Transformer overheating fault state evaluation method and device based on temperature inversion, storage medium and electronic equipment

Abstract

The invention relates to a transformer overheating fault state evaluation method and device based on temperature inversion, a storage medium and electronic equipment, and the method specifically comprises the steps: collecting a transformer overheating fault state quantity and a characteristic parameter measurement value in a transformer operation process, and constructing an evaluation index system comprising a target layer, a scheme layer and an index layer based on an analytic hierarchy process; obtaining the actual value of the fault state quantity of each index in the evaluation index system corresponding to the target transformer, and calculating the membership degree of each index of the target transformer through the membership degree; calculating the comprehensive weight of an index layer and the comprehensive weight of a scheme layer in the evaluation index system based on the discrimination matrix of the evaluation index system through a combined weighting method; and based on the membership degree of each index of the target transformer, the comprehensive weight of the index layer and the comprehensive weight of the scheme layer, calculating to obtain a state quantity score of the target transformer.

Description

Technical Field

[0001] The present invention relates to the technical field of transformer operation and maintenance, and particularly relates to a method and device for evaluating the overheating fault state of a transformer based on temperature inversion, a storage medium, and an electronic device. Background Art

[0002] As an indispensable and important device in the power system, the operating state of a transformer directly affects the stability and security of the power system. Therefore, in-depth research on the evaluation technology of transformer state has important theoretical and practical significance. With the continuous increase of voltage levels, the increasing capacity of transformers, and their wider application, the imbalance between the growth of power supply capacity and structural size makes the occurrence of faults more frequent. Among transformer faults, overheating faults account for a relatively large proportion, and the overheating faults of transformer windings are the most concerned part. Therefore, carrying out the state evaluation of transformer thermal faults is of great significance for improving the reliability of the power system.

[0003] Evaluating the state of a transformer means combining various test data such as the factory test data, handover test data, annual pretest data, and annual on-line monitoring data of a certain transformer, and using a certain method (artificial intelligence, data mining, etc.) to scientifically analyze these data and their changing trends, and at the same time considering the influence of non-test data such as the family defect data and operating environment data of the transformer to comprehensively judge the overall operating state of the transformer. At present, most of the research on the operating state evaluation of power transformers is based on the analysis of dissolved gas in oil (DGA) data, but it still has problems such as difficult extraction and analysis of gas components, complex chemical reaction relationships between gases, lack of theoretical basis for threshold selection, slow reaction speed, and high false alarm rate. The related evaluation methods have great limitations. Summary of the Invention

[0004] The present invention provides a method and device for evaluating the overheating fault state of a transformer based on temperature inversion, a storage medium, and an electronic device, focusing on the overheating fault situation of the transformer, carrying out the state evaluation and abnormal state identification of the transformer, constructing a multi-level state evaluation model, using the method of membership function to confirm the internal temperature threshold of the transformer, and calculating the balance of subjective weight and objective weight through the Nash equilibrium theory, so as to improve the accuracy of the overheating state evaluation of the transformer.

[0005] The solution of the present invention to the above technical problems is as follows: A method for evaluating the overheating fault state of a transformer based on temperature inversion, comprising the following steps:

[0006] Collect the state quantities of the overheating fault of the transformer and the measured values of the characteristic parameters in its operation process, and construct an evaluation index system based on the analytic hierarchy process. The evaluation index system includes a scheme layer and an index layer;

[0007] By using the combined weighting method, calculate the comprehensive weights of the index layer and the comprehensive weights of the scheme layer in the evaluation index system;

[0008] Obtain the actual operation data of each overheating fault state quantity in the evaluation index system corresponding to the target transformer, and calculate the membership degree of each fault state quantity corresponding to the target transformer;

[0009] Based on the membership degree of each fault state quantity of the target transformer, the comprehensive weight of the index layer and the comprehensive weight of the scheme layer, calculate the probability result of the overheating fault state of the target transformer.

[0010] Preferably, the evaluation index system further includes a target layer; collect the overheating fault state quantities of the transformer and the measured values of the characteristic parameters during its operation, and construct an evaluation index system based on the analytic hierarchy process, including:

[0011] Collect the overheating fault state quantities of the transformer, classify them according to the characteristics of the overheating fault state quantities of the transformer and the evaluation purpose, and obtain the classification result of the overheating fault state;

[0012] Set the fault state levels of the overheating fault state quantities, and based on the measured values of the characteristic parameters of the overheating fault state quantities during the operation of the transformer, calculate the boundary values of the fault state levels of each overheating fault state quantity through differential thresholds;

[0013] Combined with the hierarchical division method, construct an evaluation index system with the overheating fault state evaluation result as the target layer, the overheating fault state classification result as the scheme layer, and the overheating fault state quantity and the boundary values of its fault state levels as the index layer.

[0014] The purpose of adopting this step is that the existing indicators are not classified in an organized manner, lack hierarchy and multi-dimensionality, and the mapping relationship between the indicators and the states is relatively chaotic and fuzzy. Establishing a hierarchical evaluation index system can intuitively and clearly establish the association between the fault types and the fault state quantities (and indicators).

[0015] Preferably, the setting of the fault state levels of the overheating fault state quantities and the calculation of the boundary values of each fault state level through differential thresholds based on the measured values of the characteristic parameters of the overheating fault state quantities during the operation of the transformer include:

[0016] Set that the overheating fault state quantity has 4 types of fault state levels: normal, attention, abnormal, and severe;

[0017] Calculate the average value of the measured values of the characteristic parameters of each fault state quantity at the initial stage of the transformer operation as the boundary value of the normal state;

[0018] Based on the measured values of the characteristic parameters of the fault state quantities during the operation of the transformer, calculate the boundary values of the attention state, abnormal state, and severe state through differential thresholds;

[0019] The probability density function of the Weibull model is as follows:

[0020]

[0021] The cumulative distribution function is as follows:

[0022]

[0023] Where: x is the measured value of the characteristic parameter; β is the shape parameter, reflecting the trend of the measured value of the characteristic parameter of the fault state quantity changing with time; α is called the scale parameter, reflecting the typical value or average level of the measured value of the characteristic parameter of the fault state quantity.

[0024] Use the maximum likelihood estimation method to determine the parameters α and β, and establish a system of likelihood function equations based on the measured values of the characteristic parameters during the operation of the transformer; initialize the parameters α0 and β0, and substitute the characteristic parameter data X = (x0, x1, …, x n ) into the system of likelihood equations, and obtain α and β after optimization.

[0025] Statistically calculate the probabilities of multiple transformers of the same type being in 4 states: normal, attention, abnormal, and severe during operation.

[0026] p = [p1, p2, p3, p4]; p1, p2, p3, p4 are the probabilities of a single transformer being in the 4 states of normal, attention, abnormal, and severe during operation, respectively.

[0027] Accumulate p successively to obtain the cumulative distribution probability F = [F1, F2, F3, 1].

[0028] Substitute F1, F2, and F3 into the above formula successively.

[0029] X p = α[-In(1 - F)] 1 / β

[0030] Calculate to obtain Xp = [x1, x2, x3], where x1, x2, x3 are the boundary values of the attention state, abnormal state, and severe state, respectively.

[0031] It should be noted that for the range of status variables specified in existing standards, direct reference can be made. For example, as described in the IEC standard, the thermal fault status is divided into low-temperature overheating, medium-temperature overheating, and high-temperature overheating. The threshold can be used to define the low-temperature overheating below 150°C as the normal value. In this case, the fault is judged by other status variables. The low-temperature overheating between 150 - 300°C is defined as the attention value, and the transformer status is jointly judged by combining other status variables in this state; the medium-temperature overheating between 300 - 700°C is defined as the warning value, and the high-temperature overheating above 700°C is defined as the outage value. For the top oil temperature, referring to the current and temperature limits in GB / T 1904.7 "Power Transformers - Part 7: Loading Guide for Oil-Immersed Power Transformers", it is required that the hot-spot temperature of the winding does not exceed 120°C and the top oil temperature does not exceed 105°C during the rated load operation of the transformer. This value can be used as the attention value of this status variable, and combined with the Weibull model, its abnormal value and severe value are calculated.

