Application for improving insulation performance of electrical equipment based on electrical conduction optimization protective agent

Abstract

The invention discloses an application for improving the insulation performance of electrical equipment based on an electrical conduction optimization protective agent, relates to the technical field of electrical insulation, and aims to dynamically acquire temperature, electric field distribution and humidity parameters, quickly respond to the equipment operation state and environment change, guarantee data accuracy in a high-load or extreme environment and improve the insulation performance of the electrical equipment in the aspects of state evaluation. Constructing a data-driven state model by combining a support vector machine (SVM) and a long short-term memory (LSTM) network algorithm, and performing equipment state trend prediction; in cooperation with an electric field distribution diagram generated based on finite element analysis, a high-risk area is accurately marked, and the comprehensiveness and accuracy of state evaluation are remarkably improved; in terms of protection measures, multiple types of electrical conduction optimization protective agents are applied to a high-risk area, heat-resistant, hydrophobic and high-dielectric-strength protective agents are selected for different problems, the coating thickness is dynamically adjusted in combination with the change of the electric field strength, and insulation degradation caused by local hot spots, electric field concentration and humidity is effectively relieved.

Description

Technical Field

[0001] The present invention relates to the technical field of electrical insulation, and particularly to an application for improving the insulation performance of electrical equipment based on an electro-conduction optimized protective agent. Background Art

[0002] The insulation performance of high-voltage electrical equipment is crucial for the safe and stable operation of the power system. In the actual operating environment, the insulation performance is vulnerable to various factors such as temperature, humidity, uneven electric field distribution, and aging. During long-term operation, the dynamic changes in insulation performance are difficult to monitor in a timely manner, and fault maintenance has a lag.

[0003] Some traditional solutions use high-performance insulation materials with low losses to reduce energy consumption, reduce the risk of electric field concentration by increasing the thickness of the insulation layer, and capture temperature anomalies and partial discharge phenomena through infrared thermal imagers and electric field sensors, which improve the monitoring and maintenance capabilities of the equipment insulation performance to a certain extent. However, such solutions are mostly passive responses, only providing surface data of the equipment status, lacking the dynamic feedback and prediction capabilities for the deep changes in insulation performance, and unable to effectively improve the insulation performance of electrical equipment. Therefore, there is an urgent need for a method that uses an electro-conduction optimized protective agent to adaptively insulate different regions of electrical equipment to improve the insulation performance of electrical equipment. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] The present invention provides an application for improving the insulation performance of electrical equipment based on an electro-conduction optimized protective agent to solve the problem that traditional insulation solutions for electrical equipment are mostly passive responses, lacking the dynamic feedback and prediction capabilities for the deep changes in insulation performance, and unable to effectively improve the insulation performance of electrical equipment.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] An embodiment of the present invention provides an application for improving the insulation performance of electrical equipment based on an electro-conduction optimized protective agent. The application steps include:

[0008] Step S1: Install sensors at the connection points and high electric field strength parts of the electrical equipment to monitor insulation parameters, including temperature, electric field distribution, humidity, and partial discharge conditions; perform preliminary data processing at the sensor end and mark normal events;

[0009] Step S2: Combine the real-time monitoring data with historical insulation parameters, and use the support vector machine (SVM) and long short-term memory network (LSTM) algorithms to generate a state model, including: temperature distribution prediction, humidity change trend, and the intensity and position of discharge signals;

[0010] Step S3: The maintenance personnel use a portable coating device to apply the electro-conduction optimization protective agent.

[0011] The coating method in Step S3 is as follows:

[0012] Select the type of protective agent suitable for the position of the electrical equipment according to the electric field distribution map, including:

[0013] Heat-resistant protective agent, used for local hot spot areas,

[0014] Hydrophobic protective agent, used for high-humidity environments,

[0015] High dielectric strength protective agent, used for electric field concentration areas,

[0016] Step S4: Based on the long-term insulation parameter data, analyze the performance changes of different protective agents under different environmental conditions, and use the Bayesian optimization algorithm to calculate the protective agent adjustment suggestions.

[0017] As a preferred solution of the application of a method for improving the insulation performance of electrical equipment based on an electro-conduction optimization protective agent according to the present invention, wherein: the sampling frequency at the sensor end is dynamically adjusted, and the sampling frequency is increased during high-load operation or drastic environmental changes;

[0018] The high electric field strength parts include cable joints and terminals, the edge areas of high-voltage transformer windings, and the electric field concentration points in insulator strings;

[0019] The abnormal events include sudden temperature increase, abnormal electric field, or excessive humidity.

