Transformer performance evaluation method and system based on improved AHP considering scene applicability

Abstract

The application discloses a transformer performance evaluation method and system based on improved AHP considering scene applicability, which comprises the following steps: firstly, a comprehensive performance evaluation index system is constructed; then, a judgment matrix is corrected to improve the analytic hierarchy process, and the comprehensive subjective weight of each index is obtained; then, the variable weight theory of random field scene is considered, the index state vector is calculated, and the final variable weight vector is obtained through re-clustering based on the specification cutting thought; finally, the weighted deviation degree is calculated based on the cloud centroid model and the final variable weight vector, so that the comprehensive performance of the three-phase oil-immersed transformer is comprehensively evaluated; the application constructs a comprehensive three-phase oil-immersed transformer comprehensive performance evaluation index system, investigates the transformer performance from multiple dimensions, and improves the adaptability in different scenes by means of the specification cutting thought and the random field scene; finally, the cloud centroid model effectively processes the uncertainty of information, so that a more accurate and effective basis is provided for the performance evaluation of the transformer.

Description

TECHNICAL FIELD

[0001] The present application relates to transformer performance evaluation means, and belongs to the field of transformer comprehensive performance evaluation, in particular to a transformer performance evaluation method and system based on improved AHP considering scene applicability. BACKGROUND

[0002] The establishment of new power systems and power spot markets causes power loads to present characteristics such as diversification, variable space-time, increased peak-valley difference, and load surge. However, due to the rated load constraint of power equipment, the planning, scheduling and operation strategies of conventional systems cannot meet the actual requirements, and characteristics such as decreased adequacy, lack of flexibility, and weak investment appear. Therefore, to slow down system investment and ensure the safety of power systems, it is urgent to develop comprehensive performance evaluation of power lines and transformers.

[0003] In the prior art, the difficulty of transformer performance evaluation mainly lies in how to determine the evaluation indexes and the weight distribution of the evaluation indexes, and most conventional methods need to determine the weight of each index according to expert opinions, and subjective weighting methods are used, which excessively rely on expert experience. Even if there are methods of using group decision-making and fuzzy mathematics to obtain the index weight, the problem of fuzzy uncertainty in constructing the comparison judgment matrix is solved, but the importance of each index is still determined by expert opinions, and the index weight is determined from a single aspect, so that the weight cannot simultaneously consider subjectivity and objectivity, and further leads to low accuracy of the evaluation result. SUMMARY

[0004] The present application aims to overcome the above-mentioned defects and problems in the prior art, and provides a transformer performance evaluation method and system based on improved AHP considering scene applicability, which has high accuracy.

[0005] To achieve the above-mentioned purpose, the technical solution of the present application is as follows: a transformer performance evaluation method based on improved AHP considering scene applicability, comprising:

[0006] S1, constructing a three-phase oil-immersed transformer comprehensive performance evaluation index system;

[0007] S2, correcting the judgment matrix based on the zero deviation matrix, to improve the analytic hierarchy process, and obtain the comprehensive subjective weight of each index in the evaluation index system;

[0008] S3, based on the variable weight theory considering random scenes, calculating the index state vector of the comprehensive subjective weight, and re-clustering the index state vector through the standard cut idea, to obtain the final variable weight vector;

[0009] S4, based on the cloud centroid model and the final variable weight vector, a weighted deviation degree is calculated to comprehensively evaluate the comprehensive performance of the three-phase oil-immersed transformer.

[0010] The step S2 specifically comprises:

[0011] S21, by expert scoring and using a 1-9 scale method, the importance of each index in the comprehensive performance evaluation index system of the three-phase oil-immersed transformer is determined, and a judgment matrix A is constructed, A=[a ij ] n×n ;

[0012] The expression of the judgment matrix is as follows:

[0013]

[0014] Wherein: a ij is the scale value between the index and the index;

[0015] S22, the consistency index CI is calculated, and the expression is as follows:

[0016]

[0017] Wherein: λ max is the maximum eigenvalue of the judgment matrix A, and n is the order of the judgment matrix;

[0018] S23, the average random consistency index RI is found and the consistency ratio CR is calculated, and the expression is as follows:

[0019]

[0020] If CR<0.1, step S29 is performed; if CR≥0.1, step S24 is performed to modify the judgment matrix;

[0021] S24, the judgment matrix is expressed as:

[0022]

[0023] W=[W ij ]=[w i / w j ],w=(w1...w i ...w n ) T ;D=[d ij ];

[0024] Wherein: A is the judgment matrix, W is the maximum eigenvector of the judgment matrix A, D is the deviation matrix, is the symbol of Hadamard product;

[0025] When A satisfies the consistency matrix of AHP, all d ij = 1 and λ max (D) = λ max (A) = n; a specific deviation matrix with all d ij = 1 is called a zero deviation matrix, denoted as D I;

[0026] S25, let the judgment matrix A = [a * ] be the initial judgment matrix, the consistency ratio standard CR * = 0.1, the iteration number k = 0, and the constant γ is a positive number less than 1 but close to 1;

[0027] S26, calculate the maximum eigenvalue λ (k) (A max ) of the judgment matrix A (k) after the kth iteration; wherein the maximum eigenvalue vector w (k) and the priority vector W (k) are as follows:

[0028]

[0029] and the deviation matrix D (k) after the kth iteration is obtained according to ;

[0030] S27, calculate the consistency ratio CR (k) after the kth iteration, and compare CR (k) with CR * ;

[0031] If CR (k) > CR * , step S28 is performed; if CR (k) ≤ CR * , the judgment matrix A (k) satisfying the consistency requirement is output, and the iteration ends;

[0032] S28, let be the modified judgment matrix; wherein the modified deviation matrix is as follows:

[0033]

[0034] and step S22 is performed until the judgment matrix satisfying the consistency requirement is obtained;

[0035] S29, based on the judgment matrix A' = [a' ij ] satisfying the consistency requirement, the comprehensive index weight of each index is calculated using the geometric mean method, and an initial weight matrix is constructed based on the comprehensive index weight;

[0036] The expression of the comprehensive index weight is as follows:

[0037]

[0038] wherein: ω i is the comprehensive index weight, a kj is the element of the judgment matrix in the kth row and jth column.

