System and method for safety evaluation of substation main grounding grid

Abstract

The application relates to the technical field of transformer substations, and discloses a system and method for safety evaluation of a main grounding grid of a transformer substation. The method comprises the following steps: a multi-point potential gradient sensor is arranged on a main grounding grid of a transformer substation of a power transmission and transformation project, and synchronous measurement is performed to obtain an original potential gradient data set; effective signal and interference signal are distinguished from the original potential gradient data set to obtain potential distribution information of the grounding grid of the transformer substation; a dynamic impedance model of the grounding grid of the transformer substation is established and used for fault current analysis to obtain a whole-process potential distribution graph containing an initial stage, a steady state stage and a recovery stage; and safety risk evaluation is performed based on the whole-process potential distribution graph to output a safety risk area report. The application effectively distinguishes effective measurement signals from interference signals, significantly improves the reliability and accuracy of the potential measurement data of the grounding grid in a complex electromagnetic environment, and realizes accurate positioning of a risk area and effective tracking of a hazard source.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of transformer substations, and in particular to a system and method for safety evaluation of a main grounding grid of a transformer substation. BACKGROUND

[0002] The main grounding grid of a transformer substation is a critical safety facility in power transmission and transformation projects, and its health status is directly related to the safe operation of the power system and personal safety protection. With the expansion of the power grid scale and the growth of the load, the short-circuit fault current of the transformer substation is increasing, and the safety evaluation of the grounding grid is facing unprecedented challenges. Traditional grounding grid evaluation methods mainly include two technical routes: frequency domain measurement method and time domain impulse method. However, these methods have obvious limitations in complex soil conditions. Due to the limited number of measurement points, the data is sparse, and the actual state of the grounding grid cannot be fully reflected. At the same time, it is difficult to effectively distinguish the effective measurement signal from the interference signal, and the background noise is high, the waveform distortion is serious, especially under the condition of high load operation of the transformer substation equipment, the interference signal is more complex and variable.

[0003] More importantly, existing grounding grid safety evaluation methods focus on static or steady-state characteristics, ignoring the transient potential variation law in the entire fault process. In particular, the dangerous potential region that may occur during the short-circuit fault recovery phase has been long ignored, and these regions often lead to excessive contact voltage and step voltage. In addition, traditional research lacks comprehensive consideration of the seasonal variation of soil resistivity, the influence of temperature and humidity, and the skin effect and proximity effect of conductors, resulting in a large deviation between the evaluation results and the actual situation, making it difficult to provide reliable technical basis for the safety improvement of the grounding grid. SUMMARY

[0004] The present application provides a system and method for safety evaluation of a main grounding grid of a transformer substation, which effectively distinguishes between effective measurement signals and interference signals, significantly improves the reliability and accuracy of grounding grid potential measurement data in complex electromagnetic environments, and realizes accurate positioning of risk areas and effective tracking of dangerous sources.

[0005] In a first aspect, the present application provides a method for safety evaluation of a main grounding grid of a transformer substation, which comprises:

[0006] A multi-point potential gradient sensor is arranged on the main grounding grid of the transformer substation of the power transmission and transformation project, and synchronous measurement is performed by the multi-point potential gradient sensor to obtain an original potential gradient data set;

[0007] The original potential gradient data set is subjected to effective signal and interference signal distinguishing processing to obtain transformer substation grounding grid potential distribution information;

[0008] A dynamic impedance model of the transformer substation grounding grid is established according to the transformer substation grounding grid potential distribution information;

[0009] The substation grounding grid dynamic impedance model is used for fault current analysis to obtain a whole-process potential distribution diagram including an initial stage, a steady state stage and a recovery stage;

[0010] Based on the whole-process potential distribution diagram, safety risk assessment is performed to output a safety risk area report.

[0011] In a second aspect, the present application provides a substation main grounding grid safety evaluation system, which comprises:

[0012] A setting module is configured to set a multi-point potential gradient sensor for a substation main grounding grid of a power transmission and transformation project, and perform synchronous measurement through the multi-point potential gradient sensor to obtain an original potential gradient data set;

[0013] A distinguishing processing module is configured to perform effective signal and interference signal distinguishing processing on the original potential gradient data set to obtain substation grounding grid potential distribution information;

[0014] An establishing module is configured to establish a substation grounding grid dynamic impedance model according to the substation grounding grid potential distribution information;

[0015] A fault current analysis module is configured to use the substation grounding grid dynamic impedance model for fault current analysis to obtain a whole-process potential distribution diagram including an initial stage, a steady state stage and a recovery stage;

[0016] A safety risk assessment module is configured to perform safety risk assessment based on the whole-process potential distribution diagram to output a safety risk area report.

[0017] In the technical scheme provided by the present application, through multi-point potential gradient synchronous measurement data acquisition technology, high-density and high-time-precision monitoring of substation grounding grid potential distribution is realized, the limitation of sparse data of traditional measurement methods is overcome, and through potential gradient signal recognition and screening methods based on deep learning, effective measurement signals and interference signals are effectively distinguished, and the reliability and accuracy of grounding grid potential measurement data in a complex electromagnetic environment are significantly improved. The constructed layered soil resistivity dynamic impedance model considers the influence of temperature, humidity and seasonal change on soil resistivity, as well as the conductor skin effect and proximity effect, so that the grounding grid impedance characteristic representation is more accurate. The grounding grid fault current distribution transient analysis method first focuses on the secondary potential peak value phenomenon in the fault recovery stage, reveals the safety risk points ignored by traditional methods, and comprehensively evaluates the potential distribution change in the whole fault process. The risk assessment method based on a three-dimensional safety risk assessment matrix realizes accurate positioning of the risk area and effective tracking of the hazard source. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative effort on the basis of these drawings.

[0019] Figure 1 An embodiment of the method for safety evaluation of the main grounding grid of the substation in the present application is shown in the figure.

[0020] Figure 2 An embodiment of the system for safety evaluation of the main grounding grid of the substation in the present application is shown in the figure. DETAILED DESCRIPTION

[0021] The present application provides a system and method for safety evaluation of the main grounding grid of the substation. The terms "first", "second", "third", "fourth" and the like (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily mean a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the term "comprise" or "have" and any variation thereof is intended to cover non-exclusive inclusion, for example, a process, method, system, product or device comprising a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0022] For the sake of understanding, the specific process of the embodiments of the present application will be described below. Please refer to Figure 1 An embodiment of the method for safety evaluation of the main grounding grid of the substation in the present application includes:

[0023] Step S101, a multi-point potential gradient sensor is arranged on the main grounding grid of the substation of the power transmission and transformation project, and a synchronous measurement is performed through the multi-point potential gradient sensor to obtain an original potential gradient data set.

[0024] It can be understood that the execution subject of the present application can be a system for safety evaluation of the main grounding grid of the substation, and can also be a terminal or a server, which is not limited here. The embodiments of the present application take the server as the execution subject for example.

