Power grid equipment safety analysis model optimization method based on intelligent test tool

Abstract

The invention belongs to the technical field of power grids, and particularly relates to a power grid equipment safety analysis model optimization method based on an intelligent test tool. According to the method, historical data and real-time data are collected and analyzed, existing information resources are fully utilized, the accuracy of fault and risk identification is improved, a Bayesian optimization method is utilized to construct and optimize an equipment safety analysis model, the fault and abnormity prediction capability is improved, and through risk assessment and an intelligent test tool, the fault and abnormity prediction efficiency is improved. Key equipment needing to be optimized is accurately positioned, and the utilization rate of maintenance resources is increased; and through a Bayesian optimization method, the model is continuously updated and perfected based on new data, and it is ensured that the prediction capability of the model is continuously improved along with time lapse.

Description

Technical Field

[0001] The present application belongs to the field of power grid technology, and more specifically, to a method for optimizing a power grid equipment safety analysis model based on an intelligent testing tool. Background Art

[0002] The safety of power grid equipment is an important guarantee for the stable operation of the power system. Ensuring the normal operation of power grid equipment is crucial to preventing large-scale power outages and ensuring the stability of power supply. However, with the expansion of the scale of the power grid and the increase in the complexity of equipment, the safety analysis and maintenance of power grid equipment are facing more and more challenges.

[0003] Grid equipment includes transformers, circuit breakers, cables, switchgear, etc. These devices play a key role in the power system. Any failure or abnormality of any equipment may lead to instability of the power system and even cause large-scale power outages. Therefore, effective analysis and monitoring of the safety of grid equipment is crucial to ensure the stable operation of the power system.

[0004] Traditional power grid equipment maintenance methods mainly rely on regular inspections and empirical judgments, which have the following limitations:

[0005] High periodic maintenance costs: Regular inspection and maintenance require a lot of manpower and material resources, and the maintenance cycle is long.

[0006] Fault warning is not timely: Traditional methods make it difficult to detect potential faults and abnormalities of equipment in a timely manner, and maintenance is often performed only after obvious problems have occurred in the equipment.

[0007] Insufficient data utilization: The large amount of operating data generated by power grid equipment is not fully utilized, and potential failures and risks of equipment cannot be identified and predicted in a timely manner. Summary of the invention

[0008] The present invention provides a power grid equipment safety analysis model optimization method based on an intelligent testing tool, which is intended to solve the problems of waste of material resources and long maintenance cycle caused by the current traditional power grid equipment maintenance method using manual methods.

[0009] The power grid equipment safety analysis model optimization method based on intelligent testing tools includes the following steps:

[0010] Step 1: Collect historical operation data of power grid equipment, analyze the operation data of equipment using data analysis and statistical methods, and preliminarily identify existing safety hazards and risk points;

[0011] Step 2: Based on the preliminary identification results, a safety analysis model for power grid equipment is constructed to predict equipment failures and anomalies;

[0012] Step 3: Based on the output of the safety analysis model, evaluate the identified potential risk points and determine the key equipment and issues that need to be optimized;

[0013] Step 4: Deploy intelligent testing tools on the devices that need to be optimized, test the devices based on the intelligent testing tools, collect and analyze real-time data, and verify the risk points identified in the security analysis model;

[0014] Step 5: Based on the test data and historical data, the Bayesian optimization method is used to update and improve the power grid equipment safety analysis model.

[0015] The present invention collects and analyzes historical data and real-time data, makes full use of existing information resources, improves the accuracy of fault and risk identification, uses Bayesian optimization methods to build and optimize equipment safety analysis models, improves the ability to predict faults and anomalies, and accurately locates key equipment that needs to be optimized through risk assessment and intelligent testing tools, thereby improving the utilization rate of maintenance resources; and through the Bayesian optimization method, continuously updates and improves the model based on new data, ensuring that the model's predictive ability continues to improve over time.

[0016] Preferably, the step 1 comprises the following steps:

[0017] Step 1.1: Collect historical operation data of power grid equipment, including equipment operation data, fault records and maintenance records;

[0018] Step 1.2: Preprocess the collected data, including data cleaning, detecting and presenting outliers through the 3σ principle or IQR method, and filling missing data by linear interpolation; normalize the cleaned data;

[0019] Step 1.3: Time-align and integrate the normalized equipment operation data, fault records, and maintenance records to form a unified data set;

[0020] Step 1.4: Based on statistical analysis, time series analysis, and failure mode analysis, identify the safety hazards and risk points of the equipment, and obtain preliminary identification of safety hazards and risk points.