[0039] The step S3 specifically comprises:

[0040] S31, the initial weight matrix obtained by the improved analytic hierarchy process is W={w1, w2, …wn}, and the index data is X={x1, x2, …xn}; n}; n

[0041] S32, the index information of the index data is normalized to obtain the normalized index vector Y={y1, y2, …yn}; n};

[0042] Then the positive index is as follows:

[0043]

[0044] Then the negative index is as follows:

[0045]

[0046] wherein: x i is the ith index data, y i is the normalized ith index data;

[0047] S33, the index state vector U i (Y) is calculated by combining the variable weight balancing function of the incentive factor and the penalty factor; the expression of the index state vector is as follows:

[0048]

[0049] wherein: β is the incentive factor, γ is the penalty factor, C is the through factor, and k is the adjustment factor;

[0050] S34, based on the initial weight matrix W and the index state vector U i (Y), the variable weight vector G i is calculated; the expression of the variable weight vector G i is as follows:

[0051]

[0052] S35, the variable weight vector G i ​A difference vector between the initial weight matrix W and the weight matrix W, and taking absolute value of each data of the difference vector, an absolute difference vector V is obtained;

[0053] S36, taking the absolute difference vector V as a graph, and based on the minimum Ncut function, the indicators with relatively intense weight changes and other indicators are partitioned; the expression of the partition is as follows:

[0054]

[0055]

[0056] Wherein: A, B are two subgraphs of the partition, the nodes in the graph are represented by the absolute difference of the indicators; W u，v The similarity between nodes u and v; u, v, p are nodes of graphs A, B and V respectively; partition A is the indicator with relatively intense weight changes, and partition B is other indicators;

[0057] S37, for the partition A of the indicator with relatively intense weight changes, the corresponding indicator state vector is corrected, and then step S34 is performed to obtain the final variable weight vector;

[0058] The expression of the correction is as follows:

[0059]

[0060] Wherein: U i ′ is the correction vector of the indicator with relatively intense weight changes, U i A , W i A The indicator variable weight and constant weight of the i-th indicator of partition A, λ B The fluctuation indicator of other indicators in partition B.

[0061] The step S4 specifically comprises:

[0062] S41, determine the standard cloud of the cloud centroid model, and the expression of the digital features of the standard cloud is as follows:

[0063]

[0064] Wherein: Ex i Is the cloud expectation, En i Is the cloud entropy, C i，max Is the maximum value of the score interval corresponding to the evaluation grade, C i，min Is the minimum value of the score interval corresponding to the evaluation grade;

[0065] S42, the ideal state of the comprehensive performance of the three-phase oil-immersed transformer is divided into positive ideal state and negative ideal state, and the centroid position vector and cloud centroid vector are both in positive ideal state; and define the centroid position vector and cloud centroid vector are both in negative ideal state;

[0066] Wherein:

[0067] T + = L + · G i ; T - = L - · G i ;

[0068]

[0069] In the above formula: x + is a positive index, x - is a negative index, N is the number of positive indexes, Ex (i) is the standard cloud of the secondary index in the evaluation index system;

[0070] S43, the cloud centroid vector is normalized; the expression of the normalization is as follows:

[0071]

[0072] Wherein: T i p is the normalized cloud centroid vector value;

[0073] S44, based on the normalized cloud centroid vector value T i p of each secondary index and the variable weight vector G i , the weighted deviation degree θ is obtained; the expression of the weighted deviation degree θ is as follows:

[0074]

[0075] S45, the weighted deviation degree θ is taken as the score value of the comprehensive performance of the three-phase oil-immersed transformer, and the score value is matched with the preset evaluation level, to obtain the comprehensive performance evaluation of the three-phase oil-immersed transformer.

[0076] The three-phase oil-immersed transformer comprehensive performance evaluation index system comprises:

[0077] Primary index: operation reliability, electrical experiment index, oil chemical experiment information;

[0078] Secondary index: hydrogen content, acetylene content, total hydrocarbon content, methane content, iron core grounding current, winding direct current resistance, winding dielectric loss, winding polarization index, oil breakdown voltage, oil micro water, oil dielectric loss, oil sugar aldehyde.

[0079] An improved AHP-based transformer performance evaluation system considering scene applicability, the system comprising:

[0080] An evaluation index system construction module for constructing a three-phase oil-immersed transformer comprehensive performance evaluation index system;

[0081] A comprehensive subjective weight acquisition module for correcting the judgment matrix based on the zero deviation matrix to improve the analytic hierarchy process and obtain the comprehensive subjective weight of each index in the evaluation index system;

[0082] A variable weight vector acquisition module for calculating the index state vector of the comprehensive subjective weight based on the variable weight theory considering the random scene, and re-clustering the index state vector through the specification cut idea to obtain the final variable weight vector;

[0083] A comprehensive evaluation module for calculating the weighted deviation degree based on the cloud centroid model and the final variable weight vector to comprehensively evaluate the comprehensive performance of the three-phase oil-immersed transformer.

[0084] The comprehensive subjective weight acquisition module is used to obtain the comprehensive subjective weight according to the following steps:

[0085] S21, determine the importance of each index in the three-phase oil-immersed transformer comprehensive performance evaluation index system by expert scoring and using the 1-9 scale method, construct the judgment matrix A, A=[a ij ] n×n ;

[0086] The expression of the judgment matrix is as follows:

[0087]

[0088] Wherein: a ij is the scale value between the index and the index;

[0089] S22, calculate the consistency index CI, the expression is as follows:

[0090]

[0091] Wherein: λ max is the maximum eigenvalue of the judgment matrix A, and n is the order of the judgment matrix;

[0092] S23, find the average random consistency index RI and calculate the consistency ratio CR, the expression is as follows:

[0093]

[0094] If CR < 0.1, then step S29 is performed; if CR ≥ 0.1, then step S24 is performed to modify the judgment matrix;

[0095] S24, the judgment matrix is expressed as:

[0096]

[0097] W = [W ij ] = [w i / w j ], w = (w1...w i ...w n ) T ; D = [d ij ] ;

[0098] wherein A is the judgment matrix, W is the largest eigenvector of the judgment matrix A, D is the deviation matrix, is the symbol of Hadamard product;

[0099] When A satisfies the consistency matrix of the analytic hierarchy process, all d ij = 1 and λ max (D) = λ max (A) = n; a specific deviation matrix with all d ij = 1 is called a zero deviation matrix, denoted as DI;

[0100] S25, let the judgment matrix The consistency ratio standard CR * = 0.1, the iteration number k = 0, and the constant γ is a positive number less than 1 but close to 1;

[0101] S26, calculate the largest eigenvalue λ max (A (k) ) of the judgment matrix A (k) after the kth iteration; wherein the largest eigenvector w (k) and the priority vector W (k) are as follows:

[0102]

[0103] and the deviation matrix D (k) after the kth iteration is obtained according to ;

[0104] S27, calculate the consistency ratio CR (k) after the kth iteration, and compare CR (k) with CR * ;

[0105] If CR (k) > CR *If CR (k) ≤ CR * , output the judgment matrix A satisfying the consistency requirement (k) , and end the iteration.

[0106] S28, let correct the judgment matrix; wherein the correction deviation matrix is as follows:

[0107]

[0108] and proceed to step S22 until a judgment matrix satisfying the consistency requirement is obtained.

[0109] S29, based on the judgment matrix A' = [a' ij ] satisfying the consistency requirement, calculate the comprehensive index weight of each index by using the geometric mean method; the expression of the comprehensive index weight is as follows:

[0110]

[0111] wherein ω i is the comprehensive index weight, and a kj is the element of the kth row and jth column of the judgment matrix.

[0112] The variable weight vector acquisition module is used to obtain a variable weight vector according to the following steps:

[0113] S31, the initial weight matrix obtained by the improved analytic hierarchy process is W = {w1, w2, … w m}, and the index data is X = {x1, x2, … x n};

[0114] S32, normalize the index information of the index data to obtain the normalized index vector Y = {y1, y2… y n};

[0115] If the index is positive, the forward index is as follows:

[0116]

[0117] If the index is negative, the negative index is as follows:

[0118]

[0119] wherein x i is the ith index data, and y i is the normalized ith index data.

