Transformer comprehensive safety assessment method and system based on symbol regression

Abstract

The invention relates to a transformer comprehensive safety assessment method and system based on symbol regression. The method comprises the following steps: acquiring monitoring variables of an electrical signal dimension, an equipment state dimension and an operation environment dimension of a transformer; based on the maximum mutual information coefficient, screening the dimension monitoring variables to obtain a plurality of monitoring variable sets with closely associated variables; performing sign regression on the plurality of monitoring variable sets to obtain a nonlinear function expression between any reference monitoring variable and other monitoring variables in the monitoring variable sets, namely a mutual check function, and realizing sign regression by using a genetic algorithm; establishing a single-variable fault probability function according to the influence of the multi-dimensional monitoring variable on the transformer equipment; and building a system dynamics model based on the mutual verification function and the fault probability function to realize comprehensive safety evaluation of the transformer. Compared with the prior art, the method has the advantages of improving the safety assessment level of the transformer, being high in interpretability and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of transformer safety assessment, and in particular to a transformer comprehensive safety assessment method and system based on symbolic regression. Background Art

[0002] With the continuous installation of high-precision monitoring devices in power systems, digital intelligent electrical equipment safety assessment has gradually become possible. With the help of a large amount of monitoring data, the comprehensive analysis of electrical variables and non-electrical variables can effectively improve the level of transformer safety assessment. However, the existing transformer safety assessment only evaluates the equipment operation safety based on a single monitoring variable, and fails to explore the internal relationship between multiple monitoring variables.

[0003] Chinese patent application CN202211466260 discloses a multi-parameter fusion method for defect identification, state classification and active safety protection of power transformers. This method integrates multi-source data such as the operating environment conditions and operating parameters of power transformers to form a knowledge graph, conducts safety assessment on the transformer, and takes corresponding safety regulation and protection measures based on the assessment results. However, this method fails to find the specific functional relationship between multi-source data and has poor interpretability.

[0004] In summary, there is currently a lack of a safety assessment method that fully considers the mutual influence between multi-dimensional monitoring variables. Summary of the invention

[0005] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and provide a transformer comprehensive safety assessment method and system based on symbolic regression to solve or partially solve the problem that the variables considered are not comprehensive and the relationship between variables cannot be fully established.

[0006] The purpose of the present invention can be achieved by the following technical solutions:

[0007] One aspect of the present invention provides a transformer comprehensive safety assessment method based on symbolic regression, comprising the following steps:

[0008] Step S1, obtaining monitoring variables of the transformer's electrical signal dimension, equipment status dimension, and operating environment dimension;

[0009] Step S2, based on the maximum mutual information coefficient, screening the dimensional monitoring variables to obtain a plurality of monitoring variable sets with close correlations between the variables;

[0010] Step S3, performing symbolic regression on the plurality of monitoring variable sets to obtain a nonlinear function expression between any reference monitoring variable in the monitoring variable set and other monitoring variables, i.e., a mutual verification function, and implementing symbolic regression using a genetic algorithm;

[0011] Step S4, establishing a single variable fault probability function according to the influence of the multi-dimensional monitoring variables on the transformer equipment;

[0012] Step S5, building a system dynamics model based on the mutual verification function and the fault probability function to achieve a comprehensive safety assessment of the transformer.

[0013] As a preferred technical solution, in step S2, the process of acquiring the monitoring variable set includes the following steps:

[0014] Step S201, calculating the maximum mutual information coefficient and weighted correlation between every two monitoring variables;

[0015] Step S202, arranging the monitoring variables in descending order based on the weighted correlation degree, and calculating the feature optimization weight;

[0016] Step S203: construct multiple monitoring variable sets based on the feature optimization weights.

[0017] As a preferred technical solution, the weighted correlation is calculated using the following formula:

[0018]

[0019] Among them, J i is the weighted correlation of the ith mutual verification monitoring variable, n is the total number of mutual verification monitoring variables, MIC i (A,B i ) is the mutual verification dependent variable A and the mutual verification monitoring variable B i The maximum mutual information coefficient between .