[0021] Preferably, the step 1.4 comprises the following steps:

[0022] The statistical analysis uses mean and variance to identify anomalies and hidden dangers in equipment operation data:

[0023]

[0024] Where: μ represents the average value of the corresponding observations; n represents the number of corresponding observations; x i represents the i-th observation value;

[0025]

[0026] Where: 2 It represents variance, which indicates the degree of dispersion of data;

[0027] Correlation analysis:

[0028]

[0029] Where: XY is the correlation coefficient, which indicates the degree of correlation between two variables, and its value range is [-1,1]; x i and i Respectively represent the variables X and Y corresponding to the i-th observation value;

[0030] Through mean, variance and correlation coefficient analysis, abnormal fluctuations in operating parameters and hidden dangers can be identified;

[0031] Use time series analysis methods to identify trends and periodic characteristics in equipment operation data and discover potential risk points:

[0032] Trend Analysis:

[0033]

[0034] Where: MA t represents the moving average of time t; k is the window size, which means taking k data points before and after; x t+i represents the observation value at time t+i;

[0035] Seasonality Analysis:

[0036]

[0037] Where: ACF(k) represents the autocorrelation coefficient of lag k; x t and x t+k Represent the observation values ​​at time t and time t+k respectively;

[0038] Through trend and seasonal analysis, identify long-term change trends and cyclical fluctuations in equipment operation data and discover potential risk points;

[0039] Through failure mode analysis, common failure modes and high-frequency failure types in equipment operation are identified to determine potential safety hazards:

[0040] Count the number of occurrences of each fault type in the historical data:

[0041]

[0042] Where: Freq(i) represents the frequency of fault type i; Count(Fault i ) represents the number of occurrences of fault type i; ∑Count(Fault j ) represents the total number of occurrences of all fault types;

[0043] Fault tree analysis: Based on the equipment's fault records, a fault tree analysis model is constructed to identify the cause of the fault:

[0044] Top event: major failure of equipment;

[0045] Intermediate event: the intermediate fault cause that leads to the top event;

[0046] Basic event: the lowest level of fault cause;

[0047] Through frequency analysis and fault tree analysis, common failure modes and high-frequency failure types in equipment operation can be identified to determine potential safety hazards.

[0048] Preferably, the safety analysis model adopts a random forest model.

[0049] Preferably, step 3 comprises the following steps:

[0050] Risk score calculation: Based on the failure and abnormal data predicted by the model, a risk score is calculated for each potential risk point:

[0051] Risk Score i =P i ×S i ;

[0052] Where: Risk Score i represents the risk score of the ith potential risk point; P i represents the failure probability of the i-th risk point, which is predicted by the model; S i Indicates the fault severity of the i-th risk point, which is scored based on the historical fault data of the equipment;

[0053] Where P i The calculation method is as follows:

[0054]

[0055] Where: Number of trees predicting failure represents the number of trees in the random forest that predict the equipment failure; Total number of trees represents the total number of trees in the random forest model;

[0056] Where S i The calculation method is as follows:

[0057] S i =W t +T t +W c ;

[0058] Where: W t Indicates the repair time score; T t represents the downtime score; W c represents the repair cost score;

[0059] Summarize all risk points for each device and calculate the total risk score of the device:

[0060]

[0061] Where: Total Risk Score j represents the total risk score of the jth device;

[0062] The devices are sorted according to the total risk score of each device to determine the devices that need to be optimized; the top k devices in the sorting are selected as the optimized devices.

[0063] Preferably, step 4 comprises the following steps:

[0064] For the faults predicted in step 3, conduct targeted tests on the corresponding equipment, impose specific fault conditions on the corresponding equipment, and monitor the equipment response;

[0065] During the test, the intelligent test tool collects the operating data of the device under different fault conditions in real time, and verifies whether the device has the predicted fault by comparing the real-time data under the fault condition with the operating data:

[0066] The authenticity of the fault is determined by the fault characteristic value and fault threshold of the operating data collected in real time by the test tool. If the fault characteristic value is greater than the fault threshold, the fault is considered to be authentic.

