An integrated test method and system for boosting voltage and current of GIS equipment

Abstract

The present invention provides a GIS equipment voltage and current boosting integrated test method and system, which relates to the technical field of electrical equipment testing. The method comprises: collecting the discharge voltage of the GIS equipment and recording the discharge phase of the discharge voltage; calculating the average value of the discharge voltage at the same time, and combining the average values ​​of the discharge voltage in a sampling order to obtain a discharge signal time series; reconstructing the discharge signal time series in phase space by a time delay embedding method to obtain a multi-dimensional phase feature; evenly dividing the phase feature according to a preset phase angle, and calculating the maximum value of the Lyapunov exponent of the phase feature within each phase angle range after division; taking the phase feature corresponding to the maximum value of the Lyapunov exponent as the chaotic feature within the corresponding phase angle range; constructing a classification model; collecting real-time chaotic features in each operation loading process; inputting the real-time chaotic features into the classification model as input data, and outputting the insulation defect type.

Description

Technical Field

[0001] The present invention relates to the technical field of electrical equipment testing, and in particular to a GIS equipment voltage and current boosting integrated testing method and system. Background Art

[0002] GIS equipment (Gas Insulated Switchgear) is a switchgear for high-voltage current in power systems. Its characteristic is that it uses gas (usually sulfur hexafluoride SF6) as an insulating medium instead of traditional air or oil. This makes GIS equipment smaller than traditional air-insulated switchgear and more suitable for space-constrained environments, such as urban or industrial applications. It is mainly used in the field of power transmission and distribution to control, protect and isolate electrical equipment to ensure the safe and stable operation of the power grid. Compared with traditional air or oil-insulated switchgear, it has significant advantages. First, due to the excellent insulation performance of SF6 gas, GIS equipment can be more compact, greatly saving installation space, and is particularly suitable for use in cities or space-constrained occasions. Secondly, this type of equipment has higher reliability and lower maintenance requirements, because SF6 gas has stable performance and good sealing in extreme environments, reducing the frequency of routine maintenance. In addition, GIS equipment has a long design life and can provide long-term and stable service in harsh operating environments, effectively supporting the safe and efficient operation of power systems. These advantages make GIS equipment an indispensable part of modern power transmission and distribution systems.

[0003] The integrated voltage and current increasing test method for GIS equipment (gas insulated switchgear) is specially designed to test the performance and durability of GIS equipment under high voltage and high current conditions. This test method covers the operation of increasing voltage and current on GIS equipment at the same time to simulate the extreme conditions that it may encounter in actual operation. The test can not only ensure the high performance and reliability of GIS equipment, but also help to discover and solve potential problems before the product is put on the market, thereby improving the overall safety and efficiency of the power system.

[0004] However, the existing process of testing partial discharge phenomena in GIS equipment often ignores the electromagnetic interference that often exists in the dielectric window, and the signal acquisition equipment can only collect signals in a single direction. The unified processing of the collected signals causes the slight magnetic field changes to be ignored. In addition, direct partial discharge judgment based on the collected partial discharge signals is prone to misjudgment, which leads to a low recognition rate of partial discharge phenomena and affects the safe operation of GIS equipment. Summary of the invention

[0005] In order to solve the technical problems that the process of testing the partial discharge phenomenon of GIS equipment in the prior art often ignores the electromagnetic interference of the dielectric window, and the signal acquisition equipment can only collect signals in one direction, and the unified processing of the collected signals causes the slight magnetic field changes to be ignored, and the direct partial discharge judgment based on the collected partial discharge signals is prone to misjudgment, which leads to a low recognition rate of the partial discharge phenomenon and affects the safe operation of the GIS equipment, the present invention provides a GIS equipment voltage and current boosting integrated test method and system.

[0006] The technical solution provided by the embodiment of the present invention is as follows:

[0007] First aspect

[0008] An embodiment of the present invention provides an integrated test method for boosting voltage and current of GIS equipment, which is used for a double semicircular magnetic antenna, wherein the double semicircular magnetic antenna includes a first antenna and a second antenna that are symmetrically arranged, and the first antenna and the second antenna are both composed of a plurality of coaxial shield coils arranged in layers and connected to an operational amplifier, and the method includes:

[0009] S1: Collecting the discharge voltage of the GIS device within a preset time through the first antenna and the second antenna respectively;

[0010] S2: Calculate the average discharge voltage at the same time, and combine the average discharge voltage in the sampling order to obtain the discharge signal time series;

[0011] S3: Reconstruct the phase space of the discharge signal time series through the time delay embedding method to obtain multi-dimensional phase characteristics;

