the root mean square error. When the root mean square error reaches a preset threshold, the training of the data correction model is stopped to optimize the data correction model.

[0143] Further, the lightning characteristics used to describe the characteristics of lightning current can also be extracted from the lightning current waveform data based on the trained data correction model, that is:

[0144] The data correction model extracts the amplitude feature according to the amplitude of the lightning current in the lightning current waveform data.

[0145] According to the wave head time in the lightning current waveform data, a wave head time feature is extracted;

[0146] According to the wave tail time in the lightning current waveform data, a wave tail time feature is extracted;

[0147] The amplitude feature, the wave head time feature, and the wave tail time feature are all taken as lightning features.

[0148] Further, after the amplitude feature, the wave head time feature, and the wave tail time feature are obtained, a correlation feature can also be obtained based on the amplitude feature, the wave head time feature, the wave tail time feature, and the soil conductivity of the environmental parameter; illustratively, the correlation feature includes a first correlation feature and a second correlation feature; then:

[0149] The generating, according to the lightning features and the environmental parameter, of the correlation feature for representing the correlation between the environment and the lightning current characteristics includes:

[0150] Generating, according to the amplitude feature and the soil conductivity, of a first correlation feature for quantifying the influence degree of the soil conductivity on the lightning current amplitude;

[0151] Generating, according to the wave head time feature and the soil conductivity, of a second correlation feature for quantifying the influence degree of the soil conductivity on the lightning current rising rate.

[0152] Illustratively, after receiving the lightning current amplitude in the lightning current waveform data, the data correction model can process the lightning current amplitude through the neuron network structure inside the model. According to the pattern and weight learned by its training, the amplitude feature capable of representing the characteristics of the lightning current amplitude is extracted from the lightning current amplitude, such as analyzing the size, change trend, and correlation with other parameters of the amplitude, so as to obtain the amplitude feature.

[0153] Similarly, the neuron network of the model is used to analyze and process the wave head time, extract the wave head time feature reflecting the characteristics of the wave head time, and extract the wave tail time feature from the wave tail time.

[0154] Further, the data correction model combines the extracted amplitude feature with the soil conductivity in the environmental parameter to generate the first correlation feature for quantifying the influence degree of the soil conductivity on the lightning current amplitude, so as to determine how the soil conductivity specifically affects the lightning current amplitude. The wave head time feature and the soil conductivity are also taken as inputs, so that the second correlation feature for quantifying the influence degree of the soil conductivity on the lightning current rising rate can be generated.

[0155] The amplitude feature reflects the intensity of the lightning current, the wave head time feature embodies the rising speed and steepness of the lightning current, and the wave tail time feature relates to the attenuation of the lightning current. Therefore, the embodiment of the present application can comprehensively describe the characteristics of the lightning current from multiple dimensions by extracting the amplitude feature, the wave head time feature and the wave tail time feature respectively.

[0156] Moreover, the generated first and second correlation features can clearly reveal the relationship between soil conductivity and lightning current amplitude and lightning current rising rate, so that the model can deeply understand how environmental factors affect the characteristics of lightning current. By considering the complex relationship between environmental parameters and lightning current characteristics, the model can better adapt to different environmental conditions and improve its prediction and analysis capabilities in various situations.

[0157] For step s4, in a preferred embodiment, the corrected lightning current waveform data includes: corrected lightning current amplitude, corrected wave head time and corrected wave tail time; and the lightning current time-domain waveform also corresponds to a preset lightning current time-domain reference waveform.

[0158] The corrected lightning current model includes:

[0159]

[0160] Wherein, i(t) is the instantaneous value of the lightning current at time t, a is the corrected lightning current amplitude, β is the corrected wave head time, γ is the corrected wave tail time, I m is the peak current of the corrected lightning current model, I p is the peak current of the lightning current time-domain reference waveform, T1 is the wave head time of the lightning current time-domain reference waveform, T2 is the wave tail time of the lightning current time-domain reference waveform, σ is the soil conductivity, τ1 is the wave head time constant, τ2 is the wave tail time constant, and s is a preset parameter for affecting the speed of the size of the lightning current rising over time.