[0067] Preferably, the step 5 comprises the following steps:

[0068] Construct an initial Gaussian process model, and gradually update the security analysis model based on the Bayesian optimization method:

[0069] Define the objective function f(x), which represents the error and loss between the model prediction and the actual test results;

[0070] Prior distribution: Assume that the objective function f(x) follows a Gaussian process. Initially, the value of the objective function is set to zero, the mean is 0, and the covariance function is as follows:

[0071]

[0072] Where: 2 represents variance; l represents the length scale parameter; x represents the eigenvector of the measured data point; x ′ The feature vector representing the predicted data point;

[0073] Bayesian Updating: Observation Point Data Includes n observed data points, where x i represents the feature vector, y i represents the corresponding actual observed value;

[0074] Posterior distribution:

[0075] p(f(x * )|D,x * )=N(μ * ,σ 2 );

[0076] Where: x * Represents a new data point for prediction, with the feature vector x * ;

[0077]

[0078] Where: k * represents the covariance vector between the new data point and all observed data points; T represents transpose; K represents the covariance matrix between the observed data points; y represents the actual value vector of the observed data;

[0079]

[0080] Where: k(x * ,x * ) represents the covariance of the new data point itself;

[0081] Acquisition function: Select an acquisition function for an optimal data point:

[0082]

[0083] Where: f(x + ) represents the current optimal value;

[0084] Test: Test the device based on the acquisition function to obtain new data points (x * ,y * );

[0085] Update model: Update the posterior distribution of the Gaussian process model based on new data points;

[0086] Repeat the acquisition function to update the model step until convergence or reaching the preset number of iterations to obtain the hyperparameters of the security analysis model.

[0087] The beneficial effects of the present invention include:

[0088] The present invention collects and analyzes historical data and real-time data, makes full use of existing information resources, improves the accuracy of fault and risk identification, uses Bayesian optimization methods to build and optimize equipment safety analysis models, improves the ability to predict faults and anomalies, and accurately locates key equipment that needs to be optimized through risk assessment and intelligent testing tools, thereby improving the utilization rate of maintenance resources; and through the Bayesian optimization method, continuously updates and improves the model based on new data, ensuring that the model's predictive ability continues to improve over time. BRIEF DESCRIPTION OF THE DRAWINGS

[0089] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0090] Figure 1 The overall system architecture diagram provided for the embodiment of the present invention. DETAILED DESCRIPTION

[0091] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0092] See also Figure 1 As shown, the best embodiment of the present invention is further described;

[0093] The power grid equipment safety analysis model optimization method based on intelligent testing tools includes the following steps:

[0094] Step 1: Collect historical operation data of power grid equipment, analyze the operation data of equipment using data analysis and statistical methods, and preliminarily identify existing safety hazards and risk points;

[0095] The step 1 comprises the following steps:

[0096] Step 1.1: Collect historical operation data of power grid equipment, including equipment operation data, fault records and maintenance records;

[0097] Step 1.2: Preprocess the collected data, including data cleaning, detecting and presenting outliers through the 3σ principle or IQR method, and filling missing data by linear interpolation; normalize the cleaned data;

[0098] Step 1.3: Time-align and integrate the normalized equipment operation data, fault records, and maintenance records to form a unified data set;

[0099] Step 1.4: Based on statistical analysis, time series analysis, and failure mode analysis, identify the safety hazards and risk points of the equipment, and obtain preliminary identification of safety hazards and risk points.

[0100] Preferably, the step 1.4 comprises the following steps:

[0101] The statistical analysis uses mean and variance to identify anomalies and hidden dangers in equipment operation data:

[0102]

[0103] Where: μ represents the average value of the corresponding observations; n represents the number of corresponding observations; x i represents the i-th observation value;

[0104]

[0105] Where: 2 It represents variance, which indicates the degree of dispersion of data;

[0106] Correlation analysis:

[0107]

[0108] Where: XY is the correlation coefficient, which indicates the degree of correlation between two variables, and its value range is [-1,1]; x i and i Respectively represent the variables X and Y corresponding to the i-th observation value;

[0109] Through mean, variance and correlation coefficient analysis, abnormal fluctuations in operating parameters and hidden dangers can be identified;

[0110] Use time series analysis methods to identify trends and periodic characteristics in equipment operation data and discover potential risk points:

[0111] Trend Analysis:

[0112]

[0113] Where: MA trepresents the moving average of time t; k is the window size, which means taking k data points before and after; x t+i represents the observation value at time t+i;

[0114] Seasonality Analysis:

[0115]

[0116] Where: ACF(k) represents the autocorrelation coefficient of lag k; x t and x t+k Represent the observation values ​​at time t and time t+k respectively;

[0117] Through trend and seasonal analysis, identify long-term change trends and cyclical fluctuations in equipment operation data and discover potential risk points;

[0118] Through failure mode analysis, common failure modes and high-frequency failure types in equipment operation are identified to determine potential safety hazards:

[0119] Count the number of occurrences of each fault type in the historical data:

[0120]

[0121] Where: Freq(i) represents the frequency of fault type i; Count(Fault i ) represents the number of occurrences of fault type i; ∑Count(Fault j ) represents the total number of occurrences of all fault types;

[0122] Fault tree analysis: Based on the equipment's fault records, a fault tree analysis model is constructed to identify the cause of the fault:

[0123] Top event: major failure of equipment;

[0124] Intermediate event: the intermediate fault cause that leads to the top event;

[0125] Basic event: the lowest level of fault cause;

[0126] Through frequency analysis and fault tree analysis, common failure modes and high-frequency failure types in equipment operation can be identified to determine potential safety hazards.

[0127] Step 2: Based on the results of the preliminary identification, a safety analysis model for power grid equipment is constructed to predict equipment failures and anomalies; the safety analysis model uses a random forest model.

[0128] Step 3: Based on the output of the safety analysis model, evaluate the identified potential risk points and determine the key equipment and issues that need to be optimized;

[0129] The step 3 comprises the following steps:

[0130] Risk score calculation: Based on the failure and abnormal data predicted by the model, a risk score is calculated for each potential risk point:

[0131] Risk Score i =P i ×S i ;

[0132] Where: Risk Score i represents the risk score of the ith potential risk point; P i represents the failure probability of the i-th risk point, which is predicted by the model; S i Indicates the fault severity of the i-th risk point, which is scored based on the historical fault data of the equipment;

[0133] Where P i The calculation method is as follows:

[0134]

[0135] Where: Number of trees predicting failure represents the number of trees in the random forest that predict the equipment failure; Total number of trees represents the total number of trees in the random forest model;

[0136] Where S i The calculation method is as follows:

[0137] S i =W t +T t +W c ;

[0138] Where: W t Indicates the repair time score; T t represents the downtime score; W c represents the repair cost score;

[0139] Summarize all risk points for each device and calculate the total risk score of the device:

[0140]

[0141] Where: Total Risk Score j represents the total risk score of the jth device;

[0142] The devices are sorted according to the total risk score of each device to determine the devices that need to be optimized; the top k devices in the sorting are selected as the optimized devices.

[0143] Step 4: Deploy intelligent testing tools on the devices that need to be optimized, test the devices based on the intelligent testing tools, collect and analyze real-time data, and verify the risk points identified in the security analysis model;

[0144] The step 4 comprises the following steps:

[0145] For the faults predicted in step 3, conduct targeted tests on the corresponding equipment, apply specific fault conditions to the corresponding equipment, and monitor the equipment response; for example, if voltage anomalies are predicted, apply different voltage levels through the test tool and observe the response data of the equipment;

[0146] During the test, the intelligent test tool collects the operating data of the device under different fault conditions in real time, and verifies whether the device has the predicted fault by comparing the real-time data under the fault condition with the operating data:

[0147] The authenticity of the fault is determined by the fault characteristic value and fault threshold of the operating data collected in real time by the test tool. If the fault characteristic value is greater than the fault threshold, the fault is considered to be authentic.

[0148] Step 5: Based on the test data and historical data, the Bayesian optimization method is used to update and improve the power grid equipment safety analysis model.