[0020] As a preferred technical solution, the feature optimization weight is calculated using the following formula:

[0021]

[0022] Among them, J s is the sum of the weighted correlations of the mutual verification features in the optimal set, ω j is the preferred weight of the jth feature.

[0023] As a preferred technical solution, in step S3, the calculation process of the nonlinear function expression includes:

[0024] By constructing a random function expression, the population in the genetic algorithm is initialized, the accuracy of the function expression is used as the fitness function, the individuals in the population are evaluated, and new individuals are generated through selection, crossover and mutation processing according to probability to achieve population update. After repeated iterations, the nonlinear function expression between the reference monitoring variable set and other monitoring variable sets is obtained.

[0025] As a preferred technical solution, the symbolic regression problem is modeled as:

[0026]

[0027] Among them, L(f(x), D) is the fitness function for judging the accuracy of f(x), X is the independent variable from the data set D, F is the function set, F = {+, -, ×, ÷, sin, cos}, and M is the symbolic expression space composed of independent variables, functions and constant terms.

[0028] As a preferred technical solution, in the step S1, the monitoring variables in the electrical signal dimension include load current and operating voltage, the monitoring variables in the equipment status dimension include core grounding current, clamp grounding current, SF6 gas pressure, partial discharge, main transformer oil level height, main transformer oil level ratio, main transformer oil temperature and main transformer temperature, and the monitoring variables in the operating environment indicator dimension include water level, humidity and temperature.

[0029] As a preferred technical solution, the step S4 specifically includes:

[0030] The fault probability functions of the electrical signal dimension, equipment status dimension and operating environment dimension are constructed respectively to obtain a comprehensive fault probability function.

[0031] As a preferred technical solution, the step S5 specifically includes:

[0032] A system dynamics flow diagram is constructed based on the nonlinear function expression and the comprehensive fault probability function to form a transformer safety assessment model.

[0033] As a preferred technical solution, when the difference between the predicted reference monitoring variable set value and the measured reference monitoring variable set value is greater than a preset test threshold, an alarm is triggered for the monitoring device corresponding to the monitoring variable.

[0034] Another aspect of the present invention provides a transformer comprehensive safety assessment system based on symbolic regression, comprising:

[0035] A monitoring variable set generation module is used to obtain monitoring variables of the transformer's electrical signal dimension, equipment status dimension, and operating environment dimension, and to screen the monitoring variables of the dimensions based on the maximum mutual information coefficient to obtain multiple monitoring variable sets with close correlations between the variables;

[0036] A nonlinear function relationship modeling module, used for modeling the nonlinear function relationship between the multiple monitoring variable sets as a symbolic regression problem, and solving the nonlinear function expression between any reference monitoring variable set and other monitoring variable sets in the multiple monitoring variable sets by a genetic algorithm;

[0037] The transformer abnormality monitoring module is used to calculate the predicted reference monitoring variable set value based on the nonlinear function expression, and realize abnormal monitoring of the transformer based on the predicted reference monitoring variable set value and the measured reference monitoring variable set value.

[0038] As a preferred technical solution, it also includes:

[0039] The safety assessment model construction module is used to construct the fault probability functions of the electrical signal dimension, the equipment status dimension and the operating environment dimension respectively, obtain the comprehensive fault probability function, and construct the system dynamics flow chart based on the nonlinear function expression and the comprehensive fault probability function to form a safety assessment model of the transformer.

[0040] Compared with the prior art, the present invention has at least one of the following beneficial effects:

[0041] (1) Fully consider the impact of multi-dimensional variables on the comprehensive safety of the transformer: The present invention comprehensively considers the monitoring variables of the electrical signal dimension, the equipment status dimension and the operating environment dimension, and issues early warnings based on the relationship between the various variables, which can more comprehensively evaluate the comprehensive safety of the transformer.

