Three-level power box electrical variable monitoring method and system

Abstract

The invention discloses a three-level power box electrical variable monitoring method and system, and relates to the technical field of electrical variable monitoring. A three-level power box electrical variable monitoring system comprises an electrical variable environment disturbance judgment module, an electrical variable disturbance compensation updating module and an electrical variable dynamic threshold division module. By monitoring and analyzing the environment variables and the power box electrical variables in real time and utilizing the improved random forest algorithm and the group optimization algorithm, accurate prediction and compensation of the electrical variables are achieved, the influence of environment disturbance on the electrical variables can be effectively recognized and compensated, and the accuracy and reliability of monitoring of the power box electrical variables are improved; meanwhile, through dynamic threshold division, a monitoring standard can be adjusted according to a real-time environment variable and an electric variable interference prediction result, and quantitative evaluation of an electric variable risk is realized, so that the operation state of the power box is optimized, and safety and stability of a power system are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of electric variable monitoring, and in particular to a method and system for monitoring electric variables of a three-level power supply box. Background Art

[0002] In temporary construction sites such as civil excavation, building construction and electrical maintenance, the three-level power supply box is the main power supply equipment. Due to the complex on-site environment and uneven distribution of various equipment, the construction process is often accompanied by a large number of lead wires and frequent cross-operations, which increases the risk of electrical failures. In particular, high-frequency equipment operation and line migration make the electrical lines in the power box prone to insulation aging, leakage and short circuit problems, leading to safety hazards such as electric shock and fire. In order to improve the electrical safety of construction sites, electrical variable monitoring methods have gradually become an important protective technical means. Electrical variable monitoring can promptly identify the possible risks of leakage and short circuit in the line through real-time detection of key parameters such as voltage, current, grounding resistance and leakage current. For example, through dynamic monitoring of current and leakage current, abnormal fluctuations in the line can be quickly captured, and an early warning can be issued before leakage or short circuit occurs to ensure the safety of operators. However, electrical variable monitoring of the power box in a complex environment may be affected by temperature, humidity, dirt, equipment aging, sensor accuracy, poor grounding, wiring problems, data processing delays and human operational errors, resulting in measurement errors and safety hazards. It is necessary to make adaptive adjustments according to the on-site environmental conditions to make the monitoring results closer to the actual electrical variable parameters, and be able to adjust the monitoring threshold in real time to achieve the effect of early warning. Summary of the invention

[0003] The present invention aims to provide a method and system for monitoring electric variables of a three-level power supply box, which can accurately monitor electric variables and issue early warnings in various complex environments.

[0004] A method for monitoring electric variables of a three-level power supply box comprises the following steps: Step S1: Determination of electrical variable environmental disturbance Obtain the real-time environmental variables and the real-time power box electrical variable set D by monitoring the initial frequency, D={D i |i=1,2,…,I};D i D is the real-time power box electrical variable data item i , I is the total number of types of data items in the power box electrical variable set; the real-time environmental variables are input into the power box environmental prediction and judgment model for analysis to obtain the electrical variable interference prediction results; the power box environmental prediction and judgment model includes an environmental variable processing layer, an environmental feature analysis layer, an environmental prediction and judgment layer, an environmental auxiliary judgment layer and a result output layer. The environmental feature analysis layer is constructed based on the improvement of the random forest algorithm and is used to predict and analyze the real-time environmental variables; Step S2: Update of electrical variable disturbance compensation Based on the electric variable interference prediction results and the power box electric variable disturbance judgment model, the real-time power box electric variable set D is judged to obtain the updated real-time power box electric variable set D', D'={D i '|i=1, 2, ..., I}; The power box electrical variable disturbance judgment model is improved and constructed based on wavelet transform and swarm optimization algorithm; The power box electric variable disturbance judgment model includes a disturbance electric variable calculation layer, a disturbance compensation analysis layer and an electric variable re-output layer; The disturbance electric variable calculation layer is used to perform disturbance detection on all real-time power box electric variable data items Di in the real-time power box electric variable set D, and obtain the disturbance electric variable data item R n , n=1, 2, ..., N; N is all real-time power box electrical variable data items D i The total number of data items affected by environmental variables; The interference compensation analysis layer is used to predict the disturbance of the electric variable based on the disturbance electric variable data item R n The electric variable compensation calculation is performed with the electric variable disturbance compensation function to obtain the compensated electric variable data item R n ', the compensation electric variable data item R n 'Corresponding to the real-time power supply box electrical variable data item D i Match and get the updated power box electrical variable data item D i '; The electric variable disturbance compensation function is constructed based on the swarm optimization algorithm after improving the iterative diversity speed of the swarm optimization algorithm; The electrical variable re-output layer is used to update all the power box electrical variable data items D i ' and the original real-time power box electrical variable data item D i Combine and obtain updated real-time power box electrical variable set D'; Step S3: Dynamic threshold division of electrical variables Based on the updated real-time power box electric variable set D' and the power box electric variable risk quantification model, threshold judgment is performed to obtain the electric variable risk quantification result; the power box electric variable risk quantification model includes an electric variable data sub-item calculation layer, a quantitative risk assessment layer and an electric variable risk quantification output layer; based on the electric variable risk quantification result and the dynamic power box electric variable threshold judgment standard, an assessment is performed to obtain the real-time power box electric variable monitoring result, and the dynamic power box electric variable threshold judgment standard is determined by the real-time environmental variables and the electric variable interference prediction results; subsequent operations are performed based on the real-time power box electric variable monitoring results.

