Remote monitoring method and system for electrostatic potential of mobile box-type transformer based on symbolic regression

Abstract

The present invention discloses a remote monitoring method and system for the electrostatic potential of a mobile box transformer based on symbolic regression, and relates to the technical field of power systems. It includes preprocessing multi-dimensional electrostatic monitoring data to obtain standardized data; inputting the standardized data into a preset symbolic regression machine learning model; optimizing the parameters of the symbolic regression machine learning model based on an evolutionary algorithm; and realizing remote monitoring of the electrostatic potential according to the output results of the optimized model. The present invention realizes intelligent monitoring of the electrostatic potential of a mobile box transformer through a symbolic regression machine learning algorithm, which significantly improves the real-time and accuracy of monitoring. The use of a distributed sensor array can comprehensively and real-timely collect multi-dimensional electrostatic data around the mobile box transformer, and improve data quality and availability through a series of data preprocessing and standardization steps.

Description

Technical Field

[0001] The present invention relates to the technical field of electric power systems, and in particular to a method and system for remotely monitoring the electrostatic potential of a mobile box-type transformer vehicle based on symbolic regression. Background Art

[0002] In the power system, static electricity will accumulate in mobile box transformers due to equipment operation and environmental factors. This static electricity will not only interfere with the normal operation of the equipment, but may also cause safety accidents. The existing electrostatic potential monitoring methods mainly have the following technical deficiencies: Single monitoring method: Traditional monitoring methods mainly rely on single sensors arranged at fixed points, which cannot achieve comprehensive monitoring of the electrostatic field around the mobile box transformer, resulting in insufficient representativeness and integrity of the monitoring data; Low data processing efficiency: The existing technology uses a simple data processing method, lacks the ability to effectively integrate and analyze multi-dimensional electrostatic monitoring data, and is difficult to quickly extract effective feature information from massive data; Limited prediction ability: Traditional methods are mainly based on threshold judgment or simple statistical analysis, and cannot establish a mathematical model of electrostatic potential changes, lacking the ability to predict the trend of electrostatic potential changes; Insufficient real-time performance: Due to the slow response speed of the monitoring system, large data transmission and processing delays, it is difficult to meet the needs of real-time monitoring of the electrostatic potential of mobile box transformers; High system cost: The existing technology requires the deployment of a large number of sensors to improve monitoring accuracy, resulting in high system deployment and maintenance costs, and increased system complexity.

[0003] In view of the above problems, the present invention proposes an electrostatic potential monitoring method based on symbolic regression machine learning. This method collects multi-dimensional electrostatic monitoring data, combines data preprocessing and standardization technology, constructs a symbolic regression machine learning model, and uses evolutionary algorithms to achieve adaptive optimization of model parameters. Compared with the prior art, the present invention has the following technical innovations:

[0004] It realizes collaborative analysis of multi-dimensional data and improves the integrity and reliability of monitoring data. It adopts machine learning methods to establish an electrostatic potential prediction model and enhances the prediction ability of the system. It realizes automatic mapping of data to model through symbolic regression algorithm and reduces the dependence on expert experience. It introduces evolutionary algorithm to optimize model parameters and improves the adaptability and accuracy of the model. It can generate a high-precision electrostatic potential model based on fewer sensor data points, which significantly reduces the system deployment and maintenance costs. Summary of the invention

[0005] In view of the problems existing in the prior art, the present invention is proposed.

[0006] Therefore, the present invention provides a method and system for intelligently detecting construction waste based on drones, which can solve the problems mentioned in the background technology.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] In a first aspect, an embodiment of the present invention provides a method for remotely monitoring electrostatic potential of a mobile box-type transformer vehicle based on symbolic regression, which comprises collecting multi-dimensional electrostatic monitoring data;

[0009] Preprocess the multi-dimensional electrostatic monitoring data to obtain standardized data;

[0010] Input the standardized data into the preset symbolic regression machine learning model;

[0011] Parameter optimization of symbolic regression machine learning models based on evolutionary algorithms;

[0012] Remote monitoring of electrostatic potential is achieved based on the optimized model output results.