[0161] Illustratively, each parameter (such as the corrected lightning current amplitude, the wave head time and the wave tail time) in the corrected lightning current model is predicted by the data correction model, and the training samples of the data correction model are also obtained by multi-objective optimization, so that the data correction model can obtain the corrected data under the condition of minimizing the time-domain root mean square error and the frequency domain energy relative error. Therefore, the initial lightning current model is corrected according to the corrected data predicted by the data correction model, so that the corrected lightning current model can better fit the actual lightning current waveform in the time domain and the frequency domain.

[0162] As Figure 4 shown, on the basis of the embodiments of the above-mentioned lightning current time-domain waveform correction method, the present application correspondingly provides device item embodiments;

[0163] An embodiment of the present application provides a lightning current time domain waveform correction device, comprising a data acquisition module, a lightning current model fitting module, a data correction module and a lightning current model correction module.

[0164] The data acquisition module is used for acquiring a current lightning current time domain waveform and environmental parameters; wherein the lightning current time domain waveform corresponds to a lightning current waveform data.

[0165] The lightning current model fitting module is used for fitting the lightning current time domain waveform to generate an initial lightning current model; wherein the lightning current model is used for representing a curve of lightning current change over time.

[0166] The data correction module is used for inputting the lightning current waveform data and the environmental parameters into a preset data correction model, so that the data correction model extracts lightning characteristics for describing lightning current characteristics according to the lightning current waveform data; generates an association characteristic for representing an association relationship between the environment and the lightning current characteristics according to the lightning characteristics and the environmental parameters; and outputs corrected lightning current waveform data according to the association characteristic and the lightning characteristics.

[0167] The lightning current model correction module is used for correcting the initial lightning current model according to the environmental parameters and the corrected lightning current waveform data to generate a corrected lightning current model.

[0168] It should be noted that the above-described device embodiments are only illustrative, wherein the modules described as separate components can or can not be physically separated, and the components displayed as modules can or can not be physical modules, which can be located in one place or distributed on multiple network modules. Part or all of the modules can be selected to achieve the purpose of the present embodiment according to actual needs. In addition, the connection relationship between the modules in the device embodiment provided by the present application indicates that there is a communication connection between them, which can be implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement it without creative labor.

[0169] Those skilled in the art can clearly understand that, in order to facilitate and be brief, the specific working process of the above-described device can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0170] On the basis of the above-described various lightning current time domain waveform correction method embodiments, the present application correspondingly provides terminal device embodiments.

[0171] An embodiment of the present application provides a terminal device, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and the processor implements a lightning current time-domain waveform correction method according to any one of the method embodiments of the present application when executing the computer program.

[0172] The terminal device can be a desktop computer, a notebook computer, a palm computer, a cloud server, and the like. The terminal device can include, but is not limited to, a processor and a memory.

[0173] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or the like. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor. The processor is a control center of the terminal device, and is connected to various parts of the terminal device through various interfaces and lines.

[0174] The memory can be used to store the computer program, and the processor realizes various functions of the terminal device by running or executing the computer program stored in the memory and calling data stored in the memory. The memory can mainly include a program storage area and a data storage area. The program storage area can store an operating system, at least one application required by a function, and the like; and the data storage area can store data created according to use of the terminal device, and the like. In addition, the memory can include a high-speed random access memory, and can also include a nonvolatile memory, for example, a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state memory devices.

[0175] On the basis of the above-mentioned various lightning current time-domain waveform correction method embodiments, the present application correspondingly provides a storage medium embodiment.

[0176] An embodiment of the present application provides a storage medium, the storage medium comprising a stored computer program, wherein the computer readable storage medium is controlled to perform a method for modifying a lightning current time-domain waveform according to any one of the method embodiments of the present application when the computer program is run.