[0012] S4: evenly divide the phase characteristics according to the preset phase angle, and calculate the maximum value of the Lyapunov exponent of the phase characteristics within each phase angle range after division;

[0013] S5: The phase feature corresponding to the maximum value of the Lyapunov exponent is taken as the chaotic feature within the corresponding phase angle range;

[0014] S6: construct a classification model based on an extreme gradient boosting machine, wherein the classification model includes an input layer, multiple decision trees arranged in a stacked manner, and an output layer, wherein each decision tree corrects the residual of the previous decision tree connected thereto, and the classification model is provided with a multi-class logarithmic loss function;

[0015] S7: Loading is performed in the order of voltage boosting operation, current boosting operation and parallel voltage boosting and current boosting operation. In the process of loading each operation, real-time chaotic features are collected, wherein adjacent operations are separated by a preset time.

[0016] S8: The real-time chaotic features are fed as input data into the classification model, and the insulation defect type is output.

[0017] Second aspect

[0018] An embodiment of the present invention provides a GIS equipment voltage and current boosting integrated test system, comprising: a processor;

[0019] A memory having computer-readable instructions stored thereon, which, when executed by a processor, implements the integrated test method for boosting voltage and current of GIS equipment as in the first aspect.

[0020] The third aspect

[0021] An embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the integrated test method for boosting voltage and current of GIS equipment as described in the first aspect is implemented.

[0022] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0023] (1) In the present invention, a double semicircular antenna including a first antenna and a second antenna composed of a plurality of coaxial shield coils symmetrically stacked and connected to an operational amplifier is provided to collect local discharge signals. No matter in which direction the signal source is in the window, the symmetrical installation method ensures uniform collection of the magnetic field from the dielectric window of the GIS equipment. This symmetry enables the antenna to receive signals from two directions, thereby improving the accuracy of the measurement and the redundancy of the system, which can not only reduce the noise of the collected signal, but also increase the sensitivity to the signal, effectively capture weak signals, and increase the accuracy of the collected signal. In the process of processing the collected signals, the discharge voltage collected by the two antennas is averaged to ensure the accuracy and coverage of the collected signals. Then the Lyapunov exponent is calculated to obtain the chaotic characteristics, which not only enhances the understanding of the system sensitivity and predictive ability, but also improves the accuracy of fault identification. By converting local discharge signals into chaotic characteristics for pattern recognition, chaos theory emphasizes the sensitivity of the system to initial conditions, which enables chaotic characteristics to capture tiny changes, which is crucial for early diagnosis and identification of weak discharge activities. This method greatly improves the sensitivity, accuracy and robustness of insulation defect detection, and effectively handles the randomness and complexity of signals.

[0024] (2) After obtaining the chaotic features, constructing a classification model based on the extreme gradient boosting machine and performing orderly operation tests has significant advantages. The model uses multiple decision trees, each of which corrects the residual of the previous tree, so that the model can gradually optimize and improve the accuracy of diagnosis. This incremental learning method is particularly suitable for processing complex data structures, such as multivariate and nonlinear problems encountered in insulation defect detection of GIS equipment. Through multi-class logarithmic loss functions, the model can effectively handle multi-classification problems. The sequential loading method of boosting operation, current boosting operation and parallel boosting and current boosting operation can not only effectively shield the equipment from premature heating caused by the first current boost, which affects the accuracy of subsequent tests, but also can comprehensively and accurately test the operation of GIS equipment under different loading methods to ensure the normal operation of the equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0026] Figure 1 A schematic flow chart of a GIS equipment pressure and current boosting integrated test method provided by an embodiment of the present invention;

[0027] Figure 2 A schematic structural diagram of a double semicircular magnetic antenna provided by an embodiment of the present invention;

[0028] Figure 3 A schematic diagram of a circuit structure of a first antenna provided by an embodiment of the present invention;

[0029] Figure 4 A schematic diagram of the structure of a classification model provided by an embodiment of the present invention;

[0030] Figure 5 A schematic structural diagram of a GIS equipment voltage and current boosting integrated test system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0031] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0032] Reference Manual Attached Figure 1 , showing a flow chart of a GIS equipment pressure and current boosting integrated test method provided by an embodiment of the present invention.

[0033] Reference Manual Attached Figure 2 , showing a schematic structural diagram of a double semicircular magnetic antenna provided in an embodiment of the present invention.

[0034] Reference Manual Attached Figure 3 , shows a schematic diagram of the circuit structure of a first antenna provided by an embodiment of the present invention.

