Method and system for detecting state of zinc oxide arrester valve plate

Abstract

The invention relates to a state detection method and system for a valve plate of a zinc oxide arrester, belongs to the technical field of zinc oxide arresters, and solves the technical problem of inaccurate detection caused by an inflection point effect of an existing detection method. The method comprises the following steps: carrying out impact current test on an undegraded lightning arrester valve plate in different environments, and obtaining test environment data and degradation state parameters of the undegraded lightning arrester valve plate; taking the test environment data as input, and taking the degradation state parameters of the undegraded lightning arrester valve plate as labels to train a threshold optimization model; performing an impact current test on the target lightning arrester valve plate to obtain current environment data and a degradation state parameter of the target lightning arrester valve plate; inputting the current environment data into the trained threshold optimization model to obtain a degradation threshold; and judging the degradation state of the target lightning arrester valve plate according to the degradation state parameter and the degradation threshold value of the target lightning arrester valve plate. The deterioration state parameters have the advantages of being advanced, and the detection accuracy is improved.

Description

Technical Field

[0001] The invention relates to the technical field of zinc oxide lightning arresters, and in particular to a state detection method and system for a valve plate of a zinc oxide lightning arrester. Background Art

[0002] As the core device of lightning protection, zinc oxide arrester valve plates have the advantages of excellent nonlinear volt-ampere characteristics, fast response time, low residual voltage, no follow current and large current discharge capacity. They are widely used in precision electrical appliances, power supplies, information systems, rail transit, communications and other power-related lightning protection. Therefore, the effect of overvoltage protection directly depends on the quality of zinc oxide arrester valve plates. During operation, zinc oxide arrester valve plates will inevitably deteriorate due to lightning shocks and the continuous application of system AC and DC voltages. This phenomenon will seriously affect line safety. At the least, the arrester will completely lose its protective function, and at worst, it will cause power system failures and even explosions and fires. Therefore, it is usually necessary to detect the status of zinc oxide arresters.

[0003] The existing detection methods of arrester valve plates are usually based on the two parameters of varistor voltage or leakage current. However, during the deterioration of zinc oxide arrester valve plates, these two parameters have an obvious inflection point effect, which may lead to the following situation: the monitoring values ​​of varistor voltage and / or leakage current meet the numerical range required by the specification, but the zinc oxide arrester valve plate itself has been fully deteriorated. Due to the inflection point effect, the existing detection methods based on varistor voltage or leakage current have low accuracy and are prone to safety problems. Summary of the invention

[0004] In view of the above analysis, the embodiments of the present invention aim to provide a method and system for detecting the status of a zinc oxide arrester valve plate, so as to solve the technical problem that the existing detection method causes inaccurate detection due to the inflection point effect.

[0005] On the one hand, an embodiment of the present invention provides a method for detecting the state of a zinc oxide lightning arrester valve plate, the method comprising:

[0006] When performing impulse current test, the product of the capacitance and varistor voltage of the arrester valve after being subjected to impulse current is defined as the degradation state parameter;

[0007] Performing an impulse current test on a non-degraded arrester valve plate under different environments to obtain test environment data and degradation state parameters of the non-degraded arrester valve plate;

[0008] The test environment data is used as input, and the degradation state parameters of the valve plate of the non-degraded arrester are used as labels to train a pre-established threshold optimization model;

[0009] Performing an impulse current test on a target arrester valve plate to obtain current environmental data and degradation state parameters of the target arrester valve plate;

[0010] Inputting the current environmental data into a trained threshold optimization model to obtain a degradation threshold;

[0011] Determine whether the degradation state parameter of the target arrester valve plate exceeds the degradation threshold; if not, the target arrester valve plate is in the early stage of degradation; if so, the arrester valve plate is in the middle and late stage of degradation.

[0012] Based on a further improvement of the above method, the environmental data includes environmental temperature and environmental humidity.

[0013] Based on the further improvement of the above method, the threshold optimization model is a binary linear regression model, and the formula of the threshold optimization model is:

[0014] f(x)=k1x1+k2x2+b;

[0015] X = [x1, x2, 1];

[0016]

[0017] In the formula, x1 is the ambient humidity, x2 is the ambient humidity, is the degradation threshold of the model output, k1, k2 and b are all model parameters.

[0018] Based on the further improvement of the above method, the loss function of the threshold optimization model is the mean square error function, and the formula is:

[0019]

[0020] Where y is the degradation state parameter.

[0021] Based on the further improvement of the above method, when the surge current test is performed on the arrester valve plate, the current value of the surge current used is greater than the breakdown current of the arrester valve plate.

[0022] Based on the further improvement of the above method, after the arrester valve plate is struck by the impulse current, the capacitance of the arrester valve plate is detected first, and then the varistor voltage of the arrester valve plate is detected.

[0023] Based on the further improvement of the above method, when the surge current test is performed on the arrester valve plate, an 8 / 20μs surge current is applied to the unused arrester valve plate or the target arrester valve plate.