[0005] As a preferred technical solution of the present invention, the power box environment prediction and judgment model in step S1 includes an environmental variable processing layer, an environmental feature analysis layer, an environmental prediction and judgment layer, an environmental auxiliary judgment layer and a result output layer; The environment variable processing layer is used to preprocess the real-time environment variables to obtain preprocessed real-time environment variables; the preprocessed real-time environment variables include several preprocessed environment sub-item variables; The environmental feature analysis layer is used to perform cyclic feature segmentation on the preprocessed real-time environmental variables using the random forest algorithm until the environmental variable strong correlation features and environmental variable weak correlation features are obtained; The specific steps for performing cyclic feature segmentation include: Step A1: Calculate the Gini index of each preprocessed environmental sub-item variable in the preprocessed real-time environmental variable to obtain the initial feature importance value of each environmental sub-item variable; Step A2: Perform confidence-weighted allocation on the initial feature importance value of each environmental sub-item variable according to the corresponding pre-processed environmental sub-item variable to obtain the reallocated feature importance value of each environmental sub-item variable; Step A3: randomly remove the pre-processed environment sub-item variable with the lowest feature importance value from the current environment sub-item variable redistribution feature set according to the cyclic removal feature set to obtain the remaining pre-processed environment sub-item variables, and record the performance parameters of the current environment sub-item variable; Step A4: If the decrease in the performance parameter of the current environmental sub-item variable is within the preset performance range, the removed pre-processed environmental sub-item variable is classified as a weakly associated feature of the environmental variable, and the process proceeds to step A1 based on the remaining pre-processed environmental sub-item variables; otherwise, the process returns to step A3; Repeat steps A1 to A4 until all preprocessing environment sub-item variables are divided into environment variable strong correlation features and environment variable weak correlation features; the environment variable strong correlation features are other preprocessing environment sub-item variables excluding the environment variable weak correlation features; The environmental prediction and judgment layer is used to perform prediction analysis based on the strong correlation characteristics of environmental variables to obtain the prediction results of strong interference of electrical variables; The environment auxiliary judgment layer is used to perform auxiliary analysis based on the weak correlation characteristics of environmental variables to obtain the prediction results of weak interference of electrical variables; The result output layer is used to perform feature fusion on the prediction results of strong interference of electric variables and the prediction results of weak interference of electric variables to obtain the prediction results of electric variable interference.

[0006] As a preferred technical solution of the present invention, the specific steps of training the environment prediction judgment layer and the environment auxiliary judgment layer include: Collecting several groups of environmental variable feature training samples, each group of environmental variable feature training samples contains environmental variable features and corresponding annotated label values; combining several groups of environmental variable feature training samples to obtain an environmental variable feature training set; The specific steps of the training environment prediction and judgment layer are: Based on the label requirements of the environment prediction and judgment layer, the environment variable feature training set is updated to obtain the environment prediction and judgment training set; the environment prediction and judgment training set is used to perform model training on the environment prediction and judgment layer to obtain the initial environment prediction and judgment layer; the initial environment prediction and judgment layer is model evaluated to obtain the initial environment prediction and judgment layer model evaluation result; if the initial environment prediction and judgment layer model evaluation result is passed, the initial environment prediction and judgment layer is used as the environment prediction and judgment layer in the power box environment prediction and judgment model; otherwise, the environment variable feature training set is used to continue model training; Specific steps for training the environment-assisted judgment layer: Based on the label requirements of the environment auxiliary judgment layer, the environmental variable feature training set is updated to obtain the environment auxiliary judgment training set; the environment auxiliary judgment layer is trained with the environment auxiliary judgment training set to obtain the initial environment auxiliary judgment layer; the model of the initial environment auxiliary judgment layer is evaluated to obtain the model evaluation result of the initial environment auxiliary judgment layer; if the model evaluation result of the initial environment auxiliary judgment layer is passed, the initial environment auxiliary judgment layer is used as the environment auxiliary judgment layer in the power box environment prediction judgment model; otherwise, the model training is continued with the environment auxiliary judgment training set.

[0007] As a preferred technical solution of the present invention, the specific steps of calculating the disturbance electric variable calculation layer in the power box electric variable disturbance judgment model in step S2 include: For real-time power box electrical variable data item D i Calculate the wavelet coefficients of the maximum stacked discrete wavelet transform and obtain the wavelet coefficients B to be analyzed i ; The wavelet coefficient B to be analyzed i Calculate the wavelet energy and obtain the energy E of the wavelet coefficient to be analyzed i ; Calculate the absolute value of the energy difference in the wavelet coefficient energy Ei to be analyzed and obtain the electrical variable data item feature T i ; If the electrical variable data item characteristic T i If the electric variable fluctuation range is exceeded, the real-time power box electric variable data item D i As the disturbance electrical variable data item R n ; Otherwise, retain the real-time power box electrical variable data item D i unchanged, as the original real-time power box electrical variable data item D i .

[0008] As a preferred technical solution of the present invention, the specific steps of constructing the electric variable disturbance compensation function include: Collecting a number of verified electric variable disturbance training samples; combining the number of verified electric variable disturbance training samples to obtain an electric variable disturbance compensation training set; Construct K electrical variable disturbance compensation function individuals H based on swarm optimization algorithm k , each electrical variable disturbance compensation function individual H k is a set of solutions for calculating the function of electrical variable disturbance compensation; the K electrical variable disturbance compensation function individuals H k Combination, get the iterative population of electrical variable disturbance compensation; set the maximum number of iterations; electrical variable disturbance compensation function individual H k The fitness of k ; Calculate the fitness Y k Specific steps: Using the electrical variable disturbance compensation function individual H k The electric variable disturbance compensation training set is calculated to obtain the electric variable compensation accuracy; the electric variable compensation accuracy is used as the fitness Y k ; Introduce population iterative learning factor Z1 and population iterative learning factor Z2; set Z1=Z max +(Z max -Z min )*sin[(P*Π) / P max ],Z2=Z max -(Z max -Z min )*cos[(P*Π) / P max ]; among them, Z max =2*q,Z min =q, q is the preset parameter, P is the current number of iterations, P max is the maximum number of iterations; In the iterative population iteration process of the electric variable disturbance compensation, the electric variable disturbance compensation function individual H is calculated based on the population iteration learning factor Z1 and the population iteration learning factor Z2. k Conduct a search update; When the maximum number of iterations is reached, the individual electric variable disturbance compensation function corresponding to the maximum output fitness is the optimal individual electric variable disturbance compensation function; the electric variable disturbance compensation function is constructed based on the optimal individual electric variable disturbance compensation function.

[0009] As a preferred technical solution of the present invention, the specific steps of determining the threshold judgment standard of the electric variable of the dynamic power supply box in step S3 include: Set the basic threshold judgment standard of electric variables; perform feature extraction based on real-time environmental variables and electric variable interference prediction results to obtain the electric variable interference label value; match the electric variable threshold dynamic change library based on the electric variable interference label value to obtain the electric variable dynamic threshold adjustment factor; adjust the basic threshold judgment standard of electric variables according to the dynamic threshold adjustment factor of electric variables to obtain the dynamic power supply box electric variable threshold judgment standard.