[0035] An embodiment of the present invention provides an integrated test method for boosting voltage and current of GIS equipment, which is used for a double semicircular magnetic antenna, wherein the double semicircular magnetic antenna includes a first antenna and a second antenna symmetrically arranged, and the first antenna and the second antenna are both composed of a plurality of coaxial shield coils stacked and connected to an operational amplifier.

[0036] Among them, the shield coil is an electromagnetic coil with a metal shield, which is used to reduce or prevent electromagnetic interference. The design purpose of this coil is to resist external electromagnetic field interference by wrapping a layer of conductive material, while also preventing the electromagnetic field generated by the coil itself from affecting surrounding equipment.

[0037] Specifically, the structure of the antenna mainly includes two shielded pickup coils, each of which is connected to its own high input impedance operational amplifier. Each pickup coil is composed of a multi-turn semicircular coil and is made of coaxial cable to provide electric field shielding. This design allows the antenna to effectively capture the magnetic field passing through the window in the dielectric window. The advantages of this antenna design are its high sensitivity and low noise performance. First, the use of coaxial cable to make the coil can effectively shield external electromagnetic interference and improve the accuracy of measurement. Secondly, the multi-turn semicircular design increases the sensing area of ​​the coil, enabling it to capture weaker magnetic field changes, which is very important for detecting low-amplitude partial discharge signals. In addition, the symmetrical installation of the antenna enables it to uniformly sense the magnetic field in the dielectric window of the GIS device, regardless of the direction of the signal source in the window, that is, the symmetrical installation ensures the uniform collection of the magnetic field from the dielectric window of the GIS device. This symmetry allows the antenna to receive signals from two directions, thereby improving the accuracy of the measurement and the redundancy of the system, which is not available in the overall signal acquisition coil. Through this structural design, the new magnetic field antenna not only improves the detection sensitivity of partial discharge signals, but also effectively reduces the impact of background noise, enabling the antenna to play a greater role in the health monitoring and fault diagnosis of electrical equipment.

[0038] In a possible implementation, the double semicircular magnetic antenna further includes: an op amp non-reverse terminal resistor, an op amp feedback resistor, an op amp input gate resistor, a DC decoupling capacitor, an op amp output terminal resistor, and an op amp power supply.

[0039] In a possible implementation, the circuit structures of the first antenna and the second antenna of the dual semicircular magnetic antenna are the same, wherein the circuit structure of the first antenna is specifically:

[0040] The op amp power supply is connected to the op amp, the op amp is connected to the first coil through the op amp non-reverse terminal resistor, the op amp feedback resistor is connected in parallel to the reverse input terminal and the output terminal of the op amp, one end of the op amp input gate resistor is grounded, and the other end is connected between the feedback resistor and the reverse input terminal of the op amp, one end of the DC decoupling capacitor is connected between the output terminal of the op amp and the feedback resistor, and the other end outputs a detection signal, i.e., a discharge voltage, through the op amp output terminal resistor.

[0041] It should be noted that the magnetic antenna uses two semicircular coils installed at the top and bottom of the dielectric window. This symmetrical layout greatly enhances the antenna's ability to capture magnetic field changes and improves the sensitivity and accuracy of detection. Secondly, this design enables the antenna to capture magnetic field signals generated by partial discharge from multiple directions, enhances signal reliability, and reduces interference from environmental noise. In addition, the new antenna is easy to install, compatible with the dielectric window of existing GIS equipment, does not require additional structural changes, is easy to maintain and replace, and provides strong technical support for the safe operation of the power system.

[0042] Among them, R1 is the feedback resistor of the op amp, and R2 is the input gate resistor of the op amp. Both are resistors connected to the operational amplifier, representing the feedback resistor and the gain setting resistor, respectively, which are used to determine the gain and input impedance of the operational amplifier. R3 is the non-reverse terminal resistor of the op amp, which is used to stabilize the signal and prevent oscillation caused by the resonance characteristics of the circuit. C represents the DC decoupling capacitor, which is located at the output end of the operational amplifier and is used to block the DC component to ensure that the output signal contains only the AC component. R4 is the output terminal resistor of the op amp to set the output impedance to adapt to the downstream circuit, ensure the effective transmission of the signal, and reduce reflections and losses. Coil represents the coaxial shield coil. OPA represents the op amp. Vout represents the output terminal of the op amp. V+ and V- represent the positive and negative poles of the power supply, respectively, for the normal operation of the op amp.

[0043] The processing flow of the integrated test method for boosting pressure and current of GIS equipment may include the following steps:

[0044] S1: The discharge voltage of the GIS device within a preset time period is collected through the first antenna and the second antenna respectively.