[0120] S33, calculate the index state vector U i by combining the variable weight balancing function of the incentive factor and the penalty factor.(Y); the expression of the index state vector is as follows:

[0121]

[0122] Wherein: β is the incentive factor, γ is the penalty factor, C is the through factor, k is the adjustment factor;

[0123] S34, based on the initial weight matrix W and the index state vector U i (Y), the variable weight vector G is obtained by calculation i ; the expression of the variable weight vector G i is as follows:

[0124]

[0125] S35, the variable weight vector G i The difference vector between the initial weight matrix W is calculated, and the absolute difference vector V is obtained after taking the absolute value of each data of the difference vector;

[0126] S36, the absolute difference vector V is regarded as a graph, and the indexes with relatively large weight changes and other indexes are partitioned based on the minimization of the Ncut function; the expression of the partition is as follows:

[0127]

[0128]

[0129] Wherein: A, B are two subgraphs of the partition, the nodes in the graph are represented by the index absolute difference; W u，v is the similarity between nodes u and v; u, v, p are nodes of graphs A, B and V respectively; partition A is the index with relatively large weight changes, and partition B is other indexes;

[0130] S37, for the partition A of the index with relatively large weight changes, the corresponding index state vector is corrected, and then step S34 is performed to obtain the final variable weight vector;

[0131] The expression of the correction is as follows:

[0132]

[0133] Wherein: U i ' is the correction vector of the index with relatively large weight changes, U i A , W i A is the index variable weight and constant weight of the i-th index of partition A, λ B is the fluctuation index of other indexes in partition B.

[0134] The comprehensive evaluation module is used to obtain the comprehensive evaluation according to the following steps:

[0135] S41, determine the standard cloud of the cloud centroid model, and the expression of the digital features of the standard cloud is as follows:

[0136]

[0137] Wherein: Ex i is the cloud expectation, En i is the cloud entropy, C i，max is the maximum value of the evaluation grade corresponding score value interval, C i，min is the minimum value of the evaluation grade corresponding score interval.

[0138] S42, the ideal state of the comprehensive performance of the three-phase oil-immersed transformer is divided into positive ideal state and negative ideal state, and the centroid position vector and the cloud centroid vector are defined to be in the positive ideal state; and the centroid position vector and the cloud centroid vector are defined to be in the negative ideal state.

[0139] Wherein:

[0140] T+=L+·G i ; T-=L-·G i ;

[0141]

[0142] In the above formula: x + is a positive index, x - is a negative index, N is the number of positive indexes, Ex (i) is the standard cloud of the secondary index in the evaluation index system.

[0143] S43, the cloud centroid vector is normalized; the expression of the normalization is as follows:

[0144]

[0145] Wherein: T i p is the normalized cloud centroid vector value.

[0146] S44, based on the normalized cloud centroid vector value T i p of each secondary index and the variable weight vector G i , a weighted deviation degree θ is obtained; the expression of the weighted deviation degree θ is as follows:

[0147]

[0148] S45, the weighted deviation degree theta is taken as the score value of the comprehensive performance of the three-phase oil-immersed transformer, and the score value is matched with a preset evaluation level to obtain the comprehensive performance evaluation of the three-phase oil-immersed transformer.

[0149] An improved AHP-based transformer performance evaluation device considering scene applicability, comprising a processor 5 and a memory 6;

[0150] The processor 6 is used for storing computer program code 61 and transmitting the computer program code 61 to the processor 5;

[0151] The processor 5 is used for executing the above-mentioned improved AHP-based transformer performance evaluation method considering scene applicability according to the instructions in the computer program code 61.

[0152] Compared with the prior art, the beneficial effects of the present application are:

[0153] In the improved AHP-based transformer performance evaluation method and system considering scene applicability, the method first constructs a comprehensive performance evaluation index system of a three-phase oil-immersed transformer, then corrects the judgment matrix based on the zero deviation matrix to improve the analytic hierarchy process and obtain the comprehensive subjective weight of each index in the evaluation index system, then considers the variable weight theory of the random field scene, calculates the index state vector, and re-clusters through the specification cut idea to obtain the final variable weight vector, and finally calculates the weighted deviation degree based on the cloud centroid model and the final variable weight vector to comprehensively evaluate the comprehensive performance of the three-phase oil-immersed transformer. In the application, a more comprehensive comprehensive performance evaluation index system of the three-phase oil-immersed transformer is constructed, the transformer performance is investigated from multiple dimensions, and the specification cut idea and the random field scene are used to improve the adaptability in different scenes. Finally, the cloud centroid model effectively handles the uncertainty of information, thereby providing more accurate and effective basis for the performance evaluation of the transformer. BRIEF DESCRIPTION OF DRAWINGS

[0154] Figure 1 It is the method step flow chart of the present application.

[0155] Figure 2 It is the transformer performance evaluation method flow chart in embodiment 1 of the present application.

[0156] Figure 3 It is the evaluation index system schematic diagram in embodiment 1 of the present application.

[0157] Figure 4 It is the average random consistency index table in embodiment 1 of the present application.

[0158] Figure 5is a comparative analysis table of evaluation results of different evaluation models in embodiment 1 of the present application.

[0159] Figure 6 is a schematic diagram of the system structure of the present application.

[0160] Figure 7 is a schematic diagram of the device structure of the present application.

[0161] In the figure: evaluation index system construction module 1, comprehensive subjective weight acquisition module 2, variable weight vector acquisition module 3, comprehensive evaluation module 4, processor 5, memory 6, computer program code 61. DETAILED DESCRIPTION

[0162] The present application is further described in detail below in conjunction with the accompanying drawings, description and specific embodiments.

[0163] Embodiment 1

[0164] Referring to Figure 1 A transformer performance evaluation method based on improved AHP considering scene applicability, comprising:

[0165] S1, constructing a three-phase oil-immersed transformer comprehensive performance evaluation index system;

[0166] Further, the evaluation index system in the present application includes 3 first-level indexes and 12 second-level indexes;

[0167] Referring to Figure 3 The three first-level indexes include operation reliability, electrical experiment index and oil chemical experiment information to jointly reflect the comprehensive performance of the three-phase oil-immersed transformer; the 12 second-level indexes belong to the first-level indexes and include hydrogen content, acetylene content, total hydrocarbon content, methane content, core grounding current, winding DC resistance, winding dielectric loss, winding polarization index, oil breakdown voltage, oil micro water, oil dielectric loss, and oil sugar aldehyde.

[0168] The operation reliability mainly reflects the real-time state working condition of the transformer by obtaining state evaluation indexes of the monitoring device under the condition that the oil-immersed transformer is not powered off, and the main state evaluation indexes are hydrogen content in oil, acetylene content, total hydrocarbon content, and methane content.

[0169] Hydrogen content: when the oil-immersed transformer is internally damp, water is decomposed into a large amount of hydrogen under the action of high temperature and high pressure, so the hydrogen content mainly reflects the damp condition or insulation state of the oil-immersed transformer during operation.