[0149] The step 5 comprises the following steps:

[0150] Construct an initial Gaussian process model, and gradually update the security analysis model based on the Bayesian optimization method:

[0151] Define the objective function f(x), which represents the error and loss between the model prediction and the actual test results;

[0152] Prior distribution: Assume that the objective function f(x) follows a Gaussian process. Initially, the value of the objective function is set to zero, the mean is 0, and the covariance function is as follows:

[0153]

[0154] Where: 2 represents variance; l represents the length scale parameter; x represents the eigenvector of the measured data point; x ′ The feature vector representing the predicted data point;

[0155] Bayesian Updating: Observation Point Data Includes n observed data points, where x i represents the feature vector, y i represents the corresponding actual observed value;

[0156] Posterior distribution:

[0157] p(f(x * )|D,x * )=N(μ * ,σ 2 );

[0158] Where: x * Represents a new data point for prediction, with the feature vector x * ;

[0159]

[0160] Where: k * represents the covariance vector between the new data point and all observed data points; T represents transpose; K represents the covariance matrix between the observed data points; y represents the actual value vector of the observed data;

[0161]

[0162] Where: k(x * ,x * ) represents the covariance of the new data point itself;

[0163] Acquisition function: Select an acquisition function for an optimal data point:

[0164]

[0165] Where: f(x + ) represents the current optimal value;

[0166] Test: Test the device based on the acquisition function to obtain new data points (x * ,y * );

[0167] Update model: Update the posterior distribution of the Gaussian process model based on new data points;

[0168] Repeat the acquisition function to update the model step until convergence or reaching the preset number of iterations to obtain the hyperparameters of the security analysis model.

[0169] The present invention collects and analyzes historical data and real-time data, makes full use of existing information resources, improves the accuracy of fault and risk identification, uses Bayesian optimization methods to build and optimize equipment safety analysis models, improves the ability to predict faults and anomalies, and accurately locates key equipment that needs to be optimized through risk assessment and intelligent testing tools, thereby improving the utilization rate of maintenance resources; and through the Bayesian optimization method, continuously updates and improves the model based on new data, ensuring that the model's predictive ability continues to improve over time.

[0170] The above are only preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A power grid equipment safety analysis model optimization method based on intelligent testing tools, characterized in that: The following steps are involved: Step 1: Collect historical operation data of power grid equipment, analyze the operation data of equipment using data analysis and statistical methods, and preliminarily identify existing safety hazards and risk points; Step 2: Based on the preliminary identification results, a safety analysis model for power grid equipment is constructed to predict equipment failures and anomalies; Step 3: Based on the output of the safety analysis model, evaluate the identified potential risk points and determine the key equipment and issues that need to be optimized; Step 4: Deploy intelligent testing tools on the devices that need to be optimized, test the devices based on the intelligent testing tools, collect and analyze real-time data, and verify the risk points identified in the security analysis model; Step 5: Based on the test data and historical data, the Bayesian optimization method is used to update and improve the power grid equipment safety analysis model.

2. The method for optimizing the power grid equipment safety analysis model based on the intelligent testing tool according to claim 1, characterized in that: The step 1 comprises the following steps: Step 1.1: Collect historical operation data of power grid equipment, including equipment operation data, fault records and maintenance records; Step 1.2: Preprocess the collected data, including data cleaning, detecting and presenting outliers through the 3σ principle or IQR method, and filling missing data by linear interpolation; normalize the cleaned data; Step 1.3: Time-align and integrate the normalized equipment operation data, fault records, and maintenance records to form a unified data set; Step 1.4: Based on statistical analysis, time series analysis, and failure mode analysis, identify the safety hazards and risk points of the equipment, and obtain preliminary identification of safety hazards and risk points.

3. The method for optimizing the power grid equipment safety analysis model based on the intelligent testing tool according to claim 2 is characterized in that: The step 1.4 comprises the following steps: The statistical analysis uses mean and variance to identify anomalies and hidden dangers in equipment operation data: Where: μ represents the average value of the corresponding observations; n represents the number of corresponding observations; x i represents the i-th observation value; Where: 2 It represents variance, which indicates the degree of dispersion of data; Correlation analysis: Where: XY is the correlation coefficient, which indicates the degree of correlation between two variables, and its value range is [-1,1]; x i and i Respectively represent the variables X and Y corresponding to the i-th observation value; Through mean, variance and correlation coefficient analysis, abnormal fluctuations in operating parameters and hidden dangers can be identified; Use time series analysis methods to identify trends and periodic characteristics in equipment operation data and discover potential risk points: Trend Analysis: Where: MA t represents the moving average of time t; k is the window size, which means taking k data points before and after; x t+i represents the observation value at time t+i; Seasonality Analysis: Where: ACF(k) represents the autocorrelation coefficient of lag k; x t and x t+k Represent the observation values ​​at time t and time t+k respectively; Through trend and seasonal analysis, identify long-term change trends and cyclical fluctuations in equipment operation data and discover potential risk points; Through failure mode analysis, common failure modes and high-frequency failure types in equipment operation are identified to determine potential safety hazards: Count the number of occurrences of each fault type in the historical data: Where: Freq(i) represents the frequency of fault type i; Count(fault i ) represents the number of occurrences of fault type i; ∑Count(Fault j ) represents the total number of occurrences of all fault types; Fault tree analysis: Based on the equipment's fault records, a fault tree analysis model is constructed to identify the cause of the fault: Top event: major failure of equipment; Intermediate event: the intermediate fault cause that leads to the top event; Basic event: the lowest level of fault cause; Through frequency analysis and fault tree analysis, common failure modes and high-frequency failure types in equipment operation can be identified to determine potential safety hazards.