[0045] It should be noted that those skilled in the art can set the preset duration according to actual needs, and the present invention is not limited here. Specifically, the preset duration can be two minutes.

[0046] S2: Calculate the average discharge voltage at the same time, and combine the average discharge voltage in the sampling order to obtain the discharge signal time series.

[0047] It should be noted that, compared with the traditional acquisition method, this method can more accurately measure the local discharge signal and avoid the measurement result deviation caused by the insensitivity of the unified measurement to the discharge voltage. By calculating the average value of the discharge voltage at the same time and sequentially combining these average values ​​into a time series, the present invention can effectively reduce the impact of occasional errors at a single measuring point and enhance the stability and reliability of the data. This method not only improves the measurement accuracy of the discharge signal, but also helps to capture the dynamic changes of the entire discharge process more continuously and systematically, providing a more balanced and comprehensive data foundation for subsequent in-depth analysis and fault diagnosis.

[0048] S3: The phase space of the discharge signal time series is reconstructed by the time delay embedding method to obtain multi-dimensional phase characteristics.

[0049] Among them, the time delay embedding method is a mathematical technique used to reconstruct the multidimensional phase space of a system from a univariate time series. This method is based on the Takens embedding theorem, which states that the dynamics of a system can be reconstructed from a single observation time series. Phase space reconstruction is the process of reconstructing the multidimensional dynamic phase space of a system from a one-dimensional time series using the time delay embedding method. In this reconstructed phase space, each point represents the state of the system at a certain moment, and the trajectory from one point to another reveals the evolution of the system state over time. This reconstruction makes the dynamic characteristics of the system originally hidden in the single-dimensional time series visible and can be further analyzed. Through phase space reconstruction, the multidimensional phase features obtained from the original time series can more comprehensively describe the dynamic behavior of the system. These features include but are not limited to phase features that represent dynamic properties such as attractors, singularities, and periodicity of the system. Multidimensional phase features, that is, multidimensional phase features, provide deeper system dynamic information and help understand the inherent laws of complex systems, especially in prediction, classification, and system behavior analysis. In the field of electrical engineering and signal processing, it makes it possible to extract complex system dynamic information from simple voltage or current measurements. In this way, possible failures or abnormal behaviors of the system can be diagnosed and predicted more accurately, thereby improving the reliability and efficiency of the system in practical applications.

[0050] In a possible implementation, S3 specifically includes:

[0051] S301: Determine the embedding dimension of the time-delay embedding method based on Takkens embedding theorem:

[0052] m≥2d+1

[0053] Wherein, m represents the embedding dimension, and d represents the system degrees of freedom of the GIS equipment, wherein the system degrees of freedom include the discharge voltage and the discharge phase.

[0054] Among them, Takkens embedding theorem is generally an important concept in dynamical system theory. It provides a theoretical basis for reconstructing the state space of a dynamic system from time series data, which is especially critical when dealing with systems that may exhibit chaotic behavior. Using Takkens embedding theorem to determine the embedding dimension and time delay allows us to effectively reconstruct the complete state space of a dynamic system from only time series data. This approach not only ensures that the dimension of the data is sufficient to capture the full dynamic behavior of the system, but also helps us reveal the complex dynamics that may exist in the system, such as periodicity, chaos, or nonlinear behavior.

[0055] S302: Determine the time delay of the time delay embedding method by using the autocorrelation function method:

[0056]

[0057] Where R(τ) represents the autocorrelation function value at time delay τ, x t represents the discharge voltage at time t, represents the voltage average of the discharge signal time series, and N represents the total length of the discharge signal time series.

[0058] S303: Select a time delay where the autocorrelation function value is less than a preset autocorrelation function value:

[0059] R(τ)≤R'(τ)

[0060] Wherein, R'(τ) represents the preset autocorrelation function value.

[0061] Among them, the time delay is the time interval between the elements in the phase space vector. Its selection is crucial for the accurate reconstruction of the phase space. It is determined by analyzing the autocorrelation function of the original data. A time interval can be found in which the information overlap between data points is minimized, thereby maximizing the independence between system states.

[0062] It should be noted that those skilled in the art can set the size of the preset autocorrelation function value according to actual needs, and the present invention is not limited here.

[0063] Optionally, R'(τ) = 1 - e -1 .

[0064] S304: Reconstruct the phase space of the discharge signal time series according to the embedding dimension and time delay:

[0065] X i =(x i ,x i+τ ,x i+2τ ,…,x i+(m-1)τ ),i=1,2,…,n

[0066] n=N-(m-1)τ

[0067] Among them, X i represents the i-th phase feature in the phase space, n represents the total number of phase features, x i represents the i-th discharge voltage in the discharge signal time series, and N represents the total length of the discharge signal time series.