[0025] Specifically, the potential gradient sensors of the bipolar probe structure are arranged around and inside the grounding grid of the power transmission and transformation project substation. The probe structure is made of stainless steel alloy material, with a platinum gold layer plated on the surface to ensure the stability and corrosion resistance of the measurement. The probe spacing is set to 1 meter, and a reasonable topological layout is formed according to the structural characteristics of the grounding grid. The measurement point arrangement density in the vicinity of the grounding downlead area is not less than 2 points per square meter, while in other areas it is maintained at not less than 0.5 points per square meter, thereby forming a high-density, multi-point coverage potential gradient measurement network. The potential gradient sensors in the measurement network are connected to the central data acquisition unit, which uses a 24-bit high-precision analog-to-digital converter with a sampling rate set to 10 kHz, much higher than the traditional 100 Hz sampling rate, ensuring the capture of microsecond-level potential changes. At the same time, the entire system is synchronized with the clock calibration through GPS time service technology, so that the measurement time error of all sensors is controlled within 50 nanoseconds, forming a multi-point synchronous measurement system. After completing the synchronous clock calibration, the multi-point synchronous measurement system injects a 77Hz, 5A test excitation current signal at the preset position of the substation for 2 seconds to excite the substation grounding grid, and the potential gradient sensors capture the potential gradient changes of each measurement point during this period. The potential gradient change signal generated by the test excitation signal is collected by the analog-to-digital converter, and during the collection process, a complete data set is automatically stored every 0.2 seconds to ensure the integrity and accuracy of the measurement data. In order to accurately record the corresponding external environmental conditions of the measurement data, the atmospheric temperature, humidity, ground temperature, and operation status of the key target equipment inside the substation are recorded simultaneously during the measurement process. These data serve as environmental parameter matching data, which are matched and integrated with the original measurement data. By combining the original measurement data with the environmental parameter matching data, the original potential gradient data set containing time tags, position coordinates, potential gradient amplitude, phase information, and environmental information is formed.

[0026] Step S102, distinguish the effective signal and the interference signal from the original potential gradient data set to obtain the potential distribution information of the substation grounding grid;

[0027] Specifically, the original potential gradient data set is input into a fully trained multi-layer convolutional neural network model, which includes multiple convolutional layers, pooling layers and fully connected layers. Local features of the data are extracted through convolution operation, and the features are reduced in dimension through the pooling layer to reduce redundant information. After integrating the global features through the fully connected layer, the model accurately identifies the effective signals and interference signals in the potential gradient data and generates identification result data. For the interference signals identified in the identification result data, an adaptive wavelet threshold filtering technology is used for suppression. This technology automatically adjusts the filtering threshold according to the local statistical characteristics of the signal. At the same time, for long-time scale low-frequency interference signals, a high-order sliding median filter is used for smoothing processing, so that the low-frequency noise in the data is effectively suppressed, thereby retaining the detailed features of the original signal to the greatest extent. The potential gradient data after noise reduction is obtained after the interference signal suppression. The potential gradient data after noise reduction is converted to the frequency domain, the amplitude and phase information of the test signal are extracted therefrom, and the phase difference and amplitude change relationship between each measurement point are calculated to obtain a spatial distribution matrix representing the potential gradient change relationship of each measurement point. Through feature vector analysis on the spatial distribution matrix, the main feature vectors and corresponding feature information are extracted, and are sorted according to the importance of the features. According to the feature value contribution rate, the top several feature vectors with a proportion exceeding a preset target value are retained. These retained feature vectors reflect the core information of the potential gradient distribution of the grounding grid, and the potential distribution information of the grounding grid of the transformer substation is obtained.

[0028] Step S103, establishing a dynamic impedance model of the grounding grid of the transformer substation according to the potential distribution information of the grounding grid of the transformer substation;

[0029] Specifically, according to the substation grounding grid potential distribution information, the soil parameters are inversed by the method of data fitting and model solving, and the thickness and resistivity parameters of the soil are inversed, wherein the soil is divided into three layers of surface layer, middle layer and deep layer, and the thickness and resistivity parameters of each layer are solved by fitting the measured data to reflect the electrical characteristics of the soil at different depths. In order to improve the accuracy of the model, the temperature coefficient and humidity coefficient compensation calculation is performed on the inversed soil thickness and resistivity parameters, the influence of environmental temperature and humidity changes on soil resistivity is introduced into the model by introducing the temperature and humidity compensation model, and a dynamic soil resistivity model is obtained. Based on the soil resistivity model compensated by temperature and humidity, a three-dimensional model of the substation grounding grid area is established, and the grounding grid area is divided into a large number of hexahedral grid units by using the finite element mesh division technology, wherein the grid size near the grounding body is set to be smaller, i.e. 0.1 meters, to improve the calculation accuracy, and the grid size gradually increases to 1 meter away from the grounding body, forming a fine gridding three-dimensional model, and a complex impedance representation equation is established according to the gridding three-dimensional model, so as to consider the impedance characteristics at different frequencies. Since the impedance characteristics of the substation grounding grid are affected by the skin effect and the proximity effect, in order to more accurately describe the characteristics of the conductor under high frequency current, a grounding grid metal conductor refinement model considering the skin effect and the proximity effect is constructed according to the complex impedance representation equation, which subdivides the conductor section into multiple units to simulate the current distribution characteristics, thereby improving the modeling accuracy. In order to improve the adaptability of the model, a seasonal change correction coefficient is introduced to reflect the change law of soil resistivity and grounding impedance under different seasonal conditions, and the value range of the correction coefficient is between 0.8 and 1.2, the minimum value is taken in summer, and the maximum value is taken in winter, so as to obtain a grounding grid model dynamically reflecting the seasonal change. The parameter optimization iteration of the seasonal correction grounding grid model is performed by genetic algorithm, the genetic algorithm simulates the process of natural evolution, constantly adjusts the model parameters to approximate the optimal solution, and the iteration number is set to 1000 times, and the convergence condition is that the parameter change rate of continuous multiple iterations is less than 0.1%, so as to finally obtain an accurate substation grounding grid dynamic impedance model reflecting the grounding grid impedance characteristics under different frequencies and different seasonal conditions.

