Transformer state evaluation method based on fuzzy mathematical theory

Abstract

The invention relates to the technical field of transformers, in particular to a transformer state evaluation method based on a fuzzy mathematical theory. The method comprises the following steps: firstly, collecting original test data of a transformer, and performing normalization processing; and calculating the support degree and the lifting degree of the feature index based on the association rule, and judging the rationality of the feature index based on the lifting degree. And then, establishing a transformer state evaluation system according to the selected characteristic indexes, and dividing comment set grades of the transformer. Then, the basic probability of the feature index is determined using a dynamic fuzzy membership adjustment mechanism. And finally, using an improved D-S evidence theory based on a Spearman grade correlation coefficient to fuse the basic probabilities of different characteristic indexes to evaluate the comprehensive state of the transformer. The method can effectively improve the accuracy of transformer state evaluation, and can be used for guiding the actual maintenance arrangement work of the transformer.

Description

Technical Field

[0001] The invention relates to the technical field of transformers, and in particular to a transformer state evaluation method based on fuzzy mathematics theory. Background Art

[0002] The main function of a power transformer is to transmit high-voltage electricity to a place far away from the power plant and reduce the voltage to a safe level when it reaches the end user. Power transformers achieve voltage rise and fall through the principle of electromagnetic induction and are widely used in power transmission, distribution and power conversion. The reliable operation of power transformers is very important, which directly affects the stability of the power system and the continuity of power supply. Failure or performance degradation of power transformers may cause power outages, equipment damage or grid instability, thus affecting industrial production, commercial operations and daily life. In order to avoid these conditions, power transformers need to be maintained and monitored regularly to ensure normal operation, thereby ensuring the reliability of power supply and the safety of the system.

[0003] Traditional methods for assessing the status of power transformers rely mainly on regular physical inspections and preventive maintenance. Regular inspections generally include testing the insulating oil of power transformers, such as determining its insulation strength, acid value, and water content, and conducting electrical tests and on-site inspections. The advantage of traditional methods is that they can detect and deal with obvious problems in a timely manner, which helps prevent failures. However, there are also obvious disadvantages, such as regular inspections often have a lag and cannot reflect the actual operating status of power equipment in real time, resulting in delays in discovering and dealing with problems. In addition, a lot of time and manpower are required for regular maintenance and inspections, and the maintenance cost is high.

[0004] With the advancement of technology, modern methods of power transformer maintenance are gradually turning to online monitoring systems and intelligent diagnosis. Online monitoring systems use sensors to monitor key indicators such as temperature, pressure, and oil quality of power transformers in real time, supporting early fault warning and remote monitoring. Intelligent diagnostic systems use data analysis and machine learning technologies to conduct in-depth analysis of historical and real-time data to predict potential faults, thereby optimizing maintenance strategies and improving maintenance efficiency and accuracy. However, the large amount of data generated by online monitoring requires complex analysis and processing, especially when dealing with nonlinear relationships between indicators, and the accuracy of predictions still has room for improvement.

[0005] Therefore, a transformer state assessment method based on fuzzy mathematics theory is proposed. Summary of the invention

[0006] The purpose of the present invention is to provide a transformer state assessment method based on fuzzy mathematics theory, which is used to improve the accuracy of transformer state assessment. First, the original test data of the transformer is collected and normalized; the support and lift of the characteristic indicators are calculated based on the association rules, and the rationality of the characteristic indicators is judged based on the lift. Then, a transformer state assessment system is established based on the selected characteristic indicators, and the comment set level of the transformer is divided. Next, the basic probability of the characteristic indicators is determined using a dynamic fuzzy membership adjustment mechanism. Finally, the basic probabilities of different characteristic indicators are integrated using the improved DS evidence theory based on the Spearman rank correlation coefficient to perform a comprehensive transformer state assessment.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A transformer state assessment method based on fuzzy mathematics theory, comprising:

[0009] Collecting the original test data of the transformer and performing normalization processing; calculating the support and lift of the characteristic index based on the association rules, and judging the rationality of the characteristic index based on the lift;

[0010] Establishing a transformer status evaluation system according to the selected characteristic indicators, and classifying the transformer comment set levels;

[0011] Determine the basic probability of the characteristic indicator using a dynamic fuzzy membership adjustment mechanism;

[0012] The improved DS evidence theory based on Spearman rank correlation coefficient is used to integrate the basic probabilities of different characteristic indicators for comprehensive transformer condition assessment.