[0068] It should be noted that, first, the embedding dimension is determined using the Takkens embedding theorem. This step is to ensure that the dimension of the time series data is sufficient to reflect the entire dynamic behavior of the system, thereby revealing possible complex dynamics such as chaos or nonlinear behavior in the system. Next, the time delay is determined by the autocorrelation function method to help find a suitable time interval so that the information of each point in the phase space is as independent as possible and the information overlap between data points is reduced, which is crucial for the subsequent reconstruction of the system state space. Finally, according to the determined embedding dimension and time delay, the phase space of the discharge signal time series is reconstructed. This step allows the construction of a multidimensional space describing the system state from the original one-dimensional time series, thereby more comprehensively understanding the behavior of the system. The advantage of the whole process is that it can reconstruct the complete dynamic picture of the system from limited observation data through mathematical and statistical methods, enhance the understanding and prediction of the system behavior, especially when dealing with complex power systems such as GIS equipment. This method can provide deeper insights and provide a scientific basis for subsequent analysis, monitoring and fault diagnosis. The application of this method improves the accuracy and efficiency of discharge analysis, helps to optimize the maintenance and operation of equipment, and thus improves the overall performance and safety of the system.

[0069] S4: Evenly divide the phase characteristics according to the preset phase angles, and calculate the maximum value of the Lyapunov exponent of the phase characteristics within each phase angle range after division.

[0070] Among them, the preset phase angle refers to the pre-set, evenly distributed angle division of the phase within the discharge cycle of the electrical equipment. The phase angle is a measure of the position of each point in a cycle in the electrical waveform. For the AC system, a complete cycle is 360 degrees. The Lyapunov index is a measure of the speed at which a dynamic system diverges or converges from the initial conditions. It can be used to analyze whether the system has chaotic behavior. As long as the maximum Lyapunov index is calculated, it can be determined whether the system is chaotic. The positive and negative and size of the Lyapunov index can reveal the sensitivity of the system to the initial state and whether the evolution of the system state over time is stable.

[0071] It should be noted that those skilled in the art can set the size of the preset phase angle according to actual needs, and the present invention is not limited here.

[0072] In a possible implementation, S4 specifically includes:

[0073] S401: Determine the nearest neighbor point of each phase feature:

[0074]

[0075] Among them, X ip For any phase feature X j , and the i-th phase feature X i The distances are all the phase characteristics corresponding to the nearest neighbor point ip, |||| represents the calculation of the Euclidean distance, represents, Δt represents the sampling interval of the discharge voltage, τ w Represents the preset time window length, τ w / Δt represents the number of sampling points in a preset time window length, and ji represents the index interval.

[0076] It should be noted that the ratio is used to determine the minimum and maximum index intervals between data points in the analysis. This helps ensure that the time intervals between data points in the analysis are neither too short (causing the data points to be too interdependent and affected by autocorrelation) nor too long (causing a loss of temporal correlation). w / Δt is the key parameter that connects the actual data acquisition settings with the dynamic system analysis requirements, which ensures the validity and reliability of time series analysis.

[0077] It should be noted that those skilled in the art can set the size of the preset time window length according to actual needs, and the present invention is not limited here.

[0078] S402: Calculate the local maximum value of the Lyapunov exponent of each phase feature point at each phase angle at a preset step size:

[0079]

[0080] Where λ1(i) represents the local maximum value of the Lyapunov exponent calculated at the i-th phase characteristic point, c represents the number of iterations, h represents the time step, and X ip+ch represents the phase characteristics of the nearest neighbor point ip after the ch time step, X i+ch Represents the phase characteristics of the i-th phase characteristic point after ch time steps.

[0081] It should be noted that those skilled in the art can set the size of the preset step size according to actual needs, and the present invention is not limited here.

[0082] It should be noted that in the analysis of dynamic systems, the local maximum of the Lyapunov exponent is a key indicator used to measure the sensitivity of the system to the initial conditions and the degree of chaos in its dynamics. Through the formula, we can quantify the divergence speed of the system state from the time period. The separation speed of the system state is evaluated by calculating the ratio of the change in the distance between the state vector at the initial time point and a future time point. The local maximum of the Lyapunov exponent with a positive value indicates that the system exhibits chaotic behavior, indicating that the system state is extremely sensitive to the initial conditions, while the negative value or the exponent close to zero shows the stability or periodicity of the system.