[0032] Preferably, by the combined weighting method, the comprehensive weights of the index layer and the scheme layer in the evaluation index system are calculated, including:

[0033] Based on the evaluation index system, combined with expert experience, discrimination matrices are constructed for the scheme layer and the index layer, and the consistency of the discrimination matrices is tested. By solving the eigenvectors of the discrimination matrices, the subjective weights of each scheme layer and index layer are obtained;

[0034] Based on the measured values of the characteristic parameters of each overheating fault status variable and the corresponding fault types during the operation of the transformer, the objective weights of the scheme layer and the index layer are calculated by the objective weighting method;

[0035] The subjective weights and objective weights are combined to form a weight set, and a linear combination is performed on the weight set. Through an optimization method, the deviation between the combined weight and each subjective weight and objective weight is minimized, and the optimized weight is normalized to obtain the final comprehensive weight.

[0036] Preferably, the above-mentioned based on the evaluation index system, combined with expert experience, constructing discrimination matrices for the scheme layer and the index layer, testing the consistency of the discrimination matrices, and obtaining the subjective weights of each scheme layer and index layer by solving the eigenvectors of the discrimination matrices includes:

[0037] The status variables at the same level in the evaluation index system are compared pairwise. Among the two status variables, the status variable with a greater impact on the upper level is more important, and a greater importance value is assigned to it, thereby obtaining the discrimination matrix A, and the expression is as follows:

[0038]

[0039] where, a ij represents the relative importance value of status variables i and j;

[0040] To conduct a consistency test on the discrimination matrix to make A a consistent matrix, the parameter for determining whether the discrimination matrix is a consistent matrix is CI, and the expression is as follows:

[0041]

[0042] n is the order of the discrimination matrix, that is, the number of state variables, and λ is the largest eigenvalue corresponding to the eigenvector of the discrimination matrix A. When CI = 0, the matrix A is a consistent matrix; the larger CI is, the more inconsistent the matrix A is;

[0043] When the consistency of the discrimination matrix A cannot be satisfied, the discrimination matrix cannot be used to determine the importance. The values in A should be adjusted until the consistency requirement is met. The consistency ratio CR is usually used to measure the consistency situation, and its definition is:

[0044]

[0045] In the formula, RI is called the random consistency index. The value of RI is determined according to the number of elements in the discrimination matrix A. Usually, 0.1 is considered the boundary for the matrix A to meet the consistency. If CR < 0.1, it is considered that the consistency test is satisfied. If the value of CR exceeds 0.1, it is considered that the consistency test is not satisfied; after normalizing the eigenvector of the discrimination matrix, the corresponding weights are obtained;

[0046] Based on the discrimination matrix, calculate the subjective weights of the state variables in the scheme layer, and the calculation result is denoted as w 2i ; Based on the discrimination matrix, calculate the basic weights of the state variables in the index layer, and the calculation result is denoted as w 1i , and combine the subjective weights of the state variables in the corresponding scheme layer to calculate the subjective weight w i of the i-th state variable in the index layer. The calculation formula is: w i = w 1i ·w 2i .

[0047] Preferably, based on the measured values of the characteristic parameters of each overheating fault state variable during the operation of the transformer and the corresponding fault types, calculate the objective weights of the scheme layer and the index layer through the objective weighting method, including:

[0048] Organize the fault types corresponding to the measured values of the characteristic parameters of each overheating fault state variable during the operation of the transformer into a transaction database. Each record represents the operating state of a transformer, specifically as follows:

[0049] 1) Transaction database D = {any comprehensive state variable exceeds the standard};

[0050] 2) Determine the item set X ij, = {the j-th single status quantity in the i-th comprehensive status quantity exceeds the standard};

[0051] 3) Determine the item set F i = {the i-th type of fault occurs};

[0052] When a power transformer fails, there is at least 1 type of comprehensive fault; the comprehensive fault type is presented by several single status quantities exceeding the standard; the confidence level of the transformer status assessment is calculated as follows:

[0053]

[0054] The weight coefficient is the ratio of the confidence level of the single event of the transformer to the total single confidence level, as follows:

[0055]

[0056] In the formula: θ ij is the weight coefficient of X i in F ij ; c ij is the confidence level of X i in F ij ; n is the total number of i in F i .

[0057] The present invention also provides a transformer overheating fault status assessment device based on temperature inversion, including:

[0058] An evaluation index system construction module, configured to collect the overheating fault status quantities of the transformer and the measured values of the characteristic parameters during its operation, and construct an evaluation index system based on the analytic hierarchy process. The evaluation index system includes a scheme layer and an index layer;

[0059] A comprehensive weight calculation module, configured to calculate the comprehensive weight of the index layer and the comprehensive weight of the scheme layer in the evaluation index system by the combined weighting method;

[0060] A membership degree calculation module, configured to obtain the actual operation data of each overheating fault status quantity of the target transformer corresponding to the evaluation index system, and calculate the membership degree of the target transformer corresponding to each fault status quantity;

[0061] A comprehensive evaluation module, configured to calculate the overheating fault status probability result of the target transformer based on the membership degrees of each fault status quantity of the target transformer, the comprehensive weight of the index layer, and the comprehensive weight of the scheme layer.

[0062] Preferably, the evaluation index system construction module is specifically configured to:

[0063] Collect the overheating fault status quantities of the transformer, classify them according to the characteristics of the overheating fault status quantities of the transformer and the evaluation purpose, and obtain the overheating fault status classification result;

[0064] Set the fault state levels of overheating fault state variables, and based on the measured values of the characteristic parameters of the overheating fault state variables during the operation of the transformer, calculate the boundary values of the fault state levels of each overheating fault state variable through differential thresholds;

[0065] Combined with the hierarchical division method, construct an evaluation index system with the overheating fault state evaluation result as the target layer, the overheating fault state classification result as the scheme layer, and the overheating fault state variables and the boundary values of their fault state levels as the index layer.

[0066] Preferably, the evaluation index system construction module is specifically used to set the fault state levels of overheating fault state variables, and based on the measured values of the characteristic parameters of the overheating fault state variables during the operation of the transformer, calculate the boundary values of the fault state levels of each overheating fault state variable through differential thresholds:

[0067] Set that the overheating fault state variable has 4 types of fault state levels: normal, attention, abnormal, and severe;

[0068] Calculate the average value of the measured values of the characteristic parameters of each fault state variable at the initial stage of the transformer operation as the boundary value of the normal state;

[0069] Based on the measured values of the characteristic parameters of the fault state variables during the operation of the transformer, calculate the boundary values of the attention state, abnormal state, and severe state through differential thresholds;

[0070] The probability density function of the Weibull model is as follows:

[0071]

[0072] The cumulative distribution function is as follows:

[0073]

[0074] Where: x is the measured value of the characteristic parameter; β is the shape parameter, reflecting the trend of the measured value of the characteristic parameter of the fault state variable changing with time; α is called the scale single number, reflecting the typical value or average level of the measured value of the characteristic parameter of the fault state variable;

[0075] Use the maximum likelihood estimation method to determine the parameters α and β, and establish a system of likelihood function equations according to the measured values of the characteristic parameters during the operation of the transformer; initialize the parameters α0 and β0, and substitute the characteristic parameter data X = (x0, x1,..., x n ) into the likelihood equation system, and obtain α and β after optimization;

[0076] Statistically calculate the probabilities of multiple transformers of the same type being in the normal, attention, abnormal, and severe 4 states during operation;

[0077] p = [p1, p2, p3, p4]; p1, p2, p3, and p4 are the probabilities of a single transformer being in the normal, attention, abnormal, and severe states during operation, respectively.