[0030] According to the complex impedance characterization equation, the metal conductor of the grounding grid is meshed, the conductor cross section is divided into multiple segmented units, the non-uniform distribution of current on the conductor cross section under the action of high-frequency current is described through the fine section of the segmented unit model. The electromagnetic field is calculated on the refined conductor cross section model, and the current density distribution function reflecting the skin effect is obtained by using the electromagnetic field solving method. The function reflects the distribution characteristics that the current is gradually concentrated on the surface layer of the conductor with the increase of the current frequency, thereby forming the skin phenomenon. Based on the current density distribution function, the electromagnetic coupling effect between adjacent conductors is calculated, and the conductor mutual impedance matrix reflecting the proximity effect is constructed, reflecting the impedance change characteristics between different conductors caused by mutual coupling. The seasonal factor analysis is performed on the conductor mutual impedance matrix, and the soil resistivity will fluctuate significantly with the change of temperature and humidity, so the seasonal correction parameter set is introduced to dynamically correct the conductor mutual impedance characteristics in different seasons. In order to improve the model accuracy, according to the seasonal correction parameter set and the conductor mutual impedance matrix, the soil resistivity is corrected and calculated to obtain a seasonal soil resistivity model reflecting the influence of seasonal change. The model accurately describes the impedance change characteristics of the soil under different temperature and humidity conditions. After the seasonal soil resistivity model is comprehensively processed with the segmented unit model of the conductor cross section, the current density distribution function reflecting the skin effect and the conductor mutual impedance matrix reflecting the proximity effect, a seasonal corrected grounding grid model is obtained, reflecting the electromagnetic characteristics of the grounding grid under different seasonal conditions, and describing the change of conductor current distribution caused by skin effect and proximity effect.

[0031] In step S104, the dynamic impedance model of the substation grounding grid is used for fault current analysis to obtain a whole-process potential distribution diagram including an initial stage, a steady-state stage and a recovery stage.

[0032] Specifically, based on the constructed dynamic impedance model, the fault current is decomposed and processed, which is disassembled into the superposition form of DC component and AC component, wherein the DC component reflects the large current transient process in the initial stage of fault, and the AC component reflects the AC characteristics gradually decaying over time, and the fault current function is obtained after decomposition. The fault current function is decomposed by time step, and the entire analysis time period is divided into a plurality of equal interval time step sequences, and the length of each time step is not more than 100 microseconds, so as to ensure accurate capture of the transient change characteristics of the fault current in each stage. For each time point in the decomposed time step sequence, the node potential method is used to establish the grid equation of the grounding grid, which connects the potential change of each node in the grounding grid with the injected current through the matrix form, and in order to improve the calculation efficiency, the fast convolution algorithm is used to convert the frequency domain impedance data into time domain response, and the grounding grid transient response equation set is obtained. According to the obtained transient response equation set, the transient potential values of each node of the grounding grid and the potential difference between adjacent nodes are calculated step by step, and the distribution data of each component of the fault current in the grounding grid is obtained. Based on the distribution data of the fault current components, the ground surface electric field intensity is calculated, and the spatial distribution of the electric field intensity directly reflects the change of the potential gradient, and in order to accurately identify the secondary potential peak phenomenon in the fault recovery stage, the potential gradient maximum value monitoring algorithm is used for dynamic tracking, when the potential gradient is redistributed due to electromagnetic induction or impedance change in the system recovery process, the monitoring algorithm effectively captures this key phenomenon, and records the corresponding potential gradient distribution data. Based on the obtained potential gradient distribution data, the whole process of fault is analyzed, the potential distribution characteristics in the initial stage, the steady state stage and the recovery stage are studied, and the migration law of the position of the maximum potential gradient is analyzed combined with the change of the node potential, and the whole process potential distribution map is formed.

[0033] Step S105, based on the whole process potential distribution map, safety risk assessment is carried out, and a safety risk area report is output.

[0034] Specifically, the potential variation of the main grounding grid of the substation in the whole process of fault is analyzed by the whole-process potential distribution diagram, combined with the potential gradient distribution characteristics, the potential difference of adjacent nodes and the electric field intensity data, the safety risk assessment is performed, and the space-time risk distribution data covering the whole substation area is obtained, which reflects the risk distribution of different spatial positions and contains the risk evolution trend at each time step. The spatial-time risk distribution data is analyzed by three-dimensional K-means clustering, and the whole risk data set is divided into several risk clustering areas, different clustering areas represent the risk characteristic areas with similar spatial and time change rules, and the determination of K value is automatically adjusted by the contour coefficient, so as to ensure that the clustering result reflects the real risk distribution characteristics to the greatest extent. For the identified risk clustering area set, the space-time risk density of each clustering area is calculated, and by comparing the space-time risk density with the preset density threshold, the high-risk area set is identified, and these high-risk areas correspond to the path through which the fault current flows, the weak point of the grounding grid or the area with sharp potential gradient change. In order to clarify the causes of the formation of high-risk areas and the potential risk sources, the hazard source tracking analysis is performed on the identified high-risk area set, the improved path backtracking algorithm is adopted to track the propagation path of the fault current in the grounding grid, and the fault current target path and the key target node information passing through it which cause the risk of the area to rise are identified. These target nodes are nodes with abnormal increase of current density or nodes with sharp change of potential gradient, and the key monitoring and analysis of these nodes can help to clarify the root cause of fault risk. Based on the identified high-risk area set, the fault current target path and the target node information, the key information such as the spatial distribution, risk level, exceeding degree and duration of each risk area is comprehensively analyzed, the safety risk area report is generated, the specific position, risk level, fault path characteristics and key node information of each high-risk area are recorded, and the corresponding safety protection suggestions are put forward for different types of risk areas.

[0035] In the embodiments of the present application, by using the multi-point potential gradient synchronous measurement data acquisition technology, high-density and high-time-precision monitoring of the potential distribution of the grounding grid of the substation is realized, the limitation of sparse data of the traditional measurement method is overcome, the potential gradient signal recognition and screening method based on deep learning is used to effectively distinguish the effective measurement signal and the interference signal, and the reliability and accuracy of the grounding grid potential measurement data in the complex electromagnetic environment are significantly improved. The constructed layered soil resistivity dynamic impedance model considers the influence of temperature, humidity and seasonal change on soil resistivity, as well as the conductor skin effect and proximity effect, so that the impedance characteristic representation of the grounding grid of the substation is more accurate. The grounding grid fault current distribution transient analysis method first focuses on the secondary potential peak phenomenon in the fault recovery stage, reveals the safety risk point ignored by the traditional method, and comprehensively evaluates the potential distribution change in the whole fault process. The risk assessment method based on the three-dimensional safety risk assessment matrix realizes the accurate positioning of the risk area and the effective tracking of the hazard source.

[0036] In a specific embodiment, the process of step S101 can specifically include the following steps:

[0037] The potential gradient sensors of the bipolar probe structure are arranged around and inside the target nodes of the grounding grid of the substation according to the topological structure to obtain a measurement network;

[0038] The potential gradient sensors in the measurement network are connected to a central data acquisition unit, and a synchronous clock calibration is performed through GPS time service technology to obtain a multi-point synchronous measurement system;

[0039] The test excitation condition is injected at the preset position of the substation through the multi-point synchronous measurement system;

[0040] Under the test excitation condition, the potential gradient changes of all measurement points are collected by using an analog-to-digital converter to obtain original measurement data;

[0041] The original measurement data are time-labeled, and the atmospheric temperature, humidity, ground temperature and operating state of the target equipment of the substation at the measurement time are recorded synchronously to obtain environment parameter matching data;

[0042] The original measurement data and the environment parameter matching data are combined and processed to generate an original potential gradient data set.