[0010] As a preferred technical solution of the present invention, the power box electric variable risk quantification model includes an electric variable data item calculation layer, a quantitative risk assessment layer and an electric variable risk quantification output layer; The electric variable data sub-item calculation layer is used to compare the updated real-time power box electric variable set D' with the basic parameter standard of the power box electric variable to obtain the power box electric variable difference parameter C i ; The quantitative risk assessment layer is used to calculate the power box electrical variable difference parameter C i Match the risk labels to obtain the power box electrical variable risk quantification value L i ; The quantitative risk assessment layer is constructed based on the BP neural network, and the electric variable difference risk matching training set is collected; the BP neural network model is trained using the electric variable difference risk matching training set to obtain the quantitative risk assessment layer; The output layer of the electrical variable risk quantification is used to quantify the electrical variable risk value L of all power boxes i Combine them to obtain the quantification results of electrical variable risk.

[0011] An electric variable environmental disturbance judgment module includes a data acquisition unit and an environmental disturbance judgment unit; The data acquisition unit is used to acquire the real-time environmental variables and the real-time power box electrical variable set D at the monitoring initial frequency, D={D i |i=1,2,…,I};D i D is the real-time power box electrical variable data item i , I is the total number of types of data items in the power box electrical variable concentration; The environmental disturbance judgment unit is used to input the real-time environmental variables into the power box environmental prediction judgment model for analysis to obtain the power variable interference prediction results; the power box environmental prediction judgment model includes an environmental variable processing layer, an environmental feature analysis layer, an environmental prediction judgment layer, an environmental auxiliary judgment layer and a result output layer. The environmental feature analysis layer is constructed based on the improvement of the random forest algorithm and is used to predict and analyze the real-time environmental variables; An electric variable disturbance compensation updating module includes a compensation analysis unit; The compensation sub-item unit is used to judge the real-time power supply box electric variable set D based on the electric variable interference prediction result and the power supply box electric variable disturbance judgment model, and obtain the updated real-time power supply box electric variable set D', D'={D i '|i=1, 2, ..., I}; The power box electrical variable disturbance judgment model is improved and constructed based on wavelet transform and swarm optimization algorithm; The electric variable dynamic threshold division module includes a risk quantification unit and a threshold judgment unit; The risk quantification unit is used to perform threshold judgment based on the updated real-time power box electric variable set D' and the power box electric variable risk quantification model to obtain the electric variable risk quantification result; the power box electric variable risk quantification model includes an electric variable data sub-item calculation layer, a quantitative risk assessment layer and an electric variable risk quantification output layer; The threshold judgment unit is used to evaluate based on the electric variable risk quantification results and the dynamic power box electric variable threshold judgment standard to obtain the real-time power box electric variable monitoring results. The dynamic power box electric variable threshold judgment standard is determined by the real-time environmental variables and the electric variable interference prediction results; subsequent operations are performed based on the real-time power box electric variable monitoring results.

[0012] The present invention has the following advantages: 1. The present invention realizes accurate prediction and compensation of electric variables by real-time monitoring and analysis of environmental variables and electric variables of power boxes, and utilizes improved random forest algorithm and swarm optimization algorithm. It can effectively identify and compensate for the influence of environmental disturbance on electric variables, and improve the accuracy and reliability of electric variable monitoring of power boxes. At the same time, through dynamic threshold division, it can adjust the monitoring standard according to the real-time environmental variables and electric variable interference prediction results, realize quantitative assessment of electric variable risks, thereby optimizing the operating state of the power box and ensuring the safety and stability of the power system.

[0013] 2. The present invention conducts in-depth analysis through a multi-level power box environment prediction and judgment model. The feature analysis layer adopts an improved random forest algorithm. By calculating the Gini index and confidence weighted distribution, it effectively screens out strong correlation features, eliminates noise interference, and improves the accuracy and robustness of the model; the iterative feature selection method of cyclic feature segmentation further streamlines the feature set to ensure the high correlation of features; the environmental prediction and judgment layer and the environmental auxiliary judgment layer analyze the strong correlation and weak correlation features respectively, and finally the prediction results are integrated through the result output layer to obtain a comprehensive prediction of electrical variable interference, thereby improving the accuracy and reliability of electrical variable monitoring of the power box.

[0014] 3. The present invention realizes accurate disturbance detection and compensation of the real-time power box electrical variable set by combining wavelet transform and swarm optimization algorithm; the introduction of wavelet transform enhances the model's time-frequency analysis capability of signals, while the improved swarm optimization algorithm improves the computational efficiency and accuracy of the disturbance compensation function, and can more effectively identify and compensate for electrical variable disturbances caused by environmental variables, thereby improving the stability and reliability of the electrical variable data of the power box; through the combination of electrical variable re-output layers, the updated real-time power box electrical variable set D' can more accurately reflect the actual operating status of the power box, providing a solid data foundation for subsequent electrical variable risk quantification and monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1A schematic diagram of the structure of a three-stage power supply box electric variable monitoring system adopted in an embodiment of the present invention.

[0016] Figure 2 A flow chart of a method for monitoring electrical variables of a three-stage power supply box adopted in an embodiment of the present invention. DETAILED DESCRIPTION