[0177] The storage medium is a computer readable storage medium, and the computer program is stored in the computer readable storage medium. When the computer program is executed by a processor, the steps of each method embodiment described above can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable files or some intermediate forms. The computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content included in the computer readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer readable medium does not include electrical carrier signals and telecommunication signals.

[0178] The above is the preferred embodiment of the present application. It should be pointed out that for ordinary skilled persons in the technical field, some improvements and refinements can be made without departing from the principles of the present application, and these improvements and refinements are also considered to be within the protection scope of the present application.

Claims

1. A method for correcting a lightning current time domain waveform, characterized in that: include: Obtaining a current lightning current time domain waveform and environmental parameters; wherein the lightning current time domain waveform corresponds to a lightning current waveform data; Fitting the lightning current time domain waveform to generate an initial lightning current model; wherein the lightning current model is used to characterize a curve of lightning current changing with time; Inputting the lightning current waveform data and environmental parameters into a preset data correction model, so that the data correction model extracts lightning characteristics for describing lightning current characteristics based on the lightning current waveform data; generating correlation features for characterizing the correlation relationship between the environment and the lightning current characteristics based on the lightning characteristics and the environmental parameters; and outputting corrected lightning current waveform data based on the correlation features and the lightning characteristics; The initial lightning current model is corrected according to the environmental parameters and the corrected lightning current waveform data to generate a corrected lightning current model.

2. A method for correcting a lightning current time domain waveform according to claim 1, characterized in that: The lightning current waveform data includes: lightning current amplitude, wave head time and wave tail time; The data correction model extracts lightning features for describing lightning current characteristics based on lightning current waveform data, including: The data correction model extracts amplitude features based on the lightning current amplitude in the lightning current waveform data; Extracting the wave front time feature according to the wave front time in the lightning current waveform data; According to the tail time in the lightning current waveform data, the tail time feature is extracted; The amplitude characteristics, wave head time characteristics and wave tail time characteristics are all used as lightning characteristics.

3. The method for correcting a lightning current time domain waveform according to claim 2, wherein: The environmental parameters include soil conductivity; the associated features include a first associated feature and a second associated feature; The generating, based on the lightning characteristics and the environmental parameters, a correlation feature for characterizing the correlation relationship between the environment and the lightning current characteristics includes: generating a first correlation feature for quantifying the influence of soil conductivity on the lightning current amplitude based on the amplitude feature and soil conductivity; According to the wave front time characteristic and the soil conductivity, a second correlation feature is generated for quantifying the influence of the soil conductivity on the lightning current rise rate.

4. A method for correcting a lightning current time domain waveform according to claim 3, characterized in that: The generation process of the preset data correction model includes: Acquire a plurality of lightning current time-domain waveform samples and an environmental parameter sample corresponding to each lightning current time-domain waveform sample; wherein each lightning current time-domain waveform sample corresponds to a lightning current waveform data sample; For each lightning current time domain waveform sample, a corresponding lightning current model sample is constructed according to the lightning current waveform data sample; Based on the lightning current model samples, a multi-objective optimization model is constructed with the goals of minimizing the time-domain root mean square error and minimizing the frequency-domain energy relative error; wherein the time-domain root mean square error is used to characterize the degree of fit of the lightning current model samples to the lightning current waveform energy distribution in the time domain; the frequency-domain energy relative error is used to characterize the degree of fit of the lightning current model samples to the lightning current waveform energy distribution in the frequency domain; The multi-objective optimization model is solved to generate a modified lightning current model sample when the time domain root mean square error and the frequency domain energy relative error are minimized; wherein the modified lightning current model sample includes: a modified lightning current waveform data sample; The lightning current waveform data samples and environmental parameter samples corresponding to the lightning current time domain waveform samples are used as training samples; Taking each training sample and the actual corrected lightning current waveform data sample of each training sample as input, and the corrected prediction result of the lightning current waveform data sample of each training sample as output, the data correction model to be trained is iteratively trained until the model converges to generate a preset data correction model.