[0024] Based on the further improvement of the above method, the method further includes:

[0025] When performing an impulse current test on a target arrester valve plate, monitoring and recording the highest temperature of the target arrester valve plate after being subjected to the impulse current;

[0026] Performing an offline leakage current test on the target arrester valve plate in the middle and late stages of degradation to obtain the leakage current of the target arrester valve plate; the middle and late stages of degradation include the middle stage of degradation and the late stage of degradation;

[0027] Whether the target arrester valve plate is in the middle or late stage of degradation is determined based on the leakage current of the arrester valve plate and the maximum temperature after being subjected to the impact current; if the arrester valve plate is in the middle stage of degradation, the arrester valve plate can continue to be put into use; if the arrester valve plate is in the late stage of degradation, the arrester valve plate cannot continue to be put into use.

[0028] Based on the further improvement of the above method, judging whether the state of the arrester valve plate is in the middle stage of degradation or in the late stage of degradation according to the leakage current of the target arrester valve plate and the highest temperature after being subjected to the impact current includes:

[0029] Build a state detection model based on machine learning;

[0030] Obtaining leakage currents of arrester valve plate samples at an early stage of degradation and arrester valve plate samples at a late stage of degradation and the highest temperatures after being subjected to impact currents, respectively;

[0031] The leakage current of the arrester valve plate sample and the maximum temperature after being subjected to the impact current are used as inputs, and the degradation state is used as a label to train the state detection model;

[0032] The leakage current of the target arrester valve plate in the middle and late stages of degradation and the maximum temperature after being subjected to the impact current are input into the trained state detection model to obtain whether the arrester valve plate is in the middle or late stages of degradation.

[0033] Based on the further improvement of the above method, the state detection model is a support vector classifier.

[0034] Based on the further improvement of the above method, the method comprises:

[0035] If the arrester valve plate is in the early stage of degradation, then after the arrester valve plate is put into use, the working state parameter of the arrester valve plate is obtained once every first online monitoring sampling time;

[0036] If the arrester valve plate is in the middle stage of degradation, then after the arrester valve plate is put into use, the working state parameter of the arrester valve plate is obtained once every second online monitoring sampling time;

[0037] Wherein, the second online monitoring sampling time is less than the first online monitoring sampling time;

[0038] The working state parameters of the arrester valve plate are used to monitor the health state of the arrester valve plate online, and the working state parameters include the leakage current and / or temperature of the arrester valve plate.

[0039] On the other hand, an embodiment of the present invention further provides a state detection system for a zinc oxide lightning arrester valve plate, the system comprising:

[0040] An impulse current generator is used to apply an impulse current to the valve plate of the target lightning arrester to perform an impulse current test;

[0041] A capacitance detection device, used to obtain the capacitance of the target arrester valve plate after being subjected to an impact current;

[0042] A voltage detection device, used to obtain the varistor voltage of the target arrester valve after being subjected to the impact current;

[0043] A calculation module, used for calculating the product of the capacitance and the varistor voltage of the target arrester valve plate after receiving the impulse current, and outputting the product as the degradation state parameter of the arrester valve plate;

[0044] A threshold optimization model is used to obtain the degradation threshold based on the current environmental data;

[0045] The judgment module judges whether the degradation state parameter of the target arrester valve plate exceeds the degradation threshold; if not, the target arrester valve plate is in the early stage of degradation; if so, the arrester valve plate is in the middle and late stage of degradation.

[0046] Based on the further improvement of the above system, the system further includes:

[0047] The offline leakage current detection circuit performs an offline leakage current test on the arrester valve plate to obtain the leakage current of the arrester valve plate.

[0048] Based on the further improvement of the above system, the offline leakage current detection circuit includes: a high voltage power supply generator and a current detection device,

[0049] Among them, one end of the lightning arrester valve plate is grounded, and the other end is connected to the high-voltage power supply generator. The high-voltage power supply generator applies normal working voltage to the lightning arrester valve plate. The current detection device is connected to the ground end of the lightning arrester valve plate to detect the leakage current of the lightning arrester valve plate.

[0050] Based on further improvement of the above system, the offline leakage current detection circuit further includes: a transformer, which is connected between the high-voltage power supply generator and the arrester valve plate.

[0051] Based on further improvement of the above system, the system further comprises a temperature detection element, and the temperature detection element is arranged on the arrester valve plate to detect the maximum temperature of the arrester valve plate after being subjected to the impact current.

[0052] Based on the further improvement of the above system, the system further includes:

[0053] The state detection model is used to judge whether the arrester valve plate is in the middle or late stage of degradation according to the leakage current of the arrester valve plate in the middle and late stages of degradation and the maximum temperature output after the impact current; the middle and late stages of degradation include the middle and late stages of degradation.

[0054] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects:

[0055] 1. In the present invention, the product of the capacitance of the arrester valve plate and the varistor voltage is used as the degradation state parameter, and the degradation threshold is combined to judge the degradation state of the arrester valve plate, so as to achieve a preliminary division into the early stage of degradation and the middle and late stages of degradation. Compared with parameters such as varistor voltage or leakage current, the degradation state parameter of the present invention has the advantage of being predictive and there is no inflection point effect, thereby improving the detection accuracy and facilitating the safe and stable operation of the power system.