[0013] The association rule is defined as follows: if the transaction database is D, and A and B are transaction subsets of D, then and but It is represented as an association rule, A and B are the antecedent and consequent of the association rule;

[0014] Define the following variables: freq AB Indicates the number of times A and B appear at the same time; freq A Indicates the number of times A appears; freq B represents the number of times B appears; total represents the total number of transactions in the transaction database D; the calculation formula for support is support AB =freq AB / total;support A =freq A / total;supportB =freq B / total; the calculation formula for lift is Simplified to

[0015] The relevance of the feature index is determined based on the lift: It means that there is a positive correlation between the antecedent A and the consequent B; when Indicates that there is no significant correlation between the antecedent A and the consequent B; when It means there is a negative correlation between the antecedent A and the consequent B.

[0016] Furthermore, the characteristic index is measured according to the improvement degree, and three indexes, namely, dissolved gas analysis, electrical test and insulating oil test, are selected as subsets of the factor set of transformer failure, namely, F = {F1, F2, F3}; a subset consists of multiple factors, namely, F1 = {F 11 ,F 12 ,F 13 ,...}; A transformer condition assessment system is constructed based on the factor set.

[0017] Furthermore, the review set is divided into five levels, namely excellent, good, general, failure, and severe failure.

[0018] Furthermore, it is also necessary to use the hierarchical analysis method to calculate the weight coefficient θ of the characteristic index p , p=1,2,...,P.

[0019] Furthermore, in the dynamic fuzzy membership adjustment mechanism, the level in the comment set is equivalent to the state quantity S, and each state quantity S uses a triangular membership function to represent the corresponding fuzziness degree. The triangular membership function μ S (x) is expressed in the form of:

[0020]

[0021] Among them, a i , b i and c i is a set of parameters defining the triangle membership function, i=1, 2, 3, 4, 5, that is, each state quantity has a corresponding set of parameters, a total of 5 sets;

[0022] Five groups of parameters are dynamically adjusted and the particle swarm optimization algorithm is used for parameter optimization. The specific steps are as follows: the first step is to represent a group of parameters corresponding to each membership function as a particle and initialize the particle group; the second step is to define the error function of minimizing the fuzzy control system as the objective function; the third step is to update the position and velocity formula of each particle in the solution space; the fourth step is to repeat the parameter updating process and perform iterative optimization until the error function converges to the minimum value;

[0023] By substituting the measured index data into the defined membership function, the membership value of each state quantity is calculated, and the fuzzy evaluation matrix R of the membership distribution is generated. p , p=1,2,...,P.

[0024] Furthermore, the implementation process of the improved DS evidence theory based on Spearman rank correlation coefficient is as follows: the identification framework of the improved DS evidence theory based on Spearman rank correlation coefficient is divided into five levels H1, H2, H3, H4, H5 and uncertain Right now Given a body of evidence E p and E q ,and Sort the basic probability distribution of each evidence body to obtain a level sequence; p and E q Each pair of sample points (x p ,x q ) and (y p ,y q ), p=1,2,...,P, q=1,2,...,Q, calculate the level difference d p =rank(x p )-rank(y p );Calculate the evidence body E p and E q Spearman rank correlation coefficient Here, n represents the number of samples.

[0025] Further, using the Spearman rank correlation coefficient ρ pq To adjust the weight of the evidence, the implementation process is:

[0026] Assume there are P evidence bodies, weight ω p The calculation formula is ω p =1 / (1+∑ q≠p 1-ρ pq ), where ω p is the body of evidence E qThe weight of each evidence body is adjusted by weight, that is, m' p =ω p ·m p ; Among them, m p represents the original basic probability distribution; m' p represents the modified basic probability distribution.

[0027] Furthermore, the modified basic probability distribution m' p and m' q Dempster combination rule is used for fusion. p and E q , the combination rules are:

[0028]

[0029] Among them, m'(A) represents the basic probability distribution value after fusion; m' p (B) and m' q (C) represents the modified basic probability distribution; B and C represent all possible sets of evidence bodies; It represents the amount of evidence conflict. If the amount of evidence conflict is close to 1, the conflict is relatively large, and it is necessary to review the weight of the evidence body and the correction of the basic probability distribution.

[0030] Furthermore, the final rating value of the transformer is obtained based on the result m'(A) obtained by the combination rule. Wherein, H(i) represents the five levels of degradation of the transformer, i=1, 2, 3, 4, 5.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] 1. The present invention judges the rationality of feature indicators based on the degree of lift, which helps to identify and prioritize the most influential features for transformer condition assessment. The degree of lift measures the strong correlation between the features and the assessment results, and can eliminate irrelevant or redundant features, thereby improving the accuracy and effectiveness of the assessment system. This method ensures that the selected feature indicators have practical significance and reference value in data analysis, which helps to optimize the transformer condition assessment model.

[0033] 2. The present invention uses a dynamic fuzzy membership adjustment mechanism to determine the basic probability of characteristic indicators, which can dynamically adjust and optimize the weight of each characteristic in the evaluation to reflect actual changes and uncertainties. This method improves the adaptability and accuracy of the evaluation model, making the transformer status evaluation more realistic and reliable, and can more accurately capture subtle changes in the equipment status, thereby improving the accuracy and credibility of the comprehensive evaluation.