[0078] Cumulatively sum p successively to obtain the cumulative distribution probability F = [F1, F2, F3, 1].

[0079] Substitute F1, F2, and F3 into the above formula successively.

[0080] X p = α[-In(1 - F)] 1 / β

[0081] Calculate Xp = [x1, x2, x3], where x1, x2, and x3 are the boundary values between the normal state, attention state, abnormal state, and severe state, respectively.

[0082] Preferably, the comprehensive weight calculation module is specifically configured to:

[0083] Based on the evaluation index system, construct a discrimination matrix for the scheme layer and index layer in combination with expert experience, conduct a consistency test on the discrimination matrix, and obtain the subjective weights of each scheme layer and index layer by solving the eigenvector of the discrimination matrix.

[0084] Based on the measured values of the characteristic parameters of each overheating fault state quantity and the corresponding fault types during the operation of the transformer, calculate the objective weights of the scheme layer and index layer through the objective weighting method.

[0085] Form a weight set with the subjective weights and objective weights, perform a linear combination on the weight set, and through an optimization method, minimize the deviation between the combined weight and each subjective weight and objective weight, and normalize the optimized weight to obtain the final comprehensive weight.

[0086] Preferably, the comprehensive weight calculation module is specifically configured to construct a discrimination matrix for the scheme layer and index layer based on the evaluation index system in combination with expert experience, conduct a consistency test on the discrimination matrix, and obtain the subjective weights of each scheme layer and index layer by solving the eigenvector of the discrimination matrix:

[0087] Compare the state quantities at the same level in the evaluation index system in pairs. Among the two state quantities, the state quantity that has a greater impact on the upper layer is more important, and a greater importance value is assigned to it, thereby obtaining the discrimination matrix A, and the expression is as follows:

[0088]

[0089] Among them, a ij represents the relative importance value of state quantities i and j;

[0090] To conduct a consistency test on the discrimination matrix and make \(A\) a consistent matrix, the parameter for determining whether the discrimination matrix is a consistent matrix is \(CI\), and the expression is as follows:

[0091]

[0092] \(n\) is the order of the discrimination matrix, that is, the number of state variables, and \(\lambda\) is the largest eigenvalue corresponding to the eigenvector of the discrimination matrix \(A\). When \(CI = 0\), the matrix \(A\) is a consistent matrix; the larger \(CI\) is, the more inconsistent the matrix \(A\) is;

[0093] When the consistency of the discrimination matrix \(A\) cannot be satisfied, this discrimination matrix cannot be used for importance determination. The values in \(A\) should be adjusted until the consistency requirement is met. The consistency ratio \(CR\) is usually used to measure the consistency situation, and its definition is:

[0094]

[0095] In the formula, \(RI\) is called the random consistency index. The value of \(RI\) is determined according to the number of elements included in the discrimination matrix \(A\). Generally, it is considered that \(0.1\) is the boundary for the matrix \(A\) to meet the consistency. If \(CR < 0.1\), it is considered that the consistency test is satisfied. If the value of \(CR\) exceeds \(0.1\), it is considered that the consistency test is not satisfied; the corresponding weights are obtained after normalizing the eigenvector of the discrimination matrix;

[0096] Calculate the subjective weights of the state variables in the solution layer based on the discrimination matrix, and the calculation result is denoted as \(w\) 2i ; Calculate the basic weights of the state variables in the index layer based on the discrimination matrix, and the calculation result is denoted as \(w\) 1i , and combine the subjective weights of the state variables in the corresponding solution layer to calculate the subjective weight \(w\) i of the \(i\)-th state variable in the index layer. The calculation formula is: \(w\) i = \(w\) 1i · \(w\) 2i .

[0097] Preferably, the comprehensive weight calculation module is specifically configured to calculate the objective weights of the solution layer and the index layer based on the measured values of the characteristic parameters of each overheating fault state variable during the operation of the transformer and the corresponding fault types through the objective weighting method:

[0098] Organize the measured values of the characteristic parameters of each overheating fault state variable during the operation of the transformer and the corresponding fault types into a transaction database. Each record represents the operating state of a transformer, specifically as follows:

[0099] 1) Transaction database \(D=\) {any comprehensive state variable exceeds the standard};

[0100] 2) Determine the item set \(X\) ij , = {the \(j\)-th single state variable in the \(i\)-th comprehensive state variable exceeds the standard};

[0101] 3) Determine the item set F i = {The i-th type of fault occurs};

[0102] When a power transformer fails, there is at least one comprehensive fault type; the comprehensive fault type is presented by several single state variables exceeding the standard; the confidence level of the transformer state assessment is calculated as follows:

[0103]

[0104] The weight coefficient is the ratio of the confidence level of the single event of the transformer to the total single confidence level, as follows:

[0105]

[0106] In the formula: θ ij is the weight coefficient of X in F i ; c ij is the confidence level of X in F ij ; n is the total number of i in F i ; ij ; i ;

[0107] The present invention also provides a computer storage medium, the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned transformer overheating fault state assessment method based on temperature inversion are realized.

[0108] The present invention also provides an electronic device, including a memory and a processor: the memory is used to store computer-executable instructions, the processor is used to execute the computer-executable instructions, and when the computer-executable instructions are executed by the processor, the steps of the above-mentioned transformer overheating fault state assessment method based on temperature inversion are realized.

[0109] The beneficial effects of the present invention are as follows: By constructing a multi-level evaluation system including an objective layer, a scheme layer, and an index layer, the present invention comprehensively covers various state variables related to transformer overheating faults, ensuring the comprehensiveness and systematicness of the evaluation. Decomposing complex evaluation problems into multiple levels makes the evaluation process more organized and structured, facilitating understanding and operation. Using a differential threshold calculation method based on the Weibull distribution and combining with the equipment state distribution probability to scientifically determine the thresholds of each state variable, improving the rationality and accuracy of the thresholds. Using a combined weighting method, combining subjective weighting methods (such as the analytic hierarchy process) and objective weighting methods, not only considering the expert's experience judgment but also making full use of actual data, making the determination of weights more scientific and reliable. Combining the membership degree and weights to comprehensively evaluate the overall state of the transformer, not only considering the influence of a single index but also integrating the weight effects of multiple indexes, making the evaluation results more comprehensive and accurate.

[0110] The above description is only an overview of the technical solution of the present invention. In order to be able to understand the technical means of the present invention more clearly and implement it in accordance with the content of the specification, the following takes the preferred embodiments of the present invention and combines with the drawings to describe in detail as follows. The specific implementation manners of the present invention are given in detail by the following embodiments and their drawings. Brief Description of the Drawings

[0111] The drawings described herein are used to provide a further understanding of the present invention, and constitute a part of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation to the present invention. In the drawings:

[0112] Figure 1 A flowchart of a method for evaluating the overheating fault state of a transformer based on temperature inversion provided for Embodiment 1;

[0113] Figure 2 A schematic diagram of the evaluation index system provided for Embodiment 1;

[0114] Figure 3 A module diagram of a device for evaluating the overheating fault state of a transformer based on temperature inversion provided for Embodiment 2. Detailed Description of the Preferred Embodiments

[0115] The principles and features of the present invention are described below in conjunction with the drawings. The examples cited are only used to explain the present invention and are not intended to limit the scope of the present invention.