[0056] 2. In the present invention, environmental factors are taken into consideration, mainly the influence of ambient temperature and ambient humidity on the degradation state parameters of the arrester valve plate. In order to avoid misjudgment of the degradation state of the arrester valve plate due to environmental changes, impulse current tests are performed on non-degraded arrester valve plates in different environments to obtain the degradation state parameters of the non-degraded arrester valve plates in different environments, and the above data are used to train the constructed threshold optimization model, so that the trained threshold optimization model can output the degradation thresholds in different environments, thereby more accurately realizing the division of the arrester into the early and late degradation stages.

[0057] 3. In the present invention, the middle and late stages of degradation are subdivided into the middle and late stages of degradation. From the perspective of practical application, the valve disc of the lightning arrester in the middle stage of degradation still has good performance and can continue to be put into use, while the performance of the valve disc of the lightning arrester in the late stage of degradation is greatly reduced and can no longer be put into use. If all the valve discs of the lightning arrester in the middle and late stages of degradation are replaced and no longer used, it will lead to frequent replacement of the valve discs of the lightning arrester, low efficiency of use, and increased costs. Taking the above into consideration, in the present invention, after determining that the target lightning arrester valve disc is in the middle and late stages of degradation, a leakage current test is performed on the valve disc of the lightning arrester, and the leakage current and the highest temperature after being subjected to the impact current are used to further determine whether the valve disc of the lightning arrester is in the middle or late stages of degradation, thereby achieving accurate subdivision of the middle and late stages of degradation.

[0058] 4. In the present invention, considering that the arrester valve plates in the late stage of degradation are still put into use, the arrester valve plates in the middle stage of degradation are more prone to failure than those in the early stage of degradation, and more attention needs to be paid to their working status parameters and performance conditions, so a shorter online monitoring sampling time is set, thereby improving the safety of the power system.

[0059] In the present invention, the above-mentioned technical solutions can also be combined with each other to achieve more preferred combination solutions. Other features and advantages of the present invention will be described in the subsequent description, and some advantages can become obvious from the description, or can be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained through the contents particularly pointed out in the description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] The drawings are only for the purpose of illustrating particular embodiments and are not to be considered limiting of the present invention. Like reference symbols denote like components throughout the drawings.

[0061] Figure 1 It is a flow chart of a method for detecting the state of a zinc oxide lightning arrester valve plate according to an embodiment of the present invention;

[0062] Figure 2 It is a schematic diagram showing the variation of degradation state parameters of the valve plate of the lightning arrester with the number of impacts in an embodiment of the present invention;

[0063] Figure 3 Schematic diagram of the change of the varistor voltage of the arrester valve plate under different environments in the embodiment of the present invention.

[0064] Figure 4 Schematic diagram of the change of the capacitance of the arrester valve plate under different environments in the embodiment of the present invention. DETAILED DESCRIPTION

[0065] The present invention provides a method and system for detecting the state of a zinc oxide arrester valve plate, which are used to accurately detect the degradation state of an arrester valve plate in a smart power grid. In practical applications, each step of the arrester state detection method of the present invention is implemented by corresponding hardware devices or software in the system. Among them, the threshold optimization model and state detection model of the present invention are both artificial intelligence models, which can adopt traditional machine learning models, reinforcement learning (enhanced learning) models, and deep learning models. In practical applications, they are installed in computer devices such as industrial control computer systems, cloud platforms, or Internet of Things platforms of smart power distribution systems.

[0066] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0067] A specific embodiment of the present invention discloses a method for detecting the state of a zinc oxide arrester valve plate, such as Figure 1 The method comprises:

[0068] When performing impulse current test, the product of the capacitance and varistor voltage of the arrester valve after being subjected to impulse current is defined as the degradation state parameter;

[0069] Performing an impulse current test on a non-degraded arrester valve plate under different environments to obtain test environment data and degradation state parameters of the non-degraded arrester valve plate;

[0070] The test environment data is used as input, and the degradation state parameters of the valve plate of the non-degraded arrester are used as labels to train a pre-established threshold optimization model;

[0071] Performing an impulse current test on a target arrester valve plate to obtain current environmental data and degradation state parameters of the target arrester valve plate;

[0072] Inputting the current environmental data into a trained threshold optimization model to obtain a degradation threshold;

[0073] Determine whether the degradation state parameter of the target arrester valve plate exceeds the degradation threshold; if not, the target arrester valve plate is in the early stage of degradation; if so, the arrester valve plate is in the middle and late stage of degradation.

[0074] In the embodiment of the present invention, the capacitance of the arrester valve plate refers to its macroscopic performance in the circuit, that is, how much charge the valve plate can store under a given voltage. Specifically, an LCR tester is used to detect the capacitance of the arrester valve plate.

[0075] In a zinc oxide lightning arrester, the valve plate is a varistor made of zinc oxide material. The varistor voltage of the arrester valve plate refers to the voltage threshold at which the zinc oxide varistor valve plate begins to significantly reduce its resistance value, thereby allowing a larger current to pass. Specifically, a CJ1001 varistor DC parameter meter is used to detect the varistor voltage of the arrester valve plate.