[0013] As a preferred solution of the method for remote monitoring electrostatic potential of a mobile box-type transformer vehicle based on symbolic regression of the present invention, the multi-dimensional electrostatic monitoring data includes the first type of electrostatic field data, the second type of potential real value data, the third type of time domain feature data and the fourth type of environmental parameter data;

[0014] Data preprocessing includes data preprocessing operations and data enhancement processing. The data preprocessing operations include time alignment operations and outlier processing operations. The data enhancement processing is used to expand the multi-dimensional electrostatic monitoring data after data preprocessing.

[0015] As a preferred solution of the method for remote monitoring electrostatic potential of a mobile box-type transformer vehicle based on symbolic regression of the present invention, the symbolic regression machine learning model includes:

[0016] For the first type of electrostatic field data and the second type of potential real value data, the sensor nodes and data collection points in the monitoring area are used as nodes of the calculation graph;

[0017] Based on the third type of time domain feature data and the fourth type of environmental parameter data, determine the weight of the data transmission edge between nodes;

[0018] Set any node of the computational graph as the initial point, select other nodes according to the optimization order of the symbolic regression machine learning model, and add all nodes to the set to be processed.

[0019] As a preferred solution of the method for remote monitoring of electrostatic potential of mobile box-type transformers based on symbolic regression of the present invention, the standardized data is input into a preset symbolic regression machine learning model, including preliminary feature extraction of the first type of electrostatic field data and the second type of potential true value data, and feature fusion is performed on the third type of time domain feature data and the fourth type of environmental parameter data. When a feature node in the set to be processed is processed, the node is deleted from the set to be processed. When the set to be processed is an empty set, the data processing is completed.

[0020] As a preferred solution of the method for remote monitoring of electrostatic potential of a mobile box-type transformer vehicle based on symbolic regression of the present invention, the first type of electrostatic field data is acquired by collecting the electrostatic sensor array, the second type of potential true value data is acquired by the contact measurement system, the third type of time domain feature data is obtained by signal analysis, and the fourth type of environmental parameter data includes environmental factors such as temperature and humidity.

[0021] As a preferred solution of the method for remote monitoring electrostatic potential of a mobile box-type transformer vehicle based on symbolic regression of the present invention, the optimization process of the symbolic regression machine learning model includes:

[0022] Construct a chromosome population containing multiple genes. Each gene represents a mathematical expression. A symbolic binary tree structure is used to represent each gene. The tree nodes contain basic operators and variables. The gene output is Chromosome output is a linear combination of genes:

[0023]

[0024] Among them, b0 is the bias parameter, b1…b G is the scaling parameter; the linear combination can be expressed in matrix form:

[0025]

[0026] Among them, G is the gene output matrix, B is the linear combination parameter vector;

[0027] Solve for the linear combination parameters using the least squares method:

[0028] B=(G T G) -1 G T y.

[0029] As a preferred solution of the method for remote monitoring electrostatic potential of a mobile box-type transformer based on symbolic regression of the present invention, the specific implementation of the evolutionary algorithm includes using the NRMSE indicator to evaluate the chromosome fitness:

[0030]

[0031] The specific optimization process includes chromosome-level crossover when the complete binomial tree is exchanged, gene-level crossover when the subtrees of the binary tree are exchanged, changing the nodes of the binary tree according to a predetermined mutation rate, and the chromosomes of both the parent and offspring are passed to the next generation. When the predefined accuracy target is reached, the evolution process is terminated and the relevant mathematical model is output.

[0032] In a second aspect, an embodiment of the present invention provides a remote monitoring system for electrostatic potential of a mobile box-type transformer based on symbolic regression, which includes an electrostatic collection module, an electrostatic parameter analysis module, and a remote monitoring module;

[0033] The electrostatic collection module contains multiple EPS sensors, which are used to collect electrostatic field data, potential real value data, time domain characteristic data, and environmental parameter data around the box-type transformer, realizing multi-dimensional electrostatic data collection;

[0034] The electrostatic parameter analysis module processes and analyzes the collected data based on the symbolic regression machine learning algorithm, builds an electrostatic potential prediction model, and achieves parameter optimization;

[0035] The remote monitoring module is used to receive and process the output results of the electrostatic parameter analysis module in real time, monitor the electrostatic potential through the established prediction model, and warn and handle abnormal situations.