[0034] 3. The present invention uses the improved DS evidence theory based on the Spearman rank correlation coefficient to fuse different characteristic indicators, which can effectively integrate the relevant information of each indicator and handle the nonlinear relationship between indicators. This method improves the accuracy and robustness of the comprehensive evaluation, reduces the influence of subjective bias and data noise by considering the ranking correlation of characteristic indicators, and thus provides a more accurate and reliable evaluation result for the transformer status. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A method flow chart of a transformer state assessment method based on fuzzy mathematics theory of the present invention;

[0036] Figure 2 is a schematic diagram of a transformer status assessment system of the present invention;

[0037] Figure 3 It is a flow chart of the improved DS evidence theory based on Spearman rank correlation coefficient of the present invention. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] In order to improve the accuracy of transformer status assessment, the present invention provides a transformer status assessment method based on fuzzy mathematics theory. In order to illustrate the function of the present invention, the effectiveness of the present invention will be described from the following examples.

[0040] Embodiment 1

[0041] The status of transformer A is evaluated, and other evaluation methods are compared to verify the effectiveness of the transformer status evaluation method based on fuzzy mathematics theory used in this embodiment.

[0042] Reference Figure 1 , showing a method flow chart of transformer condition assessment method based on fuzzy mathematics theory. Figure 1 In step S01, original test data of the transformer is collected and normalized; support and lift of characteristic indicators are calculated based on association rules, and rationality of the characteristic indicators is judged based on the lift.

[0043] Among them, since the dimension and scale of each indicator in the transformer evaluation system are different, normalization processing is required.

[0044] Furthermore, the association rule is defined as: if the transaction database is D, and A and B are transaction subsets of D, then and but It is expressed as an association rule, A and B are the antecedent and consequent of the association rule;

[0045] Define the following variables: freq AB Indicates the number of times A and B appear at the same time; freq A Indicates the number of times A appears; freq B represents the number of times B appears; total represents the total number of transactions in the transaction database D; the calculation formula for support is support AB =freq AB / total;support A =freq A / total;support B =freq B / total; the calculation formula for lift is Simplified to

[0046] The correlation of the feature indicators is determined according to the lift:

[0047] when It means that there is a positive correlation between the antecedent A and the consequent B, that is, the appearance of A increases the possibility of the appearance of B, indicating that there is an actual correlation between the characteristic indicators A and B, that is, the appearance of A and B is independent;

[0048] when It means that there is no significant correlation between the antecedent A and the consequent B;

[0049] when It means that there is a negative correlation between the antecedent A and the consequent B, that is, the appearance of A reduces the possibility of the appearance of B, indicating that there is no actual connection between A and B or there is a negative correlation.

[0050] The support and lift of feature indicators are calculated based on association rules, which can effectively identify the most influential features for transformer condition assessment.

[0051] Furthermore, the characteristic index is measured according to the improvement degree, and three indexes, namely, dissolved gas analysis, electrical test and insulating oil test, are selected as subsets of the factor set of transformer failure, namely, F = {F1, F2, F3}; a subset consists of multiple factors, namely, F1 = {F 11 ,F 12 ,F 13 ,...}. The improvement ensures that the selected indicators have practical significance, thereby improving the accuracy and effectiveness of the evaluation.

[0052] Further, refer to Figure 1 In step S02, a transformer condition evaluation system is established according to the selected characteristic index, and the transformer comment set is graded. Figure 2 The evaluation set is divided into five levels, namely excellent, good, general, fault, and severe fault. The transformer operation status level is shown in Table 1.

[0053] Table 1 Transformer operating status level

[0054]

[0055] Dividing the comment set into five levels can describe the equipment status in more detail, help with accurate diagnosis and evaluation, and thus formulate more effective maintenance and management measures.

[0056] Further, refer to Figure 1 Step S03 in the embodiment uses a dynamic fuzzy membership adjustment mechanism to determine the basic probability of the characteristic index. First, the weight coefficient θ of the characteristic index needs to be calculated using the hierarchical analysis method. p , p=1,2,...,P; the steps of the hierarchical analysis method are: the first step is to construct a comparison matrix; the second step is to calculate the importance ranking of the characteristic indicators; the third step is to perform a consistency check to ensure that the comparison matrix is ​​close to the actual operation of the transformer. The hierarchical analysis method used in this embodiment is a prior art, and the specific process of its method steps will not be described in detail here.

[0057] In this embodiment, the transformer A state index weight coefficient is shown in Table 2.

[0058] Table 2 Transformer A status index weight coefficient

[0059]

[0060] By using the analytic hierarchy process to calculate the weight coefficients of characteristic indicators, the importance of each indicator can be systematically evaluated, thereby more accurately reflecting its impact on the transformer status.