[0076] Specifically, the environmental data includes ambient temperature and ambient humidity. Among them, ambient temperature and ambient humidity have a greater impact on the capacitance and varistor voltage of the arrester valve plate, such as Figure 3 and Figure 4 as shown in .

[0077] Compared with the prior art, in the embodiment of the present invention, the product of the capacitance of the arrester valve plate and the varistor voltage is used as the degradation state parameter, and the degradation threshold is combined to judge the degradation state of the arrester valve plate, thereby realizing a preliminary division into the early stage of degradation and the middle and late stages of degradation. Compared with parameters such as varistor voltage or leakage current, the degradation state parameter of the present invention has the advantage of being predictive and there is no inflection point effect, thereby improving the detection accuracy and facilitating the safe and stable operation of the power system.

[0078] At the same time, in the embodiment of the present invention, environmental factors are taken into consideration, mainly the influence of ambient temperature and ambient humidity on the degradation state parameters of the arrester valve plate. In order to avoid misjudgment of the degradation state of the arrester valve plate due to environmental changes, impulse current tests are performed on non-degraded arrester valve plates in different environments to obtain the degradation state parameters of the non-degraded arrester valve plates in different environments, and the above data are used to train the constructed threshold optimization model, so that the trained threshold optimization model can output the degradation thresholds in different environments, thereby more accurately realizing the division of the arrester into the early and late degradation stages.

[0079] It should be noted that after the arrester valve plate is put into use, it will gradually deteriorate with time and use conditions. That is, after the arrester valve plate is put into use, its degradation process begins. Therefore, the non-degraded arrester valve plate in the embodiment of the present invention is the arrester valve plate that has not been put into use / has not been subjected to current shock.

[0080] In the initial stage of degradation, the performance of the arrester valve is good. In the middle and late stages of degradation, the performance of the arrester valve will drop significantly. At this time, regular monitoring and maintenance are particularly needed to promptly detect and handle the faults of the arrester to ensure the stable operation of the power system. In other words, the arrester degradation state in the embodiment of the present invention is used to indicate the degree of degradation of the arrester valve, and is not used for the health state of the arrester valve. The health state of the arrester valve is used to indicate whether the arrester valve is faulty. Generally speaking, the arrester valve in the early and middle stages of degradation is in a normal state, while the arrester valve in the late stage of degradation is in a faulty state.

[0081] In addition, the method of the embodiment of the present invention is an off-line testing method, which requires the arrester and its valve plate to be disassembled from the high-voltage line for testing.

[0082] When implementing, Figure 2As shown in , after the arrester valve plate is subjected to the impulse current (for the first time), the product of its capacitance and varistor voltage (deterioration state parameter) drops rapidly; then, with the increase of the number of impulses, the degradation state parameter decreases. At this stage, the depletion layer width t of the arrester increases. At this time, the Schottky barrier distortion plays a leading role in the degradation of the arrester valve plate, proving that the arrester valve plate is still in the early stage of degradation and has good performance; then, the degradation state parameter begins to gradually increase and even exceeds the degradation state parameter after the first current impulse. At this stage, the depletion layer width t is gradually decreasing, indicating that the dominant position of Schottky barrier distortion in the degradation of the zinc oxide arrester valve plate is gradually declining, which means that the zinc oxide arrester valve plate has passed the early stage of degradation and entered the middle and late stages of degradation.

[0083] Specifically, the current value of the impulse current is greater than the breakdown current of the arrester valve. The selection of the impulse current should be combined with the characteristics of the arrester valve. If the current is too small, the effect on the deterioration of the arrester valve is too small, and even the zinc oxide arrester valve cannot be broken down, so good experimental results cannot be obtained.

[0084] Preferably, after the arrester valve plate is struck by the impulse current, the capacitance of the arrester valve plate is detected first, and then the varistor voltage of the arrester valve plate is detected. In this way, the internal test circuit of the DC parameter meter is prevented from causing the capacitance change effect on the tested arrester valve plate when testing the current and voltage, resulting in inaccurate experimental data.

[0085] Specifically, when performing an impulse current test on a lightning arrester valve plate, an impulse current of 8 / 20 μs is applied to an unused lightning arrester valve plate or a target lightning arrester valve plate, wherein the impulse current of 8 / 20 μs is used to simulate a lightning current.

[0086] In one embodiment, the threshold optimization model is a binary linear regression model, and the formula of the threshold optimization model is:

[0087] f(x)=k1x1+k2x2+b;

[0088] X = [x1, x2, 1];

[0089]

[0090] In the formula, x1 is the ambient humidity, x2 is the ambient humidity, is the degradation threshold of the model output, k1, k2 and b are all model parameters.

[0091] Among them, binary linear regression is a complex model in regression analysis, which can comprehensively consider the impact of two input variables such as ambient temperature and ambient humidity on the output variable-degradation threshold.

[0092] Specifically, the loss function of the threshold optimization model is the mean square error function, and the formula is:

[0093]

[0094] Where y is the degradation state parameter.