[0170] Acetylene content: The oil-immersed transformer itself and the new oil will not produce acetylene, but when the oil-immersed transformer oil is injected into the transformer oil tank, the process is imperfect; the sealed test tank has a leak and the transformer test process has defects that cause discharge to produce acetylene, so the acetylene content mainly reflects the safety performance of the oil-immersed transformer during operation.

[0171] Total hydrocarbon content: The generation of total hydrocarbons is related to the pyrolysis, oxidation or degradation of the insulating oil of the oil-immersed transformer at high temperature, and the total hydrocarbon content can reflect the influence of the thermal load on the insulating oil during the operation of the oil-immersed transformer and the aging level of the insulating oil.

[0172] Methane content: Methane can be produced by electric arc discharge, thermal decomposition or oxidation of insulating oil at high temperature, and the methane content of the oil-immersed transformer reflects the level of the insulating system during operation.

[0173] The electrical test indicators mainly include four test items of core grounding current, winding direct current resistance, winding dielectric loss and winding polarization index.

[0174] Core grounding current: When the core grounding current is too large, there is a large circulating current inside the equipment, which may cause the core to heat up, thereby affecting the operation of the equipment, therefore, timely and accurate judgment of whether the core exists multi-point grounding phenomenon is of great significance to ensure the safe operation of the power transformer.

[0175] Winding direct current resistance: The winding direct current resistance is used to test the welding quality of the transformer winding, to judge whether there are abnormal phenomena such as broken strands and short circuits between the winding layers and turns, and to reflect the stability of the winding longitudinal insulation and the connection state of the current loop in the evaluation of the oil-immersed transformer.

[0176] Winding dielectric loss: The winding dielectric loss factor mainly reflects whether the oil-immersed transformer exists moisture, oil aging failure, through-discharge channel and serious local damage, etc. in the evaluation of the oil-immersed transformer.

[0177] Winding polarization index: The winding polarization index mainly reflects whether the winding of the oil-immersed transformer exists insulation moisture, aging and the main insulation level in the evaluation of the oil-immersed transformer.

[0178] The oiling experiment information is mainly to judge the insulation condition and physicochemical properties of the oil in the oil-immersed transformer. The transformer oil as an internal insulation medium is the main factor to ensure the insulation state between the windings, between the winding and the core and between the winding and the shell; the oil breakdown voltage, the micro water in the oil, the oil dielectric loss and the sugar aldehyde in the oil are selected to judge the degradation degree of the performance of the insulating oil of the oil-immersed transformer.

[0179] Oil breakdown voltage: The oil breakdown voltage test of power transformer oil is mainly to determine the capacity of the internal withstand voltage of the device; when the quality of the oil-immersed transformer oil deteriorates, the withstand voltage capacity will weaken. Therefore, the size of the oil breakdown voltage can comprehensively reflect the insulation performance of the oil-immersed transformer oil.

[0180] Micro water in oil: In order to ensure that the insulation performance of the transformer oil meets the stable operation requirements, the oil should not contain water. When the water content in the oil exceeds a certain limit, the insulation oil withstand voltage capacity will decrease, the dielectric loss will increase, and other oil quality deterioration problems may occur, which may threaten the stable operation of the oil-immersed transformer.

[0181] Oil dielectric loss: It can reflect the degree of oil quality deterioration and pollution, and is an important indicator affecting the performance of transformer oil.

[0182] Glycol in oil: When affected by external electric field, mechanical stress and the like, the cellulose chain of solid insulating material is broken to produce glycol dissolved in transformer oil. Therefore, the glycol concentration in the oil-immersed transformer oil can reveal the state of the transformer insulation system.

[0183] S2, based on the zero deviation matrix correction judgment matrix, to improve the analytic hierarchy process, and obtain the comprehensive subjective weight of each index in the evaluation index system;

[0184] Referring to Figure 2 , the present application proposes a heuristic algorithm, which can automatically generate a consistent matrix from the original inconsistent judgment matrix, and can avoid the need for experts to re-score due to inconsistent judgment matrix. The main steps of improving AHP to calculate the subjective weight of each index are as follows:

[0185] S21, through expert scoring and using 1-9 scale method, the importance between each first-level index in the evaluation index system and the second-level index under each first-level index is determined. Assuming that a1, a2......a n , the judgment matrix A is constructed, A=[a ij ] n×n ;

[0186] The expression of the judgment matrix is as follows:

[0187]

[0188] Wherein: a ij is the scale value between the indexes, and

[0189] S22, calculate the consistency index CI, its expression is as follows:

[0190]

[0191] where λ max is the maximum eigenvalue of the judgment matrix A, and n is the order of the judgment matrix;

[0192] S23, find the average random consistency index RI and calculate the consistency ratio CR; the consistency index RI corresponds to the order of the judgment matrix, and can be found by looking up the average random consistency index table; see Figure 4 .

[0193] The expression of the consistency ratio CR is as follows:

[0194]

[0195] If CR < 0.1, proceed to step S29; if CR ≥ 0.1, proceed to step S24 to modify the judgment matrix;

[0196] S24, express the judgment matrix as:

[0197]

[0198] W = [W ij ] = [w i / w j ], w = (w1...w i ...w n ) T ; D = [d ij ];

[0199] where A is the judgment matrix, W is the maximum eigenvector of the judgment matrix A, D is the deviation matrix, is the symbol of Hadamard product;

[0200] When A satisfies the consistency matrix of the analytic hierarchy process, all d ij = 1 and λ max (D) = λ max (A) = n; otherwise, there exists d ij ≠ 1 and λ max (D) = λ max (A) > n; all specific deviation matrices with d ij = 1 are called zero deviation matrices, denoted as DI;

[0201] Then the modified deviation matrix is used to improve the consistency ratio, and the main steps are as follows:

[0202] S25, initialize parameters: let the judgment matrix The consistency ratio standard CR * = 0.1, the iteration number k = 0, and the constant γ is a positive number less than 1 but close to 1;

[0203] S26, variable calculation: calculate the judgment matrix A after the kth iteration (k) of the maximum eigenvalue λ max (A (k) ); wherein the maximum eigenvalue vector w (k) and the priority vector W (k) are as follows:

[0204]

[0205] and according to , the deviation matrix D (k) after the kth iteration is obtained;

[0206] S27, judgment matrix consistency: calculate the consistency ratio CR (k) after the kth iteration, and compare CR (k) with CR * ;

[0207] If CR (k) > CR * , proceed to step S28; if CR (k) ≤ CR * , output the judgment matrix A (k) that meets the consistency requirement, and the iteration ends;

[0208] S28, modify the judgment matrix: let be the modified judgment matrix; wherein the modified deviation matrix is as follows:

[0209]

[0210] and proceed to step S22 until the judgment matrix that meets the consistency requirement is obtained;

[0211] S29, based on the judgment matrix A' = [a' ij ] that meets the consistency requirement, calculate the comprehensive index weight of each index using the geometric mean method; the expression of the comprehensive index weight is as follows:

[0212]

[0213] wherein: ω i is the comprehensive index weight, and a kj is the element of the kth row and jth column of the judgment matrix.