[0061] Furthermore, in the dynamic fuzzy membership adjustment mechanism, the level in the comment set is equivalent to the state quantity S, and each state quantity S uses a triangular membership function to represent the corresponding fuzziness degree. The triangular membership function μ S (x) is expressed in the form of:

[0062]

[0063] Among them, a i , b iand c i is a set of parameters defining the triangle membership function, i=1, 2, 3, 4, 5, that is, each state quantity has a corresponding set of parameters, a total of 5 sets;

[0064] Dynamically adjust the five groups of parameters and use the particle swarm optimization algorithm to optimize the parameters. The specific steps are as follows:

[0065] The first step is to represent a set of parameters corresponding to each membership function as a particle, that is, p = [p1, p2, p3, ..., p i ]; where p i ={a i ,b i ,c i}; Initialize particle group p;

[0066] The second step is to define the error function of minimizing the fuzzy control system as the objective function:

[0067]

[0068] in, represents the predicted output calculated based on the parameter p.

[0069] The third step is to update the position and velocity formula of each particle in the solution space:

[0070] The speed update formula is:

[0071]

[0072] The position update formula is:

[0073]

[0074] in, represents the velocity of the i-th particle in the t-th round; represents the position of the i-th particle in the t-th round, i.e., the parameter value; ω represents the inertia weight; c1 and c2 represent the learning factors; r1 and r2 represent random numbers; p best,i represents the best historical position of the i-th particle; g best represents the global optimal position.

[0075] In the fourth step, the particle velocity and position are repeatedly updated until the error function f(p) converges to the minimum value or reaches the preset number of iterations.

[0076] By substituting the measured index data into the defined membership function, the membership value of each state quantity is calculated, and the fuzzy evaluation matrix R of the membership distribution is generated. p , p=1,2,...,P.

[0077] The dynamic fuzzy membership adjustment mechanism is combined with the particle swarm optimization algorithm to optimize the parameters of the triangular membership function, which can accurately adjust the fuzzy evaluation matrix and improve the accuracy of the fuzzy measurement of the state quantity. The optimized membership function can more accurately reflect the actual state of the transformer and support more effective decision-making and management. This method enables the evaluation system to adapt to different transformer states more flexibly, thereby improving the reliability and accuracy of the evaluation results.

[0078] Further, refer to Figure 1 In step S04, the improved DS evidence theory based on the Spearman rank correlation coefficient is used to fuse the basic probabilities of the different characteristic indicators to perform a comprehensive transformer status assessment. Figure 3 As shown, a flowchart of the improved DS evidence theory based on the Spearman rank correlation coefficient is presented.

[0079] The implementation process of the improved DS evidence theory based on Spearman rank correlation coefficient is as follows: the identification framework of the improved DS evidence theory based on Spearman rank correlation coefficient is divided into five levels H1, H2, H3, H4, H5 and uncertain H6, which represent the degree of transformer degradation. Right now Given a body of evidence E p and E q , and E p , Sort the basic probability distribution of each evidence body to obtain a level sequence; p and E q Each pair of sample points (x p ,x q ) and (y p ,y q ), p=1,2,...,P, q=1,2,...,Q, calculate the level difference d p =rank(x p )-rank(y p );Calculate the evidence body E p and E q Spearman rank correlation coefficient Here, n represents the number of samples.

[0080] In this embodiment, sorting, calculating rank difference and Spearman rank correlation coefficient help to quantify the correlation between feature indicators. These steps can improve the consistency and accuracy of data, provide a solid foundation for weight calculation and basic probability allocation, thereby enhancing the effect of evidence fusion and conflict handling, and improving the reliability of the final decision.

[0081] Further, using the Spearman rank correlation coefficient ρ pq To adjust the weight of the evidence, the implementation process is:

[0082] Assume there are P evidence bodies, weight ω p The calculation formula is ω p =1 / (1+∑ q≠p 1-ρ pq ), where ω p is the body of evidence E q The weight of each evidence body is adjusted by weight, that is, m' p =ω p ·m p ; Among them, m p Represents the original basic probability distribution, that is; m' p represents the modified basic probability distribution.

[0083] The Spearman rank correlation coefficient is used to assign the relative importance of each indicator, which enhances the accuracy of the evaluation. The distribution of evidence is adjusted according to the weight, thereby correcting the basic probability distribution, making the final evidence fusion result more in line with the actual situation, which improves the reliability and accuracy of decision-making.

[0084] Furthermore, the modified basic probability distribution m' p and m' q Dempster combination rule is used for fusion. p and E q , the combination rules are:

[0085]

[0086] Among them, m'(A) represents the basic probability distribution value after fusion; m' p (B) and m' q (C) represents the modified basic probability distribution; B and C represent all possible sets of evidence bodies; It represents the amount of evidence conflict. If the amount of evidence conflict is close to 1, the conflict is relatively large, and it is necessary to review the weight of the evidence body and the correction of the basic probability distribution.