[0042] (2) Fully explore the implicit nonlinear relationship between various variables: The present invention establishes the implicit nonlinear functional relationship between multidimensional monitoring variables based on the symbolic regression method, effectively improves the utilization efficiency of the monitoring variables, and performs double safety protection on the transformer system based on the data during normal operation of the transformer, effectively improving the safety assessment level of the transformer.

[0043] (3) Strong interpretability: The present invention uses a symbolic regression method to associate electrical monitoring variables with non-electrical monitoring variables to obtain a specific functional relationship expression, which makes the mutual verification evaluation have good interpretability.

[0044] (4) Overcoming the limitations of the single monitoring variable evaluation method: The present invention combines the single monitoring variable fault assessment with the multi-detection variable mutual verification safety assessment, thus overcoming the limitations of the single monitoring variable evaluation method. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 Schematic diagram of the transformer comprehensive safety assessment process framework in the embodiment;

[0046] Figure 2 It is a schematic flow chart of the characteristic optimization steps in the embodiment;

[0047] Figure 3 It is a symbolic regression flow graph based on genetic algorithm in the embodiment;

[0048] Figure 4This is a schematic diagram of the device alarm process in the embodiment;

[0049] Figure 5 It is a schematic diagram of the dynamics model of the safety assessment system in the embodiment;

[0050] Figure 6 Schematic diagram of a transformer comprehensive safety assessment system based on symbolic regression in an embodiment. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0052] Example 1

[0053] In view of the problems existing in the above-mentioned prior art, this embodiment provides a comprehensive transformer safety assessment method based on symbolic regression. First, important monitoring variables of the transformer are selected to establish a single variable fault probability function for the transformer, and a functional relationship between multi-dimensional monitoring variables is established based on symbolic regression for transformer mutual verification safety assessment; and the single variable fault probability function and the mutual functional relationship are converted into a system dynamics equation, and a transformer system safety protection relationship model is constructed to achieve a comprehensive safety assessment of the transformer system. Figure 1 , the method comprises the following steps:

[0054] Step S1, selecting transformer multi-dimensional monitoring variables.

[0055] The key characteristics of the transformer cover three dimensions: electrical signals, equipment status indicators, and operating environment indicators. The key characteristics of the electrical signal dimension are load current and operating voltage; the key characteristics of the equipment status indicator dimension are core grounding current, clamp grounding current, partial discharge, main transformer oil temperature, main transformer oil level height, main transformer oil level ratio, main transformer temperature, and SF6 gas pressure; the key characteristics of the operating environment indicator dimension are water level, humidity, and temperature. As shown in Table 1.

[0056] Table 1 Multi-dimensional monitoring variables

[0057]

[0058] Step S2, determining a set of monitoring variables with a larger correlation based on the monitoring variable optimization method of the maximum mutual information coefficient. This step includes the following sub-steps:

[0059] Step S201, calculating the maximum mutual information coefficient between each monitoring variable.

[0060] The maximum mutual information coefficient measures the influence of the monitoring variable on the mutual verification dependent variable, and its weight reflects the effectiveness of the monitoring variable. The maximum mutual information coefficient between the multidimensional monitoring variables of transformer monitoring is calculated respectively. i The maximum mutual information coefficient between i (A,B i ); The preferred steps of mutual verification monitoring variables based on weighted correlation are as follows Figure 2 As shown in Figure 2, the calculation formula for weighted association is:

[0061]

[0062] Where: J i is the weighted correlation of the ith mutual verification monitoring variable; n is the total number of mutual verification monitoring variables.

[0063] Step S202, the calculated weighted correlations are arranged in descending order, and a feature with the largest weighted correlation is selected each time to be added to the preferred set, and it is removed from the set to be preferred. The weight formula for feature selection is:

[0064]

[0065] Where: J s is the sum of the weighted correlations of the mutual verification features in the preferred set; ω j is the weight of the i-th feature optimization. When this value is less than the threshold α, it is considered that the mutual verification feature screening is completed. In this paper, α is taken as 0.12.