[0214] S3, based on the variable weight theory considering the random field scene, calculate the index state vector of the comprehensive subjective weight, and through the standard cut idea, re-cluster the index state vector to reduce the degree of drastic change of the weight change, and obtain the final variable weight vector;

[0215] Normalized Cut is an algorithm framework for image segmentation and clustering, which finds the optimal segmentation by minimizing the ratio of inter-class similarity and intra-class similarity; the core of this idea is to transform the clustering problem into an optimization problem, by finding a partition to maximize the ratio of intra-class similarity and inter-class similarity.

[0216] Further, considering that three-phase oil-immersed transformers have different adaptabilities in different scenarios, a variable weight theory considering random scenarios is proposed, and the steps are as follows:

[0217] S31, the initial weight matrix obtained by the improved analytic hierarchy process is W={w1, w2, … w n} and the index data is X={x1, x2, … x n};

[0218] S32, the index information of the index data is normalized to obtain the normalized index vector Y={y1, y2… y n};

[0219] Then the positive and negative indicators are as follows:

[0220]

[0221] Where: x i is the i-th index data, and y i is the normalized i-th index data.

[0222] S33, the index state vector U i (Y) is calculated by combining the variable weight balancing function of the incentive factor and the penalty factor; the expression of the index state vector is as follows:

[0223]

[0224] Where: β is the incentive factor, γ is the penalty factor, C is the through factor, and k is the adjustment factor; they all satisfy α, β, γ, C∈[0, 1], without loss of generality, in this embodiment, α=0.2, β=0.6, γ=0.9, and C=0.2.

[0225] S34, based on the initial weight matrix W and the index state vector U i (Y), the variable weight vector G i is calculated; the expression of the variable weight vector G i is as follows:

[0226]

[0227] S35, the variable weight vector Gi A difference vector between the initial weight matrix W and the final weight matrix W is obtained, and each data of the difference vector is taken as an absolute value to obtain an absolute difference vector V;

[0228] S36, taking the absolute difference vector V as a graph, and based on the minimum Ncut function, the indicators with relatively intense weight changes and other indicators are partitioned; the expression of the partition is as follows:

[0229]

[0230]

[0231] Wherein: A, B are two subgraphs of the partition, the nodes in the graph are represented by the absolute difference of the indicators; W u，v is the similarity between nodes u and v; u, v, p are nodes of graphs A, B and V respectively; partition A is the indicator with relatively intense weight changes, and partition B is other indicators;

[0232] S37, for the partition A of the indicators with relatively intense weight changes, the corresponding indicator state vector is corrected, and then step S34 is performed to obtain the final variable weight vector;

[0233] The expression of the correction is as follows:

[0234]

[0235] Wherein: U i ′ is the correction vector of the indicators with relatively intense weight changes, U i A , W i A is the indicator variable weight and constant weight of the i-th indicator of the partition A, λ B is the fluctuation indicator of other indicators in the partition B.

[0236] S4, based on the cloud centroid model and the final variable weight vector, the weighted deviation degree is calculated to obtain a comprehensive evaluation of the comprehensive performance of the three-phase oil-immersed transformer.

[0237] Further, the cloud centroid model is used in the technical solution for comprehensive evaluation, and the specific steps are as follows:

[0238] S41, determine the standard cloud of the cloud centroid model, and the expression of the digital features of the standard cloud is as follows:

[0239]

[0240] Wherein: Ex i is the cloud expectation, En i is the cloud entropy, and C i，maxTo evaluate the maximum value of the grading corresponding to the score value interval, C i，min To evaluate the minimum value of the grading corresponding to the score interval;

[0241] S42, the ideal state of the comprehensive performance of the three-phase oil-immersed transformer is divided into positive ideal state and negative ideal state, and the centroid position vector and the cloud centroid vector are in the positive ideal state; and the centroid position vector and the cloud centroid vector are in the negative ideal state;

[0242] Wherein:

[0243] T+=L+·G i ; T-=L-·G i ;

[0244]

[0245] In the above formula: x + is a positive index, x - is a negative index, N is the number of positive indexes, Ex (i) is the standard cloud of the secondary index in the evaluation index system;

[0246] S43, the cloud centroid vector is normalized; the expression of the normalization is as follows:

[0247]

[0248] Wherein: T i p is the normalized cloud centroid vector value;

[0249] S44, based on the normalized cloud centroid vector value T i p of each secondary index and the variable weight vector G i , the weighted deviation degree θ is obtained; the expression of the weighted deviation degree θ is as follows:

[0250]

[0251] S45, the weighted deviation degree θ is taken as the score value of the comprehensive performance of the three-phase oil-immersed transformer, and the score value is matched with the preset evaluation grade to obtain the comprehensive performance evaluation of the three-phase oil-immersed transformer.

[0252] In the technical solution, the quality level of the oil-immersed transformer is divided into I level (0.75≤θ<1), II level (0.5≤θ<0.75), III level (0.25≤θ<0.5) and IV level (0≤θ≤0.25) from high to low according to the grade discrimination based on the score results. The preset evaluation grades are {I, II, III, IV}; wherein, I is the highest level, II is the high level, III is the middle level, and IV is the low level.

[0253] In the embodiment, the evaluation method of the scheme is compared with the accuracy of other existing methods. Referring to Figure 5 The evaluation method of the scheme has the highest accuracy, reaching 86.2%, and the evaluation accuracy is increased by 13.7%, 8% and 15.9% compared with the entropy weight method, the fuzzy comprehensive evaluation method and the grey correlation degree method. It can be known from the figure that the method proposed in the scheme effectively improves the accuracy of the comprehensive performance of the three-phase oil-immersed transformer.

[0254] Embodiment 2:

[0255] Referring to Figure 6 A transformer performance evaluation system based on an improved AHP considering scene applicability, the system comprises:

[0256] An evaluation index system construction module 1 is configured to construct an evaluation index system for the comprehensive performance of a three-phase oil-immersed transformer.

[0257] Further, the evaluation index system constructed by the evaluation index system construction module 1 comprises:

[0258] Primary indicators: operation reliability, electrical experiment indicators, oil and chemical experiment information;

[0259] Secondary indicators: hydrogen content, acetylene content, total hydrocarbon content, methane content, core grounding current, winding DC resistance, winding dielectric loss, winding polarization index, oil breakdown voltage, oil micro water, oil dielectric loss, oil sugar aldehyde.

[0260] A comprehensive subjective weight acquisition module 2 is configured to correct a judgment matrix based on a zero deviation matrix, to improve the analytic hierarchy process, and to obtain comprehensive subjective weights of each indicator in the evaluation index system.