[0087] By applying the Dempster combination rule, the evidence from different sources is integrated and the degree of support of each piece of evidence is comprehensively considered, which improves the comprehensiveness and accuracy of the overall evaluation. In addition, handling evidence conflicts optimizes the evidence integration process, resolves inconsistencies between evidence, and enhances the credibility and effectiveness of the final decision.

[0088] Finally, the final score of the transformer is obtained based on the result m'(A) obtained by the combination rule. Wherein, H(i) represents the five levels of degradation of the transformer, i=1, 2, 3, 4, 5.

[0089] The final rating value of the transformer is calculated by combining the rank correlation coefficient and DS evidence theory, which helps to improve the accuracy of evidence fusion in the presence of conflicts.

[0090] In this embodiment, after collecting the test data of three indicators of dissolved gas analysis, electrical test and insulating oil test of transformer A, the method used in this embodiment described above is compared with the DS classic method and the Murphy method. The comparative experimental results are shown in Table 3.

[0091] Table 3 Comparative experimental results for transformer A test data

[0092] method Rating value DS Classic Method 78.65 Murphy Method 74.08 The method used in this example 84.26

[0093] As can be seen from Table 3, the final score obtained by the method used in this embodiment is 84.26, which is a serious fault state. When the transformer A is inspected after power outage, it is found that the winding has a trace of fuse, which shows that the method used in this embodiment can better meet the requirements of maintenance equipment.

[0094] The transformer status assessment method based on fuzzy mathematics theory used in this embodiment can significantly improve the assessment accuracy and reliability. By dynamically adjusting and integrating different characteristic indicators, the method effectively reflects the actual operating status of the transformer, reduces the possibility of misjudgment and missed judgment, and thus optimizes maintenance decisions and prevents faults.

[0095] Embodiment 2

[0096] State assessments were performed on several transformers to verify the effectiveness of a transformer state assessment method based on fuzzy mathematics theory used in this embodiment.

[0097] The transformer state assessment method based on fuzzy mathematics theory used in this embodiment is the same as the method used in the previous embodiment.

[0098] Firstly, the original test data of the transformer is collected and normalized; the support and lift of the characteristic index are calculated based on the association rules, and the rationality of the characteristic index is judged based on the lift.

[0099] The association rule is defined as: if the transaction database is D, and A and B are transaction subsets of D, then and but It is represented as an association rule, A and B are the antecedent and consequent of the association rule;

[0100] Define the following variables: freqAB Indicates the number of times A and B appear at the same time; freq A Indicates the number of times A appears; freq B represents the number of times B appears; total represents the total number of transactions in the transaction database D; the calculation formula for support is support AB =freq AB / total;support A =freq A / total;support B =freq B / total; the calculation formula for lift is Simplified to

[0101] The correlation of the feature indicators is determined according to the lift:

[0102] when It means that there is a positive correlation between the antecedent A and the consequent B, that is, the appearance of A increases the possibility of the appearance of B, indicating that there is an actual correlation between the characteristic indicators A and B, that is, the appearance of A and B is independent;

[0103] when It means that there is no significant correlation between the antecedent A and the consequent B;

[0104] when It means that there is a negative correlation between the antecedent A and the consequent B, that is, the appearance of A reduces the possibility of the appearance of B, indicating that there is no actual connection between A and B or there is a negative correlation.

[0105] According to the improvement degree, the characteristic index is measured, and three indexes, namely, dissolved gas analysis, electrical test and insulating oil test, are selected as a subset of the factor set of transformer failure, namely, F = {F1, F2, F3}; a subset is composed of multiple factors, namely, F1 = {F 11 ,F 12 ,F 13 ,...}.

[0106] Secondly, a transformer status evaluation system is established based on the selected characteristic indicators, and the transformer comment set is graded into five levels, namely, excellent, good, general, fault, and severe fault.

[0107] Then, the basic probability of the characteristic index is determined using the dynamic fuzzy membership adjustment mechanism. First, the weight coefficient θ of the characteristic index needs to be calculated using the hierarchical analysis method. p, p=1,2,...,P; the steps of the hierarchical analysis method are: the first step is to construct a comparison matrix; the second step is to calculate the importance ranking of the characteristic indicators; the third step is to perform a consistency check to ensure that the comparison matrix is ​​close to the actual operation of the transformer. The hierarchical analysis method used in this embodiment is a prior art, and the specific process of its method steps will not be described in detail here. The weight coefficients of several transformer status indicators are shown in Table 4.