[0116] The following embodiments are all evaluated using the relevant data of a certain substation. The content of hydrogen and total hydrocarbons in the oil of the main transformer in the substation shows a continuous growth trend. From June 8, 2020 to June 26, 2020, the total hydrocarbons increased from 469.72 μL / L to 720 μL / L, and acetylene increased from 0.1 μL / L to 0.4 μL / L. The growth rate of total hydrocarbons accelerated. During this period, the valve group load was relatively high, about 1318.9 - 1559.84 MW, and the load rate of the converter transformer was at most about 94%. The chromatographic three-ratio code was 021 or 022, and there was a trend of the defect developing from medium-temperature overheating to high-temperature overheating. By analyzing the relationship between the oil chromatogram data and the load change, it was found that the contents of total hydrocarbons, H2, C2H2, C2H4, C2H6, CH4, and CO2 were positively correlated with the load.

[0117] Embodiment 1

[0118] As Figure 1 shown, this embodiment provides a method for evaluating the overheating fault state of a transformer based on temperature inversion, including the following steps:

[0119] S1. Collect the overheating fault state variables of the transformer and the measured values of the characteristic parameters during its operation, and construct an evaluation index system including an objective layer, a scheme layer, and an index layer based on the analytic hierarchy process. Specifically, it includes:

[0120] 1.1 Collect the overheating fault state variables of the transformer, classify them according to the characteristics of the overheating fault state variables of the transformer and the evaluation purpose, and obtain the classification results of the overheating fault states, as shown in Table 1:

[0121] Table 1 State variables and classification results of the transformer

[0122]

[0123] 1.2 Set the fault state levels of the overheating fault state variables, and based on the measured values of the characteristic parameters of the overheating fault state variables during the operation of the transformer, calculate the boundary values of the fault state levels of each overheating state variable through differential thresholds. Specifically, it includes:

[0124] Set that the overheating fault state variables have 4 types of fault state levels: normal, attention, abnormal, and serious, as shown in Table 2:

[0125] Table 2 Classification of the levels of the overheating fault state variables of the transformer

[0126] Serial number Status level Status description 1 H1 Normal The status quantity is stable and can continue to operate stably 2 H2 Attention The status quantity is close to or exceeds the attention threshold, and maintenance can be carried out according to the plan 3 H3 Abnormal The status quantity is close to or exceeds the abnormal threshold, and power outage maintenance should be planned in advance with priority 4 H4 Severe The status quantity is close to or exceeds the severe threshold, and power outage detection and maintenance should be completed as soon as possible

[0127] Calculate the average value of the measured values of the characteristic parameters of each fault state variable at the initial stage of the transformer operation as the boundary value of the normal state;

[0128] Based on the measured values of characteristic parameters of fault state variables during the operation of the transformer, calculate the boundary values of the attention state, abnormal state, and severe state through differential thresholds;

[0129] The probability density function of the Weibull model is as follows:

[0130]

[0131] The cumulative distribution function is as follows:

[0132]

[0133] Where: x is the measured value of the characteristic parameter; β is the shape parameter, reflecting the trend of the measured value of the characteristic parameter of the fault state variable changing with time; α is called the scale parameter, reflecting the typical value or average level of the measured value of the characteristic parameter of the fault state variable;

[0134] Use the maximum likelihood estimation method to determine the parameters α and β, and establish a system of likelihood function equations based on the measured values of the characteristic parameters during the operation of the transformer; Initialize the parameters α0 and β0, and substitute the characteristic parameter data X=(x0,x1,…,x n ) into the likelihood equation system, and obtain α and β after optimization;

[0135] Statistically calculate the probabilities of multiple transformers of the same type being in the normal, attention, abnormal, and severe states during operation;

[0136] p = [p1, p2, p3, p4]; p1, p2, p3, p4 are the probabilities of a single transformer being in the normal, attention, abnormal, and severe states during operation, respectively;

[0137] Cumulatively sum p in turn to obtain the cumulative distribution probability F = [F1, F2, F3, 1];

[0138] Substitute F1, F2, and F3 into the above formula in turn;

[0139] X p = α[-In(1 - F)] 1 / β

[0140] Calculate to obtain Xp = [x1, x2, x3], where x1, x2, and x3 are the boundary values of the attention state, abnormal state, and severe state, respectively.

[0141] It should be noted that the range of status variables specified in existing standards can be directly referred to. For example, as described in the IEC standard, the thermal fault status is divided into low-temperature overheating, medium-temperature overheating, and high-temperature overheating. The threshold can be used to define the low-temperature overheating below 150°C as the normal value, and the faults in this case are judged by other status variables. The low-temperature overheating of 150 - 300°C is defined as the attention value, and the transformer status is jointly judged with other status variables in this state; the medium-temperature overheating of 300 - 700°C is defined as the warning value, and the high-temperature overheating above 700°C is defined as the shutdown value. For the top oil temperature, referring to the current and temperature limits in GB / T 1904.7 "Power Transformers - Part 7: Loading Guide for Oil-Immersed Power Transformers", it is required that the hot spot temperature of the winding does not exceed 120°C and the top oil temperature does not exceed 105°C when the transformer is operating at rated load. This value can be used as the attention value of this status variable. Combining with the Weibull model, its abnormal value and severe value are calculated. The boundary values of each fault status level are shown in Table 3:

[0142] Table 3 Boundary Values of Each Fault Status Level of the Transformer

[0143]

[0144] Combined with the hierarchical division method, a hierarchical structure is constructed with the evaluation result of the overheating fault status as the target layer, the classification result of the overheating fault status as the scheme layer, and the boundary values of the overheating fault status variables and their fault status levels as the index layer. As Figure 2 shown, each fault status variable and the classification result of the fault status are indicators.

[0145] S2. By using the combined weighting method, calculate the comprehensive weights of the index layer and the scheme layer in the evaluation index system, specifically including:

[0146] 2.1 Based on the evaluation index system, combined with expert experience, construct discriminant matrices for the scheme layer and the index layer, and conduct consistency tests on the discriminant matrices. By solving the eigenvectors of the discriminant matrices, obtain the subjective weights of each scheme layer and index layer;

[0147] Compare the status variables at the same level in the evaluation index system in pairs. Among the two status variables, the one that has a greater impact on the upper layer is more important, and a greater importance value is assigned to it, thereby obtaining the discriminant matrix A. The expression is as follows:

[0148]

[0149] where a ij represents the relative importance value of status variables i and j;

[0150] Conduct a consistency test on the discriminant matrix to make A a consistent matrix. The parameter for whether the discriminant matrix is a consistent matrix is CI, and the expression is as follows:

[0151]

[0152] When CI = 0, matrix A is a consistent matrix; the larger CI is, the more inconsistent matrix A becomes;

[0153] n is the order of the discrimination matrix, that is, the number of state variables, λ is the largest eigenvalue corresponding to the eigenvector of the discrimination matrix A. When the consistency of the discrimination matrix A cannot be satisfied, this discrimination matrix cannot be used for importance determination. The values in A should be adjusted until the consistency requirement is met. The consistency ratio CR is usually used to measure the consistency situation, and its definition is:

[0154]

[0155] In the formula, RI is called the random consistency index. The value of RI is determined according to the number of elements in the discrimination matrix A. Usually, 0.1 is considered the boundary for matrix A to meet the consistency. If CR < 0.1, it is considered that the consistency test is passed; if the value of CR exceeds 0.1, it is considered that the consistency test is not passed; the corresponding weights are obtained after normalizing the eigenvector of the discrimination matrix.

[0156] Calculate the subjective weights of the state variables in the solution layer based on the discrimination matrix, and the calculation result is denoted as w 2i ; Calculate the basic weights of the state variables in the index layer based on the discrimination matrix, and the calculation result is denoted as w 1i , and combine the subjective weights of the state variables in the corresponding solution layer to calculate the subjective weight w i of the i-th state variable in the index layer. The calculation formula is: w i = w 1i · w 2i .