[0017] In order to enable persons skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0018] Example 1, a method for monitoring electric variables of a three-level power supply box, see Figure 2 As shown, the following steps are included: Step S1: Determination of electrical variable environmental disturbance Obtain the real-time environmental variables and the real-time power box electrical variable set D by monitoring the initial frequency, D={D i |i=1, 2, …, I}; D i D is the real-time power box electrical variable data item i , I is the total number of types of data items in the power box electrical variable set; the real-time environmental variables are input into the power box environmental prediction and judgment model for analysis to obtain the electrical variable interference prediction results; the power box environmental prediction and judgment model includes an environmental variable processing layer, an environmental feature analysis layer, an environmental prediction and judgment layer, an environmental auxiliary judgment layer and a result output layer. The environmental feature analysis layer is constructed based on the improvement of the random forest algorithm and is used to predict and analyze the real-time environmental variables; The initial monitoring frequency is set by professional technicians according to the actual situation, and the real-time environmental variables and real-time power box electrical variables are acquired by special sensors installed on the three-level power box; The power box environment prediction and judgment model in step S1 includes an environment variable processing layer, an environment feature analysis layer, an environment prediction and judgment layer, an environment auxiliary judgment layer and a result output layer; The environment variable processing layer is used to preprocess the real-time environment variables to obtain preprocessed real-time environment variables; the preprocessed real-time environment variables include several preprocessed environment sub-item variables; The environmental feature analysis layer is used to perform cyclic feature segmentation on the preprocessed real-time environmental variables using the random forest algorithm until the environmental variable strong correlation features and environmental variable weak correlation features are obtained; The specific steps for performing cyclic feature segmentation include: Step A1: Calculate the Gini index of each preprocessed environmental sub-item variable in the preprocessed real-time environmental variable to obtain the initial feature importance value of each environmental sub-item variable; Step A2: Perform confidence-weighted allocation on the initial feature importance value of each environmental sub-item variable according to the corresponding pre-processed environmental sub-item variable to obtain the reallocated feature importance value of each environmental sub-item variable; Step A3: randomly remove the pre-processing environment sub-item variable with the lowest feature importance value of the current environment sub-item variable redistribution feature according to the cyclic removal feature set, obtain the remaining pre-processing environment sub-item variable, and record the performance parameters of the current environment sub-item variable; the cyclic removal feature set is a numerical value of the pre-processing environment sub-item variable randomly removed, for example, randomly selected within the range of [1, G], and the size of G is set by professional and technical personnel according to actual conditions; Step A4: If the decrease in the performance parameter of the current environmental sub-item variable is within the preset performance range, the removed pre-processed environmental sub-item variable is classified as a weakly associated feature of the environmental variable, and the process proceeds to step A1 based on the remaining pre-processed environmental sub-item variables; otherwise, the process returns to step A3; Repeat steps A1 to A4 until all preprocessing environment sub-item variables are divided into environment variable strong correlation features and environment variable weak correlation features; the environment variable strong correlation features are other preprocessing environment sub-item variables excluding the environment variable weak correlation features; By calculating the Gini index of each preprocessed environmental variable, variables with high feature importance can be preliminarily screened out, and unimportant features can be eliminated, which can ensure that the features that the algorithm focuses on are highly representative and reduce noise interference; in step A2, by weighting the initial feature importance value with confidence, the contribution of each variable to the model can be more accurately evaluated. This dynamic weighting method can better reflect the fluctuation of real-time environmental data and improve the robustness and adaptability of the model; steps A3 and A4 of cyclically removing features can ensure that variables with low feature importance are gradually eliminated without affecting the overall performance of the model. This iterative feature selection method ensures the streamlining of the feature set, so that the features finally retained have a high correlation with the prediction target, that is, the strong correlation features of environmental variables; The environmental prediction and judgment layer is used to perform prediction analysis based on the strong correlation characteristics of environmental variables to obtain the prediction results of strong interference of electrical variables; The environment auxiliary judgment layer is used to perform auxiliary analysis based on the weak correlation characteristics of environmental variables to obtain the prediction results of weak interference of electrical variables; The result output layer is used to fuse the features of the prediction results of strong interference of electric variables and the prediction results of weak interference of electric variables to obtain the prediction results of electric variable interference; By using the improved random forest algorithm in the environmental feature analysis layer, the model can accurately extract environmental features that are strongly correlated with the power box electrical variables, forming "environmental variable strong correlation features", which can help the model identify the key features in the environment that really affect the electrical variables, making the prediction results more explanatory and targeted; as the importance of environmental variable features is gradually allocated and screened, weakly correlated features and unimportant features are removed or classified as auxiliary features; by reducing noise, the negative impact of weakly correlated features on model prediction is reduced, the robustness and stability of the model are improved, and the prediction results are more reliable; the model structure adopts a multi-level analysis: the strongly correlated features of environmental variables are mainly analyzed in the environmental prediction judgment layer, while the weakly correlated features are supplemented in the auxiliary judgment layer. Through this hierarchical structure, the model can more comprehensively capture the impact of environmental changes on electrical variables, making the prediction results more accurate; through real-time monitoring and dynamic input of environmental variable data by sensors, combined with the Gini index and dynamic weighted allocation method, the model can adapt to the fluctuation of environmental variables at any time. The model can quickly adjust when facing different environmental conditions, and its adaptability to the real-time environment is enhanced; The specific steps of training the environment prediction judgment layer and the environment auxiliary judgment layer include: Collecting several groups of environmental variable feature training samples, each group of environmental variable feature training samples contains environmental variable features and corresponding annotated label values; combining several groups of environmental variable feature training samples to obtain an environmental variable feature training set; The specific steps of the training environment prediction and judgment layer are: Based on the label requirements of the environment prediction and judgment layer, the environment variable feature training set is updated to obtain the environment prediction and judgment training set; the environment prediction and judgment training set is used to perform model training on the environment prediction and judgment layer to obtain the initial environment prediction and judgment layer; the initial environment prediction and judgment layer is model evaluated to obtain the initial environment prediction and judgment layer model evaluation result; if the initial environment prediction and judgment layer model evaluation result is passed, the initial environment prediction and judgment layer is used as the environment prediction and judgment layer in the power box environment prediction and judgment model; otherwise, the environment variable feature training set is used to continue model training; Specific steps for training the environment-assisted judgment layer: Based on the label requirements of the environment auxiliary judgment layer, the environment variable feature training set is updated to obtain the environment auxiliary judgment training set; the environment auxiliary judgment layer is trained with the environment auxiliary judgment training set to obtain the initial environment auxiliary judgment layer; the model of the initial environment auxiliary judgment layer is evaluated to obtain the model evaluation result of the initial environment auxiliary judgment layer; if the model evaluation result of the initial environment auxiliary judgment layer is passed, the initial environment auxiliary judgment layer is used as the environment auxiliary judgment layer in the power box environment prediction judgment model; otherwise, the model training is continued with the environment auxiliary judgment training set; The label requirements of the environmental prediction and judgment layer and the environmental auxiliary judgment layer are set by professional technicians according to actual conditions. The accuracy and applicability of the model are ensured by iteratively updating and training the environmental prediction and judgment layer and the environmental auxiliary judgment layer. By updating the training set based on label requirements, the model can more accurately capture the characteristics of environmental variables and improve the accuracy of prediction, ensuring the efficiency and accuracy of the power box environment prediction and judgment model, and providing strong support for the power box electrical variable monitoring. Step S2: Electrical variable disturbance compensation update Based on the electric variable interference prediction results and the power box electric variable disturbance judgment model, the real-time power box electric variable set D is judged to obtain the updated real-time power box electric variable set D', D'={D i '|i=1, 2, ..., I}; The power box electrical variable disturbance judgment model is improved and constructed based on wavelet transform and swarm optimization algorithm; The power box electric variable disturbance judgment model includes a disturbance electric variable calculation layer, a disturbance compensation analysis layer and an electric variable re-output layer; The disturbance electric variable calculation layer is used to perform disturbance detection on all real-time power box electric variable data items Di in the real-time power box electric variable set D, and obtain the disturbance electric variable data item R n , n=1, 2, ..., N; N is all real-time power box electrical variable data items D i The total number of data items that are disturbed by environmental variables; The interference compensation analysis layer is used to predict the disturbance of the electric variable based on the disturbance electric variable data item R n The electric variable compensation calculation is performed with the electric variable disturbance compensation function to obtain the compensated electric variable data item R n ', the compensation electric variable data item R n 'Corresponding real-time power box electrical variable data item D i Match and get the updated power box electrical variable data item D i '; The electric variable disturbance compensation function is constructed based on the swarm optimization algorithm after improving the