[0108] Table 4 Weight coefficients of several transformer status indicators

[0109]

[0110] In the dynamic fuzzy membership adjustment mechanism, the level in the comment set is equivalent to the state quantity S, and each state quantity S uses a triangular membership function to represent the corresponding fuzziness degree. The triangular membership function μ S (x) is expressed in the form of:

[0111]

[0112] Among them, a i , b i and c i is a set of parameters defining the triangle membership function, i=1, 2, 3, 4, 5, that is, each state quantity has a corresponding set of parameters, a total of 5 sets;

[0113] Dynamically adjust the five groups of parameters and use the particle swarm optimization algorithm to optimize the parameters. The specific steps are as follows:

[0114] The first step is to represent a set of parameters corresponding to each membership function as a particle, that is, p = [p1, p2, p3, ..., p i ]; where p i ={a i ,b i ,c i}; Initialize particle group p;

[0115] The second step is to define the error function of minimizing the fuzzy control system as the objective function:

[0116]

[0117] in, represents the predicted output calculated based on the parameter p.

[0118] The third step is to update the position and velocity formula of each particle in the solution space:

[0119] The speed update formula is:

[0120]

[0121] The position update formula is:

[0122]

[0123] in, represents the velocity of the i-th particle in the t-th round; represents the position of the i-th particle in the t-th round, i.e., the parameter value; ω represents the inertia weight; c1 and c2 represent the learning factors; r1 and r2 represent random numbers; p best,i represents the best historical position of the i-th particle; g best represents the global optimal position.

[0124] In the fourth step, the particle velocity and position are repeatedly updated until the error function f(p) converges to the minimum value or reaches the preset number of iterations.

[0125] By substituting the measured index data into the defined membership function, the membership value of each state quantity is calculated, and the fuzzy evaluation matrix R of the membership distribution is generated. p , p=1,2,...,P.

[0126] Then, the improved DS evidence theory based on Spearman rank correlation coefficient is used to integrate the basic probabilities of different characteristic indicators to perform comprehensive transformer status assessment. Figure 3 As shown, a flowchart of the improved DS evidence theory based on the Spearman rank correlation coefficient is presented.

[0127] The implementation process of the improved DS evidence theory based on Spearman rank correlation coefficient is as follows: the identification framework of the improved DS evidence theory based on Spearman rank correlation coefficient is divided into five levels H1, H2, H3, H4, H5 and uncertain H6, which represent the degree of transformer degradation. Right now Given a body of evidence E p and E q ,and Sort the basic probability distribution of each evidence body to obtain a level sequence; p and E q Each pair of sample points (x p ,x q ) and (y p ,y q ), p=1,2,...,P, q=1,2,...,Q, calculate the level difference d p =rank(x p )-rank(y p );Calculate the evidence body Ep and E q Spearman rank correlation coefficient Here, n represents the number of samples.

[0128] Using the Spearman rank correlation coefficient ρ pq To adjust the weight of the evidence, the implementation process is:

[0129] Assume there are P evidence bodies, weight ω p The calculation formula is ω p =1 / (1+∑ q≠p 1-ρ pq ), where ω p is the body of evidence E q The weight of each evidence body is adjusted by weight, that is, m' p =ω p ·m p ; Among them, m p Represents the original basic probability distribution, that is; m' p represents the modified basic probability distribution.

[0130] The modified basic probability distribution m' p and m' q Dempster combination rule is used for fusion. p and E q , the combination rules are:

[0131]

[0132] Among them, m'(A) represents the basic probability distribution value after fusion; m' p (B) and m' q (C) represents the modified basic probability distribution; B and C represent all possible sets of evidence bodies; It represents the amount of evidence conflict. If the amount of evidence conflict is close to 1, the conflict is relatively large, and it is necessary to review the weight of the evidence body and the correction of the basic probability distribution.

[0133] Finally, the final score of the transformer is obtained based on the result m'(A) obtained by the combination rule. Wherein, H(i) represents the five levels of degradation of the transformer, i=1, 2, 3, 4, 5.

[0134] Furthermore, in the dynamic fuzzy membership adjustment mechanism, the level in the comment set is equated with the state quantity S, and each state quantity S uses a triangular membership function to represent the corresponding fuzziness degree. The triangular membership function μ S (x) is expressed in the form of:

[0135]

[0136] Among them, a i , b i and c i It is a set of parameters that define the triangle membership function, i = 1, 2, 3, 4, 5, that is, each state quantity has a corresponding set of parameters, a total of 5 groups;

[0137] Dynamically adjust the five groups of parameters and use the particle swarm optimization algorithm to optimize the parameters. The specific steps are as follows:

[0138] The first step is to represent a set of parameters corresponding to each membership function as a particle, that is, p = [p1, p2, p3, ..., p i ]; where p i ={a i ,b i ,c i}; Initialize particle group p;

[0139] The second step is to define the error function of minimizing the fuzzy control system as the objective function:

[0140]

[0141] in, represents the predicted output calculated based on the parameter p.