[0036] In a third aspect, an embodiment of the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program instructions are executed by the processor, the steps of the remote monitoring method of the electrostatic potential of a mobile box-type transformer based on symbolic regression as in the first aspect of the present invention are implemented.

[0037] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program instructions are executed by a processor, the steps of the method for remotely monitoring the electrostatic potential of a mobile box-type transformer based on symbolic regression as in the first aspect of the present invention are implemented.

[0038] The beneficial effects of the present invention are as follows: the present invention proposes a remote monitoring method and system for the electrostatic potential of a mobile box transformer based on symbolic regression, and realizes intelligent monitoring of the electrostatic potential of the mobile box transformer through a symbolic regression machine learning algorithm, which significantly improves the real-time and accuracy of the monitoring. The use of a distributed sensor array can comprehensively and in real time collect multi-dimensional electrostatic data around the mobile box transformer, and improve the data quality and availability through a series of data preprocessing and standardization steps. Subsequently, the symbolic regression machine learning model is used to perform intelligent analysis on the processed data, and the adaptive optimization of the model parameters is realized by combining the evolutionary algorithm, and a high-precision electrostatic potential prediction model is established. The mathematical inversion model is trained using field test data, without relying on the structural information of a specific application scenario, which greatly improves the versatility and adaptability of the method. At the same time, a new EPS is designed and implemented through a commercial amplifier, which has high sensitivity and improves the reliability of data acquisition. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0040] Figure 1 It is a flow chart of a method and system for remote monitoring of electrostatic potential of a mobile box-type transformer vehicle based on symbolic regression;

[0041] Figure 2 It is a schematic diagram of symbolic regression modeling of a remote monitoring method and system of electrostatic potential of a mobile box-type transformer based on symbolic regression; DETAILED DESCRIPTION

[0042] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0043] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0044] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0045] Example 1

[0046] Reference Figure 1-2 , which is the first embodiment of the present invention, and provides a method for remote monitoring of electrostatic potential of a mobile box transformer based on symbolic regression, comprising:

[0047] S1, collect multi-dimensional electrostatic monitoring data;

[0048] In an embodiment of the present application, the multi-dimensional electrostatic monitoring data includes a first type of electrostatic field data, a second type of potential real value data, a third type of time domain feature data, and a fourth type of environmental parameter data;

[0049] The data preprocessing includes data preprocessing operations and data enhancement processing, the data preprocessing operations include time alignment operations and outlier processing operations, and the data enhancement processing is used to expand the multi-dimensional electrostatic monitoring data after data preprocessing.

[0050] It should be noted that in terms of electrical and mechanical aspects, there is no contact between the EPS (electrostatic sensor) and the box-type transformer, and the distance is more than 1 meter. Therefore, the equivalent coupling capacitance between the two is very small. In order to remotely measure the electrostatic potential of the box-type transformer, the electrostatic sensor needs to have a very small input capacitance and an extremely high input resistance to improve the sensitivity. The design of this EPS integrates three typical positive feedback analog circuit design methods: neutralization, guidance, and active protection.

[0051] It should also be noted that multiple EPSs are first deployed in actual application scenarios. Then, the electrostatic potential of the mobile box transformer is tested simultaneously using contact and non-contact methods.

[0052] It should be noted that C e It is the equivalent coupling capacitance, which is determined by the size of the induction electrode and the distance from the box-type transformer. in ,C x and R in are input capacitance, stray capacitance and input resistance respectively. e With C in and C x The capacitor voltage divider network, C e and R in The first-order high-pass filter is formed. Both methods will reduce the sensitivity of the sensor to electrostatic signals. The neutralization circuit part feeds the output signal back to the input end, reducing the equivalent input capacitance C in The pilot circuit part provides an adjustable DC leakage path to limit the leakage current at the amplifier input to prevent the amplifier from saturating. The active protection circuit part uses a buffer amplifier to drive the shielding shell of the sensing electrode to reduce the stray capacitance C xIn addition, due to the low-frequency characteristics of the electrostatic induction signal, it is very sensitive to electromagnetic interference from the power grid, and a 50Hz notch filter and programmable amplification technology are also used.