[0095] In the embodiment of the present invention, the performance of the model is evaluated by the mean square error. In other words, the optimization goal of the threshold optimization model is to minimize the mean square error, that is, By optimizing Make Minimum. The minimum model parameters are the model parameters when model training is completed.

[0096] Specifically, when the threshold optimization model is trained, the least squares method is used to estimate the model parameters.

[0097] When the least squares method is used to estimate the model parameters, the mean square error function is applied to the matrix Derivative, the formula is:

[0098]

[0099] This formula can be used to solve the matrix Then through Calculate the mean square error at this time.

[0100] In one embodiment, after obtaining the degradation state of the arrester valve plate, the method further includes:

[0101] When performing an impulse current test on a target arrester valve plate, monitoring and recording the highest temperature of the target arrester valve plate after being subjected to the impulse current;

[0102] Performing an offline leakage current test on the target arrester valve plate in the middle and late stages of degradation to obtain the leakage current of the target arrester valve plate; the middle and late stages of degradation include the middle stage of degradation and the late stage of degradation;

[0103] Whether the target arrester valve plate is in the middle or late stage of degradation is determined based on the leakage current of the arrester valve plate and the maximum temperature after being subjected to the impact current; if the arrester valve plate is in the middle stage of degradation, the arrester valve plate can continue to be put into use; if the arrester valve plate is in the late stage of degradation, the arrester valve plate cannot continue to be put into use.

[0104] In the embodiment of the present invention, the middle and late stages of degradation are subdivided into the middle stage of degradation and the late stage of degradation. From the perspective of practical application, the arrester valve plates in the middle stage of degradation still have good performance and can continue to be put into use, while the performance of the arrester valve plates in the late stage of degradation is greatly reduced and can no longer be put into use. If all the arrester valve plates in the middle and late stages of degradation are replaced and no longer used, the arrester valve plates will be frequently replaced, with low efficiency and increased costs.

[0105] Taking the above situation into consideration, in the present invention, after determining that the target arrester valve plate is in the middle or late stage of degradation, a leakage current test is performed on the arrester valve plate, and based on its leakage current and the highest temperature after being subjected to the impact current, it is further determined whether the arrester valve plate is in the middle or late stage of degradation, thereby achieving accurate subdivision of the middle and late stages of degradation.

[0106] Specifically, judging whether the state of the valve plate of the target arrester is in the middle stage of degradation or in the late stage of degradation according to the leakage current of the valve plate of the target arrester and the highest temperature after being subjected to the impact current includes:

[0107] Build a state detection model based on machine learning;

[0108] Obtaining leakage currents and maximum temperatures after being subjected to impulse currents of arrester valve plate samples of arrester valve plate samples in the middle stage of degradation and arrester valve plate samples in the late stage of degradation respectively;

[0109] The leakage current of the arrester valve plate sample and the maximum temperature after being subjected to the impact current are used as inputs, and the degradation state is used as a label to train the state detection model;

[0110] The leakage current of the target arrester valve plate in the middle and late stages of degradation and the maximum temperature after being subjected to the impact current are input into the trained state detection model to obtain whether the arrester valve plate is in the middle or late stages of degradation.

[0111] In an embodiment of the present invention, a state detection model is constructed based on machine learning, and the leakage current and the maximum temperature after being subjected to the impact current are used as input data of the model, so as to more accurately judge whether the lightning arrester is in the middle or late stage of degradation, so as to determine whether the lightning arrester valve plate can be put into use.

[0112] Specifically, the state detection model is a support vector classifier.

[0113] Specifically, a particle swarm algorithm is used to adjust the hyperparameters of the support vector classifier to obtain the optimal hyperparameters of the support vector classifier.

[0114] In the embodiment of the present invention, a support vector classifier is used, and the hyperparameters of the training model are optimized in combination with a particle swarm algorithm. Support vector classification (SVC) is a form of support vector machine (SVM) for classification problems. The goal in a binary classification problem is to determine an optimal decision boundary that can separate the data points of the two categories as much as possible according to the categories and make the model have the strongest generalization ability.

[0115] In support vector classification, it is assumed that the data set is linearly separable, which means that points of different categories can be distinguished by a linear boundary or hyperplane. The data set consists of point pairs {(x1,y1),(x2,y2),…,(xN,yN)}, where xi is the feature vector and yi is its corresponding binary category label.

[0116] The hyperplane can be defined by a weight vector w and a bias term b, and its formula is:

[0117] w T x+b=0;

[0118] In the formula, w T x represents the dot product of vector w and vector x.

[0119] The goal of the support vector classifier is to maximize the minimum distance, or margin, from the data point to the decision boundary.

[0120] The decision function finally generated by the support vector classifier is: f(x) = w T x+b.

[0121] Hyperparameter adjustment is an important step in machine learning. This paper uses the particle swarm optimization (PSO) algorithm to obtain the optimal hyperparameters of SVC. The core idea of ​​the PSO algorithm is to simulate the assistance of individuals in a group to explore the location of the optimal solution. In PSO, each solution represents a particle in the search space, and each particle contains a position and a velocity, where the position refers to a solution, and the velocity measures the direction and speed of the particle's movement in the search space.