[0066] Step S203, establishment of a monitoring variable set.

[0067] In view of the importance of the monitoring variables to the transformer and the result of the correlation calculation in step S202, five monitoring variables, namely, the operating temperature, oil temperature, ambient temperature, oil level, and core grounding current of the transformer, are selected for mutual verification. Feature optimization is performed by calculating the weighted correlation between each monitoring variable. According to the actual operation data calculation results of a transformer, the weight of the weighted correlation is taken as 0.12, that is, when the weight of the feature optimization is less than 0.12, the correlation between the two monitoring variables can be ignored.

[0068] From the feature optimization results, the optimal set of monitoring variables for each variable is: S1 = {X4, X5, X8, X2, X3}; S2 = {X4, X5, X8, X1, X3}; S3 = {X8, X2, X1, X4, X5}; S4 = {X2, X1, X9, X5, X3}; S5 = {X2, X1, X9, X 12 ,X8}.

[0069] The symbols represent: X1 operating temperature, X2 oil temperature, X3 ambient temperature, X4 oil level height, X5 oil level ratio, X6 clamp grounding current, X7 core grounding current, X8 on-load ventilator temperature, X9 main ventilator temperature, X 10 High current value, X 11 Main ventilator humidity, X 12 12 variables including on-load ventilator humidity; and a feature optimization set of 5 mutually verified variables including S1 transformer operating temperature, S2 oil temperature, S3 ambient temperature, S4 oil level, and S5 core grounding current.

[0070] Step S3, using symbolic regression method to establish a nonlinear functional relationship between the selected monitoring variable sets. This step includes the following sub-steps:

[0071] Step S301, using genetic algorithm to solve the symbolic regression problem, such as Figure 3 As shown. First, a set of random formulas is established to represent this relationship, that is, the population is initialized, and all individuals are evaluated using the fitness function; new individuals are generated using selection, crossover, and mutation operations according to probability to obtain a new population; it is determined whether the new population is optimal or the maximum number of iterations is set. If found, the loop is terminated, otherwise the above process is continued.

[0072] Step S302, symbolic regression predicts the target variable through symbolic operations and function fitting, establishes the relationship between the monitoring variables in the set, and finally obtains a clear nonlinear function expression with good interpretability. Assume that the symbolic regression expression is f(x), and its L(f(x), D) is the fitness function for judging the accuracy of f(x). Symbolic regression is to find an f(x) so that L(f(x), D) achieves the maximum value in this case, as shown in the following formula.

[0073]

[0074] Where X is the independent variable from the data set D, X = {x1, x2, ..., x k}, F is a function set, F = {+, -, ×, ÷, sin, cos, …}, and M is a symbolic expression space consisting of independent variables, functions, and constant terms.

[0075] Step S303: Establish a functional relationship between monitoring variables according to the optimal result of step S203. Taking S1 as an example, the established function is:

[0076] S′1=f1(S2,S3,S4,S5,S8) 4)

[0077] According to the function expression, the estimated value S of the kth monitoring variable during normal operation is obtained by analyzing the historical data. k′ and the actual monitoring value S k The maximum difference of is used to determine the difference threshold, and the expression of the safety mutual verification error is:

[0078] ξ k =|S k -S' k | 5)

[0079] Where ξ k is the kth variable check difference, S k is the monitoring value of the kth variable, S' k is the estimated value of the kth variable, when ξ k If the value is greater than the test threshold, the monitoring variable is judged to be abnormal, and an alarm is issued to the monitoring device corresponding to the monitoring variable. The specific process is as follows: Figure 4 shown.

[0080] Step S4, establishing a single variable to equipment failure probability function relationship based on the failure principle. This step includes the following sub-steps:

[0081] Step S401, constructing a fault probability function in the electrical signal dimension.