[0053] S2, performing data preprocessing on the multi-dimensional electrostatic monitoring data to obtain standardized data;

[0054] In an embodiment of the present application, the inputting of the standardized data into a preset symbolic regression machine learning model includes preliminary feature extraction of the first type of electrostatic field data and the second type of potential true value data, and feature fusion combined with the third type of time domain feature data and the fourth type of environmental parameter data. When a feature node in the set to be processed is processed, the node is deleted from the set to be processed. When the set to be processed is an empty set, the data processing is completed.

[0055] It should be noted that the electrostatic sensor obtains the electrostatic field around the box-type transformer as the model input. At the same time, the contact test system obtains the actual electrostatic potential of the box-type transformer as the expected model output. After data cleaning and alignment, the raw data is provided to the algorithm.

[0056] It should also be noted that the symbolic regression method uses evolutionary algorithms to search in the space of mathematical expressions to improve the accuracy of model predictions. It does not require a predetermined model form and can search for the form and parameters of mathematical expressions, reducing reliance on artificial prior knowledge. Moreover, symbolic models are easy to deploy and do not require specialized hardware or software environments.

[0057] S3, inputting the standardized data into a preset symbolic regression machine learning model;

[0058] In an embodiment of the present application, the symbolic regression machine learning model includes:

[0059] For the first type of electrostatic field data and the second type of potential real value data, the sensor nodes and data collection points in the monitoring area are used as nodes of the calculation graph;

[0060] Based on the third type of time domain feature data and the fourth type of environmental parameter data, determine the weight of the data transmission edge between nodes;

[0061] Set any node of the computational graph as the initial point, select other nodes according to the optimization order of the symbolic regression machine learning model, and add all nodes to the set to be processed.

[0062] Furthermore, the inputting of the standardized data into a preset symbolic regression machine learning model includes preliminary feature extraction of the first type of electrostatic field data and the second type of potential true value data, feature fusion combined with the third type of time domain feature data and the fourth type of environmental parameter data, when a feature node in the set to be processed is processed, the node is deleted from the set to be processed, and when the set to be processed is an empty set, the data processing is completed.

[0063] Randomly initialize the chromosome population. Each chromosome contains many genes, and each gene is a mathematical expression represented by a symbolic binary tree. Chromosomes are alternative mathematical models. Through elementary operations and electrostatic sensor output data, a symbolic binary tree with a maximum depth of 10 is constructed. To increase genetic diversity, repeated gene fragments in the first generation must be avoided.

[0064] Gene output is The chromosome output is Linear combinations of genes:

[0065]

[0066] Among them, b0 is the bias parameter, b1…b G is the scaling parameter; since all data have the same length, the above equation can be expressed as a matrix:

[0067]

[0068] Among them, G is the gene output matrix, B is the linear combination parameter vector; in the formula, that is:

[0069]

[0070] B=[b0,b1,…,b G ] T

[0071] S4, optimizing parameters of the symbolic regression machine learning model based on an evolutionary algorithm;

[0072] Under the premise of satisfying several assumptions, the optimal value of B can be obtained by the least squares method from the target data y and G:

[0073] B=(G T G) -1 G T y

[0074] It should be noted that in order to evaluate the fitness of candidate chromosomes, the NRMSE indicator is used. It is calculated by the output data and target data of the mathematical model corresponding to the chromosome. The selection process always selects the chromosome with the highest fitness. If the predefined accuracy target is achieved, the evolution process is terminated and the relevant mathematical model is output; otherwise, crossover and mutation operations are performed. The NRMSE indicator is used to evaluate the chromosome fitness:

[0075]

[0076] The specific optimization process includes chromosome-level crossover when the complete binomial tree is exchanged, gene-level crossover when the subtrees of the binary tree are exchanged, changing the nodes of the binary tree according to a predetermined mutation rate, and the chromosomes of both the parent and offspring are passed to the next generation. When the predefined accuracy target is reached, the evolution process is terminated and the relevant mathematical model is output.