[0043] Specifically, the sensor arrangement is carried out in the grounding grid area of the power transmission and transformation project substation. The potential gradient sensors of the bipolar probe structure are arranged around the substation grounding grid and the key internal target nodes. The sensors are made of stainless steel alloy material and plated with platinum gold layer on the probe surface to ensure high corrosion resistance and excellent measurement stability during long-term operation. The probe spacing of each bipolar probe is set to 1 meter to ensure that the measurement accuracy meets the engineering requirements. At the same time, according to the topological structure of the grounding grid, the sensor arrangement is laid out so that the measurement point density in the grounding downlead area is not less than 2 points per square meter, and in other areas, it is maintained at not less than 0.5 points per square meter, forming a high-density potential gradient measurement network covering the entire substation grounding grid. The potential gradient sensors in the measurement network are connected with the central data acquisition unit. The data acquisition unit uses a 24-bit high-precision analog-to-digital converter, has higher measurement accuracy and anti-interference ability, and the sampling rate is set to 10 kHz, so that the microsecond-level potential gradient change data is captured. In order to ensure that the data acquisition of each measurement point remains synchronized, the entire system is clock-synchronized and calibrated through GPS time service technology. The GPS time service system controls the clock error of all sensors within 50 nanoseconds, realizing a multi-point synchronous measurement system. After completing the clock synchronization calibration, the multi-point synchronous measurement system injects test excitation conditions at the preset positions of the substation. During the excitation process, a 77Hz frequency, 5 ampere test current signal is injected into different positions of the substation through a preset program, and the injection duration is 2 seconds. The excitation signal forms a stable potential gradient change field in the grounding grid, so that the potential change of each measurement point during this period is captured by the arranged potential gradient sensors. Under the test excitation conditions, the analog-to-digital converter is used to collect the potential gradient change of all measurement points in real time. The original measurement data is stored at a high precision of 10 kHz, and during the collection process, a complete data set is automatically stored every 0.2 seconds, so as to ensure that no key measurement information is missed during the test process. In order to more accurately analyze the influence of environmental factors in the potential gradient change process and effectively compensate the measurement data, the original measurement data is time-stamped, and the atmospheric temperature, humidity, ground temperature and operating state of the key target equipment of the substation at the measurement time are recorded synchronously. These environmental parameters are bound as matching data with the original measurement data to generate an environmental parameter matching data set. The original measurement data and the environmental parameter matching data are combined and processed to generate an original potential gradient data set, which contains time label, position coordinates, potential gradient amplitude, phase information and corresponding environmental parameter information.

[0044] In a specific embodiment, the process of performing step S102 can specifically include the following steps:

[0045] The original potential gradient data set is input into a multi-layer convolutional neural network model for signal feature extraction, to obtain recognition result data;

[0046] Interference signals in the recognition result data are suppressed to obtain potential gradient data after noise reduction;

[0047] Fast Fourier transform is performed on the potential gradient data after noise reduction to obtain frequency domain feature data;

[0048] According to the frequency domain feature data, the phase difference and amplitude ratio between each measurement point are calculated to obtain a spatial distribution matrix representing the potential gradient relationship of each measurement point;

[0049] The spatial distribution matrix is subjected to eigenvector analysis to obtain a characteristic decomposition result;

[0050] Based on the characteristic decomposition result, the first k eigenvectors with a contribution rate exceeding a preset target value are retained to obtain the potential distribution information of the substation grounding grid.

[0051] Specifically, the original potential gradient data set is input into a fully trained multi-layer convolutional neural network model, which is trained with a large number of samples, including various typical potential gradient signal sample data, including normal signals, power frequency interference signals, lightning interference signals, and switching operation interference signals, etc. After model training and parameter optimization, the model can identify and extract the main signal features in the original data. The convolutional neural network model includes 5 convolutional layers, 3 pooling layers and 2 fully connected layers. The convolutional layer extracts multi-scale features from the input data, the pooling layer reduces the dimension to reduce the amount of calculation, and the fully connected layer integrates the features and classifies the signals to generate recognition result data. The recognition result data contains feature information of the effective signal and can also mark different types of interference signals. For the interference signals in the recognition result data, interference suppression processing is performed, and the adaptive wavelet threshold filtering technology is used to effectively suppress the identified interference signals. This technology automatically adjusts the filtering threshold according to the local characteristics of the signal, thereby suppressing the interference while preserving the information of the effective signal to the greatest extent. For long-time scale low-frequency interference signals, a high-order sliding median filter is used for smoothing processing, so that the low-frequency noise in the data is effectively suppressed, and the denoised potential gradient data is generated. The denoised potential gradient data is subjected to fast Fourier transform to convert the time domain signal to the frequency domain signal. The Fourier transform extracts the amplitude and phase information of different frequency components, and calculates the phase difference and amplitude ratio between the measurement points. The frequency domain feature data can more clearly represent the spatial variation of the potential gradient signal. By analyzing the frequency domain feature data, a spatial distribution matrix representing the potential gradient relationship between the measurement points is constructed. The elements of the spatial distribution matrix represent the potential gradient relationship between different measurement points. The rows and columns of the matrix correspond to different measurement points, and the data in the matrix reflect the distribution of the potential gradient amplitude ratio and phase difference between the measurement points, forming a spatial feature description. In order to extract the key feature information hidden in the spatial distribution matrix, feature vector analysis is performed on the spatial distribution matrix. A group of feature vectors and corresponding eigenvalues are extracted through matrix eigenvalue decomposition. These feature vectors reflect the main mode and trend of the potential distribution of the grounding grid, and the size of the eigenvalue is used to measure the importance of the corresponding feature vector. In order to optimize the data and retain the feature vectors with high contribution rate, the top k feature vectors with a contribution rate exceeding a preset target value are retained according to the eigenvalue decomposition result. These feature vectors accurately describe the potential gradient distribution information of the grounding grid, thereby forming a reduced feature matrix and generating an accurate potential distribution information data set.

[0052] In a specific embodiment, the process of step S103 can specifically include the following steps:

[0053] According to the potential distribution information of the grounding grid of the transformer substation, the soil parameters are inversed to obtain the soil thickness and resistivity parameters;

[0054] The temperature coefficient and humidity coefficient calculation is performed on the soil thickness and resistivity parameters to obtain a soil resistivity model compensated for temperature and humidity;

[0055] Based on the soil resistivity model compensated for temperature and humidity, a three-dimensional modeling is performed to obtain a grid-based three-dimensional model of the substation grounding grid area, and a complex impedance representation equation is established according to the grid-based three-dimensional model;

[0056] According to the complex impedance representation equation, a grounding grid metal conductor refinement model considering the skin effect and proximity effect is constructed, and a seasonal change correction coefficient is introduced to obtain a seasonally corrected grounding grid model;

[0057] The seasonally corrected grounding grid model is iteratively optimized by a genetic algorithm to obtain a dynamic impedance model of the substation grounding grid.