[0083] S403: Taking the average of the local maximum values ​​of the Lyapunov exponent in units of phase angle, the maximum value of the Lyapunov exponent at each phase angle is obtained:

[0084]

[0085] Among them, λ1 represents the maximum value of Lyapunov exponent, λ1(i) represents the mth local maximum value of Lyapunov exponent, m=1,2,…M, and M represents the total number of local maximum values ​​of Lyapunov exponent obtained at the phase angle.

[0086] It should be noted that by systematically dividing and analyzing the discharge phases, the maximum value of the Lyapunov exponent within each phase range is calculated to evaluate whether the system exhibits chaotic behavior. The key to this process is to accurately quantify the speed at which the system state changes with a small change in the initial conditions, using the Lyapunov exponent as a measure of this change. This index is a core indicator in dynamic system analysis, which can reveal the system's sensitivity to the initial state and whether its evolution over time is stable.

[0087] S5: The phase feature corresponding to the maximum value of the Lyapunov exponent is taken as the chaotic feature within the corresponding phase angle range.

[0088] Among them, the number of chaotic features is related to the number of pre-divided phase angles. In practical applications, it can be divided into 18 or 36 uniform phase angles to capture the discharge voltage characteristics from a coarse and fine granularity, providing an accurate data basis for the subsequent accurate judgment of the insulation defect type. Chaotic features are usually associated with the chaotic behavior exhibited by the system in the analysis of dynamic systems, that is, the system state evolves over time and shows complex dynamics that are highly sensitive and dependent on initial conditions. In electrical engineering, especially in discharge analysis, chaotic features can be used to identify and predict under what conditions the system may exhibit irregular discharge behaviors, which may be triggered by defects, damage or changes in external environmental factors within the system. As an indicator of the sensitivity of a dynamic system, the maximum value of the Lyapunov exponent reflects the most sensitive aspect of the system to changes in the initial state. In the context of discharge voltage, a positive Lyapunov exponent indicates that even a small voltage change can lead to a significant change in the system behavior. This property is a typical feature of a chaotic system. Therefore, considering the phase feature corresponding to the maximum value of the Lyapunov exponent as a chaotic feature helps to locate potential unstable areas in the discharge process, thereby providing a basis for further analysis and intervention.

[0089] Refer to the instruction manual Figure 4 , showing a structural schematic diagram of a classification model provided by an embodiment of the present invention.

[0090] S6: Build a classification model based on extreme gradient boosting machine.

[0091] The classification model includes an input layer, multiple decision trees arranged in a stacked manner, and an output layer, wherein each decision tree corrects the residual of the previous decision tree connected thereto, and the classification model is provided with multiple types of logarithmic loss functions.

[0092] Among them, the extreme gradient boosting machine is mainly used to improve the performance and speed of tree models (especially decision trees). This algorithm systematically uses the gradient boosting framework to generate a new tree at each step, and then tries to correct the residual of the previous step (that is, the error predicted by the previous model). This process is repeated, and each step reduces the loss function of the model (that is, the error evaluation index). In this way, the model can gradually learn the complex structure in the data and effectively improve the accuracy of the prediction. The main advantage of the multi-class logarithmic loss function is that it can directly use the probability of the model output for optimization, so that the model not only focuses on the correct classification during training, but also focuses on the accuracy of the predicted probability. This is very useful for many application scenarios that require probabilistic explanations. For example, in classification tasks with high uncertainty, providing probabilistic information in addition to the final classification can help decision makers better understand the prediction results of the model.

[0093] In a possible implementation, S6 specifically includes:

[0094] S601: Collect chaotic feature sets under various insulation defect types.

[0095] S602: With the multi-class logarithmic loss function value being less than a preset multi-class logarithmic loss function value as a goal, the classification model is trained using the chaotic feature set as input data, and a predicted value is output.

[0096] S603: Calculate the residual between the true value and the predicted value:

[0097]

[0098] Among them, y i represents the true value of sample i in the chaotic feature set, Represents the predicted value of the current classification model after t-1 rounds of iteration.

[0099] S604: Taking the prediction residual as the goal, the next decision tree connected to each decision tree is trained using the chaotic feature set.

[0100] S605: Connect the next decision tree to the last layer of the current model and update the classification model.

[0101] S606: Repeat S602-S605 until the residual is smaller than the preset residual.

[0102] It can be understood that by using a classification model based on extreme gradient boosting machine combined with chaotic feature sets to identify insulation defects of GIS equipment, through iterative training, the model continuously optimizes the prediction residual and gradually improves the prediction accuracy. This method greatly improves the accuracy and efficiency of diagnosis, helps to detect potential faults in advance, and thus enhances the reliability and safety of the system.