[0077] S5, realizing remote monitoring of electrostatic potential according to the output results of the optimized model;

[0078] It should be noted that the evolution process is terminated when the predefined accuracy target is reached, and the relevant mathematical model is output to use the electrostatic signal measured on site and the symbolic regression machine learning algorithm to remotely monitor the electrostatic potential of the box-type transformer, without the need for specific application scenario structure information.

[0079] Furthermore, this embodiment also provides a remote monitoring system for electrostatic potential of a mobile box-type transformer based on symbolic regression, comprising:

[0080] Static electricity collection module, static electricity parameter analysis module, remote monitoring module;

[0081] The electrostatic collection module includes multiple EPS sensors for collecting electrostatic field data, potential real value data, time domain feature data, and environmental parameter data around the box-type transformer, thereby realizing multi-dimensional electrostatic data collection;

[0082] The electrostatic parameter analysis module processes and analyzes the collected data based on the symbolic regression machine learning algorithm, constructs an electrostatic potential prediction model, and realizes parameter optimization;

[0083] The remote monitoring module is used to receive and process the output results of the electrostatic parameter analysis module in real time, monitor the electrostatic potential through the established prediction model, and warn and handle abnormal situations.

[0084] This embodiment also provides a computer device, which is suitable for the case of a remote monitoring method for the electrostatic potential of a mobile box-type transformer substation based on symbolic regression, and includes a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the remote monitoring method for the electrostatic potential of a mobile box-type transformer substation based on symbolic regression as proposed in the above embodiment.

[0085] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0086] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for remotely monitoring the electrostatic potential of a mobile box-type transformer based on symbolic regression as proposed in the above embodiment is implemented.

[0087] In summary, the present invention proposes a remote monitoring method and system for electrostatic potential of a mobile box transformer based on symbolic regression, which collects multi-dimensional electrostatic monitoring data; performs data preprocessing on the multi-dimensional electrostatic monitoring data to obtain standardized data; inputs the standardized data into a preset symbolic regression machine learning model; optimizes the parameters of the symbolic regression machine learning model based on an evolutionary algorithm; and realizes remote monitoring of electrostatic potential according to the output result of the optimized model. The present invention realizes intelligent monitoring of electrostatic potential of a mobile box transformer through a symbolic regression machine learning algorithm, which significantly improves the real-time and accuracy of monitoring. The distributed sensor array can comprehensively and real-timely collect multi-dimensional electrostatic data around the mobile box transformer, and improves data quality and availability through a series of data preprocessing and standardization steps. Subsequently, the symbolic regression machine learning model is used to perform intelligent analysis on the processed data, and the adaptive optimization of model parameters is realized by combining the evolutionary algorithm, and a high-precision electrostatic potential prediction model is established. The mathematical inversion model is trained using field test data, without relying on the structural information of a specific application scenario, which greatly improves the versatility and adaptability of the method. At the same time, a new EPS is designed and implemented through a commercial amplifier, which has high sensitivity and improves the reliability of data acquisition.

[0088] Example 2

[0089] This is the second embodiment of the present invention. This embodiment provides a remote monitoring method for the electrostatic potential of a mobile box-type transformer based on symbolic regression. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0090] In an electrostatic monitoring experiment of a mobile box transformer in a 110kV substation, a group of comparative experiments were designed to address the problems of low monitoring accuracy and slow response speed in the prior art. The traditional single-point method, conventional array method and the three optimization schemes of the present invention were used for a 30-day field test.

[0091] Basic conditions setting: The experiment uses Texas Instruments OPA129U operational amplifier to design the input stage of the EPS sensor, with an input impedance of 1015Ω and an input capacitance of less than 1pF. The sensor housing is made of 304 stainless steel, and the internal circuit adopts a multi-layer shielding structure. The reference potential is collected using the HVEM-2000 high-voltage electrostatic meter (measuring range 0-20kV, accuracy 0.1%). Environmental parameter monitoring range: temperature (-20℃~60℃, accuracy ±0.1℃), relative humidity (0-100%RH, accuracy ±2%RH) and atmospheric pressure (800-1100hPa, accuracy ±0.3hPa).