[0058] Specifically, based on the substation grounding grid potential distribution information, soil parameter inversion is carried out. In the inversion process, the measured potential gradient change data and the theoretical calculation model are matched, the soil in the grounding grid area is divided into three layers of surface layer, middle layer and deep layer structure, and the thickness and corresponding resistivity parameters of each layer of soil are inversed. These parameters reflect the electrical characteristics of soil at different depths. In order to improve the accuracy of the soil resistivity model, the temperature coefficient and humidity coefficient compensation calculation of the inversed soil thickness and resistivity parameters are carried out. By introducing the temperature and humidity correction model, the influence of environmental temperature and humidity change on soil resistivity is included in the calculation range, forming a dynamic soil resistivity model. With the increase of temperature, the soil resistivity shows a downward trend, and when the humidity increases, the soil resistivity will also decrease accordingly. By combining the temperature coefficient and humidity coefficient, the soil resistivity change law under different environmental conditions is accurately reflected. Based on the soil resistivity model compensated by temperature and humidity, a three-dimensional model of the substation grounding grid area is established. The finite element mesh division technology is used to divide the entire grounding grid area into multiple hexahedral grid units. The grid division density is higher near the grounding body, and the unit size reaches 0.1 meters. In the area far from the grounding body, the grid size gradually increases to 1 meter, forming a high-precision grid three-dimensional model. According to the grid three-dimensional model, a complex impedance representation equation is established. By introducing complex impedance, the impedance characteristics of the grounding grid area under different frequency conditions are represented. This equation reflects the resistance characteristics under low frequency conditions and captures the inductance and capacitance effects under high frequency conditions. In order to more accurately describe the current distribution characteristics of the metal conductor in the grounding grid, a grounding grid metal conductor refinement model considering skin effect and proximity effect is constructed according to the complex impedance representation equation. The skin effect causes high-frequency current to concentrate on the surface layer of the conductor, while the proximity effect reflects the change of current distribution between adjacent conductors due to electromagnetic coupling. In the construction of the refinement model, the conductor cross section is divided into multiple segmented units. By calculating the current density distribution of different segmented units, the electromagnetic characteristics of the conductor under different frequencies are more accurately reflected. At the same time, in order to improve the adaptability and accuracy of the model, a seasonal change correction coefficient is introduced to reflect the change law of soil resistivity and grounding impedance under different seasonal conditions. The value of this correction coefficient ranges from 0.8 to 1.2. In winter, due to the decrease of soil humidity and the increase of resistivity, the correction coefficient takes the maximum value. In summer, due to the high soil humidity, the correction coefficient takes the minimum value, so as to obtain a grounding grid model that dynamically reflects the influence of seasonal change.In order to optimize the model parameters and make the dynamic impedance model more close to the actual working condition, the genetic algorithm is used to optimize and iterate the seasonally corrected grounding grid model. The genetic algorithm optimizes the model parameters by simulating the natural evolution process. The initial population is composed of a group of random parameter combinations. Through continuous crossover, mutation and selection, the optimal solution is gradually approached. In the iteration process of each generation, the fitness function is calculated to judge the fitting degree of the current parameter combination and the actual measured data. After multiple iterations, when the parameter change rate of continuous generations is less than the set threshold, it can be judged that the model converges. Finally, an optimized dynamic impedance model of the grounding grid of the substation is obtained. The model can accurately reflect the impedance characteristics of the grounding grid under different frequencies and different seasonal conditions.

[0059] In a specific embodiment, the process of constructing a grounding grid metal conductor refinement model considering the skin effect and proximity effect according to the complex impedance representation equation and introducing a seasonal change correction coefficient to obtain a seasonally corrected grounding grid model can specifically include the following steps:

[0060] Grid subdivision is performed on the grounding grid metal conductor according to the complex impedance representation equation to obtain a segmented unit model of the conductor cross section;

[0061] The electromagnetic field of the segmented unit model of the conductor cross section is calculated to obtain a current density distribution function representing the skin effect;

[0062] Based on the current density distribution function representing the skin effect, the electromagnetic coupling effect between adjacent conductors is calculated to obtain a conductor mutual impedance matrix representing the proximity effect;

[0063] Seasonal factor analysis is performed on the conductor mutual impedance matrix representing the proximity effect to obtain a set of seasonal correction parameters;

[0064] According to the set of seasonal correction parameters and the conductor mutual impedance matrix representing the proximity effect, the soil resistivity is corrected and calculated to obtain a seasonal soil resistivity model;

[0065] The seasonal soil resistivity model, the segmented unit model of the conductor cross section, the current density distribution function representing the skin effect and the conductor mutual impedance matrix representing the proximity effect are comprehensively processed to obtain a seasonally corrected grounding grid model.

[0066] Specifically, the metal conductor in the grounding grid is finely meshed according to the complex impedance characterization equation, and the distribution characteristics of high-frequency current on the conductor cross-section are more accurately described by dividing the conductor cross-section into multiple segmented unit models, which can fully consider the current density variation at different positions and the complex variation of the conductor cross-section under different electromagnetic environments. After meshing, the conductor cross-section is divided into multiple independent units, and the current density is calculated for each unit independently, so that the non-uniformity characteristics of current distribution are more accurately captured. Based on the meshed conductor cross-section model, electromagnetic field calculation is performed, and by solving Maxwell's equations, the current density distribution function representing the skin effect is obtained. The skin effect is a phenomenon that the current gradually concentrates on the surface layer of the conductor due to the induced electromagnetic field generated by the alternating current in the conductor, resulting in a rapid decay of current density with increasing distance. This current density distribution function accurately describes the difference between the current density on the surface layer of the conductor and the current density inside the conductor. After obtaining the current density distribution function of the skin effect, the electromagnetic coupling between adjacent conductors is further considered, and by calculating the current distribution of the meshed conductor cross-section, the electromagnetic mutual inductance and impedance variation between adjacent conductors are calculated. This electromagnetic coupling forms the proximity effect, which causes the current density between adjacent conductors to change, thereby affecting the impedance characteristics of the conductors. By establishing a conductor mutual impedance matrix, the impedance variation characteristics between different conductors due to the proximity effect are described. Each element of the matrix represents the mutual impedance relationship between different conductors. Since the soil resistivity changes significantly with temperature and humidity, a seasonal factor analysis is performed on the conductor mutual impedance matrix, and a set of seasonal correction parameters is introduced to compensate for the variation of soil resistivity under different seasonal conditions. These seasonal correction parameter sets are calculated based on historical meteorological data and the variation of soil resistivity, and the correction parameter sets are dynamically adjusted according to the seasonal changes, thereby ensuring the high accuracy of the conductor mutual impedance matrix under different environmental conditions. According to the seasonal correction parameter set and the conductor mutual impedance matrix, the soil resistivity is dynamically corrected and calculated to obtain a seasonal soil resistivity model that reflects the variation of soil resistivity under different seasonal conditions. The corrected soil resistivity information is introduced into the grounding grid impedance model, so that the model maintains high calculation accuracy under different seasonal conditions. To form a seasonal corrected grounding grid model, the seasonal soil resistivity model is integrated with the segmented unit model of the conductor cross-section, the current density distribution function representing the skin effect, and the conductor mutual impedance matrix representing the proximity effect. In the integration process, the spatial coordinates of each part of the data are mapped uniformly, and the dynamic parameters are matched on the time axis, thereby ensuring the compatibility and consistency between different data sources.By integrating multi-dimensional data such as soil resistivity, conductor current distribution and electromagnetic coupling relationship, a seasonal corrected grounding grid model is obtained, which accurately reflects the electromagnetic characteristics of the grounding grid under different seasons and different frequencies, and provides accurate data support for subsequent fault current analysis, transient current distribution simulation and safety risk assessment.