[0261] Further, the comprehensive subjective weight acquisition module 2 is configured to obtain the comprehensive subjective weights according to the following steps:

[0262] S21, determine the importance of each indicator in the comprehensive performance evaluation index system of the three-phase oil-immersed transformer by expert scoring and using a 1-9 scale method, and construct a judgment matrix A, A=[a ij ] n×n ;

[0263] The expression of the judgment matrix is as follows:

[0264]

[0265] Wherein: a ij is the scale value between indexes and indexes;

[0266] S22, calculate the consistency index CI, whose expression is as follows:

[0267]

[0268] Wherein: λ max is the maximum eigenvalue of the judgment matrix A, and n is the order of the judgment matrix;

[0269] S23, find the average random consistency index RI and calculate the consistency ratio CR, whose expression is as follows:

[0270]

[0271] If CR<0.1, proceed to step S29; if CR≥0.1, proceed to step S24 to modify the judgment matrix;

[0272] S24, express the judgment matrix as:

[0273]

[0274] W=[W ij ]=[w i / w j ],w=(w1...w i ...w n ) T ;D=[d ij ];

[0275] Wherein: A is the judgment matrix, W is the maximum eigenvector of the judgment matrix A, D is the deviation matrix, and ○ is the symbol of Hadamard product;

[0276] When A satisfies the hierarchy analysis method as a consistent matrix, all d ij =1 and λ max (D)=λ max (A)=n; all specific deviation matrices with d ij =1 are called zero deviation matrices, denoted as DI;

[0277] S25, let the judgment matrix The consistency ratio standard CR * =0.1, the iteration number k=0, and the constant γ is a positive number less than 1 but close to 1;

[0278] S26, calculating the judgment matrix A after the kth iteration (k) of the maximum eigenvalue λ max (A (k) ); wherein the maximum eigenvalue vector w (k) and the priority vector W (k) are as follows:

[0279]

[0280] and according to , the deviation matrix D (k) after the kth iteration is obtained

[0281] S27, calculating the consistency ratio CR (k) after the kth iteration (k) , and comparing CP * and CR (k) ;

[0282] If CR * , then step S28 is performed; if CR (k) ≤ CR * , then the judgment matrix A (k) that meets the consistency requirement is output, and the iteration ends;

[0283] S28, setting the modified judgment matrix; wherein the modified deviation matrix is as follows:

[0284]

[0285] and performing step S22 until the judgment matrix that meets the consistency requirement is obtained;

[0286] S29, based on the judgment matrix A' = [a' ij ] that meets the consistency requirement, using the geometric mean method to calculate the comprehensive index weight of each index; the expression of the comprehensive index weight is as follows:

[0287]

[0288] wherein: ω i is the comprehensive index weight, and a kj is the element in the kth row and jth column of the judgment matrix.

[0289] The variable weight vector acquisition module 3 is used to calculate the index state vector of the comprehensive subjective weight based on the variable weight theory considering the random field scene, and to obtain the final variable weight vector by re-clustering the index state vector through the canonical cut idea.

[0290] Further, the variable weight vector acquisition module 3 is used to obtain the variable weight vector according to the following steps:

[0291] S31, the initial weight matrix obtained by the improved analytic hierarchy process is W={w1, w2, …wn}; n} and the index data is X={x1, x2, …xn}; n};

[0292] S32, the index information of the index data is normalized to obtain the normalized index vector X={y1, y2, …yn}; n};

[0293] Then the positive and negative indexes are as follows:

[0294]

[0295] Wherein: x i is the i-th index data, y i is the normalized i-th index data;

[0296] S33, the index state vector U i (Y) is calculated by combining the variable weight balance function of the incentive factor and the penalty factor; the expression of the index state vector is as follows:

[0297]

[0298] Wherein: β is the incentive factor, γ is the penalty factor, C is the through factor, and k is the adjustment factor;

[0299] S34, based on the initial weight matrix W and the index state vector U i (Y), the variable weight vector G i is calculated; the expression of the variable weight vector G i is as follows:

[0300]

[0301] S35, the difference vector between the variable weight vector G i and the initial weight matrix W is calculated, and after taking the absolute value of each data of the difference vector, the absolute difference vector V is obtained;

[0302] S36, the absolute difference vector V is regarded as a graph, and based on the minimum Ncut function, the indexes with relatively sharp weight changes and other indexes are partitioned; the expression of the partition is as follows:

[0303]

[0304]

[0305] Wherein: A, B are two sub-graphs of partition, the nodes in the graph are represented by the absolute difference of indicators; W u，v The similarity degree between nodes u, v; u, v, p are nodes of graphs A, B, V; Partition A is the indicator with relatively sharp weight change, and partition B is other indicators;

[0306] S37, for the partition A of the indicator with relatively sharp weight change, the corresponding indicator state vector is corrected, and then step S34 is performed to obtain the final variable weight vector;

[0307] The expression of the correction is as follows:

[0308]

[0309] Wherein: U i ′ is the correction vector of the indicator with relatively sharp weight change, U i A , W i A is the indicator variable weight and constant weight of the i-th indicator of partition A, λ B is the fluctuation indicator of other indicators in partition B.

[0310] The comprehensive evaluation module 4 is used to calculate and obtain the weighted deviation degree based on the cloud centroid model and the final variable weight vector, so as to comprehensively evaluate the comprehensive performance of the three-phase oil-immersed transformer.

[0311] Further, the comprehensive evaluation module 4 is used to obtain the comprehensive evaluation according to the following steps:

[0312] S41, determine the standard cloud of the cloud centroid model, and the expression of the digital features of the standard cloud is as follows:

[0313]

[0314] Wherein: Ex i is the cloud expectation, En i is the cloud entropy, C i，max is the maximum value of the evaluation grade corresponding score value interval, C i，min is the minimum value of the evaluation grade corresponding score interval;

[0315] S42, the ideal state of the comprehensive performance of the three-phase oil-immersed transformer is divided into positive ideal state and negative ideal state, and the centroid position vector and the cloud centroid vector are defined to be in the positive ideal state; and the centroid position vector and the cloud centroid vector are defined to be in the negative ideal state;

[0316] Wherein:

[0317] T+ = L+ G i ; T- = L- G i ;

[0318]

[0319] In the above formula: x + is a positive index, x - is a negative index, N is the number of positive indexes, Ex (i) is the standard cloud of the secondary index in the evaluation index system;

[0320] S43, normalize the cloud centroid vector; the expression of the normalization is as follows:

[0321]

[0322] Where: T i p is the normalized cloud centroid vector value;

[0323] S44, based on the normalized cloud centroid vector value T i p of each secondary index and the variable weight vector G i , obtain the weighted deviation degree theta; the expression of the weighted deviation degree theta is as follows:

[0324]

[0325] S45, take the weighted deviation degree theta as the score value of the comprehensive performance of the three-phase oil-immersed transformer, and match the score value with the preset evaluation level to obtain the comprehensive performance evaluation of the three-phase oil-immersed transformer.

[0326] Embodiment 3:

[0327] Referring to Figure 7 , in this embodiment, it also includes a transformer performance evaluation device based on improved AHP considering scene applicability, the device includes a processor 5 and a memory 6;

[0328] The memory 6 is used to store computer program code 61 and transmit the computer program code 61 to the processor 5;

[0329] The processor 5 is used to execute the transformer performance evaluation method based on improved AHP considering scene applicability according to the instructions in the computer program code 61.

[0330] The embodiment also includes a computer readable storage medium, wherein the computer readable storage medium stores computer executable instructions, and when the computer executable instructions are executed on a computer, the computer executable instructions realize the transformer performance evaluation method based on improved AHP considering scene applicability in embodiment 1.