iterative diversity speed of the swarm optimization algorithm; The electrical variable re-output layer is used to update all the power box electrical variable data items D i ' and the original real-time power box electrical variable data item D i Combine and obtain updated real-time power box electrical variable set D'; Through the application of the electric variable disturbance judgment model, the model can identify the electric variable data items affected by environmental interference and compensate for them; this compensation mechanism ensures that the stability of the electric variable data can be maintained under the interference of external environmental factors, reduces the abnormal fluctuations of the data caused by interference, and improves the reliability of the monitoring data; the interference compensation analysis layer combines wavelet transform and improved group optimization algorithm, calculates the compensated electric variable data items through the electric variable interference prediction results and the disturbance compensation function. This precise compensation strategy can effectively correct the errors caused by environmental interference and ensure the accuracy of the data, thereby providing reliable information for subsequent monitoring and decision-making. Basis: Through the improved swarm optimization algorithm, the model can dynamically adjust the compensation parameters to adapt to different environmental interference conditions. This dynamic adaptability enables the compensation mechanism to respond quickly to environmental changes, effectively improving the system's response speed and processing efficiency to real-time electrical variable disturbances. The electrical variable disturbance judgment model is built based on wavelet transform and swarm optimization algorithm, which can adapt to various types of environmental interference and maintain good performance in different power box application scenarios. By extracting characteristic signals and interference features through wavelet transform and combining the fast convergence characteristics of the swarm optimization algorithm, the model has a high generalization ability and can operate robustly in complex and changing environments. The specific steps of calculating the disturbance electric variable calculation layer in the power box electric variable disturbance judgment model in step S2 include: For real-time power box electrical variable data item D i Calculate the wavelet coefficients of the maximum stacked discrete wavelet transform and obtain the wavelet coefficients B to be analyzed i ; The wavelet coefficient B to be analyzed i Calculate the wavelet energy and obtain the energy E of the wavelet coefficient to be analyzed i ; Calculate the absolute value of the energy difference in the wavelet coefficient energy Ei to be analyzed and obtain the electrical variable data item feature T i ; If the electrical variable data item characteristic T i If the electric variable fluctuation range is exceeded, the real-time power box electric variable data item D i As the disturbance electrical variable data item R n ; Otherwise, retain the real-time power box electrical variable data item D i unchanged, as the original real-time power box electrical variable data item D i ; The preset electric variable fluctuation range is set by professional technicians according to actual conditions; By extracting wavelet coefficients through maximum stacked discrete wavelet transform, the model can accurately capture subtle fluctuations and disturbance information from the real-time power supply box electrical variable data; wavelet transform can decompose electrical variable signals into different frequency scales, thereby identifying characteristic changes in different frequency bands; by analyzing wavelet energy and energy difference, the model can better cope with interference of different frequencies and intensities, making it more adaptable and responsive to various environmental changes; by calculating the energy of wavelet coefficients and analyzing the absolute value of energy difference, the model can keenly detect abnormal disturbances through energy changes. The electrical variable data item feature is based on the absolute value of energy difference, which can help the model quickly identify and filter out disturbance data items, especially for identifying sudden and short-term interference signals; The specific steps of constructing the electrical variable disturbance compensation function include: Collecting a number of verified electric variable disturbance training samples; combining the number of verified electric variable disturbance training samples to obtain an electric variable disturbance compensation training set; Construct K electrical variable disturbance compensation function individuals H based on swarm optimization algorithm k , each electrical variable disturbance compensation function individual H k is a set of solutions for calculating the function of electrical variable disturbance compensation; the K electrical variable disturbance compensation function individuals H k Combination, get the iterative population of electrical variable disturbance compensation; set the maximum number of iterations; electrical variable disturbance compensation function individual H k The fitness of k ;The maximum number of iterations is set by professional technicians based on actual conditions; Calculate the fitness Y k Specific steps: Using the electrical variable disturbance compensation function individual H k The electric variable disturbance compensation training set is calculated to obtain the electric variable compensation accuracy; the electric variable compensation accuracy is used as the fitness Y k ; Introduce population iterative learning factor Z1 and population iterative learning factor Z2; set Z1=Z max +(Z max -Z min )*sin[(P*Π) / P max ],Z2=Z max -(Z max -Z min )*cos[(P*Π) / P max ]; among them, Z max =2*q,Z min =q, q is the preset parameter, P is the current number of iterations, P max is the maximum number of iterations; q is a constant, for example, it can be 1; In the iterative population iteration process of the electric variable disturbance compensation, the electric variable disturbance compensation function individual H is calculated based on the population iteration learning factor Z1 and the population iteration learning factor Z2. k Conduct a search update; When the maximum number of iterations is reached, the individual electric variable disturbance compensation function corresponding to the maximum output fitness is the optimal individual electric variable disturbance compensation function; the electric variable disturbance compensation function is constructed based on the optimal individual electric variable disturbance compensation function; By using verified electrical variable disturbance training samples to construct a disturbance compensation training set, the authenticity and representativeness of the data can be ensured, thereby improving the accuracy of the compensation function; by introducing a population iteration learning factor, the model can dynamically adjust the search process according to the current iteration state; the learning factor adjusts the weight as the number of iterations changes, so that the algorithm searches more widely in the early stage and focuses more on the optimal area in the later stage, thereby improving the search efficiency and adaptability and further optimizing the compensation function; through multiple iterative updates and adaptive adjustments, the disturbance compensation function is gradually optimized and tends to the optimal solution during the iteration; adaptive optimization enables the compensation function to not only cope with electrical variable disturbances of different degrees, but also maintain efficient and stable compensation effects under different environments, thereby enhancing the robustness of the model; Step S3: Dynamic threshold division of electrical variables Based on the updated real-time power box electric variable set D' and the power box electric variable risk quantification model, the threshold judgment is performed to obtain the electric variable risk quantification result; the power box electric variable risk quantification model includes an electric variable data sub-item calculation layer, a quantitative risk assessment layer and an electric variable risk quantification output layer; based on the electric variable risk quantification result and the dynamic power box electric variable threshold judgment standard, the evaluation is performed to obtain the real-time power box electric variable monitoring result, and the dynamic power box electric variable threshold judgment standard is determined by the real-time environmental variables and the electric variable interference prediction results; the subsequent operations are performed based on the real-time power box electric variable monitoring results; The specific steps of determining the threshold judgment standard of the dynamic power supply box electric variable in step S3 include: Set the basic threshold judgment standard for electric variables; perform feature extraction based on real-time environmental variables and electric variable interference prediction results to obtain the electric variable interference label value; perform matching in the electric variable threshold dynamic change library based on the electric variable interference label value to obtain the electric variable dynamic threshold adjustment ratio; adjust the basic threshold judgment standard for electric variables according to the electric variable dynamic threshold adjustment ratio to obtain the dynamic power supply box electric variable threshold judgment standard; the electric variable threshold dynamic change library is set by professional and technical personnel based on actual conditions and is set based on environmental conditions at the initial stage of building a three-level power supply box; The power box electric variable risk quantification model includes an electric variable data item calculation layer, a quantitative risk assessment layer and an electric variable risk quantification output layer; The electric variable data sub-item calculation layer is used to compare the updated real-time power box electric variable set D' with the basic parameter standard of the power box electric variable to obtain the power box electric variable difference parameter C i ; The quantitative risk assessment layer is used to calculate the power box electrical variable difference parameter C i Match the risk labels to obtain the power box electrical variable risk quantification value L i ; The quantitative risk assessment layer is constructed based on the BP neural network, and the electric variable difference risk matching training set is collected; the BP neural network model is trained using the electric variable difference risk matching training set to obtain the quantitative risk assessment layer; The output layer of the electrical variable risk quantification is used to quantify the electrical variable risk value L of all power boxes i Combine them to obtain the quantification result of the risk of electrical variables; By combining real-time environmental variables and electrical variable interference prediction results for feature extraction, electrical variable interference label values ​​are generated; based on the electrical variable threshold dynamic change library, the electrical variable threshold is dynamically adjusted, so that the monitoring system can adapt to the electrical variable fluctuations under different environmental conditions in real time, ensure the accuracy of the threshold, and improve the accuracy of the monitoring results; the quantitative risk assessment layer constructed by combining the electrical variable difference parameters with the BP neural network can learn the risk characteristics of different electrical variable differences through training data to achieve intelligent risk matching; the self-learning ability of the quantitative risk assessment layer enables the system to adjust the risk assessment standards in real time, so as to respond quickly when risks occur and improve the intelligence of risk monitoring; the construction of the dynamic power supply box electrical variable threshold judgment standard is highly automated, relying on the real-time data of environmental variables and interference prediction results, without the need for frequent manual settings, reducing manual intervention, reducing the operation and maintenance burden, and enabling the system to maintain stable and effective operation in a changing environment.