[0067] In a specific embodiment, the process of performing step S104 can specifically include the following steps:

[0068] Based on the dynamic impedance model of the grounding grid of the substation, the fault current is decomposed into the superposition form of the direct current component and the alternating current component, to obtain a fault current function;

[0069] The fault current function is decomposed by time step to obtain a time step sequence;

[0070] For each time point in the time step sequence, a node potential method is used to establish a grounding grid grid equation, and a fast convolution algorithm is used to convert the frequency domain impedance into a time domain response, to obtain a grounding grid transient response equation set;

[0071] According to the grounding grid transient response equation set, the transient potential value of each node of the grounding grid and the potential difference between adjacent nodes at each time step are calculated, to obtain fault current component distribution data;

[0072] Based on the fault current component distribution data, the ground electric field intensity is calculated, and a potential gradient maximum monitoring algorithm is used to detect the secondary potential peak value phenomenon in the fault recovery stage, to obtain potential gradient distribution data;

[0073] Based on the potential gradient distribution data, the potential distribution characteristics and the maximum potential gradient position migration law in the initial stage, the steady state stage and the recovery stage are analyzed, to obtain a whole process potential distribution map.

[0074] Specifically, based on the dynamic impedance model of the grounding grid of the transformer substation, the fault current is decomposed and processed, and by analyzing the current characteristics at the moment of fault occurrence, the fault current is disassembled into the superposition form of direct current component and alternating current component. The direct current component mainly reflects the rapid decay current formed after the large current injection into the grounding grid in the initial stage of fault, has a short time constant and rapidly weakens, while the alternating current component reflects the alternating current formed by the superposition of the power frequency current of the power system as the fault process advances, has a relatively long time constant, and maintains a relatively stable amplitude in the steady state stage. By decomposing the fault current into the superposition form of direct current and alternating current components, the current characteristics in different stages are reflected. The fault current function is decomposed by time step to improve the calculation accuracy and capture the details of transient potential changes. The entire analysis period is divided into a plurality of equal-interval time step sequences, and the length of the time step is set to be no more than 100 microseconds to ensure high-precision capture of the dynamic change characteristics of the fault current. A corresponding fault current component is established for each time step to form a time step sequence. For each time point in the time step sequence, the node potential method is used to establish the grid equation of the grounding grid, which couples the potential change of the grounding grid with the node admittance matrix. The injection current vector is introduced to reflect the distribution of the current in the grounding grid at different time steps. Since the dynamic impedance model is usually established in the frequency domain, in order to convert it into the data required for time domain analysis, the frequency domain impedance data is converted into time domain response by the fast convolution algorithm, which maps the complex impedance matrix in the frequency domain to the time domain to form the transient response equation set of the grounding grid, describing the transient potential change of each node of the grounding grid at different time steps. According to the transient response equation set of the grounding grid, the transient potential values of each node of the grounding grid and the potential difference between adjacent nodes are calculated step by step. By analyzing the dynamic change of the potential of each node, the distribution data of each component of the fault current in the grounding grid are obtained, and the change characteristics of the potential gradient in different stages can be accurately captured. By solving the spatial gradient of the ground surface potential distribution, the change of the ground surface electric field strength is obtained. In order to identify the key potential gradient change phenomenon in the fault recovery stage, the maximum potential gradient monitoring algorithm is used for dynamic tracking. This algorithm can accurately capture the potential redistribution phenomenon caused by electromagnetic induction and system impedance change in the fault recovery stage, so as to detect the occurrence of the "second potential peak". The second potential peak occurs in the process of gradual decay of the fault current. Due to the nonlinear characteristics of the grounding grid impedance and the electromagnetic induction effect, the potential gradient re-peaks in a short time. This phenomenon is an important characteristic that is difficult to capture by traditional fault potential analysis methods, so the use of the maximum potential gradient monitoring algorithm can significantly improve the analysis accuracy in the fault recovery stage.Based on the potential gradient distribution data, the whole fault process is analyzed, from the initial stage to the steady stage, and then to the recovery stage. The potential distribution characteristics of each stage are studied respectively, and the dynamic migration law of the maximum potential gradient position is analyzed. In the initial stage, due to the rapid injection of fault current, the potential gradient changes dramatically, and the maximum potential gradient appears in the area close to the fault point. With the fault process entering the steady stage, the potential gradient gradually tends to be stable, and the position of the maximum potential gradient migrates to the area far away from the fault point. In the recovery stage, due to the gradual attenuation of the current and the change of the system impedance, the potential gradient will change again, accompanied by the appearance of the secondary potential peak. Through the analysis of these stage characteristics, the whole process potential distribution diagram is obtained, which reflects the potential change of the grounding grid and the dynamic change characteristics of the potential gradient in the whole process.

[0075] In a specific embodiment, the process of performing step S105 can specifically include the following steps:

[0076] Performing safety risk assessment through the whole process potential distribution diagram to obtain space-time risk distribution data;

[0077] Performing three-dimensional K-means clustering analysis on the space-time risk distribution data to obtain a risk clustering region set;

[0078] Calculating the space-time risk density of each clustering region for the risk clustering region set, and comparing the space-time risk density with a preset density threshold to obtain a high-risk region set;

[0079] Performing hazard source tracking analysis on each high-risk region in the high-risk region set to obtain fault current target path and target node information;

[0080] Generating a safety risk region report according to the high-risk region set, the fault current target path and the target node information.