[0020] As a preferred solution of the application of a method for improving the insulation performance of electrical equipment based on an electro-conduction optimization protective agent according to the present invention, wherein: the steps of performing preliminary data processing at the sensor end and marking normal events are as follows:

[0021] The signal collected by the sensor is smoothed using the moving average method, and the smoothing formula is:

[0022]

[0023] where, S f (t) is the signal value after smoothing at the t-th moment, N is the size of the moving window, and S(i) is the original signal value at the i-th moment.

[0024] The abnormal detection formula is used to determine whether an abnormal event occurs, and the abnormal detection formula is:

[0025] |X t -μ|>k·σ,

[0026] where, X t$x_t$ is the sensor acquisition value at the $t$-th moment, $\mu$ is the mean of historical data, $\sigma$ is the standard deviation of historical data, and $k$ is the sensitivity factor;

[0027] In the step of dynamically adjusting the sampling frequency at the sensor end, the calculation formula of the sampling frequency is:

[0028] $f$ s $=$ base $f_0 + \Delta f \cdot I(\text{event})$,

[0029] where $f$ s is the current sampling frequency, $f_0$ base is the basic sampling frequency, $\Delta f$ is the sampling frequency increment, and $I(\text{event})$ is the event indicator function, which takes the value of 1 when an abnormal event occurs and 0 otherwise.

[0030] As a preferred solution of the application of a method for improving the insulation performance of electrical equipment based on an electroconductive optimized protective agent according to the present invention, wherein: the support vector machine algorithm is used to classify and identify potential degradation risks of different parts of the equipment;

[0031] The long short-term memory network algorithm is used to predict the time series change of insulation parameters for trend analysis;

[0032] In step S2, the electric field distribution of the equipment is also calculated based on the finite element analysis (FEA), the electric field concentration areas are marked, and an electric field distribution map is generated in combination with the results of the state model. The high-risk areas are marked in the distribution map, and the reasons for the decline in insulation performance are evaluated, including moisture, aging, and partial discharge;

[0033] The high-risk areas in the electric field distribution map are important references for the risk analysis of the state model. The state model provides an overall insulation performance trend analysis to guide the priority order of applying the protective agent; the electric field distribution map locates the electric field concentration areas to guide the application position of the protective agent.

[0034] As a preferred solution of the application of a method for improving the insulation performance of electrical equipment based on an electroconductive optimized protective agent according to the present invention, wherein: the step of combining real-time monitoring data with historical insulation parameters and generating a state model using the support vector machine (SVM) and long short-term memory network (LSTM) algorithms is as follows:

[0035] Generate a state model, and the state model formula is:

[0036] $X$ input $=$ r $\text{Concat}(X$ h $_{rt}, X$

[0037] where $X$ input is the input data matrix, including real-time monitoring data $X$ r $_{rt}$ and historical insulation parameters $X$ hThe splicing result, X r Is real-time monitoring data, including the temperature, electric field strength, and humidity values at the current moment, X h Is historical insulation parameters, including the temperature, electric field, and humidity trends accumulated over a long period

[0038] Support Vector Machine (SVM) is used for classification, and the classification formula is:

[0039]

[0040] Among them, f(x) is the classification result function, the output is +1 or -1, x is the data point to be classified, x i Is the support vector, y i Is the label of the training data, +1 indicates normal, -1 indicates potential deterioration, α i Is the weight of the support vector, K(x i , x) is the kernel function, the Gaussian kernel and the radial basis function are selected, and b is the bias term

[0041] Long Short-Term Memory (LSTM) network is used for time series prediction, and the prediction formulas include:

[0042] Forget gate calculation formula: f t = σ(W f x t + U f h t-1 + b f ),

[0043] Input gate calculation formula: i t = σ(W i x t + U i h t-1 + b i ),

[0044] Memory cell update formula: c t = f t ⊙ c t-1 + i t ⊙ tanh(W c x t + U c h t-1 + b c ),

[0045] Output gate calculation formula: o t = σ(W o x t + U o h t-1 + b o ),

[0046] The output at the current moment is yt , y t = o t ⊙ tanh(c t ),

[0047] where x t is the current input data, h t-1 is the hidden state at the previous moment, c t-1 , c t are the states of the memory cell at the previous and current moments, f t , i t , o t are the activation values of the forget gate, input gate, and output gate, W f , W i , W c , W o are the weight matrices in the corresponding calculation formulas, U f , U i , U c , U o are the hidden state weight matrices in the corresponding calculation formulas, b f , b i , b c , b o are the bias terms in the corresponding calculation formulas, σ(·) is the sigmoid function, tanh is the hyperbolic tangent function, and ⊙ is element-wise multiplication.