[0142] The third step is to update the position and velocity formula of each particle in the solution space:

[0143] The speed update formula is:

[0144]

[0145] The position update formula is:

[0146]

[0147] in, represents the velocity of the i-th particle in the t-th round; represents the position of the i-th particle in the t-th round, i.e., the parameter value; ω represents the inertia weight; c1 and c2 represent the learning factors; r1 and r2 represent random numbers; p best,i represents the best historical position of the i-th particle; g best represents the global optimal position.

[0148] In the fourth step, the particle velocity and position are repeatedly updated until the error function f(p) converges to the minimum value or reaches the preset number of iterations.

[0149] By substituting the measured index data into the defined membership function, the membership value of each state quantity is calculated, and the fuzzy evaluation matrix R of the membership distribution is generated. p , p=1,2,...,P.

[0150] Furthermore, the improved DS evidence theory based on Spearman rank correlation coefficient is used to fuse different characteristic indicators for comprehensive transformer condition assessment.

[0151] The implementation process of the improved DS evidence theory based on Spearman rank correlation coefficient is as follows: the identification framework of the improved DS evidence theory based on Spearman rank correlation coefficient is divided into five levels representing the degree of transformer degradation: H1, H2, H3, H4, H5 and uncertain Right now Given a body of evidence E p and E q ,and Sort the basic probability distribution of each evidence body to obtain a level sequence; p and E q Each pair of sample points (x p ,x q ) and (y p ,y q ), p=1,2,...,P, q=1,2,...,Q, calculate the level difference d p =rank(x p )-rank(y p );Calculate the evidence body E p and E q Spearman rank correlation coefficient Here, n represents the number of samples.

[0152] Furthermore, using the Spearman rank correlation coefficient ρ pq To adjust the weight of the evidence, the implementation process is:

[0153] Assume there are P evidence bodies, weight ω p The calculation formula is ω p =1 / (1+∑ q≠p 1-ρ pq ), where ω p It is the body of evidence E q The weight of each body of evidence is adjusted by weight, that is, m' p =ω p ·m p ; Among them, m p Represents the original basic probability distribution, that is; m' p represents the modified basic probability distribution.

[0154] Furthermore, the modified basic probability distribution m' p and m' q Using Dempster combination rule to merge, for the evidence body E p and E q , the combination rules are:

[0155]

[0156] Among them, m'(A) represents the basic probability distribution value after fusion; m' p (B) and m' q (C) represents the modified basic probability distribution; B and C represent all possible sets of evidence bodies; It represents the amount of evidence conflict. If the amount of evidence conflict is close to 1, the conflict is large, and it is necessary to review the weight of the evidence and the correction of the basic probability distribution.

[0157] Finally, the final score of the transformer is obtained based on the result m'(A) obtained based on the combination rule. Among them, H(i) represents the five levels of transformer degradation, i=1,2,3,4,5.

[0158] In this embodiment, the number of transformers is 100, and the operating status of the 100 transformers is known. The operating status evaluated by the method used in this embodiment is compared with the actual operating status to verify the accuracy of the method used in this embodiment.

[0159] The state of 100 transformer test data was evaluated by the method used in this embodiment. The evaluation results are: 40 groups of transformers are in excellent operating condition, 36 groups of transformers are in good operating condition, and 24 groups of transformers are in general operating condition. Among them, the operating state evaluation results of 98 transformers are consistent with the actual operating state, and the operating state evaluation results of 2 transformers are different from the actual operating state. Among these 3 transformers, the operating state evaluation score of transformer 1 is 20.46, which is close to the dividing line 20 of the transformer operating state degradation value; the operating state evaluation result of transformer 2 is 39.63, which is close to the dividing line 40 of the transformer operating state degradation value. Analysis shows that the difference is small. The accuracy of the method used in this embodiment is 98%, which shows the effectiveness of this transformer state evaluation method based on fuzzy mathematics theory for transformer state evaluation.

[0160] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A transformer state assessment method based on fuzzy mathematics theory, characterized in that: include: Collect the original test data of the transformer and perform normalization processing; Calculating the support and lift of the feature indicator based on the association rules, and judging the rationality of the feature indicator based on the lift; Establishing a transformer status evaluation system according to the selected characteristic indicators, and classifying the transformer comment set levels; Determine the basic probability of the characteristic indicator using a dynamic fuzzy membership adjustment mechanism; The improved DS evidence theory based on Spearman rank correlation coefficient is used to integrate the basic probabilities of different characteristic indicators for comprehensive transformer condition assessment.