[0122] The execution process of the PSO algorithm is continuous. First, the positions and velocities of a group of particles are randomly initialized, and then the fitness of each particle is calculated, that is, the quality of the solution represented by the particle is evaluated. Each particle will remember the best position it has found, that is, the individual extreme value, and the entire particle group will also track the best position found so far, that is, the global extreme value. Then, the PSO algorithm will enter the iterative update process, and the update of particle speed and position will be carried out according to certain rules.

[0123] For the update of particle velocity, it will be affected by particle inertia, individual extreme value and global extreme value. The update formula is as follows:

[0124]

[0125] Among them, w represents the inertia weight, c1 and c2 represent the acceleration constants, and r 1，i 、r 2，i represents a random number, p i，d represents the position of the individual extreme value of the particle, p g，d Indicates the location of the global extremum.

[0126] The update of the particle position comes from the sum of the particle's current position and the new velocity, expressed as:

[0127]

[0128] After each iteration, the fitness of the particle will be re-evaluated, and the individual and global extreme values ​​may be updated. The whole process will be repeated until the termination condition is met, such as reaching the preset number of iterations or the solution found reaches the expected threshold.

[0129] More specifically, the state detection model is trained using a state detection training data set, comprising the following steps:

[0130] Step 100, initialize the parameters of the state detection model; first, set the PSO algorithm parameters: number of iterations = 500, number of particles = 100; then set the Gaussian kernel SVC model parameter range: C∈[0.001,1000], γ∈[0.001,1000].

[0131] Step 200, normalizing the data in the state detection training data set;

[0132] Step 300, dividing the normalized data of the state detection training data set into a state detection training set and a state detection test set according to a preset ratio;

[0133] Step 400: Use the state detection training set to train the state detection model, and use the state detection test set to verify the performance of the trained state detection model.

[0134] Among them, dividing the training set and the test set is an important step to ensure that the machine learning model has good generalization ability, prevent overfitting, and perform accurate performance evaluation. During implementation, in an embodiment of the present invention, the state detection training data set is divided into a training set and a test set in a ratio of 9:1.

[0135] Meanwhile, in the embodiment of the present invention, considering that the units and dimensions of the characteristic data of the arrester valve plate are different, the above data are normalized before being input into the state detection model.

[0136] Specifically, the normalized data value is in the range of [0,1]. The normalization formula is:

[0137]

[0138] Among them, min(x) is the minimum value in the original data, max(x) is the maximum value in the original data, x is any value in the original data, x norm is the normalized data.

[0139] Similarly, the leakage current of the target arrester valve plate in the middle and late stages of degradation and the maximum temperature after being subjected to the impact current are normalized respectively and input into the trained state detection model to obtain whether the arrester valve plate is in the middle or late stages of degradation.

[0140] Specifically, the method further includes:

[0141] If the arrester valve plate is in the early stage of degradation, then after the arrester valve plate is put into use, the working state parameter of the arrester valve plate is obtained once every first online monitoring sampling time;

[0142] If the arrester valve plate is in the middle stage of degradation, then after the arrester valve plate is put into use, the working state parameter of the arrester valve plate is obtained once every second online monitoring sampling time;

[0143] Wherein, the second online monitoring sampling time is less than the first online monitoring sampling time;

[0144] The working state parameters of the arrester valve plate are used to monitor the health state of the arrester valve plate online, and the working state parameters include the leakage current and / or temperature of the arrester valve plate.

[0145] In the embodiment of the present invention, considering that the arrester valve plates in the late stage of degradation are still put into use, the arrester valve plates in the middle stage of degradation are more prone to failure than those in the early stage of degradation, and more attention needs to be paid to their working status parameters and performance, a shorter online monitoring sampling time is set, thereby improving the safety of the power system.

[0146] During implementation, if the arrester valve is in a healthy state after detection, it can continue to be put into use. When put into use, the arrester valve is installed on the high-voltage line, and the leakage current of the arrester valve is monitored by the online leakage current monitoring circuit, and / or the temperature of the arrester valve in the working state is detected by the temperature detection parameter.

[0147] Specifically, performing an offline leakage current test on the target arrester valve plate in the middle and late stages of degradation to obtain the leakage current of the target arrester valve plate includes the following steps:

[0148] Grounding one end of the arrester valve plate and applying a normal operating voltage to the other end of the arrester;

[0149] The current detection device is connected to the ground terminal of the arrester valve plate to detect the leakage current of the arrester valve plate.

[0150] On the other hand, an embodiment of the present invention further provides a system for detecting a deterioration state of a zinc oxide arrester valve plate, the system comprising:

[0151] An impulse current generator is used to apply an impulse current to the valve plate of the target lightning arrester to perform an impulse current test;

[0152] A capacitance detection device, used to obtain the capacitance of the target arrester valve plate after being subjected to an impact current;

[0153] A voltage detection device, used to obtain the varistor voltage of the target arrester valve plate after being subjected to the impact current;

[0154] A calculation module, used for calculating the product of the capacitance and the varistor voltage of the target arrester valve plate after the surge current is applied, and outputting the product as a degradation state parameter of the arrester valve plate;

[0155] A threshold optimization model is used to obtain the degradation threshold based on the current environmental data;

[0156] The judgment module judges whether the degradation state parameter of the target arrester valve plate exceeds the degradation threshold; if not, the target arrester valve plate is in the early stage of degradation; if so, the arrester valve plate is in the middle and late stage of degradation.