[0157] In this embodiment, 3 experts are invited to judge the weights of each index. The subjective weight results of each index (i.e., state variable) in the index layer are as follows:

[0158] X1 = [0.2733, 0.1160, 0.2733, 0.1160, 0.0418, 0.0418, 0.0689, 0.0689;

[0159] 0.2493, 0.0958, 0.6549; 0.1420, 0.4290, 0.4290];

[0160] X2 = [0.2466, 0.1504, 0.2466, 0.0906, 0.0906, 0.0555, 0.0906, 0.0291; 0.2971, 0.1629, 0.5400; 0.2971, 0.5400, 0.1629];

[0161] X3 = [0.2850, 0.1337, 0.2850, 0.1337, 0.0282, 0.0282, 0.0633, 0.0427; 0.2380, 0.1366, 0.6254; 0.6376, 0.2578, 0.1046];

[0162] The subjective weight results of each index (i.e., state variable) at the scheme layer are as follows:

[0163] Y1 = [0.6376, 0.2578, 0.1046];

[0164] Y2 = [0.1883, 0.7315, 0.0802];

[0165] Y3 = [0.3197, 0.5587, 0.1216].

[0166] 2.2 Based on the measured values of characteristic parameters of each overheating fault state variable during the operation of the transformer and the corresponding fault types, the objective weights of the scheme layer and the index layer are calculated through the objective weighting method, as follows:

[0167] The measured values of characteristic parameters of each overheating fault state variable during the operation of the transformer and the corresponding fault types are sorted into a transaction database, and each record represents the operating state of a transformer, as follows:

[0168] 1) Transaction database D = {Any comprehensive state variable exceeds the standard};

[0169] 2) Determine the item set X ij , = {The j-th single state variable in the i-th comprehensive state variable exceeds the standard};

[0170] 3) Determine the item set F i = {The i-th type of fault occurs};

[0171] When a power transformer fails, there is at least one comprehensive fault type; the comprehensive fault type is presented by several single state variables exceeding the standard; the confidence level of transformer state evaluation is calculated as follows:

[0172]

[0173] The weight coefficient is the ratio of the confidence level of a single event of the transformer to the total single confidence level, as follows:

[0174]

[0175] In the formula: θ ij is the weight coefficient of X i in F ij ; c ijis F i X in ij confidence; n is F i total number of i in F

[0176] 2.3 Combine the subjective weight and the objective weight to form a weight set, perform a linear combination on the weight set, and through an optimization method, minimize the deviation between the combined weight and each subjective weight and objective weight. Normalize the optimized weight to obtain the final comprehensive weight, as follows:

[0177] Suppose there are n different weight calculation methods to obtain the attribute weights, and n attribute weight vector sets W = {w1, w2, ……, w n}, and each weight vector wk (k = 1, 2, …, n) is a vector.

[0178] Perform an arbitrary linear combination on the n vectors to obtain possible weight vectors, and the expression is:

[0179]

[0180] In the formula, a k is the coefficient of the linear combination, and T is the transpose operation, which means interchanging the rows and columns of the matrix. Specifically, it is to transpose a row vector into a column vector or a column vector into a row vector.

[0181] Based on the optimal strategy idea of game theory, find the optimal solution w* among all possible weight vectors, and perform optimization processing on the coefficients α k of the n linear combinations, so that the deviation between w and w k is the smallest, that is:

[0182]

[0183] According to the above formula, obtain the weight coefficients (a1, a2, ……, a n ), and then normalize the weight coefficients to obtain the optimal weight of each original weight vector in the combined weight vector

[0184]

[0185] Finally, obtain the comprehensive weight:

[0186]

[0187] The results are shown in Table 4

[0188] Table 4 Comprehensive weights of the scheme layer and index layer indicators

[0189]

[0190] S3. Obtain the actual operation data of each overheating fault state quantity in the corresponding evaluation index system of the target transformer, and calculate the membership degrees of the corresponding fault state quantities of the target transformer as follows:

[0191] The membership function is usually represented in the form of a combination of a semi-trapezoid and a semi-ridge shape containing four sub-membership functions, including four situations: normal state, attention state, abnormal state, and severe state. Among them, the normal state belongs to a descending membership function, the attention state and the abnormal state belong to intermediate membership functions, and the severe state belongs to an ascending membership function. x1 to x3 are the boundary values of each membership function, and the values are shown in Table 5.

[0192] Table 5 Values corresponding to the boundary values of the membership function

[0193] Boundary value of membership function Value <![CDATA[X1]]> Boundary value of attention status <![CDATA[X2]]> Boundary value of abnormal status <![CDATA[X3]]> Boundary value of severe status

[0194] The membership function can be obtained by the following formula:

[0195]

[0196] Based on the boundary values of the fault state levels determined above, map the actual values of the fault state quantities of each index in the corresponding evaluation index system of the target transformer one by one, and calculate the membership degrees of each index of the target transformer through membership degree calculation. The results are shown in Table 6:

[0197] Table 6 Membership degrees of each index of the target transformer

[0198]

[0199]

[0200] S4. Based on the membership degrees of the various fault state quantities of the target transformer, the comprehensive weights of the index layer, and the comprehensive weights of the scheme layer, calculate the membership degrees of the corresponding fault state quantities of the target transformer as follows:

[0201] Multiply the comprehensive weights of the various indexes in the index layer by the corresponding membership degree matrix as follows:

[0202]

[0203]

[0204] Multiply the product results A1, A2, and A3 of the comprehensive weights of the various indexes in the index layer and the corresponding membership degree matrix by the comprehensive weights of the various indexes in the index layer to obtain the overall state evaluation result of the transformer as follows:

[0205]

[0206] According to the results, the probability that the transformer is in normal operation is 47.32%, and the probability that it is in a severe state is 33.9%. Combining the analysis of single-index state variables, the total hydrocarbon, the absolute increment of total hydrocarbon, and the increment of ethylene all exceed the attention values. The relevant increments are positively correlated with the load. According to the three-ratio method, it is judged that the transformer has a high-temperature overheating fault. According to the suggestions for the transformer state in the specification, it is considered that the transformer should complete the power-off maintenance as soon as possible.

[0207] Embodiment 2

[0208] As Figure 3 shown, this embodiment provides a transformer overheating fault state evaluation device based on temperature inversion, including:

[0209] An evaluation index system construction module, which is used to collect the state variables of the transformer overheating fault and the measured values of the characteristic parameters during its operation, and construct an evaluation index system based on the analytic hierarchy process. The evaluation index system includes a scheme layer and an index layer.

[0210] Specifically for:

[0211] Collect the state variables of the transformer overheating fault, classify them according to the characteristics of the state variables of the transformer overheating fault and the evaluation purpose, and obtain the classification result of the overheating fault state;

[0212] Set the fault state levels of the state variables of the overheating fault, and based on the measured values of the characteristic parameters of the state variables of the overheating fault during the operation of the transformer, calculate the boundary values of the fault state levels of each state variable of the overheating fault through differential thresholds;

[0213] Combined with the hierarchical division method, construct an evaluation index system with the overheating fault state evaluation result as the target layer, the overheating fault state classification result as the scheme layer, and the state variables of the overheating fault and the boundary values of their fault state levels as the index layer.