[0103] It should be noted that those skilled in the art can set the size of the preset residual according to actual needs, and the present invention is not limited here.

[0104] In a possible implementation, the multi-class logarithmic loss function is specifically:

[0105]

[0106] Among them, y ij represents the true label of sample i for insulation defect type j, i = 1, 2, ..., n, j = 1, 2, ..., k, p ij It represents the probability that the classification model predicts that sample i belongs to insulation defect type j.

[0107] In a possible implementation, the types of insulation defects include needle-like protrusions, free metal particles, suspended electrodes, and solid insulation air gaps.

[0108] S7: Loading is performed in the order of voltage boosting operation, current boosting operation and parallel voltage boosting and current boosting operation. During the loading process of each operation, real-time chaotic features are collected, wherein adjacent operations are spaced by a preset time.

[0109] Performing a voltage boost test first ensures that the device is electrically safe and has good insulation performance, which is the basis for subsequent current boost testing. If the current boost is performed first, the device may heat up prematurely or have poor contact, which may cause unnecessary electrical failures or insulation failures in subsequent voltage boost tests. A reasonable test sequence is essential to ensure the safety and effectiveness of the test.

[0110] It should be noted that those skilled in the art can set the preset duration according to actual needs, and the present invention is not limited thereto.

[0111] S8: The real-time chaotic features are fed as input data into the classification model, and the insulation defect type is output.

[0112] Reference Manual Attached Figure 5 , showing a structural schematic diagram of a GIS equipment voltage and current boosting integrated test system provided by the present invention.

[0113] The present invention further provides a GIS equipment voltage and current boosting integrated test system 20, which is applied to the above-mentioned GIS equipment voltage and current boosting integrated test method, comprising:

[0114] Processor 201.

[0115] The memory 202 stores computer-readable instructions. When the computer-readable instructions are executed by the processor 201, the integrated test method for boosting voltage and current of GIS equipment as described in the method embodiment is implemented.

[0116] The GIS equipment voltage and current boosting integrated test system 20 provided by the present invention can execute the above-mentioned GIS equipment voltage and current boosting integrated test method and achieve the same or similar technical effects. To avoid repetition, the present invention will not go into details.

[0117] An embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the integrated test method for boosting voltage and current of GIS equipment as described in the method embodiment.

[0118] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. The protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A GIS equipment pressure and current boosting integrated test method, characterized in that: Applied to a double semicircular magnetic antenna, wherein the double semicircular magnetic antenna comprises a first antenna and a second antenna that are symmetrically arranged, and the first antenna and the second antenna are both composed of a plurality of coaxial shield coils arranged in layers and connected to an operational amplifier, and the method comprises: S1: collecting the discharge voltage of the GIS device within a preset time period through the first antenna and the second antenna respectively; S2: Calculate the average value of the discharge voltage at the same time, and combine the average values ​​of the discharge voltage in a sampling order to obtain a discharge signal time series; S3: reconstructing the phase space of the discharge signal time series by a time delay embedding method to obtain a multi-dimensional phase feature; S4: evenly dividing the phase feature according to a preset phase angle, and calculating the maximum value of the Lyapunov exponent of the phase feature within each phase angle range after the division; S5: taking the phase feature corresponding to the maximum value of the Lyapunov exponent as the chaotic feature within the corresponding phase angle range; S6: constructing a classification model based on an extreme gradient boosting machine, wherein the classification model includes an input layer, a plurality of decision trees arranged in a stacked manner, and an output layer, wherein each decision tree corrects the residual of a connected previous decision tree, and the classification model is provided with a plurality of logarithmic loss functions; S7: Loading is performed in the order of voltage boosting operation, current boosting operation and parallel voltage boosting and current boosting operation. In the process of loading each operation, real-time chaotic features are collected, wherein adjacent operations are separated by a preset time. S8: inputting the real-time chaotic characteristics as input data into the classification model, and outputting the insulation defect type; The S3 specifically includes: S301: Determine the embedding dimension of the time delay embedding method based on Takens embedding theorem: m≥2d+1 Wherein, m represents the embedding dimension, d represents the system degree of freedom of the GIS device, wherein the system degree of freedom includes the discharge voltage and the discharge phase; S302: Determine the time delay of the time delay embedding method by using the autocorrelation function method: Where R(τ) represents the autocorrelation function value at time delay τ, x t represents the discharge voltage at time t, represents the voltage average value of the discharge signal time series, and N represents the total length of the discharge signal time series; S303: Selecting a time delay in which the autocorrelation function value is less than a preset autocorrelation function value: R(τ)≤R'(τ) Wherein, R'(τ) represents the preset autocorrelation function value; S304: Reconstruct the phase space of the discharge signal time series according to the embedding dimension and the time delay: X i =(x i ,x i+τ ,x i+2τ ,…,x i+(m-1)τ ),i=1,2,…,n n=N-(m-1)τ Among them, X i represents the i-th phase feature in the phase space, n represents the total number of phase features, x i represents the i-th discharge voltage in the discharge signal time series, and N represents the total length of the discharge signal time series; The S4 specifically includes: S401: Determine the nearest neighbor point of each phase feature: Among them, X ip For any phase feature X j , and the i-th phase feature X i The distances are all the phase characteristics corresponding to the nearest neighbor point ip, || || represents the calculation of the Euclidean distance, represents, Δt represents the sampling interval of the discharge voltage, τ w Represents the preset time window length, τ w / Δt represents the number of sampling points in the preset time window length, and ji represents the index interval; S402: Calculate the local maximum value of the Lyapunov exponent of each phase feature point at each phase angle at a preset step size: Where λ1(i) represents the local maximum value of the Lyapunov exponent calculated at the i-th phase characteristic point, c represents the number of iterations, h represents the time step, and X ip+ch represents the phase characteristics of the nearest neighbor point ip after the ch time step, X i+ch Represents the phase characteristics of the i-th phase characteristic point after ch time steps; S403: Taking the average of the local maximum values ​​of the Lyapunov exponents in units of phase angle, to obtain the maximum value of the Lyapunov exponent at each phase angle: Wherein, λ1 represents the maximum value of the Lyapunov exponent, λ1(i) represents the mth local maximum value of the Lyapunov exponent, m=1,2,…M, and M represents the total number of local maximum values ​​of the Lyapunov exponent obtained at the phase angle.