[0081] Specifically, the potential change of the substation grounding grid in the whole process of fault is analyzed by the whole-process potential distribution diagram. The whole-process potential distribution diagram dynamically captures the potential change of each node of the grounding grid in the initial stage, steady stage and recovery stage of fault, combines the potential gradient change data and the ground surface electric field intensity change information, forms a risk distribution data set covering the spatial and time dimensions of the substation grounding grid, and the data set contains the potential gradient change of different spatial positions and records the change trend of the potential gradient at different time steps, thus reflecting the potential high-risk areas and abnormal potential distribution characteristics in the whole process of fault. In order to identify the risk distribution characteristics of different areas of the substation grounding grid, three-dimensional K-means clustering analysis is performed on the spatial-time risk distribution data, the coordinate information of the spatial dimension is combined with the risk change data of the time dimension to form a three-dimensional risk matrix, and the K-means clustering algorithm is used to automatically group the risk matrix. The K-means clustering algorithm automatically divides the spatial-time data into several risk clustering areas according to the risk characteristics of different areas, and each risk clustering area represents an area with similar risk characteristics in a specific spatial position and time stage, thereby effectively dividing the entire grounding grid area into different risk areas. After completing the division of the risk clustering areas, the spatio-temporal risk density of each clustering area is calculated for the risk clustering area set. By analyzing the potential gradient change amplitude, adjacent node potential difference and ground surface electric field intensity change in each clustering area, the spatio-temporal risk density of each clustering area is obtained, and the calculated spatio-temporal risk density is compared with the preset density threshold to identify the high-risk area set. The high-risk area set is near the fault current injection point, the grounding downlead position or the area with dramatic potential gradient change. These areas are dangerous areas with potential exceeding the standard, abnormal enhancement of electric field intensity, and exceeding of step voltage and touch voltage. In order to analyze the risk source of the high-risk area, the hazard source tracking analysis is performed on the identified high-risk area set. By using the improved current path backtracking algorithm, the propagation path of the fault current in the grounding grid is tracked to identify the target path and target node information of the fault current that causes the risk of the area to rise. The algorithm is based on the node potential difference change trend and backtracks layer by layer along the current propagation direction, so as to accurately determine the propagation path of the fault current and the key node position. These key nodes are nodes with abnormal increase of current density or nodes with sudden change of potential. Focusing on the analysis and monitoring of these nodes effectively improves the safety of the grounding grid.Based on the above obtained high-risk area set, fault current target path and target node information, the spatial distribution characteristics, risk level, exceeding degree and duration of each risk area are comprehensively analyzed to generate a safety risk area report. The report records the specific location, risk characteristics, fault current propagation path and key node information of each high-risk area. By classifying the risk levels of different types of risk areas, the exceeding degree of each area is quantified, and combined with spatial-time dimension data analysis, data support is provided for the safety evaluation of the substation grounding grid. At the same time, the safety risk area report proposes corresponding safety protection suggestions for each high-risk area, including adding vertical grounding bodies, optimizing the structure of the grounding grid, increasing equipotential bonding measures, etc.

[0082] The above describes the method for safety evaluation of the main grounding grid of the substation in the embodiments of the present application. The following describes the system for safety evaluation of the main grounding grid of the substation in the embodiments of the present application. Please refer to Figure 2 An embodiment of the system for safety evaluation of the main grounding grid of the substation in the embodiments of the present application includes:

[0083] The setting module 201 is configured to set a multi-point potential gradient sensor for the main grounding grid of the substation of the power transmission and transformation project, and obtain an original potential gradient data set through synchronous measurement by the multi-point potential gradient sensor.

[0084] The distinguishing processing module 202 is configured to perform effective signal and interference signal distinguishing processing on the original potential gradient data set to obtain the potential distribution information of the grounding grid of the substation.

[0085] The establishing module 203 is configured to establish a dynamic impedance model of the grounding grid of the substation according to the potential distribution information of the grounding grid of the substation.

[0086] The fault current analysis module 204 is configured to use the dynamic impedance model of the grounding grid of the substation for fault current analysis to obtain a whole-process potential distribution diagram including an initial stage, a steady stage and a recovery stage.

[0087] The safety risk evaluation module 205 is configured to perform safety risk evaluation based on the whole-process potential distribution diagram and output a safety risk area report.

[0088] Through the synergistic cooperation of the above-mentioned various components, through the multi-point potential gradient synchronous measurement data acquisition technology, the high-density and high-time accuracy monitoring of the potential distribution of the grounding grid of the transformer substation is realized, the limitations of sparse data of the traditional measurement method are overcome, the potential gradient signal recognition and screening method based on deep learning is used to effectively distinguish the effective measurement signal and the interference signal, and the reliability and accuracy of the grounding grid potential measurement data in the complex electromagnetic environment are significantly improved. The constructed layered soil resistivity dynamic impedance model considers the influence of temperature, humidity and seasonal change on soil resistivity, as well as the conductor skin effect and proximity effect, so that the impedance characteristic representation of the grounding grid of the transformer substation is more accurate. The grounding grid fault current distribution transient analysis method first focuses on the secondary potential peak value phenomenon in the fault recovery stage, reveals the safety risk point ignored by the traditional method, and comprehensively evaluates the potential distribution change in the whole fault process. The risk assessment method based on the three-dimensional safety risk assessment matrix realizes the accurate positioning of the risk area and the effective tracking of the hazard source.

[0089] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, system and unit can refer to the corresponding process in the foregoing method embodiment, which will not be described here.

[0090] The integrated unit, if realized in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application or the whole or part of the technical solutions that make essential contributions to the prior art can be embodied in the form of a software product, which is stored in a storage medium and includes a plurality of instructions for causing a device (which can be a personal computer, a server, or a network device, etc.) for safety evaluation of a main grounding grid of a transformer substation to execute all or part of the steps of the method described in each embodiment of the present application. The foregoing storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0091] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for safety evaluation of a substation main grounding grid, characterized in that, The method comprises the following steps: a plurality of potential gradient sensors are arranged on the main grounding grid of the substation of the power transmission and transformation project, and original potential gradient data sets are obtained by synchronous measurement of the plurality of potential gradient sensors; effective signals and interference signals in the original potential gradient data sets are distinguished to obtain substation grounding grid potential distribution information; a dynamic impedance model of the substation grounding grid is established according to the substation grounding grid potential distribution information; the dynamic impedance model of the substation grounding grid is used for fault current analysis to obtain a whole-process potential distribution graph including an initial stage, a steady state stage and a recovery stage; specifically, based on the dynamic impedance model of the substation grounding grid, a fault current function is obtained by decomposing the fault current into a superposition form of a direct current component and an alternating current component; the fault current function is decomposed by time step to obtain a time step sequence; for each time point in the time step sequence, a node potential method is used to establish a grounding grid grid equation, and a fast convolution algorithm is used to convert the frequency domain impedance into a time domain response to obtain a grounding grid transient response equation set; according to the grounding grid transient response equation set, the transient potential value of each node of the grounding grid and the potential difference between adjacent nodes at each time step are calculated to obtain fault current component distribution data; based on the fault current component distribution data, the ground electric field intensity is calculated, and a potential gradient maximum value monitoring algorithm is used to detect the secondary potential peak value phenomenon in the fault recovery stage to obtain potential gradient distribution data; based on the potential gradient distribution data, the potential distribution characteristics and the maximum potential gradient position migration law of the initial stage, the steady state stage and the recovery stage are analyzed to obtain the whole-process potential distribution graph; based on the whole-process potential distribution graph, safety risk assessment is carried out, and a safety risk area report is output.