[0214] The evaluation index system construction module is specifically used to set the fault state levels of the state variables of the overheating fault, and based on the measured values of the characteristic parameters of the state variables of the overheating fault during the operation of the transformer, calculate the boundary values of the fault state levels of each state variable of the overheating fault through differential thresholds:

[0215] Set that the state variables of the overheating fault have 4 types of fault state levels: normal, attention, abnormal, and severe;

[0216] Calculate the average value of the measured values of the characteristic parameters of each fault state variable at the initial stage of the transformer operation as the boundary value of the normal state;

[0217] Based on the measured values of the characteristic parameters of the state variables of the overheating fault during the operation of the transformer, calculate the boundary values of the attention state, abnormal state, and severe state through differential thresholds;

[0218] The probability density function of the Weibull model is as follows:

[0219]

[0220] The cumulative distribution function is as follows:

[0221]

[0222] Where: x is the measured value of the characteristic parameter; β is the shape parameter, reflecting the trend of the measured value of the characteristic parameter of the fault state quantity changing with time; α is called the scale parameter, reflecting the typical value or average level of the measured value of the characteristic parameter of the fault state quantity;

[0223] The maximum likelihood estimation method is used to determine the parameters α and β. According to the measured values of the characteristic parameters during the operation of the transformer, a system of likelihood function equations is established; the parameters α0 and β0 are initialized, and the characteristic parameter data X = (x0, x1, …, x n ) is substituted into the likelihood equation system, and α and β are obtained after optimization;

[0224] Statistically calculate the probabilities of multiple transformers of the same type being in 4 states of normal, attention, abnormal, and severe during operation;

[0225] p = [p1, p2, p3, p4]; p1, p2, p3, p4 are the probabilities of a single transformer being in the 4 states of normal, attention, abnormal, and severe during operation respectively;

[0226] Accumulate p successively to obtain the cumulative distribution probability F = [F1, F2, F3, 1];

[0227] Substitute F1, F2, and F3 into the above formula successively;

[0228] X p = α[-In(1 - F)] 1 / β

[0229] Calculate Xp = [x1, x2, x3], where x1, x2, x3 are the boundary values between the normal state, attention state, abnormal state, and severe state respectively.

[0230] The comprehensive weight calculation module is used to calculate the comprehensive weights of the index layer and the scheme layer in the evaluation index system through the combined weighting method,

[0231] Specifically used for:

[0232] Based on the evaluation index system, combined with expert experience, construct a discrimination matrix for the scheme layer and the index layer, and conduct a consistency test on the discrimination matrix. By solving the eigenvector of the discrimination matrix, obtain the subjective weights of each scheme layer and index layer;

[0233] Based on the measured values of the characteristic parameters of each overheating fault state variable and the corresponding fault types during the operation of the transformer, calculate the objective weights of the scheme layer and the index layer through the objective weighting method;

[0234] Form a weight set with the subjective weight and the objective weight, perform a linear combination on the weight set, and through an optimization method, minimize the deviation between the combined weight and each subjective weight and objective weight, and perform a normalization process on the optimized weight to obtain the final comprehensive weight.

[0235] The comprehensive weight calculation module is specifically used to construct a discrimination matrix for the scheme layer and the index layer based on the evaluation index system and combined with expert experience, and perform a consistency test on the discrimination matrix. By solving the eigenvector of the discrimination matrix, obtain the subjective weights of each scheme layer and index layer:

[0236] Compare the state variables at the same level in the evaluation index system in pairs. Among the two state variables, the state variable that has a greater impact on the upper layer is more important, and a greater importance value is assigned to it, so as to obtain the discrimination matrix A, and the expression is as follows:

[0237]

[0238] Among them, a ij represents the relative importance value of state variables i and j;

[0239] It is necessary to perform a consistency test on the discrimination matrix to make A a consistent matrix. The parameter for whether the discrimination matrix is a consistent matrix is CI, and the expression is as follows:

[0240]

[0241] n is the order of the discrimination matrix, that is, the number of state variables, and λ is the largest eigenvalue corresponding to the eigenvector of the discrimination matrix A. When CI = 0, the matrix A is a consistent matrix; the larger CI is, the more inconsistent the matrix A is;

[0242] When the consistency of the discrimination matrix A cannot be satisfied, the discrimination matrix cannot be used to determine the importance, and the values in A should be adjusted until the consistency requirement is met. The consistency ratio CR is usually used to measure the consistency situation, and its definition is:

[0243]

[0244] In the formula, RI is called the random consistency index. The value of RI is determined according to the number of elements contained in the discrimination matrix A. Usually, it is considered that 0.1 is the boundary for the matrix A to meet the consistency. If CR < 0.1, it is considered to meet the consistency test. If the value of CR exceeds 0.1, it is considered not to meet the consistency test; the eigenvector of the discrimination matrix is normalized to obtain the corresponding weight;

[0245] Calculate the subjective weight of the state quantity in the solution layer based on the discrimination matrix, and record the calculation result as w 2i ; Calculate the basic weight of the state quantity in the index layer based on the discrimination matrix, and record the calculation result as w 1i , and calculate the subjective weight w i of the i-th state quantity in the index layer by combining the subjective weight of the state quantity in the corresponding solution layer. The calculation formula is: w i = w 1i · w 2i .

[0246] The comprehensive weight calculation module is specifically used to calculate the objective weights of the solution layer and the index layer based on the measured values of the characteristic parameters of each overheating fault state quantity during the operation of the transformer and the corresponding fault types through the objective weighting method:

[0247] Organize the measured values of the characteristic parameters of each overheating fault state quantity during the operation of the transformer and the corresponding fault types into a transaction database, and each record represents the operating state of a transformer, specifically as follows:

[0248] 1) Transaction database D = {any comprehensive state quantity exceeds the standard};

[0249] 2) Determine the item set X ij , = {the j-th single state quantity in the i-th comprehensive state quantity exceeds the standard};

[0250] 3) Determine the item set F i = {the i-th type of fault occurs};

[0251] When a power transformer fails, there is at least one comprehensive fault type; the comprehensive fault type is presented by several single state quantities exceeding the standard; the confidence level of transformer state evaluation is calculated as follows:

[0252]

[0253] The weight coefficient is the ratio of the confidence level of the single event of the transformer to the total single confidence level, as follows:

[0254]

[0255] In the formula: θ ij is the weight coefficient of X i in F ij ; c ij is the confidence level of X i in F ij ; n is the total number of i in F i .

[0256] The membership degree calculation module is used to obtain the actual operation data of each overheating fault state quantity in the evaluation index system corresponding to the target transformer, and calculate the membership degree of each fault state quantity corresponding to the target transformer.

[0257] The comprehensive evaluation module is used to calculate the overheating fault state probability result of the target transformer based on the membership degree of each fault state quantity of the target transformer, the comprehensive weight of the index layer, and the comprehensive weight of the solution layer.

[0258] Embodiment 3

[0259] This embodiment provides a computer storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the steps of the method for evaluating the overheating fault state of a transformer based on temperature inversion as described in Embodiment 1.

[0260] Embodiment 4

[0261] This embodiment provides an electronic device, including a memory and a processor: the memory is used to store computer-executable instructions, the processor is used to execute the computer-executable instructions, and when the computer-executable instructions are executed by the processor, it implements the steps of the method for evaluating the overheating fault state of a transformer based on temperature inversion as described in Embodiment 1.

[0262] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can adopt the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0263] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the specified functions in one Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0264] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means that implement the function specified in the flowchart(s) Figure 1 a flowchart or flowcharts and / or block(s) Figure 1 a block or blocks.

[0265] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing steps for implementing the function specified in the flowchart(s) Figure 1 a flowchart or flowcharts and / or block(s) Figure 1 a block or blocks on the computer or other programmable apparatus.