[0048] As a preferred solution of an application for improving the insulation performance of electrical equipment based on an electroconductive optimization protectant described in the present invention, wherein: the steps of calculating the electric field distribution of the device based on finite element analysis (FEA), marking the electric field concentration region, and generating an electric field distribution map by combining the results of the state model, and marking the high-risk region in the distribution map are as follows

[0049] Perform electric field distribution analysis and risk region marking. The electric field distribution calculation formula is

[0050]

[0051] where E(x, y, z) is the electric field strength in three-dimensional space, and φ(x, y, z) is the potential distribution at the spatial position (x, y, z).

[0052] Mark the risk region. The marking condition is

[0053] E(x, y, z) > E crit ,

[0054] where E crit is the critical value of the electric field strength, and the region exceeding this value is marked as high risk.

[0055] As a preferred solution of the application of improving the insulation performance of electrical equipment based on an electroconductivity-optimized protective agent according to the present invention, wherein: in step S3, after the coating is completed, based on the insulation parameters continuously collected by the sensor, the state model and the electric field distribution map before and after the protective agent is coated are compared to evaluate the actual effect of the protective agent; if it is found that the local electric field is still abnormally concentrated, the maintenance personnel shall supplement the coating of the protective agent or adjust the coating thickness.

[0056] As a preferred solution of the application of improving the insulation performance of electrical equipment based on an electroconductivity-optimized protective agent according to the present invention, wherein: the step of comparing the state model and the electric field distribution map before and after the protective agent is coated based on the insulation parameters continuously collected by the sensor to evaluate the actual effect of the protective agent is as follows:

[0057] Apply the protective agent to the high-risk area according to the electric field distribution map and adjust the coating thickness. The adjustment formula is:

[0058] T coat = T base + k·ΔE,

[0059] wherein, T coat is the coating thickness of the protective agent, T base is the basic coating thickness, k is the thickness adjustment factor, and ΔE is the amplitude of the electric field strength exceeding the critical value.

[0060] Conduct effect evaluation. The effect evaluation formula is:

[0061]

[0062] wherein, ΔP is the insulation performance change rate, P before is the insulation performance parameter value before coating, and P after is the insulation performance parameter value after coating.

[0063] As a preferred solution of the application of improving the insulation performance of electrical equipment based on an electroconductivity-optimized protective agent according to the present invention, wherein: in the Bayesian optimization algorithm,

[0064] the input variables are the type of protective agent, coating thickness, environmental temperature and humidity, and the trend of electric field distribution;

[0065] the optimization objectives are to improve insulation performance, extend the life of the protective agent, and reduce the use cost.

[0066] As a preferred solution of the application of improving the insulation performance of electrical equipment based on an electroconductivity-optimized protective agent according to the present invention, wherein: the step of analyzing the performance changes of different protective agents under different environmental conditions based on long-term insulation parameter data and calculating the adjustment suggestions for the protective agent using the Bayesian optimization algorithm is as follows:

[0067] Approximate the objective function through a surrogate model to find the optimal parameter combination. Define the objective function as the multi-objective optimization problem f(X) of the performance of the protective agent.

[0068] f(X) = w 1 S(X) + w 2 L(X) - w 3 C(X),

[0069] where f(X) is the optimization objective function, representing the comprehensive performance evaluation of the protective agent, X is the variable to be optimized, including the type of protective agent, coating thickness, and environmental temperature and humidity, S(X) is the insulation performance of the protective agent, L(X) is the service life of the protective agent, C(X) is the cost of the protective agent, and w 1 , w 2 , w 3 are the weight coefficients of the objective function, satisfying w 1 + w 2 + w 3 = 1.

[0070] Use the Gaussian process as the surrogate model of the objective function. The model formula is:

[0071]

[0072] where μ(X) is the mean function of the objective function, representing the predicted value of the surrogate model at point X, and k(X, X ′ ) is the kernel function,

[0073]

[0074] where is the variance of the function value, l is the length scale, and |X - X ′ | is the Euclidean distance between two input points;

[0075] The sampling strategy is defined as expected improvement. The improvement formula is:

[0076] EI(X) = E[max(f(X) - f * , 0)],

[0077] where EI(X) is the expected improvement value at X, f * is the current optimal objective function value, and E[·] is the mathematical expectation.

[0078] EI(X) is explicitly calculated as:

[0079] EI(X) = (f mean (X) - f * )Φ(Z) + σ(X)φ(Z),

[0080]

[0081] Among them, f mean (X) is the predicted mean value of the surrogate model at point X, σ(X) is the predicted standard deviation of the surrogate model at point X, Φ(Z) is the cumulative distribution function of the standard normal distribution, and φ(Z) is the probability density function of the standard normal distribution;

[0082] In step S4, the protective agent optimization process includes

[0083] Select the initial point set {X 1 , X 2 , …, X n} for actual simulation to construct the initial data set

[0084] Use Gaussian process to fit the initial data set Generate the initial surrogate model of the objective function

[0085] Optimize the expected improvement function EI(X), and select the sampling point for the next simulation as X next ,

[0086]

[0087] At point X next , actually evaluate the objective function value f(X next ), and update the data set

[0088]

[0089] Repeat the surrogate model update and sampling point selection until the stopping condition is met

[0090] The optimal protective agent parameters are:

[0091]

[0092] Include the type of protective agent, the best coating thickness, and the optimization strategy for adapting to different environments.