5. The method for correcting a lightning current time domain waveform according to claim 4, wherein: The data correction model includes: a plurality of input layer neurons and a plurality of hidden layer neurons; The data correction model extracts lightning features for describing lightning current characteristics based on lightning current waveform data, including: Through each input layer neuron, the parameter value signal in the lightning current waveform data is transmitted to each hidden layer neuron; For each hidden layer neuron, generating an input signal for the hidden layer neuron based on the parameter value signal, the connection weight of the hidden layer neuron, and the preset offset of the hidden layer neuron; wherein the connection weight is used to measure the importance of the signal transmitted from the input layer neuron to the hidden layer neuron; For each hidden layer neuron, the input signal of the hidden layer neuron is linearly transformed through the activation function to generate the output signal of the hidden layer neuron; The output signals of the neurons in each hidden layer are weighted and summed to generate lightning features used to describe the characteristics of lightning current.

6. A method for correcting a lightning current time domain waveform according to claim 5, characterized in that: The lightning current time domain waveform sample also corresponds to a preset lightning current time domain reference waveform sample; The multi-objective optimization model is constructed based on the lightning current model sample with the goal of minimizing the time domain root mean square error and minimizing the frequency domain energy relative error, including: The multi-objective optimization model is constructed according to the following formula: mine band 100. Among them, MSE is the root mean square error in time domain, I real (t i ) is the lightning current waveform data corresponding to the lightning current time domain reference waveform sample at the i-th sampling time, I model (t i ) is the lightning current waveform data sample corresponding to the lightning current model sample at the i-th sampling time; n is the total number of sampling time points corresponding to the lightning current model samples and the lightning current time domain reference waveform samples, ε band is the relative error of frequency domain energy, E model,band is the total energy of the lightning current waveform of the lightning current model sample, E real,band It is the total energy of the lightning current waveform of the lightning current time domain reference waveform sample.

7. A method for correcting a lightning current time domain waveform according to claim 6, characterized in that: The corrected lightning current waveform data includes: a corrected lightning current amplitude, a corrected wave front time, and a corrected wave tail time; the lightning current time domain waveform also corresponds to a preset lightning current time domain reference waveform; The modified lightning current model includes: Where i(t) is the instantaneous value of lightning current at time t, α is the corrected lightning current amplitude, β is the corrected wave head time, γ is the corrected wave tail time, I m is the peak current of the modified lightning current model, I p is the peak current of the lightning current time domain reference waveform, T1 is the wave front time of the lightning current time domain reference waveform, T2 is the wave tail time of the lightning current time domain reference waveform, σ is the soil conductivity, τ1 is the wave front time constant, τ2 is the wave tail time constant, and s is a preset parameter used to affect the speed at which the lightning current increases with time.

8. A device for correcting a lightning current time domain waveform, characterized in that: include: Data acquisition module, lightning current model fitting module, data correction module and lightning current model correction module; The data acquisition module is used to obtain the current lightning current time domain waveform and environmental parameters; wherein the lightning current time domain waveform corresponds to a lightning current waveform data; The lightning current model fitting module is used to fit the lightning current time domain waveform to generate an initial lightning current model; wherein the lightning current model is used to characterize the curve of the lightning current changing with time; The data correction module is configured to input the lightning current waveform data and environmental parameters into a preset data correction model, so that the data correction model extracts lightning characteristics used to describe lightning current characteristics based on the lightning current waveform data; generates correlation characteristics used to characterize the correlation relationship between the environment and the lightning current characteristics based on the lightning characteristics and the environmental parameters; and outputs corrected lightning current waveform data based on the correlation characteristics and the lightning characteristics; The lightning current model correction module is used to correct the initial lightning current model according to the environmental parameters and the corrected lightning current waveform data to generate a corrected lightning current model.

9. A terminal device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the method for correcting a lightning current time domain waveform according to any one of claims 1 to 7 is implemented.

10. A storage medium, characterized in that: The storage medium includes a stored computer program, wherein when the computer program is running, the device where the storage medium is located is controlled to execute the method for correcting the lightning current time domain waveform according to any one of claims 1 to 7.

Images