[0331] Generally, computer instructions to implement the method of the present application can be carried in any combination of one or more computer readable media. Non-transitory computer readable storage media can include any computer readable medium except for a signal per se passing through a medium.

[0332] The computer readable storage medium may, for example, be, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or apparatus, or any combination of the above. More specific examples (a non-exhaustive list) of the computer readable storage medium include an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, the computer readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device.

[0333] Computer program code to carry out operations of the present application can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++, or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages, and specifically Python language suitable for neural network computing and platform framework based on TensorFlow, PyTorch, etc. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0334] The above-mentioned device and non-transitory computer readable storage medium can refer to the specific description of the transformer performance evaluation method based on improved AHP considering scene applicability and its beneficial effects, which will not be repeated here.

[0335] Although the embodiments of the present application have been shown and described above, it should be understood by those having ordinary skill in the art that the above embodiments are exemplary, and cannot be interpreted as limiting the present application, and those having ordinary skill in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A transformer performance evaluation method based on improved AHP considering scenario applicability, characterized in that, include: S1. Construct a comprehensive performance evaluation index system for three-phase oil-immersed transformers; S2. Based on the zero-deviation matrix, the judgment matrix is ​​modified to improve the analytic hierarchy process and obtain the comprehensive subjective weights of each indicator in the evaluation index system. The zero deviation matrix refers to the following: when the judgment matrix satisfies the consistency matrix requirement of the analytic hierarchy process, all elements in the deviation matrix are equal to 1, and the largest eigenvalue of both the deviation matrix and the judgment matrix is ​​equal to the order of the judgment matrix. S3. Based on the variable weight theory that considers random scenarios, calculate the index state vector of comprehensive subjective weight, and re-cluster the index state vector through the idea of ​​normalization to obtain the final variable weight vector. S4. Based on the cloud centroid model and the final variable weight vector, the weighted deviation is calculated to comprehensively evaluate the overall performance of the three-phase oil-immersed transformer. Step S4 specifically includes: S41. Determine the standard cloud of the cloud centroid model. The expression for the digital characteristics of the standard cloud is as follows: ； in: For the cloud's expectation, For cloud entropy, The maximum value within the range of scores corresponding to the evaluation level. It is the minimum value of the scoring range corresponding to the evaluation level; S42. Divide the ideal state of the comprehensive performance of a three-phase oil-immersed transformer into positive ideal state and negative ideal state, and define the centroid position vector. And cloud centroid vector All are in a positive ideal state; and the centroid position vector is defined. And cloud centroid vector All are in a negative ideal state; in: ； ； ； ； In the above formula: As a positive indicator, It is a negative indicator. The number is a positive indicator. Standard cloud for evaluating secondary indicators in the indicator system; S43. Normalize the cloud centroid vector; the normalization expression is as follows: ； in: The normalized cloud centroid vector value; S44. Normalized cloud centroid vector value based on each secondary indicator. With variable weight vector Obtain the weighted bias The weighted deviation The expression is as follows: ； S45, Weighted deviation The score is used as the comprehensive performance rating of the three-phase oil-immersed transformer, and the score is matched with the preset evaluation level to obtain the comprehensive performance evaluation of the three-phase oil-immersed transformer.

2. The transformer performance evaluation method based on improved AHP considering scenario applicability according to claim 1, characterized in that: Step S2 specifically includes: S21. Through expert scoring and using the 1-9 scale method, determine the importance of each indicator in the comprehensive performance evaluation index system for three-phase oil-immersed transformers, and construct a judgment matrix. , ; The expression for the judgment matrix is ​​as follows: ； in: The scale value between indicators; S22. Calculate the consistency index Its expression is as follows: ； in: To determine the matrix The largest eigenvalue, To determine the order of a matrix; S23. Find the average random consistency index And calculate the consistency ratio. Its expression is as follows: ； like If so, proceed to step S29; If so, proceed to step S24 to correct the judgment matrix; S24. The judgment matrix is ​​represented as follows: ； ； ； in: To determine the matrix, To determine the matrix The largest eigenvector, The deviation matrix, The symbol for the Hadamard product; when When the analytic hierarchy process (AHP) is a consistency matrix, all and ;all The specific deviation matrix is ​​called the zero deviation matrix, denoted as: ; S25, Let the judgment matrix Consistency ratio standard Number of iterations ,constant It is a positive number less than 1 but close to 1; S26, Calculate the first The judgment matrix after the next iteration Maximum eigenvalue ; where the largest eigenvalue vector With preference vector as follows: ； ； And according to Calculate to obtain the first The deviation matrix after the second iteration ; S27, Calculate the... Consistency ratio after the second iteration and will and Compare; like If so, proceed to step S28; if Then the output will be a judgment matrix that satisfies the consistency requirement. The iteration ends; S28, Order The judgment matrix is ​​corrected; the correction deviation matrix is ​​as follows: ， ； Then proceed to step S22 until a judgment matrix that meets the consistency requirements is obtained; S29. Judgment matrix based on meeting consistency requirements The geometric mean method is used to calculate the comprehensive index weight of each indicator, and an initial weight matrix is ​​constructed based on the comprehensive index weight. The expression for the weight of the comprehensive index is as follows: ； in: As a comprehensive indicator weight, To determine the matrix of the first Line 1 Column elements.

3. The transformer performance evaluation method based on improved AHP considering scenario applicability according to claim 2, characterized in that: Step S3 specifically includes: S31. The initial weight matrix obtained by the improved analytic hierarchy process is: The indicator data is ; S32. Normalize the indicator information of the indicator data to obtain the normalized indicator vector. ; The positive indicators are as follows: ； The negative indicators are as follows: ； in: For the first Individual indicator data, The normalized version Individual indicator data; S33. Calculate the index state vector by combining the variable weight balance function of the incentive factor and the penalty factor. The expression for the index state vector is as follows: ； in: As a motivating factor, As a penalty factor, To pass the factor, For adjustment factors; S34, Based on the initial weight matrix With the indicator state vector Calculate the variable weight vector The variable weight vector The expression is as follows: ； S35. Calculate the variable weight vector With the initial weight matrix The difference vector between the two values ​​is obtained by taking the absolute value of each data point in the difference vector. ; S36. Transform the absolute difference vector Viewed as a graph, and based on minimization A function is used to partition indicators with relatively drastic weight changes into other indicators; the expression for the partitioning is as follows: ； ； ； in: The two subgraphs are partitioned, and the nodes in the graphs are represented by the absolute difference of the indicators; For nodes The degree of similarity between them; , , Figures are respectively , , Nodes; partitions For indicators with relatively drastic weight changes, partitioning For other indicators; S37. Partitioning of indicators with relatively drastic weight changes After correcting the corresponding index state vector, proceed to step S34 to obtain the final variable weight vector. The corrected expression is as follows: ； ； in: This is the correction vector for indicators whose weights change drastically. , For partitioning The The variable weights and constant weights of each indicator. For partitioning The volatility indicators of other indicators.