[0082] If the load current and operating voltage of the transformer exceed a certain threshold, the equipment will be damaged or even burned. The fault probability functions of the two electrical signal characteristics of load current and operating voltage are pf1(t) and pf2(t) respectively:

[0083]

[0084] In the formula are the operating thresholds of the load current and operating voltage of transformer s, are the load current and operating voltage values ​​of transformer s at time t respectively.

[0085] According to the component failure probability theory, based on the results of equations 6) and 7), the fault probability function pf(t) in the electrical signal dimension is obtained as follows:

[0086] pf(t)=1-(1-pf1(t))(1-pf2(t)))8)

[0087] Step S402, constructing a fault probability function of the transformer status indicator dimension.

[0088] The deviation between the health status index of different equipment and the standard value is used to express the impact of the index on the transformer, and HI r '(t) means:

[0089]

[0090] In the formula is the working threshold of the i-th equipment status indicator of transformer s; is the value of the i-th device status index at time t; a transformer has a total of m device status indicators, B i is the influence coefficient of the i-th equipment status indicator.

[0091] The transformer undergoes an aging process during actual operation, so the equipment failure rate l(t) is:

[0092] l(t)=Ke -100C·μ(t)·HI′(t) 10)

[0093] Where K and C are proportional coefficients, and μ(t) is the equipment aging coefficient.

[0094] According to the survey, the typical failure rate is related to the equipment health index of 0.8, while the lowest failure rate corresponds to the perfect health of the equipment, that is, the health index is 1. By analyzing historical data, the average failure rate of transformer equipment can be determined, and the lowest failure rate is set between 10% and 20% of the average failure rate. According to formula 10), the relationship between the failure rate and the probability of failure is obtained, and the failure probability function of the transformer status indicator dimension is obtained as follows:

[0095] pl(t)=1-e -l(t) 11)

[0096] Where pl(t) is the probability of equipment failure at time t.

[0097] Step S403: construct a failure probability function of the operating environment indicator dimension.

[0098] The function design of environmental factors mostly adopts low-complexity and practical exponential models, and has achieved ideal results in scientific research analysis. In order to reasonably construct the failure probability function of operating environment indicators, a piecewise function is introduced to improve the existing exponential empirical function, namely:

[0099]

[0100] Where M is the unknown coefficient of the model, which can be obtained based on objective historical data and expert experience; min is the starting value of the transformer affected; y max is the maximum value of the normal operation of the transformer; j (t) is the failure rate of the transformer caused by the jth environmental factor at time t.

[0101] The failure probability of the equipment caused by the external environment is obtained by the piecewise exponential function of the environmental factor in formula (12):

[0102]

[0103] Where: pe j (t) is the probability of equipment failure caused by the jth environmental factor on the transformer at time t.

[0104] Considering the impact of operating environment factors on the equipment, the failure probability function pe(t) of the operating environment dimension is obtained as follows:

[0105]

[0106] Where b is the number of operating environment indicators.

[0107] Step S404: comprehensive failure probability function of multiple variables.

[0108] Any abnormality in any dimension of electrical signal, equipment status index and operating environment index will cause an alarm, and the indicators are coupled with each other. The fault probability function X of the transformer multi-dimensional monitoring characteristics is obtained according to the series model based on the execution results of steps S401-S403. g1 (t):

[0109] X g1 (t) = 1-(1-pf g (t))(1-pl g (t))(1-pe g (t))15)

[0110] Step S5, building a system dynamics model for comprehensive safety assessment of the transformer system, and performing a comprehensive safety assessment on the transformer system.

[0111] According to the electrical relationship between transformer monitoring variables and between variables and equipment, a system dynamics flow chart is constructed, such as Figure 5 The functional relationship in the above steps 3 and 4 is input into the functional relationship of the system dynamics flow chart, the transformer safety assessment system dynamics equation is designed, and finally the transformer safety assessment model is formed.