[0019] Example 2, a three-level power supply box electric variable monitoring system, see Figure 1 As shown, including: An electric variable environmental disturbance judgment module includes a data acquisition unit and an environmental disturbance judgment unit; The data acquisition unit is used to acquire the real-time environmental variables and the real-time power box electrical variable set D at the monitoring initial frequency, D={D i |i=1, 2, …, I}; D i D is the real-time power box electrical variable data item i , I is the total number of types of data items in the power box electrical variable concentration; The environmental disturbance judgment unit is used to input the real-time environmental variables into the power box environmental prediction judgment model for analysis to obtain the power variable interference prediction results; the power box environmental prediction judgment model includes an environmental variable processing layer, an environmental feature analysis layer, an environmental prediction judgment layer, an environmental auxiliary judgment layer and a result output layer. The environmental feature analysis layer is constructed based on the improvement of the random forest algorithm and is used to predict and analyze the real-time environmental variables; An electric variable disturbance compensation updating module includes a compensation analysis unit; The compensation sub-item unit is used to judge the real-time power supply box electric variable set D based on the electric variable interference prediction result and the power supply box electric variable disturbance judgment model, and obtain the updated real-time power supply box electric variable set D', D'={D i '|i=1, 2, ..., I}; The power box electrical variable disturbance judgment model is improved and constructed based on wavelet transform and swarm optimization algorithm; The electric variable dynamic threshold division module includes a risk quantification unit and a threshold judgment unit; The risk quantification unit is used to perform threshold judgment based on the updated real-time power box electric variable set D' and the power box electric variable risk quantification model to obtain the electric variable risk quantification result; the power box electric variable risk quantification model includes an electric variable data sub-item calculation layer, a quantitative risk assessment layer and an electric variable risk quantification output layer; The threshold judgment unit is used to evaluate based on the electric variable risk quantification results and the dynamic power box electric variable threshold judgment standard to obtain the real-time power box electric variable monitoring results. The dynamic power box electric variable threshold judgment standard is determined by the real-time environmental variables and the electric variable interference prediction results; subsequent operations are performed based on the real-time power box electric variable monitoring results.