[0092] Experimental design:

[0093] Traditional single-point method: using 1 fixed sensor, sampling frequency 10Hz, and threshold judgment method;

[0094] Conventional array method: 12 fixed sensors are arranged, the sampling frequency is 50Hz, and the weighted average algorithm is used;

[0095] Initial optimization plan: 12 EPS sensors are arranged with a sampling frequency of 100 Hz, and a basic symbolic regression model is applied for data preprocessing and feature extraction;

[0096] Complete optimization scheme: Based on the initial scheme, the evolutionary algorithm is introduced to optimize the symbolic regression model to achieve adaptive parameter adjustment, including a complete optimization process such as multi-objective evolutionary optimization and crossover mutation operations;

[0097] Streamlined optimization scheme: The optimized sensor layout is 8 EPS, using the same algorithm as the full optimization scheme, and the optimal sensor position is determined through an intelligent layout algorithm.

[0098] Data processing flow: The original data is bandpass filtered (0.1-10Hz) to remove power frequency interference and high-frequency noise, and wavelet transform is used for time-frequency analysis. Time alignment is performed (synchronization accuracy is better than 1ms), and the improved DBSCAN algorithm is used for outlier detection and removal. The interpolation method based on Gaussian process is used to handle missing values. Data enhancement uses Gaussian white noise (SNR=20dB), random time offset (±0.5s) and environmental parameter perturbation (±5%). Model training uses a training test set division ratio of 7:3 and 5-fold cross validation.

[0099]

[0100] Through comparative analysis of experimental data, the present invention shows significant advantages in various performance indicators:

[0101] Comparison of prediction accuracy:

[0102] The NRMSE of the traditional single-point method is 8.45%, and that of the conventional array method is 5.67%

[0103] The initial optimization dropped to 3.42%, showing the fundamental advantage of the symbolic regression model.

[0104] The fully optimized solution further reduced it to 2.15%, an improvement of 74.6% over the traditional method.

[0105] The streamlined optimization solution remains at a good level of 2.38%, proving the effectiveness of the optimization algorithm

[0106] System responsiveness:

[0107] The traditional single-point method takes 150ms, and the conventional array method takes 120ms

[0108] The initial optimization solution was shortened to 85ms

[0109] The fully optimized solution achieves an optimal response time of 45ms, an increase of 70%

[0110] The streamlined optimization solution only slightly increased the time to 48ms, basically maintaining excellent performance

[0111] Monitoring reliability:

[0112] The false positive rate of the fully optimized solution is reduced to 0.7%, which is much lower than the 2.8% of the traditional single-point method.

[0113] The detection rate increased to 98.2%, 12.6 percentage points higher than the traditional method

[0114] Even in the streamlined optimization scheme, a high detection rate of 97.5% was maintained

[0115] Economic analysis:

[0116] The streamlined optimization solution reduced the number of sensors by 33% (from 12 to 8), but the performance indicators only decreased slightly:

[0117] NRMSE increased slightly by 0.23 percentage points (2.15% to 2.38%)

[0118] Response time increased by 3ms (45ms to 48ms)

[0119] The detection rate dropped by only 0.7 percentage points (98.2% to 97.5%)

[0120] This set of experimental data fully proves that the present invention significantly improves the system response speed, reduces the false alarm rate, and improves the detection rate while maintaining high-precision monitoring. In particular, the successful implementation of the streamlined optimization scheme demonstrates the innovative value of the present invention in optimizing system costs, and provides a solution that takes into account both performance and economy for engineering practice.

[0121] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A remote monitoring method for electrostatic potential of a mobile box-type transformer based on symbolic regression, characterized in that: Including collecting multi-dimensional static monitoring data; Performing data preprocessing on the multi-dimensional electrostatic monitoring data to obtain standardized data; Inputting the standardized data into a preset symbolic regression machine learning model; Optimizing parameters of the symbolic regression machine learning model based on an evolutionary algorithm; Remote monitoring of electrostatic potential is achieved based on the optimized model output results.

2. The method for remote monitoring of electrostatic potential of a mobile box-type transformer based on symbolic regression as claimed in claim 1 is characterized in that: The multi-dimensional electrostatic monitoring data includes first-class electrostatic field data, second-class potential real value data, third-class time domain characteristic data and fourth-class environmental parameter data; The data preprocessing includes data preprocessing operations and data enhancement processing, the data preprocessing operations include time alignment operations and outlier processing operations, and the data enhancement processing is used to expand the multi-dimensional electrostatic monitoring data after data preprocessing.