Claims

1. A method for evaluating the overheating fault state of a transformer based on temperature inversion, characterized in that It includes the following steps: Collect the overheating fault state variables of the transformer and the measured values of the characteristic parameters during its operation, and construct an evaluation index system based on the analytic hierarchy process. The evaluation index system includes a scheme layer and an index layer; Calculate the comprehensive weights of the index layer and the comprehensive weights of the scheme layer in the evaluation index system through the combined weighting method; Obtain the actual operation data of each overheating fault state variable in the evaluation index system corresponding to the target transformer, and calculate the membership degree of each fault state variable corresponding to the target transformer; Based on the membership degree of each fault state variable of the target transformer, the comprehensive weight of the index layer, and the comprehensive weight of the scheme layer, calculate the probability result of the overheating fault state of the target transformer.

2. The method for evaluating the overheating fault state of a transformer based on temperature inversion according to claim 1, wherein, The evaluation index system also includes a target layer; the collection of the overheating fault state variables of the transformer and the measured values of the characteristic parameters during its operation, and the construction of the evaluation index system based on the analytic hierarchy process includes: Collect the overheating fault state variables of the transformer, classify them according to the characteristics of the overheating fault state variables of the transformer and the evaluation purpose, and obtain the overheating fault state classification result; Set the fault state levels of the overheating fault state variables, and based on the measured values of the characteristic parameters of the overheating fault state variables during the operation of the transformer, calculate the boundary values of the fault state levels of each overheating fault state variable through differential thresholds; Combined with the hierarchical division method, construct an evaluation index system with the overheating fault state evaluation result as the target layer, the overheating fault state classification result as the scheme layer, and the overheating fault state variables and the boundary values of their fault state levels as the index layer.

3. The method for evaluating the overheating fault state of a transformer based on temperature inversion according to claim 2, characterized in that, The setting of the fault state levels of the overheating fault state variables and the calculation of the boundary values of each fault state level based on the measured values of the characteristic parameters of the overheating fault state variables during the operation of the transformer include: Set that the overheating fault state variables have 4 types of fault state levels: normal, attention, abnormal, and severe; Calculate the average value of the measured values of the characteristic parameters of each fault state variable at the initial stage of the transformer operation as the boundary value of the normal state; Based on the measured values of the characteristic parameters of the fault state variables during the operation of the transformer, calculate the boundary values of the attention state, abnormal state, and severe state through differential thresholds; The probability density function of the Weibull model is as follows: The cumulative distribution function is as follows: In the formula: x is the measured value of the characteristic parameter; β is the shape parameter, reflecting the trend of the measured value of the characteristic parameter of the fault state variable changing with time; α is called the scale parameter, reflecting the typical value or average level of the measured value of the characteristic parameter of the fault state variable; Determine the parameters α and β using the maximum likelihood estimation method, and establish a system of likelihood function equations based on the measured values of characteristic parameters during the operation of the transformer; initialize the parameters α0 and β0, and substitute the characteristic parameter data X = (x0, x1, …, x n ) into the system of likelihood equations, and obtain α and β after optimization; Statistically calculate the probabilities of multiple transformers of the same type being in 4 states: normal, attention, abnormal, and severe during operation; p = [p1, p2, p3, p4]; p1, p2, p3, p4 are the probabilities of a single transformer being in 4 states: normal, attention, abnormal, and severe during operation respectively; Cumulatively add p in sequence to obtain the cumulative distribution probability F = [F1, F2, F3, 1]; Substitute F1, F2, and F3 into the above formula in sequence; X p = α[-In(1 - F)] 1 / β Calculate to obtain Xp = [x1, x2, x3], where x1, x2, x3 are the boundary values of the attention state, abnormal state, and severe state respectively.

4. The method for evaluating the overheating fault state of a transformer based on temperature inversion according to claim 1, characterized in that, Calculate the comprehensive weights of the index layer and the comprehensive weights of the scheme layer in the evaluation index system through the combined weighting method, including: Based on the evaluation index system, combined with expert experience, construct discrimination matrices for the scheme layer and the index layer, conduct consistency tests on the discrimination matrices, and obtain the subjective weights of each item in the scheme layer and the index layer by solving the eigenvectors of the discrimination matrices; Based on the measured values of the characteristic parameters of each overheating fault state quantity during the operation of the transformer and the corresponding fault types, calculate the objective weights of the scheme layer and the index layer by the objective weighting method; Form a weight set with the subjective weights and the objective weights, perform a linear combination on the weight set, and through an optimization method, minimize the deviation between the combined weights and each subjective weight and objective weight, and normalize the optimized weights to obtain the final comprehensive weights.

5. The method for evaluating the overheating fault state of a transformer based on temperature inversion according to claim 4, characterized in that The method of constructing discrimination matrices for the scheme layer and the index layer based on the evaluation index system, combined with expert experience, conducting consistency tests on the discrimination matrices, and obtaining the subjective weights of each item in the scheme layer and the index layer by solving the eigenvectors of the discrimination matrices includes: Compare the state quantities at the same level in the evaluation index system in pairs, and assign a greater importance value to the state quantity that has a greater impact on the upper layer among the two state quantities, so as to obtain the discrimination matrix A, and the expression is as follows: where a ij represents the relative importance value of state variables i and j; Conduct a consistency test on the discrimination matrix to make A a consistent matrix. The parameter for determining whether the discrimination matrix is a consistent matrix is CI, and the expression is as follows: n is the order of the discrimination matrix, that is, the number of state quantities, and λ is the largest eigenvalue corresponding to the eigenvector of the discrimination matrix A. When CI = 0, the matrix A is a consistent matrix; the larger CI is, the more inconsistent the matrix A is; When the consistency of the discrimination matrix A cannot be satisfied, the discrimination matrix cannot be used to determine the importance; adjust the values in the matrix A until the set consistency ratio CR is satisfied, and its definition is: In the formula, RI is called the random consistency index. Determine the value of RI according to the number of elements contained in the discrimination matrix A, and normalize the eigenvector of the discrimination matrix A to obtain the corresponding weight; Calculate the subjective weight of the state quantity in the solution layer based on the discrimination matrix, and record the calculation result as w 2i ; Calculate the basic weight of the state quantity in the index layer based on the discrimination matrix, and record the calculation result as w 1i , and calculate the subjective weight w i of the i-th state quantity in the index layer by combining the subjective weight of the corresponding state quantity in the solution layer. The calculation formula is: w i = w 1i · w 2i .

6. The method for evaluating the overheating fault state of a transformer based on temperature inversion according to claim 4, wherein, The method of calculating the objective weights of the scheme layer and the index layer based on the measured values of the characteristic parameters of each overheating fault state quantity during the operation of the transformer and the corresponding fault types by the objective weighting method includes: Organize the measured values of the characteristic parameters of each overheating fault state quantity during the operation of the transformer and the corresponding fault types into a transaction database, and each record represents the operation state of a transformer, specifically as follows: 1) The transaction database D = {any comprehensive state quantity exceeds the standard}; 2) Determine the item set X ij , = {the j-th single state quantity in the i-th comprehensive state quantity exceeds the standard} 3) Determine the item set F i = {The occurrence of the i-th type of fault}; When a power transformer fails, there is at least 1 comprehensive fault type; the comprehensive fault type is presented by several single state quantities exceeding the standard; the confidence level of the transformer state evaluation is calculated as follows: The weight coefficient is the ratio of the confidence level of the single item of the transformer to the total confidence level of the single item, as follows: Where: θ ij is the weight coefficient of X i in F ij ; c ij is the confidence of X i in F ij ; n is the total number of i in F i .