2. The integrated test method for boosting pressure and current of GIS equipment according to claim 1 is characterized in that: The double semicircular magnetic antenna also includes: an operational amplifier non-reverse terminal resistor, an operational amplifier feedback resistor, an operational amplifier input gate resistor, a direct current decoupling capacitor, an operational amplifier output terminal resistor and an operational amplifier power supply.

3. The integrated test method for boosting pressure and current of GIS equipment according to claim 2 is characterized in that: The circuit structures of the first antenna and the second antenna of the double semicircular magnetic antenna are the same, wherein the circuit structure of the first antenna is specifically as follows: The op amp power supply is connected to the op amp, the op amp is connected to the first coil through the non-reverse terminal resistor of the op amp, the op amp feedback resistor is connected in parallel to the reverse input terminal of the op amp and the output terminal of the op amp, one end of the op amp input gate resistor is grounded, and the other end is connected between the feedback resistor and the reverse input terminal of the op amp, one end of the DC decoupling capacitor is connected between the output terminal of the op amp and the feedback resistor, and the other end outputs a detection signal, i.e., a discharge voltage, through the output terminal resistor of the op amp.

4. The integrated test method for boosting pressure and current of GIS equipment according to claim 1 is characterized in that: The S6 specifically includes: S601: Collect chaotic feature sets under various insulation defect types; S602: With the multi-class logarithmic loss function value being less than a preset multi-class logarithmic loss function value as a goal, the classification model is trained using the chaotic feature set as input data, and a predicted value is output; S603: Calculate the residual between the true value and the predicted value: Among them, y i represents the true value of sample i in the chaotic feature set, Represents the predicted value of the current classification model after t-1 rounds of iteration; S604: With the goal of predicting the residual, the next decision tree connected to each decision tree is trained using the chaotic feature set; S605: Connect the next decision tree to the last layer of the current model and update the classification model; S606: Repeat S602-S605 until the residual is smaller than a preset residual.

5. The integrated test method for boosting pressure and current of GIS equipment according to claim 1 is characterized in that: The insulation defect types include needle-like protrusions, free metal particles, suspended electrodes and solid insulation air gaps.

6. The integrated test method for boosting voltage and current of GIS equipment according to claim 1 is characterized in that: The multi-class logarithmic loss function is specifically: Among them, y ij represents the true label of sample i for insulation defect type j, i = 1, 2, ..., n, j = 1, 2, ..., k, p ij It represents the probability that the classification model predicts that sample i belongs to insulation defect type j.

7. A GIS equipment pressure and current boosting integrated test system, characterized in that: include: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the integrated test method for boosting voltage and current of GIS equipment as described in any one of claims 1 to 6 is implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the integrated test method for boosting voltage and current of GIS equipment as described in any one of claims 1 to 6 is implemented.

Images