[0093] The beneficial effects of the present invention are as follows: Firstly, the present invention dynamically collects temperature, electric field distribution, and humidity parameters. The dynamic sampling frequency can quickly respond to the operating state of the equipment and environmental changes, ensuring data accuracy in high-load or extreme environments. Secondly, in terms of state assessment, a data-driven state model is constructed by combining the support vector machine (SVM) and long short-term memory network (LSTM) algorithms to predict the equipment state trend. In cooperation with the electric field distribution map generated based on finite element analysis, high-risk areas are accurately marked, significantly improving the comprehensiveness and accuracy of state assessment. In addition, in terms of protection measures, various types of electroconductive optimization protective agents are applied to high-risk areas. For different problems, heat-resistant, hydrophobic, and high dielectric strength protective agents are selected respectively, and the coating thickness is dynamically adjusted according to the change of electric field strength, effectively alleviating insulation degradation caused by local hot spots, electric field concentration, and humidity, and the protection is more precise and efficient. Finally, the Bayesian optimization algorithm is used to optimize the types, thickness, and coating strategies of the protective agents, improving the insulation performance, significantly extending the service life of the protective agents, and reducing the maintenance frequency and cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0094] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0095] Figure 1 It is a flowchart of the application steps for improving the insulation performance of electrical equipment based on electroconductive optimization protective agents of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0096] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the following will give a detailed description of the specific embodiments of the present invention with reference to the drawings of the specification.

[0097] In the following description, many specific details are set forth in order to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0098] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that can be included in at least one implementation manner of the present invention. The "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that excludes other embodiments.

[0099] Example 1, referring to Figure 1, this embodiment provides an application for improving the insulation performance of electrical equipment based on an electroconductive optimized protective agent, including the following steps:

[0100] Step S1, install sensors at the connection points and high electric field strength parts of the electrical equipment to monitor insulation parameters, including temperature, electric field distribution, humidity, and partial discharge conditions; perform preliminary data processing at the sensor end and mark normal events;

[0101] The sensor end dynamically adjusts the sampling frequency and increases the sampling frequency during high-load operation or drastic environmental changes;

[0102] The high electric field strength parts include cable joints and terminals, the edge areas of high-voltage transformer windings, and the electric field concentration points in insulator strings;

[0103] Abnormal events include sudden temperature increase, abnormal electric field, or excessive humidity;

[0104] The step of performing preliminary data processing at the sensor end and marking normal events is as follows:

[0105] Use the moving average method to smooth the signals collected by the sensors. The smoothing formula is:

[0106]

[0107] where S f (t) is the signal value after smoothing at the t-th moment, N is the size of the moving window, and S(i) is the original signal value at the i-th moment.

[0108] Use the anomaly detection formula to determine whether an abnormal event has occurred. The anomaly detection formula is:

[0109] |X t -μ|>k·σ,

[0110] where X t is the sensor acquisition value at the t-th moment, μ is the mean of historical data, σ is the standard deviation of historical data, and k is the sensitivity factor;

[0111] In the step of the sensor end dynamically adjusting the sampling frequency, the calculation formula for the sampling frequency is:

[0112] f s =f base +Δf·I(event),

[0113] where f s is the current sampling frequency, f base is the basic sampling frequency, Δf is the sampling frequency increment, and I(event) is the event indicator function, which takes the value of 1 when an abnormal event occurs and 0 otherwise;

[0114] Specifically, a variety of sensors are deployed at the key nodes of electrical equipment to collect temperature, electric field distribution, and humidity parameters in real time. The moving average method and anomaly detection mechanism are carried out to significantly reduce the noise interference of monitoring signals. At the same time, abnormal events such as temperature surges, electric field distortions, and humidity exceeding the standard can be accurately identified. The equipment can still ensure the transparency of the operating state under complex environments and high-load operating conditions.

[0115] Step S2: Combine the real-time monitoring data with historical insulation parameters, and use the support vector machine (SVM) and long short-term memory network (LSTM) algorithms to generate a state model, including: temperature distribution prediction, humidity change trend, and the intensity and location of discharge signals.

[0116] The support vector machine algorithm is used to classify and identify potential degradation risks in different parts of the equipment.

[0117] The long short-term memory network algorithm is used to predict the time series changes of insulation parameters for trend analysis.