[0020] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all such improvements and changes should fall within the scope of protection of the appended claims of the present invention. Parts not described in detail in this specification belong to the prior art known to those skilled in the art.

Claims

1. A method for monitoring electric variables of a three-level power supply box, characterized in that: The following steps are involved: Step S1: Determination of electrical variable environmental disturbance Obtain the real-time environmental variables and the real-time power box electrical variable set D by monitoring the initial frequency, D={D i |i=1,2,…,I}; D i D is the real-time power box electrical variable data item i , I is the total number of types of data items in the power box electrical variable set; the real-time environmental variables are input into the power box environmental prediction and judgment model for analysis to obtain the electrical variable interference prediction results; the power box environmental prediction and judgment model includes an environmental variable processing layer, an environmental feature analysis layer, an environmental prediction and judgment layer, an environmental auxiliary judgment layer and a result output layer. The environmental feature analysis layer is constructed based on the improvement of the random forest algorithm and is used to predict and analyze the real-time environmental variables; Step S2: Update of electrical variable disturbance compensation Based on the electric variable interference prediction results and the power box electric variable disturbance judgment model, the real-time power box electric variable set D is judged to obtain the updated real-time power box electric variable set D', D'={D i '|i=1,2,…,I}; The power box electrical variable disturbance judgment model is improved and constructed based on wavelet transform and swarm optimization algorithm; The power box electric variable disturbance judgment model includes a disturbance electric variable calculation layer, a disturbance compensation analysis layer and an electric variable re-output layer; The disturbance electric variable calculation layer is used to perform disturbance detection on all real-time power box electric variable data items Di in the real-time power box electric variable set D, and obtain the disturbance electric variable data item R n , n=1, 2, ..., N; N is all real-time power box electrical variable data items D i The total number of data items that are disturbed by environmental variables; The interference compensation analysis layer is used to predict the disturbance of the electric variable based on the disturbance electric variable data item R n The electric variable compensation calculation is performed with the electric variable disturbance compensation function to obtain the compensated electric variable data item R n ', the compensation electric variable data item R n 'Corresponding to the real-time power supply box electrical variable data item D i Match and get the updated power box electrical variable data item D i '; The electric variable disturbance compensation function is constructed based on the swarm optimization algorithm after improving the iterative diversity speed of the swarm optimization algorithm; The electrical variable re-output layer is used to update all the power box electrical variable data items D i ' and the original real-time power box electrical variable data item D i Combine and obtain updated real-time power box electrical variable set D'; Step S3: Dynamic threshold division of electrical variables Based on the updated real-time power box electric variable set D' and the power box electric variable risk quantification model, the threshold value is judged to obtain the electric variable risk quantification result; The power box electric variable risk quantification model includes an electric variable data item calculation layer, a quantitative risk assessment layer and an electric variable risk quantification output layer; An assessment is performed based on the electric variable risk quantification results and the dynamic power box electric variable threshold judgment standard to obtain the real-time power box electric variable monitoring results. The dynamic power box electric variable threshold judgment standard is determined by the real-time environmental variables and electric variable interference prediction results; subsequent operations are performed based on the real-time power box electric variable monitoring results.

2. A method for monitoring electric variables of a three-level power supply box according to claim 1, characterized in that: The power box environment prediction and judgment model in step S1 includes an environment variable processing layer, an environment feature analysis layer, an environment prediction and judgment layer, an environment auxiliary judgment layer and a result output layer; The environment variable processing layer is used to preprocess the real-time environment variables to obtain preprocessed real-time environment variables; The preprocessing real-time environment variables include several preprocessing environment sub-item variables; The environmental feature analysis layer is used to perform cyclic feature segmentation on the preprocessed real-time environmental variables using the random forest algorithm until the environmental variable strong correlation features and environmental variable weak correlation features are obtained; The specific steps for performing cyclic feature segmentation include: Step A1: Calculate the Gini index of each preprocessed environmental sub-item variable in the preprocessed real-time environmental variable to obtain the initial feature importance value of each environmental sub-item variable; Step A2: Perform confidence-weighted allocation on the initial feature importance value of each environmental sub-item variable according to the corresponding pre-processed environmental sub-item variable to obtain the reallocated feature importance value of each environmental sub-item variable; Step A3: randomly remove the pre-processed environment sub-item variable with the lowest feature importance value from the current environment sub-item variable redistribution feature set according to the cyclic removal feature set to obtain the remaining pre-processed environment sub-item variables, and record the performance parameters of the current environment sub-item variable; Step A4: If the decrease in the performance parameter of the current environmental sub-item variable is within the preset performance range, the removed pre-processed environmental sub-item variable is classified as a weakly associated feature of the environmental variable, and the process proceeds to step A1 based on the remaining pre-processed environmental sub-item variables; otherwise, the process returns to step A3; Repeat steps A1 to A4 until all preprocessing environment sub-item variables are divided into environment variable strong correlation features and environment variable weak correlation features; the environment variable strong correlation features are other preprocessing environment sub-item variables excluding the environment variable weak correlation features; The environmental prediction and judgment layer is used to perform prediction analysis based on the strong correlation characteristics of environmental variables to obtain the prediction results of strong interference of electrical variables; The environment auxiliary judgment layer is used to perform auxiliary analysis based on the weak correlation characteristics of environmental variables to obtain the prediction results of weak interference of electrical variables; The result output layer is used to perform feature fusion on the prediction results of strong interference of electric variables and the prediction results of weak interference of electric variables to obtain the prediction results of electric variable interference.

3. A method for monitoring electric variables of a three-level power supply box according to claim 2, characterized in that: The specific steps of training the environment prediction judgment layer and the environment auxiliary judgment layer include: Collecting several groups of environmental variable feature training samples, each group of environmental variable feature training samples contains environmental variable features and corresponding annotated label values; combining several groups of environmental variable feature training samples to obtain an environmental variable feature training set; The specific steps of the training environment prediction and judgment layer are: Based on the label requirements of the environment prediction and judgment layer, the environment variable feature training set is updated to obtain the environment prediction and judgment training set; the environment prediction and judgment training set is used to perform model training on the environment prediction and judgment layer to obtain the initial environment prediction and judgment layer; the initial environment prediction and judgment layer is model evaluated to obtain the initial environment prediction and judgment layer model evaluation result; if the initial environment prediction and judgment layer model evaluation result is passed, the initial environment prediction and judgment layer is used as the environment prediction and judgment layer in the power box environment prediction and judgment model; otherwise, the environment variable feature training set is used to continue model training; Specific steps for training the environment-assisted judgment layer: Based on the label requirements of the environment auxiliary judgment layer, the environmental variable feature training set is updated to obtain the environment auxiliary judgment training set; the environment auxiliary judgment layer is trained with the environment auxiliary judgment training set to obtain the initial environment auxiliary judgment layer; the model of the initial environment auxiliary judgment layer is evaluated to obtain the model evaluation result of the initial environment auxiliary judgment layer; if the model evaluation result of the initial environment auxiliary judgment layer is passed, the initial environment auxiliary judgment layer is used as the environment auxiliary judgment layer in the power box environment prediction judgment model; otherwise, the model training is continued with the environment auxiliary judgment training set.