[0112] In summary, this method uses symbolic regression to mine the implicit relationship between monitoring variables, establishes a clear functional relationship between multi-dimensional monitoring variables, fully considers the mutual influence between multi-dimensional monitoring variables, and realizes comprehensive transformer safety assessment in combination with single variable safety assessment. This method has the following beneficial effects:

[0113] (1) Based on the symbolic regression method, the implicit nonlinear functional relationship between multidimensional monitoring variables is established, which effectively improves the utilization efficiency of the monitoring variables. Based on the data during normal operation of the transformer, double safety protection is performed on the transformer system, effectively improving the safety assessment level of the transformer.

[0114] (2) The model uses symbolic regression method to associate electrical monitoring variables with non-electrical monitoring variables and obtains a specific functional relationship expression, which makes the mutual verification evaluation highly interpretable.

[0115] (3) The method combines the single monitoring variable fault assessment with the multi-detection variable mutual verification safety assessment, overcoming the limitations of the single monitoring variable assessment method.

[0116] Example 2

[0117] See also Figure 6 Based on Example 1, this embodiment provides a transformer comprehensive safety assessment system based on symbolic regression, including:

[0118] A monitoring variable set generation module is used to obtain monitoring variables of the transformer's electrical signal dimension, equipment status dimension, and operating environment dimension, and to screen the monitoring variables of the dimensions based on the maximum mutual information coefficient to obtain multiple monitoring variable sets with close correlations between the variables;

[0119] A nonlinear function relationship modeling module, used for modeling the nonlinear function relationship between the multiple monitoring variable sets as a symbolic regression problem, and solving the nonlinear function expression between any reference monitoring variable set and other monitoring variable sets in the multiple monitoring variable sets by a genetic algorithm;

[0120] A transformer abnormality monitoring module, used to calculate the predicted reference monitoring variable set value based on the nonlinear function expression, and realize abnormality monitoring of the transformer based on the predicted reference monitoring variable set value and the measured reference monitoring variable set value;

[0121] The safety assessment model construction module is used to construct the fault probability functions of the electrical signal dimension, the equipment status dimension and the operating environment dimension respectively, obtain the comprehensive fault probability function, and construct the system dynamics flow chart based on the nonlinear function expression and the comprehensive fault probability function to form a safety assessment model of the transformer.

[0122] Example 3

[0123] This embodiment provides an electronic device, including: one or more processors and a memory, wherein the memory stores one or more programs, and the one or more programs include instructions for executing the transformer comprehensive safety assessment method based on symbolic regression as described in Example 1.

[0124] Example 4

[0125] This embodiment provides a computer-readable storage medium, including one or more programs for execution by one or more processors of an electronic device, wherein the one or more programs include instructions for executing the transformer comprehensive safety assessment method based on symbolic regression as described in Example 1.

[0126] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A transformer comprehensive safety assessment method based on symbolic regression, characterized in that: The steps include: Step S1, obtaining monitoring variables of the transformer's electrical signal dimension, equipment status dimension, and operating environment dimension; Step S2, based on the maximum mutual information coefficient, screening the dimensional monitoring variables to obtain a plurality of monitoring variable sets with close correlations between the variables; Step S3, performing symbolic regression on the plurality of monitoring variable sets to obtain a nonlinear function expression between any reference monitoring variable in the monitoring variable set and other monitoring variables, i.e., a mutual verification function, and implementing symbolic regression using a genetic algorithm; Step S4, establishing a single variable fault probability function according to the influence of the multi-dimensional monitoring variables on the transformer equipment; Step S5, building a system dynamics model based on the mutual verification function and the fault probability function to achieve a comprehensive safety assessment of the transformer.

2. A transformer comprehensive safety assessment method based on symbolic regression according to claim 1, characterized in that: In step S2, the process of acquiring the monitoring variable set includes the following steps: Step S201, calculating the maximum mutual information coefficient and weighted correlation between every two monitoring variables; Step S202, arranging the monitoring variables in descending order based on the weighted correlation degree, and calculating the feature optimization weight; Step S203: construct multiple monitoring variable sets based on the feature optimization weights.