2. A transformer state assessment method based on fuzzy mathematics theory according to claim 1, characterized in that: The association rule is defined as: if the transaction database is D, and A and B are transaction subsets of D, then and but It is expressed as an association rule, A and B are the antecedent and consequent of the association rule; Define the following variables: freq AB Indicates the number of times A and B appear at the same time; freq A Indicates the number of times A appears; freq B represents the number of times B appears; total represents the total number of transactions in the transaction database D; the calculation formula for support is support AB =freq AB / total;support A =freq A / total;support B =freq B / total; the calculation formula for lift is Simplified to The relevance of the feature index is determined based on the lift: It means that there is a positive correlation between the antecedent A and the consequent B; when It means that there is no significant correlation between the antecedent A and the consequent B; when It means there is a negative correlation between the antecedent A and the consequent B.

3. A transformer state assessment method based on fuzzy mathematics theory according to claim 2, characterized in that: According to the improvement degree, the characteristic index is measured, and three indexes, namely, dissolved gas analysis, electrical test and insulating oil test, are selected as a subset of the factor set of transformer failure, namely, F = {F1, F2, F3}; a subset is composed of multiple factors, namely, F1 = {F 11 ,F 12 ,F 13 ,...}; A transformer condition assessment system is constructed based on the factor set.

4. A transformer state assessment method based on fuzzy mathematics theory according to claim 1, characterized in that: The review set is divided into five levels, namely excellent, good, general, failure, and severe failure.

5. A transformer state assessment method based on fuzzy mathematics theory according to claim 1, characterized in that: It is also necessary to use the hierarchical analysis method to calculate the weight coefficient θ of the characteristic index p , p=1,2,...,P.

6. A transformer state assessment method based on fuzzy mathematics theory according to claim 1, characterized in that: In the dynamic fuzzy membership adjustment mechanism, the level in the comment set is equivalent to the state quantity S, and each state quantity S uses a triangular membership function to represent the corresponding fuzziness degree. The triangular membership function μ S (x) is expressed in the form of: Among them, a i 、b i and c i is a set of parameters defining the triangle membership function, i=1, 2, 3, 4, 5, that is, each state quantity has a corresponding set of parameters, a total of 5 sets; Five groups of parameters are dynamically adjusted and the particle swarm optimization algorithm is used for parameter optimization. The specific steps are as follows: the first step is to represent a group of parameters corresponding to each membership function as a particle and initialize the particle group; the second step is to define the error function of minimizing the fuzzy control system as the objective function; the third step is to update the position and velocity formula of each particle in the solution space; the fourth step is to repeat the parameter updating process and perform iterative optimization until the error function converges to the minimum value; By substituting the measured index data into the defined membership function, the membership value of each state quantity is calculated, and the fuzzy evaluation matrix R of the membership distribution is generated. p , p=1,2,...,P.

7. A transformer state assessment method based on fuzzy mathematics theory according to claim 1, characterized in that: The implementation process of the improved DS evidence theory based on Spearman rank correlation coefficient is as follows: the identification framework of the improved DS evidence theory based on Spearman rank correlation coefficient is divided into five levels H1, H2, H3, H4, H5 and uncertainty θ representing the degree of transformer degradation, that is, θ = {H1, H2, H3, H4, H5, θ}; given the evidence body E p and E q ,and Sort the basic probability distribution of each evidence body to obtain a level sequence; p and E q Each pair of sample points (x p ,x q ) and (y p ,y q ), p=1,2,...,P, q=1,2,...,Q, calculate the level difference d p =rank(x p )-rank(y p );Calculate the evidence body E p and E q Spearman rank correlation coefficient Here, n represents the number of samples.

8. A transformer state assessment method based on fuzzy mathematics theory according to claim 7, characterized in that: Using the Spearman rank correlation coefficient ρ pq To adjust the weight of the evidence, the implementation process is: Assume there are P evidence bodies, weight ω p The calculation formula is ω p =1 / (1+∑ q≠p 1-ρ pq ), where ω p is the body of evidence E q The weight of each evidence body is adjusted by weight, that is, m' p =ω p ·m p ; Among them, m p Represents the original basic probability distribution, that is; m' p represents the modified basic probability distribution.

9. A transformer state assessment method based on fuzzy mathematics theory according to claim 8, characterized in that: The modified basic probability distribution m' p and m' q Dempster combination rule is used for fusion. p and E q , the combination rules are: Among them, m'(A) represents the basic probability distribution value after fusion; m' p (B) and m' q (C) represents the modified basic probability distribution; B and C represent all possible sets of evidence bodies; It represents the amount of evidence conflict. If the amount of evidence conflict is close to 1, the conflict is relatively large, and it is necessary to review the weight of the evidence body and the correction of the basic probability distribution.

10. A transformer state assessment method based on fuzzy mathematics theory according to claim 9, characterized in that: The result m'(A) obtained based on the combination rule is used to obtain the final score of the transformer. Wherein, H(i) represents the five levels of degradation of the transformer, i=1, 2, 3, 4, 5.

Images