4. The transformer performance evaluation method based on improved AHP considering scenario applicability according to claim 1, characterized in that: The comprehensive performance evaluation index system for three-phase oil-immersed transformers includes: Primary indicators: operational reliability, electrical test indicators, and oil-chemical test information; Secondary indicators: hydrogen content, acetylene content, total hydrocarbon content, methane content, core grounding current, winding DC resistance, winding dielectric loss, winding polarization index, oil breakdown voltage, trace water in oil, oil dielectric loss, and sugar aldehydes in oil.

5. A transformer performance evaluation system based on improved AHP considering scenario applicability, characterized in that, The system includes: The evaluation index system construction module (1) is used to construct a comprehensive performance evaluation index system for three-phase oil-immersed transformers; The comprehensive subjective weight acquisition module (2) is used to modify the judgment matrix based on the zero deviation matrix in order to improve the analytic hierarchy process and obtain the comprehensive subjective weight of each indicator in the evaluation index system. The zero deviation matrix refers to the following: when the judgment matrix satisfies the consistency matrix requirement of the analytic hierarchy process, all elements in the deviation matrix are equal to 1, and the largest eigenvalue of both the deviation matrix and the judgment matrix is ​​equal to the order of the judgment matrix. The variable weight vector acquisition module (3) is used to calculate the index state vector of comprehensive subjective weight based on the variable weight theory considering random scenarios, and to re-cluster the index state vector through the idea of ​​normalization to obtain the final variable weight vector. The comprehensive evaluation module (4) is used to calculate the weighted deviation based on the cloud centroid model and the final variable weight vector, so as to comprehensively evaluate the comprehensive performance of the three-phase oil-immersed transformer. The comprehensive evaluation module (4) is used to obtain a comprehensive evaluation according to the following steps: S41. Determine the standard cloud of the cloud centroid model. The expression for the digital characteristics of the standard cloud is as follows: ； in: For the cloud's expectation, For cloud entropy, The maximum value within the range of scores corresponding to the evaluation level. It is the minimum value of the scoring range corresponding to the evaluation level; S42. Divide the ideal state of the comprehensive performance of a three-phase oil-immersed transformer into positive ideal state and negative ideal state, and define the centroid position vector. And cloud centroid vector All are in a positive ideal state; and the centroid position vector is defined. And cloud centroid vector All are in a negative ideal state; in: ； ； ； ； In the above formula: As a positive indicator, It is a negative indicator. The number is a positive indicator. Standard cloud for evaluating secondary indicators in the indicator system; S43. Normalize the cloud centroid vector; the normalization expression is as follows: ； in: The normalized cloud centroid vector value; S44. Normalized cloud centroid vector value based on each secondary indicator. With variable weight vector Obtain the weighted bias The weighted deviation The expression is as follows: ； S45, Weighted deviation The score is used as the comprehensive performance rating of the three-phase oil-immersed transformer, and the score is matched with the preset evaluation level to obtain the comprehensive performance evaluation of the three-phase oil-immersed transformer.

6. The transformer performance evaluation system based on improved AHP considering scenario applicability according to claim 5, characterized in that: The comprehensive subjective weight acquisition module (2) is used to obtain the comprehensive subjective weight according to the following steps: S21. Through expert scoring and using the 1-9 scale method, determine the importance of each indicator in the comprehensive performance evaluation index system for three-phase oil-immersed transformers, and construct a judgment matrix. , ; The expression for the judgment matrix is ​​as follows: ； in: The scale value between indicators; S22. Calculate the consistency index Its expression is as follows: ； in: To determine the matrix The largest eigenvalue, To determine the order of a matrix; S23. Find the average random consistency index And calculate the consistency ratio. Its expression is as follows: ； like If so, proceed to step S29; If so, proceed to step S24 to correct the judgment matrix; S24. The judgment matrix is ​​represented as follows: ； ； ； in: To determine the matrix, To determine the matrix The largest eigenvector, The deviation matrix, The symbol for the Hadamard product; when When the analytic hierarchy process (AHP) is a consistency matrix, all and ;all The specific deviation matrix is ​​called the zero deviation matrix, denoted as: ; S25, Let the judgment matrix Consistency ratio standard Number of iterations ,constant It is a positive number less than 1 but close to 1; S26, Calculate the first The judgment matrix after the next iteration Maximum eigenvalue ; where the largest eigenvalue vector With preference vector as follows: ； ； And according to Calculate to obtain the first The deviation matrix after the second iteration ; S27, Calculate the... Consistency ratio after the second iteration and will and Compare; like If so, proceed to step S28; if Then the output will be a judgment matrix that satisfies the consistency requirement. The iteration ends; S28, Order The judgment matrix is ​​corrected; the correction deviation matrix is ​​as follows: ， ； Then proceed to step S22 until a judgment matrix that meets the consistency requirements is obtained; S29. Judgment matrix based on meeting consistency requirements The geometric mean method is used to calculate the comprehensive index weight of each indicator, and an initial weight matrix is ​​constructed based on the comprehensive index weight. The expression for the weight of the comprehensive index is as follows: ； in: As a comprehensive indicator weight, To determine the matrix of the first Line 1 Column elements.

7. The transformer performance evaluation system based on improved AHP considering scenario applicability according to claim 6, characterized in that: The variable weight vector acquisition module (3) is used to obtain the variable weight vector according to the following steps: S31. The initial weight matrix obtained by the improved analytic hierarchy process is: The indicator data is ; S32. Normalize the indicator information of the indicator data to obtain the normalized indicator vector. ; The positive indicators are as follows: ； The negative indicators are as follows: ； in: For the first Individual indicator data, The normalized version Individual indicator data; S33. Calculate the index state vector by combining the variable weight balance function of the incentive factor and the penalty factor. The expression for the index state vector is as follows: ； in: As a motivating factor, As a penalty factor, To pass the factor, For adjustment factors; S34, Based on the initial weight matrix With the indicator state vector Calculate the variable weight vector The variable weight vector The expression is as follows: ； S35. Calculate the variable weight vector With the initial weight matrix The difference vector between the two values ​​is obtained by taking the absolute value of each data point in the difference vector. ; S36. Transform the absolute difference vector Viewed as a graph, and based on minimization A function is used to partition indicators with relatively drastic weight changes into other indicators; the expression for the partitioning is as follows: ； ； ； in: The two subgraphs are partitioned, and the nodes in the graphs are represented by the absolute difference of the indicators; For nodes The degree of similarity between them; , , Figures are respectively , , Nodes; partitions For indicators with relatively drastic weight changes, partitioning For other indicators; S37. Partitioning of indicators with relatively drastic weight changes After correcting the corresponding index state vector, proceed to step S34 to obtain the final variable weight vector. The corrected expression is as follows: ； ； in: This is the correction vector for indicators whose weights change drastically. , For partitioning The The variable weights and constant weights of each indicator. For partitioning The volatility indicators of other indicators.

8. A transformer performance evaluation device based on improved AHP considering scenario applicability, characterized in that: The device includes a processor (5) and a memory (6); The memory (6) is used to store computer program code (61) and to transmit the computer program code (61) to the processor (5). The processor (5) is used to execute the performance evaluation method for a transformer based on improved AHP based on any one of claims 1-4 according to the instructions in the computer program code (61).

Images