[0118] In step S2, the electric field distribution of the equipment is also calculated based on finite element analysis (FEA), the electric field concentration areas are marked, and an electric field distribution map is generated in combination with the results of the state model. The high-risk areas are marked in the distribution map, and the reasons for the decline in insulation performance are evaluated, including moisture, aging, and partial discharge.

[0119] The high-risk areas in the electric field distribution map are important references for the risk analysis of the state model. The state model provides an overall trend analysis of insulation performance to guide the priority order of applying protective agents; the electric field distribution map locates the electric field concentration areas to guide the application location of protective agents.

[0120] The steps of combining the real-time monitoring data with historical insulation parameters and using the support vector machine (SVM) and long short-term memory network (LSTM) algorithms to generate a state model are as follows:

[0121] Generate a state model, and the state model formula is:

[0122] X input = Concat(X r , X h ),

[0123] where X input is the input data matrix, which contains the concatenation result of the real-time monitoring data X r and the historical insulation parameters X h . X r is the real-time monitoring data, including the temperature, electric field intensity, and humidity values at the current moment. X h is the historical insulation parameter, including the long-term accumulated temperature, electric field, and humidity trends.

[0124] Classification is carried out using the support vector machine (SVM), and the classification formula is:

[0125]

[0126] Among them, f(x) is the classification result function, the output is +1 or -1, x is the data point to be classified, and x i is the support vector, and y i is the label of the training data, +1 indicates normal, -1 indicates potential deterioration, and α i is the weight of the support vector, K(x i , x) is the kernel function, the Gaussian kernel and the radial basis function are selected, and b is the bias term.

[0127] The long short-term memory network LSTM is used for time series prediction, and the prediction formula includes:

[0128] Formula for forget gate: f t = σ(W f x t + U f h t-1 + b f ),

[0129] Formula for input gate: i t = σ(W i x t + U i h t-1 + b i ),

[0130] Formula for updating memory cell: c t = f t ⊙ c t-1 + i t ⊙ tanh(W c x t + U c h t-1 + b c ),

[0131] Formula for output gate: o t = σ(W o x t + U o h t-1 + b o ),

[0132] The output at the current moment is y t , and y t = o t ⊙ tanh(c t ),

[0133] Among them, x t is the current input data, h t-1 is the hidden state at the previous moment, and ct-1 , c t is the state of the memory cell at the previous and current moments, f t , i t , o t are the activation values of the forget gate, input gate, and output gate, W f , W i , W c , W o is the weight matrix in the corresponding calculation formula, U f , U i , U c , U o is the hidden state weight matrix in the corresponding calculation formula, b f , b i , b c , b o is the bias term in the corresponding calculation formula, σ(·) is the sigmoid function, tanh is the hyperbolic tangent function, and ⊙ is element-wise multiplication;

[0134] Based on the finite element analysis FEA, calculate the electric field distribution of the device, mark the electric field concentration area, and generate an electric field distribution map in combination with the results of the state model. The steps for marking the high-risk area in the distribution map are as follows:

[0135] Conduct electric field distribution analysis and risk area marking. The electric field distribution calculation formula is:

[0136]

[0137] Among them, E(x, y, z) is the electric field strength in three-dimensional space, and φ(x, y, z) is the potential distribution at the spatial position (x, y, z).

[0138] Mark the risk area. The marking condition is:

[0139] E(x, y, z) > E crit ,

[0140] Among them, E crit is the critical value of the electric field strength, and the area exceeding this value is marked as high risk;

[0141] Specifically, adopt a state model based on the support vector machine SVM and long short-term memory network LSTM algorithms to accurately evaluate the insulation degradation risk of different parts of the device and predict the time series trend of insulation parameters; combine the electric field distribution map generated by the finite element analysis FEA to locate and mark the high-risk area, dynamically identify potential risk areas, and real-time track the effectiveness of protection measures.

[0142] Step S3: The maintenance personnel use a portable coating device to coat the electroconductive optimization protective agent;

[0143] The coating method in step S3 is as follows:

[0144] Select the type of protective agent suitable for the location of the electrical equipment according to the electric field distribution map, including:

[0145] Heat-resistant protective agent, used for local hot spot areas,

[0146] Hydrophobic protective agent, used for high-humidity environments,

[0147] High dielectric strength protective agent, used for areas with concentrated electric fields,

[0148] In step S3, after the coating is completed, based on the insulation parameters continuously collected by the sensor, compare the state model and the electric field distribution map before and after the application of the protective agent to evaluate the actual effect of the protective agent; if it is found that there is still abnormal concentration of the local electric field, the maintenance personnel shall supplement the application of the protective agent or adjust the coating thickness;

[0149] The steps for evaluating the actual effect of the protective agent by comparing the state model and the electric field distribution map before and after the application of the protective agent based on the insulation parameters continuously collected by the sensor are as follows:

[0150] Apply the protective agent to the high-risk area according to the electric field distribution map and adjust the coating thickness. The adjustment formula is:

[0151] T coat =T base +k·ΔE,

[0152] where T coat is the coating thickness of the protective agent, T base is the basic coating thickness, k is the thickness adjustment factor, and ΔE is the amplitude of the electric field strength exceeding the critical value.