7. A transformer overheating fault state evaluation device based on temperature inversion, including: An evaluation index system construction module, configured to collect the overheating fault state quantities of the transformer and the measured values of their characteristic parameters during the operation process, and construct an evaluation index system based on the analytic hierarchy process. The evaluation index system includes a scheme layer and an index layer; A comprehensive weight calculation module, configured to calculate the comprehensive weights of the index layer and the scheme layer in the evaluation index system by the combined weighting method; The membership degree calculation module is used to obtain the actual operation data of each overheating fault state quantity in the evaluation index system corresponding to the target transformer, and calculate the membership degree of each fault state quantity corresponding to the target transformer; The comprehensive evaluation module is used to calculate the overheating fault state probability result of the target transformer based on the membership degree of each fault state quantity of the target transformer, the comprehensive weight of the index layer, and the comprehensive weight of the scheme layer.

8. The device for evaluating the overheating fault state of a transformer based on temperature inversion according to claim 7, wherein The evaluation index system construction module is specifically used for: Collect the overheating fault state quantities of the transformer, classify them according to the characteristics of the overheating fault state quantities of the transformer and the evaluation purpose, and obtain the overheating fault state classification result; Set the fault state levels of the overheating fault state quantities, and based on the measured values of the characteristic parameters of the overheating fault state quantities during the operation of the transformer, calculate the boundary values of the fault state levels of each overheating fault state quantity through differential thresholds; Combined with the hierarchical division method, construct an evaluation index system with the overheating fault state evaluation result as the target layer, the overheating fault state classification result as the scheme layer, and the overheating fault state quantity and the boundary values of its fault state levels as the index layer.

9. The transformer overheating fault state evaluation device based on temperature inversion according to claim 8, characterized in that The evaluation index system construction module is specifically used to set the fault state levels of the overheating fault state quantities, and based on the measured values of the characteristic parameters of the overheating fault state quantities during the operation of the transformer, calculate the boundary values of the fault state levels of each overheating fault state quantity through differential thresholds: Set that the overheating fault state quantity has 4 types of fault state levels: normal, attention, abnormal, and severe; Calculate the average value of the measured values of the characteristic parameters of each fault state quantity at the initial stage of the transformer operation as the boundary value of the normal state; Based on the measured values of the characteristic parameters of the fault state quantities during the operation of the transformer, calculate the boundary values of the attention state, abnormal state, and severe state through differential thresholds; The probability density function of the Weibull model is as follows: The cumulative distribution function is as follows: Where: x is the measured value of the characteristic parameter; β is the shape parameter, reflecting the trend of the measured value of the characteristic parameter of the fault state quantity changing with time; α is called the scale single number, reflecting the typical value or average level of the measured value of the characteristic parameter of the fault state quantity; Use the maximum likelihood estimation method to determine the parameters α and β, and establish a system of likelihood function equations based on the measured values of characteristic parameters during the operation of the transformer; initialize the parameters α0 and β0, and substitute the characteristic parameter data X = (x0, x1, …, x n ) into the system of likelihood equations, and obtain α and β after optimization; Statistically calculate the probabilities of multiple transformers of the same type being in the normal, attention, abnormal, and severe 4 states during operation; p = [p1, p2, p3, p4]; p1, p2, p3, p4 are the probabilities of a single transformer being in the normal, attention, abnormal, and severe 4 states during operation respectively; Cumulatively add p in sequence to obtain the cumulative distribution probability F = [F1, F2, F3, 1]; Substitute F1, F2, and F3 into the above formula in sequence; X p = α[-In(1 - F)] 1 / β Calculate to obtain Xp = [x1, x2, x3], where x1, x2, x3 are the boundary values of the attention state, abnormal state, and severe state respectively.

10. The device for evaluating the overheating fault state of a transformer based on temperature inversion according to claim 8, characterized in that, The comprehensive weight calculation module is specifically used for: Based on the evaluation index system, combine expert experience to construct discrimination matrices for the scheme layer and the index layer, conduct consistency tests on the discrimination matrices, and obtain the subjective weights of each scheme layer and index layer by solving the eigenvectors of the discrimination matrices; Based on the measured values of the characteristic parameters of each overheating fault state quantity during the operation of the transformer and the corresponding fault types, calculate the objective weights of the scheme layer and the index layer through the objective weighting method; Combine the subjective weights and objective weights to form a weight set, perform a linear combination on the weight set, and through an optimization method, minimize the deviation between the combined weights and each subjective weight and objective weight. Normalize the optimized weights to obtain the final comprehensive weights.

11. The device for evaluating the overheating fault state of a transformer based on temperature inversion according to claim 10, characterized in that, The comprehensive weight calculation module is specifically configured to construct a discrimination matrix for the scheme layer and the index layer based on the evaluation index system and in combination with expert experience, and perform a consistency test on the discrimination matrix. By solving the eigenvector of the discrimination matrix, the subjective weights of each scheme layer and index layer are obtained: Compare the state quantities at the same level in the evaluation index system in pairs, and assign a greater importance value to the state quantity that has a greater impact on the upper layer among the two state quantities, so as to obtain the discrimination matrix A, and the expression is as follows: Among them, a ij represents the relative importance value of state variables i and j; Perform a consistency test on the discrimination matrix to make A a consistent matrix. The parameter for whether the discrimination matrix is a consistent matrix is CI, and the expression is as follows: n is the order of the discrimination matrix, that is, the number of state quantities, and λ is the largest eigenvalue corresponding to the eigenvector of the discrimination matrix A. When CI = 0, the matrix A is a consistent matrix; the larger CI is, the more inconsistent the matrix A is; When the consistency of the discrimination matrix A cannot be satisfied, the discrimination matrix cannot be used to determine the importance; adjust the values in the matrix A until the set consistency ratio CR is satisfied, and its definition is: In the formula, RI is called the random consistency index, and the value of RI is determined according to the number of elements included in the discrimination matrix A. After normalizing the eigenvector of the discrimination matrix A, the corresponding weights are obtained; Calculate the subjective weight of the state variables in the scheme layer based on the discrimination matrix, and denote the calculation result as w 2i ; Calculate the basic weight of the state variables in the index layer based on the discrimination matrix, and denote the calculation result as w 1i , and calculate the subjective weight w i of the i-th state variable in the index layer by combining the subjective weight of the corresponding state variable in the scheme layer. The calculation formula is: w i = w 1i · w 2i .

12. The transformer overheating fault state evaluation device based on temperature inversion according to claim 11, characterized in that The comprehensive weight calculation module is specifically configured to calculate the objective weights of the scheme layer and the index layer through an objective weighting method based on the measured values of the characteristic parameters of each overheating fault state quantity during the operation of the transformer and the corresponding fault types. Organize the measured values of the characteristic parameters of each overheating fault state quantity during the operation of the transformer and the corresponding fault types into a transaction database, and each record represents the operation state of a transformer, specifically as follows: 1) Transaction database D = {any comprehensive state quantity exceeds the standard}; 2) Determine the item set X ij = {the j-th single status quantity in the i-th comprehensive status quantity exceeds the standard} 3) Determine the item set F i = {The occurrence of the i-th type of fault}; When a power transformer fails, there is at least 1 comprehensive fault type; the comprehensive fault type is presented by several single state quantities exceeding the standard; the confidence level of the transformer state evaluation is calculated as follows: The weight coefficient is the ratio of the confidence level of a single transformer event to the total single confidence level, as follows: Where: θ ij is the weight coefficient of X i in F ij ; c ij is the confidence level of X i in F ij ; n is the total number of i i in F.

13. A computer storage medium, the computer-readable storage medium stores a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the transformer overheating fault state evaluation method based on temperature inversion according to any one of claims 1-6.

14. An electronic device, characterized in that, It includes a memory and a processor: the memory is used to store computer-executable instructions, the processor is used to execute the computer-executable instructions, and when the computer-executable instructions are executed by the processor, it implements the steps of the transformer overheating fault state evaluation method based on temperature inversion according to any one of claims 1-6.

Images