[0157] During implementation, the capacitance detection device is an LCR tester; the voltage detection device is a CJ1001 varistor DC parameter meter.

[0158] Specifically, the impulse current generator and the arrester valve plate are connected in series to form a loop through a wire to apply an impulse current to the arrester valve plate to perform an impulse current test.

[0159] It should be noted that the threshold optimization model in the system of the embodiment of the present invention is a trained threshold optimization model. Specifically, the threshold optimization model is trained using the following method:

[0160] Performing an impulse current test on a non-degraded arrester valve plate under different environments to obtain test environment data and degradation state parameters of the non-degraded arrester valve plate;

[0161] The test environment data is used as input, and the degradation state parameters of the non-degraded arrester valve plate are used as labels to train the pre-established threshold optimization model, thereby obtaining a trained threshold optimization model. Specifically, the system also includes: an offline leakage current detection circuit, which performs an offline leakage current test on the arrester valve plate to obtain the leakage current of the arrester valve plate.

[0162] Specifically, the offline leakage current detection circuit includes: a high-voltage power supply generator and a current detection device. One end of the arrester valve plate is grounded, and the other end is connected to the high-voltage power supply generator. The high-voltage power supply generator applies a normal working voltage to the arrester valve plate. The current detection device is connected to the ground terminal of the arrester valve plate to detect the leakage current of the arrester valve plate.

[0163] Among them, through the high-voltage power generator, it can provide a stable high-voltage output, thereby simulating the voltage of the high-voltage line to which the lightning arrester is connected.

[0164] Preferably, the offline leakage current detection circuit further includes: a transformer, the transformer being connected between the high-voltage power supply generator and the arrester valve plate, wherein the transformer is provided to increase the voltage output by the high-voltage power supply generator, thereby meeting the detection requirements.

[0165] Specifically, the system further includes a temperature detection element, which is disposed on the arrester valve plate to detect the maximum temperature of the arrester valve plate after being subjected to an impact current.

[0166] Specifically, the system also includes: a state detection model, which is used to judge whether the lightning arrester valve plate is in the middle or late stage of degradation based on the leakage current of the lightning arrester valve plate in the middle and late stages of degradation and the maximum temperature output after the impact current; the middle and late stages of degradation include the middle and late stages of degradation.

[0167] More specifically, the state detection model is a support vector classifier.

[0168] It should be noted that the threshold optimization model in the system of the embodiment of the present invention is a trained state detection model. Specifically, the state detection model is trained using the following method:

[0169] Obtaining leakage currents and maximum temperatures of arrester valve plate samples of arrester valve plate samples in the middle stage of degradation and arrester valve plate samples in the late stage of degradation respectively, so as to construct a state detection training data set, and using the state detection training data set to train the state detection model;

[0170] The leakage current of the target arrester valve plate in the middle and late stages of degradation and the maximum temperature after being subjected to the impact current are input into the trained state detection model to obtain whether the arrester valve plate is in the middle or late stages of degradation.

[0171] The degradation state detection system of the zinc oxide arrester valve plate of the embodiment of the present invention is used to implement the degradation state detection method of the zinc oxide arrester valve plate of the aforementioned embodiment, and the similarities are not repeated here.

[0172] Those skilled in the art will appreciate that all or part of the processes of the above-mentioned embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, wherein the computer-readable storage medium is a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0173] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for detecting the state of a zinc oxide arrester valve plate, characterized in that: The method comprises: When performing impulse current test, the product of the capacitance and varistor voltage of the arrester valve after being subjected to impulse current is defined as the degradation state parameter; Performing an impulse current test on a non-degraded arrester valve plate under different environments to obtain test environment data and degradation state parameters of the non-degraded arrester valve plate; The test environment data is used as input, and the degradation state parameters of the valve plate of the non-degraded arrester are used as labels to train a pre-established threshold optimization model; Performing an impulse current test on a target arrester valve plate to obtain current environmental data and degradation state parameters of the target arrester valve plate; Inputting the current environmental data into a trained threshold optimization model to obtain a degradation threshold; Determine whether the degradation state parameter of the target arrester valve plate exceeds the degradation threshold; if not, the target arrester valve plate is in the early stage of degradation; if so, the arrester valve plate is in the middle and late stage of degradation.

2. The method according to claim 1, characterized in that The environmental data includes environmental temperature and environmental humidity.

3. The method according to claim 2, characterized in that The threshold optimization model is a binary linear regression model, and the formula of the threshold optimization model is: f(x)=k1x1+k2x2+b; X = [x1, x2, 1]; In the formula, x1 is the ambient humidity, x2 is the ambient humidity, is the degradation threshold of the model output, k1, k2 and b are all model parameters.

4. The method according to claim 3, characterized in that The loss function of the threshold optimization model is the mean square error function, and the formula is: Where y is the degradation state parameter.