[0153] Conduct effect evaluation. The effect evaluation formula is:

[0154]

[0155] where ΔP is the insulation performance change rate, P before is the insulation performance parameter value before coating, and P after is the insulation performance parameter value after coating;

[0156] Step S4: Based on the long-term insulation parameter data, analyze the performance changes of different protective agents under different environmental conditions, and use the Bayesian optimization algorithm to calculate the protective agent adjustment suggestions;

[0157] In the Bayesian optimization algorithm,

[0158] The input variables are the type of protective agent, coating thickness, environmental temperature and humidity, and the trend of electric field distribution;

[0159] The optimization objectives are to improve the insulation performance, extend the lifespan of the protective agent, and reduce the usage cost;

[0160] Based on long-term insulation parameter data, to analyze the performance changes of different protective agents under different environmental conditions, the steps of using the Bayesian optimization algorithm to calculate the protective agent adjustment suggestions are as follows:

[0161] Approximate the objective function through a surrogate model to find the optimal parameter combination. Define the objective function as the multi-objective optimization problem f(X) of the protective agent performance,

[0162] f(X) = w 1 S(X) + w 2 L(X) - w 3 C(X),

[0163] where f(X) is the optimization objective function, representing the comprehensive performance evaluation of the protective agent, X is the variable to be optimized, including the type of protective agent, coating thickness, and environmental temperature and humidity, S(X) is the insulation performance of the protective agent, L(X) is the service life of the protective agent, C(X) is the cost of the protective agent, w 1 , w 2 , w 3 are the weight coefficients of the objective function, satisfying w 1 + w 2 + w 3 = 1,

[0164] Use the Gaussian process as the surrogate model of the objective function. The model formula is:

[0165]

[0166] where μ(X) is the mean function of the objective function, representing the predicted value of the surrogate model at point X, k(X, X ′ ) is the kernel function,

[0167]

[0168] where, is the variance of the function value, l is the length scale, |X - X ′ | is the Euclidean distance between two input points;

[0169] The sampling strategy is defined as expected improvement. The improvement formula is:

[0170] EI(X) = E[max(f(X) - f * , 0)],

[0171] where EI(X) is the expected improvement value at X, f * is the current optimal objective function value, E[·] is the mathematical expectation,

[0172] EI(X) is explicitly calculated as:

[0173] EI(X) = (f mean (X) - f * )Φ(Z) + σ(X)φ(Z),

[0174]

[0175] where, f mean (X) is the predicted mean of the surrogate model at point X, σ(X) is the predicted standard deviation of the surrogate model at point X, Φ(Z) is the cumulative distribution function of the standard normal distribution, and φ(Z) is the probability density function of the standard normal distribution;

[0176] In step S4, the protective agent optimization process includes

[0177] selecting an initial point set {X 1 , X 2 , …, X n} for actual simulation to construct an initial data set

[0178] using Gaussian process to fit the initial data set to generate an initial surrogate model of the objective function

[0179] optimizing the expected improvement function EI(X), and selecting the sampling point for the next simulation as X next ,

[0180]

[0181] actually evaluating the objective function value f(X next ) at point X next ), and updating the data set

[0182]

[0183] repeating the surrogate model update and sampling point selection until the stopping condition is met,

[0184] The optimal protective agent parameters are:

[0185]

[0186] including the type of protective agent, the best coating thickness, and the optimization strategy for adapting to different environments,

[0187] Specifically, the Bayesian optimization algorithm is introduced to establish the correlation between the protective agent performance and environmental conditions with the Gaussian process as the surrogate model, and dynamically optimize the type, thickness, and coating strategy of the protective agent.

[0188] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. An application of an electrical conductivity optimization protective agent to improve the insulation performance of electrical equipment, characterized in that: The application steps include, Step S1, installing sensors at the connection points and high electric field strength parts of the electrical equipment to monitor insulation parameters, including temperature, electric field distribution, humidity and partial discharge; performing preliminary data processing at the sensor end and marking normal events; Step S2, combining the real-time monitoring data with the historical insulation parameters, and using the support vector machine (SVM) and long short-term memory (LSTM) network algorithm to generate a state model, including: temperature distribution prediction, humidity change trend, and the intensity and location of the discharge signal; Step S3, the maintenance personnel use portable coating equipment to apply the electrical conductivity optimization protective agent; The coating method in step S3 is: Select the type of protective agent that is suitable for the location of the electrical equipment according to the electric field distribution diagram, including: Heat-resistant protective agent, used for local hot spots, Hydrophobic protective agent, used in high humidity environment, High dielectric strength protective agent, used in areas with concentrated electric fields. Step S4, based on the long-term insulation parameter data, analyze the performance changes of different protective agents under different environmental conditions, and use the Bayesian optimization algorithm to calculate the protective agent adjustment suggestions.