3. A transformer comprehensive safety assessment method based on symbolic regression according to claim 2, characterized in that: The feature optimization weight is calculated using the following formula: Among them, J i is the weighted correlation of the ith mutual verification monitoring variable, n is the total number of mutual verification monitoring variables, MIC i (A,B i ) is the mutual verification dependent variable A and the mutual verification monitoring variable B i The maximum mutual information coefficient between s is the sum of the weighted correlations of the mutual verification features in the optimal set, ω j is the preferred weight of the jth feature.

4. The transformer comprehensive safety assessment method based on symbolic regression according to claim 1 is characterized in that: In step S3, the calculation process of the nonlinear function expression includes: By constructing a random function expression, the population in the genetic algorithm is initialized, the accuracy of the function expression is used as the fitness function, the individuals in the population are evaluated, and new individuals are generated through selection, crossover and mutation processing according to probability to achieve population update. After repeated iterations, the nonlinear function expression between the reference monitoring variable set and other monitoring variable sets is obtained.

5. The transformer comprehensive safety assessment method based on symbolic regression according to claim 1 is characterized in that: The symbolic regression problem is modeled as: Among them, L(f(X), D) is the fitness function for judging the accuracy of f(x), X is the independent variable from the data set D, F is the function set, F = {+, -, ×, ÷, sin, cos}, and M is the symbolic expression space composed of independent variables, functions and constant terms.

6. The transformer comprehensive safety assessment method based on symbolic regression according to claim 1 is characterized in that: In the step S1, the monitoring variables in the electrical signal dimension include load current and operating voltage, the monitoring variables in the equipment status dimension include core grounding current, clamp grounding current, SF6 gas pressure, partial discharge, main transformer oil level height, main transformer oil level ratio, main transformer oil temperature and main transformer temperature, and the monitoring variables in the operating environment indicator dimension include water level, humidity and temperature.

7. The transformer comprehensive safety assessment method based on symbolic regression according to claim 1 is characterized in that: The step S4 specifically includes: The fault probability functions of the electrical signal dimension, equipment status dimension and operating environment dimension are constructed respectively to obtain a comprehensive fault probability function.

8. A transformer comprehensive safety assessment method based on symbolic regression according to claim 7, characterized in that: The step S5 specifically includes: A system dynamics flow diagram is constructed based on the nonlinear function expression and the comprehensive fault probability function to form a transformer safety assessment model.

9. A transformer comprehensive safety assessment system based on symbolic regression, characterized in that: include: A monitoring variable set generation module is used to obtain monitoring variables of the transformer's electrical signal dimension, equipment status dimension, and operating environment dimension, and to screen the monitoring variables of the dimensions based on the maximum mutual information coefficient to obtain multiple monitoring variable sets with close correlations between the variables; A nonlinear function relationship modeling module, used for modeling the nonlinear function relationship between the multiple monitoring variable sets as a symbolic regression problem, and solving the nonlinear function expression between any reference monitoring variable set and other monitoring variable sets in the multiple monitoring variable sets by a genetic algorithm; The transformer abnormality monitoring module is used to calculate the predicted reference monitoring variable set value based on the nonlinear function expression, and realize abnormal monitoring of the transformer based on the predicted reference monitoring variable set value and the measured reference monitoring variable set value.

10. A transformer comprehensive safety assessment system based on symbolic regression according to claim 9, characterized in that: Also includes: The safety assessment model construction module is used to construct the fault probability functions of the electrical signal dimension, the equipment status dimension and the operating environment dimension respectively, obtain the comprehensive fault probability function, and construct the system dynamics flow chart based on the nonlinear function expression and the comprehensive fault probability function to form a safety assessment model of the transformer.

Images