4. The method for optimizing the power grid equipment safety analysis model based on the intelligent testing tool according to claim 1, characterized in that: The safety analysis model adopts a random forest model.

5. The method for optimizing the power grid equipment safety analysis model based on the intelligent testing tool according to claim 1, characterized in that: The step 3 comprises the following steps: Risk score calculation: Based on the failure and abnormal data predicted by the model, a risk score is calculated for each potential risk point: Where: RiskScore i represents the risk score of the ith potential risk point; P i represents the failure probability of the i-th risk point, which is predicted by the model; S i Indicates the fault severity of the i-th risk point, which is scored based on the historical fault data of the equipment; Where P i The calculation method is as follows: Where: Number of trees predicting failure represents the number of trees in the random forest that predict the equipment failure; Total number of trees represents the total number of trees in the random forest model; Where S i The calculation method is as follows: S i =W t +T t +W c ; Where: W t Indicates the repair time score; T t represents the downtime score; W c represents the repair cost score; Summarize all risk points for each device and calculate the total risk score of the device: Where: Total Risk Score j represents the total risk score of the jth device; The devices are sorted according to the total risk score of each device to determine the devices that need to be optimized; the top k devices in the sorting are selected as the optimized devices.

6. The method for optimizing the power grid equipment safety analysis model based on the intelligent testing tool according to claim 1, characterized in that: The step 4 comprises the following steps: For the faults predicted in step 3, conduct targeted tests on the corresponding equipment, impose specific fault conditions on the corresponding equipment, and monitor the equipment response; During the test, the intelligent test tool collects the operating data of the device under different fault conditions in real time, and verifies whether the device has the predicted fault by comparing the real-time data under the fault condition with the operating data: The authenticity of the fault is determined by the fault characteristic value and fault threshold of the operating data collected in real time by the test tool. If the fault characteristic value is greater than the fault threshold, the fault is considered to be authentic.

7. The method for optimizing the power grid equipment safety analysis model based on the intelligent testing tool according to claim 1, characterized in that: The step 5 comprises the following steps: Construct an initial Gaussian process model, and gradually update the security analysis model based on the Bayesian optimization method: Define the objective function f(x), which represents the error and loss between the model prediction and the actual test results; Prior distribution: Assume that the objective function f(x) follows a Gaussian process. Initially, the value of the objective function is set to zero, the mean is 0, and the covariance function is as follows: Where: 2 represents variance; l represents the length scale parameter; x represents the eigenvector of the measured data point; x′ represents the eigenvector of the predicted data point; Bayesian Updating: Observation Point Data Includes n observed data points, where x i represents the feature vector, y i represents the corresponding actual observed value; Posterior distribution: p(f(x * )|D,x * )=N(μ * ,s 2 ); Where: x * Represents a new data point for prediction, with the feature vector x * ; Where: k * represents the covariance vector between the new data point and all observed data points; T represents transpose; K represents the covariance matrix between the observed data points; y represents the actual value vector of the observed data; Where: k(x * ,x * ) represents the covariance of the new data point itself; Acquisition function: Select an acquisition function for an optimal data point: Where: f(x + ) represents the current optimal value; Test: Test the device based on the acquisition function to obtain new data points (x * ,y * ); Update model: Update the posterior distribution of the Gaussian process model based on new data points; Repeat the acquisition function to update the model step until convergence or reaching the preset number of iterations to obtain the hyperparameters of the security analysis model.

Images