2. The application of the electrical conductivity optimization protective agent to improve the insulation performance of electrical equipment according to claim 1, characterized in that: The sensor side dynamically adjusts the sampling frequency and increases the sampling frequency when the load is high or the environment changes drastically; The high electric field strength locations include cable joints and terminals, edge areas of high voltage transformer windings, and electric field concentration points in insulator strings; The abnormal events include temperature surge, electric field abnormality or excessive humidity.

3. The application of the electrical conductivity optimization protective agent to improve the insulation performance of electrical equipment according to claim 2, characterized in that: The steps of performing preliminary data processing at the sensor end and marking normal events are: The sliding average method is used to smooth the signal collected by the sensor. The smoothing formula is: Among them, S f (t) is the signal value after smoothing at the tth moment, N is the sliding window size, S(i) is the original signal value at the ith moment, Use the anomaly detection formula to determine whether an abnormal event has occurred. The anomaly detection formula is: |X t -μ|>k·σ, Among them, X t is the sensor collection value at the tth moment, μ is the mean of historical data, σ is the standard deviation of historical data, and k is the sensitivity factor; In the step of dynamically adjusting the sampling frequency at the sensor end, the calculation formula of the sampling frequency is: f s =f base +Δf·I(event), Among them, f s is the current sampling frequency, f base is the basic sampling frequency, Δf is the sampling frequency increment, and I(event) is the event indicator function, which takes the value 1 when an abnormal event occurs and 0 otherwise.

4. The application of the electrical conductivity optimization protective agent to improve the insulation performance of electrical equipment as claimed in claim 3, characterized in that: The support vector machine algorithm is used to classify and identify potential degradation risks of different parts of the equipment; The long short-term memory network algorithm is used to predict the time series changes of insulation parameters and perform trend analysis; In step S2, the electric field distribution of the equipment is calculated based on the finite element analysis (FEA), the electric field concentration area is marked, and the electric field distribution map is generated in combination with the state model results. The high-risk areas are marked in the distribution map, and the causes of the degradation of insulation performance are evaluated, including moisture, aging and partial discharge.

5. The application of the electrical conductivity optimization protective agent to improve the insulation performance of electrical equipment as claimed in claim 4, characterized in that: The steps of combining the real-time monitoring data with the historical insulation parameters and using the support vector machine SVM and the long short-term memory network LSTM algorithm to generate the state model are: Generate a state model, the state model formula is: X input =Concat(X r ,X h ), Among them, X input is the input data matrix, containing real-time monitoring data X r and historical insulation parameter X h The splicing result, X r To monitor data in real time, including the current temperature, electric field strength and humidity value, X h Historical insulation parameters, including long-term accumulated trends in temperature, electric field, and humidity, Support vector machine SVM is used for classification, and the classification formula is: Among them, f(x) is the classification result function, the output is +1 or -1, x is the data point to be classified, x i is the support vector, y i is the label of the training data, +1 indicates normal, -1 indicates potential degradation, α i is the weight of the support vector, K(x i ,x) is the kernel function, Gaussian kernel and radial basis function are selected, b is the bias term, Long short-term memory network LSTM is used for time series prediction. The prediction formula includes: Forget gate calculation formula: f t =σ(W f x t +U f h t-1 +b f ), Input gate calculation formula: i t =σ(W i x t +U i h t-1 +b i ), Memory unit update formula: c t =f t ⊙c t-1 +i t ⊙tanh(W c x t +U c h t-1 +b c ), Output gate calculation formula: o t =σ(W o x t +U o h t-1 +b o ), The current output is y t ,y t =o t ⊙tanh(c t ), Among them, x t is the current input data, h t-1 is the hidden state of the previous moment, c t-1 、c t is the state of the memory unit at the previous moment and the current moment, f t 、i t , o t is the activation value of the forget gate, input gate, and output gate, W f ,W i ,W c ,W o is the weight matrix in the corresponding calculation formula, U f ,U i ,U c ,U o is the hidden state weight matrix in the corresponding calculation formula, b f ,b i ,b c ,b o is the bias term in the corresponding calculation formula, σ(·) is the sigmoid function, tanh is the hyperbolic tangent function, and ⊙ is element-wise multiplication.