2. The method for safety evaluation of a main grounding grid of a substation according to claim 1, characterized in that, The plurality of potential gradient sensors are arranged on the main grounding grid of the substation of the power transmission and transformation project, and original potential gradient data sets are obtained by synchronous measurement of the plurality of potential gradient sensors, which comprises: potential gradient sensors with a bipolar probe structure are arranged on the target nodes around and inside the grounding grid of the substation of the power transmission and transformation project according to the topological structure to obtain a measurement network; the potential gradient sensors in the measurement network are connected to a central data acquisition unit, and a synchronous clock is calibrated by GPS time service technology to obtain a multi-point synchronous measurement system; test excitation conditions are injected at preset positions of the substation through the multi-point synchronous measurement system; under the test excitation conditions, an analog-to-digital converter is used to collect the potential gradient changes of all measurement points to obtain original measurement data; the original measurement data are time-labeled, and the atmospheric temperature, humidity, ground temperature and operating state of the target equipment of the substation at the measurement time are recorded synchronously to obtain environment parameter matching data; the original measurement data and the environment parameter matching data are combined to generate original potential gradient data sets.

3. The method for safety evaluation of substation main grounding grid according to claim 1, characterized in that, The effective signals and interference signals in the original potential gradient data sets are distinguished to obtain substation grounding grid potential distribution information, which comprises: the original potential gradient data sets are input into a multi-layer convolutional neural network model for signal feature extraction to obtain identification result data; Inhibit the interference signal in the identification result data to obtain the potential gradient data after noise reduction; Perform fast Fourier transform on the potential gradient data after noise reduction to obtain frequency domain feature data; According to the frequency domain feature data, the phase difference and amplitude ratio between each measurement point are calculated to obtain a spatial distribution matrix representing the potential gradient relationship of each measurement point; Perform eigenvalue analysis on the spatial distribution matrix to obtain a characteristic decomposition result; Based on the characteristic decomposition result, the first k eigenvectors with a contribution rate exceeding a preset target value are retained to obtain substation grounding grid potential distribution information.

4. The method for safety evaluation of substation main grounding grid according to claim 1, characterized in that, The substation grounding grid potential distribution information is used to establish a substation grounding grid dynamic impedance model, which includes: According to the substation grounding grid potential distribution information, soil parameter inversion is performed to obtain soil thickness and resistivity parameters; Perform temperature coefficient and humidity coefficient calculation on the soil thickness and resistivity parameters to obtain a temperature and humidity compensated soil resistivity model; Based on the temperature and humidity compensated soil resistivity model, a three-dimensional modeling is performed to obtain a grid-based three-dimensional model of the substation grounding grid area, and a complex impedance representation equation is established according to the grid-based three-dimensional model; According to the complex impedance representation equation, a grounding grid metal conductor refinement model considering skin effect and proximity effect is constructed, and a seasonal change correction coefficient is introduced to obtain a seasonally corrected grounding grid model; Through genetic algorithm, parameter optimization iteration is performed on the seasonally corrected grounding grid model to obtain a substation grounding grid dynamic impedance model.

5. The method for safety evaluation of substation main grounding grid according to claim 4, characterized in that, According to the complex impedance representation equation, a grounding grid metal conductor refinement model considering skin effect and proximity effect is constructed, and a seasonal change correction coefficient is introduced to obtain a seasonally corrected grounding grid model, which includes: According to the complex impedance representation equation, the metal conductor of the grounding grid is grid subdivided to obtain a segmented unit model of the conductor cross section; Perform electromagnetic field calculation on the segmented unit model of the conductor cross section to obtain a current density distribution function representing the skin effect; Based on the current density distribution function representing the skin effect, the electromagnetic coupling effect between adjacent conductors is calculated to obtain a conductor mutual impedance matrix representing the proximity effect; Perform seasonal factor analysis on the conductor mutual impedance matrix representing the proximity effect to obtain a set of seasonal correction parameters; According to the set of seasonal correction parameters and the conductor mutual impedance matrix representing the proximity effect, the soil resistivity is corrected and calculated to obtain a seasonal soil resistivity model; The seasonal soil resistivity model, the segmented unit model of the conductor cross section, the current density distribution function representing the skin effect, and the conductor mutual impedance matrix representing the proximity effect are comprehensively processed to obtain a seasonally corrected grounding grid model.

6. The method for safety evaluation of substation main grounding grid according to claim 1, characterized in that, The safety risk assessment based on the whole-process potential distribution map outputs a safety risk area report, which includes: Perform safety risk assessment through the whole-process potential distribution map to obtain spatial-time risk distribution data; Perform three-dimensional K-means clustering analysis on the spatial-time risk distribution data to obtain a risk clustering area set; The spatio-temporal risk density of each cluster region is calculated for the risk cluster region set, and the spatio-temporal risk density is compared with a preset density threshold, to obtain a high-risk region set; Dangerous source tracking analysis is performed on each high-risk region in the high-risk region set, to obtain fault current target path and target node information; A safety risk region report is generated according to the high-risk region set, the fault current target path and the target node information.

7. A substation main grounding grid safety assessment system, characterized in that, The system for safety evaluation of a main grounding grid of a substation for implementing the method for safety evaluation of a main grounding grid of a substation according to any one of claims 1-6 comprises: A setting module is configured to set a multi-point potential gradient sensor for a main grounding grid of a substation of a power transmission and transformation project, and to perform synchronous measurement through the multi-point potential gradient sensor to obtain an original potential gradient data set; A distinguishing processing module is configured to perform effective signal and interference signal distinguishing processing on the original potential gradient data set to obtain substation grounding grid potential distribution information; An establishing module is configured to establish a substation grounding grid dynamic impedance model according to the substation grounding grid potential distribution information; A fault current analysis module is configured to use the substation grounding grid dynamic impedance model for fault current analysis to obtain a whole-process potential distribution graph including an initial stage, a steady state stage and a recovery stage; A safety risk evaluation module is configured to perform safety risk evaluation based on the whole-process potential distribution graph to output a safety risk region report.

Images