5. The method according to any one of claims 1 to 4, characterized in that When the surge current test is performed on the arrester valve plate, the current value of the adopted surge current is greater than the breakdown current of the arrester valve plate.

6. The method according to any one of claims 1 to 4, characterized in that After the arrester valve plate is struck by an impulse current, the capacitance of the arrester valve plate is detected first, and then the varistor voltage of the arrester valve plate is detected.

7. The method according to any one of claims 1 to 4, characterized in that When performing an impulse current test on a lightning arrester valve plate, an impulse current of 8 / 20μs is applied to the unused lightning arrester valve plate or the target lightning arrester valve plate.

8. The method according to any one of claims 1 to 4, characterized in that The method further comprises: When performing an impulse current test on a target arrester valve plate, monitoring and recording the highest temperature of the target arrester valve plate after being subjected to the impulse current; Performing an offline leakage current test on the valve plate of the target arrester in the middle and late stages of degradation to obtain the leakage current of the valve plate of the target arrester; the middle and late stages of degradation include the middle stage of degradation and the late stage of degradation; Whether the target arrester valve plate is in the middle or late stage of degradation is determined based on the leakage current of the arrester valve plate and the maximum temperature after being subjected to the impact current; if the arrester valve plate is in the middle stage of degradation, the arrester valve plate can continue to be put into use; if the arrester valve plate is in the late stage of degradation, the arrester valve plate cannot continue to be put into use.

9. The method according to claim 8, characterized in that The step of judging whether the state of the valve plate of the target arrester is in the middle stage of degradation or in the late stage of degradation according to the leakage current of the valve plate of the target arrester and the highest temperature after being subjected to the impact current comprises: Build a state detection model based on machine learning; Obtaining leakage currents and maximum temperatures after being subjected to impulse currents of arrester valve plate samples of arrester valve plate samples in the middle stage of degradation and arrester valve plate samples in the late stage of degradation respectively; The leakage current of the arrester valve plate sample and the maximum temperature after being subjected to the impact current are used as inputs, and the degradation state is used as a label to train the state detection model; The leakage current of the target arrester valve plate in the middle and late stages of degradation and the maximum temperature after being subjected to the impact current are input into the trained state detection model to obtain whether the arrester valve plate is in the middle or late stages of degradation.

10. The method according to claim 9, characterized in that The state detection model is a support vector classifier.

11. The method according to claim 8, characterized in that The method further comprises: If the arrester valve plate is in the early stage of degradation, then after the arrester valve plate is put into use, the working state parameter of the arrester valve plate is obtained once every first online monitoring sampling time; If the arrester valve plate is in the middle stage of degradation, then after the arrester valve plate is put into use, the working state parameter of the arrester valve plate is obtained once every second online monitoring sampling time; Wherein, the second online monitoring sampling time is less than the first online monitoring sampling time; The working state parameters of the arrester valve plate are used to monitor the health state of the arrester valve plate online, and the working state parameters include the leakage current and / or temperature of the arrester valve plate.

12. A zinc oxide arrester valve plate degradation state detection system, characterized in that: The system comprises: An impulse current generator is used to apply an impulse current to the valve plate of the target lightning arrester to perform an impulse current test; A capacitance detection device, used to obtain the capacitance of the target arrester valve plate after being subjected to an impact current; A voltage detection device, used to obtain the varistor voltage of the target arrester valve plate after being subjected to the impact current; A calculation module, used for calculating the product of the capacitance and the varistor voltage of the target arrester valve plate after the surge current is applied, and outputting the product as a degradation state parameter of the arrester valve plate; A threshold optimization model is used to obtain the degradation threshold based on the current environmental data; The judgment module judges whether the degradation state parameter of the target arrester valve plate exceeds the degradation threshold; if not, the target arrester valve plate is in the early stage of degradation; if so, the arrester valve plate is in the middle and late stage of degradation.

13. The system according to claim 12, characterized in that The system further comprises: The offline leakage current detection circuit performs an offline leakage current test on the arrester valve plate to obtain the leakage current of the arrester valve plate.

14. The system according to claim 13, characterized in that The offline leakage current detection circuit comprises: a high voltage power supply generator and a current detection device, Among them, one end of the lightning arrester valve plate is grounded, and the other end is connected to the high-voltage power supply generator. The high-voltage power supply generator applies normal working voltage to the lightning arrester valve plate. The current detection device is connected to the ground end of the lightning arrester valve plate to detect the leakage current of the lightning arrester valve plate.

15. The system according to claim 14, characterized in that The offline leakage current detection circuit further includes: a transformer connected between the high voltage power supply generator and the arrester valve plate.

16. The system according to claim 13, characterized in that The system further comprises a temperature detection element, which is arranged on the arrester valve plate and is used to detect the maximum temperature of the arrester valve plate after being subjected to an impact current.

17. The system according to claim 16, characterized in that The system further comprises: The state detection model is used to judge whether the arrester valve plate is in the middle or late stage of degradation according to the leakage current of the arrester valve plate in the middle and late stages of degradation and the maximum temperature output after the impact current; the middle and late stages of degradation include the middle and late stages of degradation.

Images