6. The application of the electrical conductivity optimization protective agent to improve the insulation performance of electrical equipment as claimed in claim 5, characterized in that: The electric field distribution of the device is calculated based on the finite element analysis FEA, the electric field concentration area is marked, and the electric field distribution map is generated in combination with the state model result. The steps of marking the high-risk area in the distribution map are as follows: Perform electric field distribution analysis and risk area marking. The electric field distribution calculation formula is: Where E(x,y,z) is the electric field intensity in three-dimensional space, φ(x,y,z) is the potential distribution at the spatial position (x,y,z), Mark risk areas. The marking conditions are: E(x,y,z)>E crit , Among them, E crit is the critical value of electric field strength, and areas exceeding this value are marked as high risk.

7. The application of the electrical conductivity optimization protective agent to improve the insulation performance of electrical equipment according to claim 6, characterized in that: In step S3, after coating is completed, based on the insulation parameters continuously collected by the sensor, the state model and the electric field distribution diagram before and after the protective agent coating are compared to evaluate the actual effect of the protective agent; if it is found that the local electric field is still abnormally concentrated, the maintenance personnel will re-apply the protective agent or adjust the coating thickness.

8. The application of the electrical conductivity optimization protective agent for improving the insulation performance of electrical equipment according to claim 7, characterized in that: The step of comparing the state model and the electric field distribution diagram before and after the protective agent is applied based on the insulation parameters continuously collected by the sensor to evaluate the actual effect of the protective agent is as follows: Apply protective agent to high-risk areas according to the electric field distribution map and adjust the coating thickness. The adjustment formula is: T coat =T base +k·ΔE, Among them, T coat is the coating thickness of the protective agent, T base is the base coating thickness, k is the thickness adjustment factor, ΔE is the magnitude by which the electric field intensity exceeds the critical value, To evaluate the effect, the formula is: Among them, ΔP is the insulation performance change rate, P before is the insulation performance parameter value before coating, P after is the insulation performance parameter value after coating.

9. The application of the electrical conductivity optimization protective agent to improve the insulation performance of electrical equipment according to claim 8, characterized in that: In the Bayesian optimization algorithm, The input variables are the type of protective agent, coating thickness, ambient temperature and humidity, and electric field distribution trend; The optimization goals are to improve insulation performance, extend the life of protective agents and reduce usage costs.

10. The application of the electrical conductivity optimization protective agent to improve the insulation performance of electrical equipment according to claim 9, characterized in that: The steps of analyzing the performance changes of different protective agents under different environmental conditions based on long-term insulation parameter data and using the Bayesian optimization algorithm to calculate the protective agent adjustment suggestions are as follows: The objective function is approximated by the proxy model to find the optimal parameter combination, and the objective function is defined as the multi-objective optimization problem f(X) of the protective agent performance. f(X)=w1S(X)+w2L(X)-w3C(X), Among them, f(X) is the optimization objective function, which represents the comprehensive performance evaluation of the protective agent. X is the variable to be optimized, including the type of protective agent, coating thickness and ambient temperature and humidity. S(X) is the insulation performance of the protective agent. L(X) is the service life of the protective agent. C(X) is the cost of the protective agent. w1, w2, and w3 are the weight coefficients of the objective function, satisfying w1+w2+w3=1. Using Gaussian process as the surrogate model of the objective function, the model formula is: Among them, μ(X) is the mean function of the objective function, which represents the predicted value of the surrogate model at point X, k(X,X′) is the kernel function, in, is the variance of the function value, l is the length scale, |XX′| is the Euclidean distance between two input points; The sampling strategy is defined as the expected improvement, and the improvement formula is: EI(X)=E[max(f(X)-f * ,0)], Where EI(X) is the expected improvement at X, f * is the current optimal objective function value, E[·] is the mathematical expectation, EI(X) is calculated explicitly as: EI(X)=(f mean (X)-f * )Φ(Z)+σ(X)φ(Z), Among them, f mean (X) is the predicted mean of the surrogate model at point X, σ(X) is the predicted standard deviation of the surrogate model at point X, Φ(Z) is the cumulative distribution function of the standard normal distribution, and φ(Z) is the probability density function of the standard normal distribution; In step S4, the protective agent optimization process includes: Select the initial point set {X1,X2,…,X n }Perform actual simulation and build initial data set Fitting the initial dataset using a Gaussian process Generate the initial objective function proxy model, Optimize the expected improvement function EI(X) and select the sampling point of the next simulation as X next , In X next The actual evaluation objective function value f(X next ), update the dataset Repeat the proxy model update and sampling point selection until the stopping condition is met. The optimal protective agent parameters are: Including types of protective agents, optimal coating thickness and optimization strategies to adapt to different environments.

Images