4. A method for monitoring electric variables of a three-level power supply box according to claim 3, characterized in that: The specific steps of calculating the disturbance electric variable calculation layer in the power box electric variable disturbance judgment model in step S2 include: For real-time power box electrical variable data item D i Calculate the wavelet coefficients of the maximum stacked discrete wavelet transform and obtain the wavelet coefficients B to be analyzed i ; The wavelet coefficient B to be analyzed i Calculate the wavelet energy and obtain the energy E of the wavelet coefficient to be analyzed i ; Calculate the absolute value of the energy difference in the wavelet coefficient energy Ei to be analyzed and obtain the electrical variable data item feature T i ; If the electrical variable data item characteristic T i If the electric variable fluctuation range is exceeded, the real-time power box electric variable data item D i As the disturbance electrical variable data item R n ; Otherwise, retain the real-time power box electrical variable data item D i unchanged, as the original real-time power box electrical variable data item D i .

5. A method for monitoring electric variables of a three-level power supply box according to claim 4, characterized in that: The specific steps of constructing the electrical variable disturbance compensation function include: Collecting a number of verified electric variable disturbance training samples; combining the number of verified electric variable disturbance training samples to obtain an electric variable disturbance compensation training set; Construct K electrical variable disturbance compensation function individuals H based on swarm optimization algorithm k , each electrical variable disturbance compensation function individual H k is a set of solutions for calculating the function of electrical variable disturbance compensation; the K electrical variable disturbance compensation function individuals H k Combination, get the iterative population of electrical variable disturbance compensation; set the maximum number of iterations; electrical variable disturbance compensation function individual H k The fitness of k ; Calculate the fitness Y k Specific steps: Using the electrical variable disturbance compensation function individual H k The electric variable disturbance compensation training set is calculated to obtain the electric variable compensation accuracy; the electric variable compensation accuracy is used as the fitness Y k ; Introduce population iterative learning factor Z1 and population iterative learning factor Z2; set Z1=Z max +(Z max -Z min )*sin[(P*Π) / P max ],Z2=Z max -(Z max -Z min )*cos[(P*Π) / P max ]; among them, Z max =2*q,Z min =q, q is the preset parameter, P is the current number of iterations, P max is the maximum number of iterations; In the iterative population iteration process of the electric variable disturbance compensation, the electric variable disturbance compensation function individual H is calculated based on the population iteration learning factor Z1 and the population iteration learning factor Z2. k Conduct a search update; When the maximum number of iterations is reached, the individual electric variable disturbance compensation function corresponding to the maximum output fitness is the optimal individual electric variable disturbance compensation function; the electric variable disturbance compensation function is constructed based on the optimal individual electric variable disturbance compensation function.

6. A method for monitoring electric variables of a three-level power supply box according to claim 5, characterized in that: The specific steps of determining the threshold judgment standard of the dynamic power supply box electric variable in step S3 include: Set the basic threshold judgment standard of electric variables; perform feature extraction based on real-time environmental variables and electric variable interference prediction results to obtain the electric variable interference label value; match the electric variable threshold dynamic change library based on the electric variable interference label value to obtain the electric variable dynamic threshold adjustment factor; adjust the basic threshold judgment standard of electric variables according to the dynamic threshold adjustment factor of electric variables to obtain the dynamic power supply box electric variable threshold judgment standard.

7. A method for monitoring electric variables of a three-level power supply box according to claim 6, characterized in that: The power box electric variable risk quantification model includes an electric variable data item calculation layer, a quantitative risk assessment layer and an electric variable risk quantification output layer; The electric variable data sub-item calculation layer is used to compare the updated real-time power box electric variable set D' with the basic parameter standard of the power box electric variable to obtain the power box electric variable difference parameter C i ; The quantitative risk assessment layer is used to calculate the power box electrical variable difference parameter C i Match the risk labels to obtain the power box electrical variable risk quantification value L i ; The quantitative risk assessment layer is constructed based on the BP neural network to collect the risk matching training set of electrical variable difference; The BP neural network model is trained using the electrical variable difference risk matching training set to obtain a quantitative risk assessment layer; The output layer of the electrical variable risk quantification is used to quantify the electrical variable risk value L of all power boxes i Combine them to obtain the quantification results of electrical variable risk.

8. A three-level power supply box electric variable monitoring system, characterized in that: The system applies a three-level power supply box electric variable monitoring method as described in any one of claims 1 to 7, including: An electric variable environmental disturbance judgment module includes a data acquisition unit and an environmental disturbance judgment unit; The data acquisition unit is used to acquire the real-time environmental variables and the real-time power box electrical variable set D at the monitoring initial frequency, D={D i |i=1,2,…,I};D i D is the real-time power box electrical variable data item i , I is the total number of types of data items in the power box electrical variable concentration; The environmental disturbance judgment unit is used to input the real-time environmental variables into the power box environmental prediction judgment model for analysis to obtain the power variable interference prediction results; the power box environmental prediction judgment model includes an environmental variable processing layer, an environmental feature analysis layer, an environmental prediction judgment layer, an environmental auxiliary judgment layer and a result output layer. The environmental feature analysis layer is constructed based on the improvement of the random forest algorithm and is used to predict and analyze the real-time environmental variables; An electric variable disturbance compensation updating module includes a compensation analysis unit; The compensation sub-item unit is used to judge the real-time power supply box electric variable set D based on the electric variable interference prediction result and the power supply box electric variable disturbance judgment model, and obtain the updated real-time power supply box electric variable set D', D'={D i '|i=1, 2, ..., I}; The power box electrical variable disturbance judgment model is improved and constructed based on wavelet transform and swarm optimization algorithm; The electric variable dynamic threshold division module includes a risk quantification unit and a threshold judgment unit; The risk quantification unit is used to perform threshold judgment based on the updated real-time power box electric variable set D' and the power box electric variable risk quantification model to obtain the electric variable risk quantification result; the power box electric variable risk quantification model includes an electric variable data sub-item calculation layer, a quantitative risk assessment layer and an electric variable risk quantification output layer; The threshold judgment unit is used to evaluate based on the electric variable risk quantification results and the dynamic power box electric variable threshold judgment standard to obtain the real-time power box electric variable monitoring results. The dynamic power box electric variable threshold judgment standard is determined by the real-time environmental variables and the electric variable interference prediction results; subsequent operations are performed based on the real-time power box electric variable monitoring results.

Images