3. The method for remote monitoring of electrostatic potential of a mobile box-type transformer based on symbolic regression as claimed in claim 2 is characterized in that: The symbolic regression machine learning model includes: For the first type of electrostatic field data and the second type of potential real value data, the sensor nodes and data collection points in the monitoring area are used as nodes of the calculation graph; Based on the third type of time domain feature data and the fourth type of environmental parameter data, determine the weight of the data transmission edge between nodes; Set any node of the computational graph as the initial point, select other nodes according to the optimization order of the symbolic regression machine learning model, and add all nodes to the set to be processed.

4. The method for remote monitoring of electrostatic potential of a mobile box-type transformer based on symbolic regression as claimed in claim 3 is characterized in that: The step of inputting the standardized data into a preset symbolic regression machine learning model includes performing preliminary feature extraction on the first type of electrostatic field data and the second type of potential true value data, and performing feature fusion on the third type of time domain feature data and the fourth type of environmental parameter data. When a feature node in the set to be processed is processed, the node is deleted from the set to be processed. When the set to be processed is an empty set, the data processing is completed.

5. The method for remote monitoring of electrostatic potential of a mobile box-type transformer based on symbolic regression as claimed in claim 4 is characterized in that: The first type of electrostatic field data is collected by an electrostatic sensor array, the second type of potential real value data is obtained by a contact measurement system, the third type of time domain characteristic data is obtained through signal analysis, and the fourth type of environmental parameter data includes environmental factors such as temperature and humidity.

6. The method for remote monitoring of electrostatic potential of a mobile box-type transformer based on symbolic regression as claimed in claim 5 is characterized in that: The optimization process of the symbolic regression machine learning model includes: Construct a chromosome population containing multiple genes. Each gene represents a mathematical expression. A symbolic binary tree structure is used to represent each gene. The tree nodes contain basic operators and variables. The gene output is Chromosome output is a linear combination of genes: Among them, b0 is the bias parameter, b 1… b G is the scaling parameter; the linear combination can be expressed in matrix form: Among them, G is the gene output matrix, B is the linear combination parameter vector; Solve for the linear combination parameters using the least squares method: B=(G T G) -1 G T y。 7. The method for remote monitoring of electrostatic potential of a mobile box-type transformer based on symbolic regression as claimed in claim 6 is characterized in that: The specific implementation of the evolutionary algorithm includes using the NRMSE indicator to evaluate the chromosome fitness: The specific optimization process includes chromosome-level crossover when the complete binomial tree is exchanged, gene-level crossover when the subtrees of the binary tree are exchanged, changing the nodes of the binary tree according to a predetermined mutation rate, and the chromosomes of both the parent and offspring are passed to the next generation. When the predefined accuracy target is reached, the evolution process is terminated and the relevant mathematical model is output.

8. A remote monitoring system for electrostatic potential of a mobile box-type transformer vehicle based on symbolic regression, based on the remote monitoring method for electrostatic potential of a mobile box-type transformer vehicle based on symbolic regression according to any one of claims 1 to 7, characterized in that: It also includes an electrostatic collection module, an electrostatic parameter analysis module, and a remote monitoring module; The electrostatic collection module includes multiple EPS sensors for collecting electrostatic field data, potential real value data, time domain feature data, and environmental parameter data around the box-type transformer, thereby realizing multi-dimensional electrostatic data collection; The electrostatic parameter analysis module processes and analyzes the collected data based on the symbolic regression machine learning algorithm, constructs an electrostatic potential prediction model, and realizes parameter optimization; The remote monitoring module is used to receive and process the output results of the electrostatic parameter analysis module in real time, monitor the electrostatic potential through the established prediction model, and warn and handle abnormal situations.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for remotely monitoring the electrostatic potential of a mobile box-type transformer based on symbolic regression described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for remotely monitoring the electrostatic potential of a mobile box-type transformer based on symbolic regression described in any one of claims 1 to 7 are implemented.

Images