Distributed adjustable resource reputation evaluation method and system considering malicious user punishment

Abstract

The invention discloses a distributed adjustable resource reputation evaluation method and system considering malicious user punishment, and the method comprises the steps: obtaining indexes affecting the reputation of a power market subject, and constructing a comprehensive rating index system; with the purpose of minimizing the deviation between the subjective weight and the objective weight, weighting the index system through a game theory combination hierarchical weighting method and an anti-entropy weight method to obtain an initial comprehensive weight; based on the initial comprehensive weight, introducing dynamic characteristics of subject reputation as an adaptive adjustment coefficient, performing adaptive adjustment on a subjective weight and an objective weight, and performing optimization by taking deviation minimization of the subjective weight and the objective weight after adaptive adjustment as a target to obtain a final comprehensive weight; based on the final comprehensive weight, combining a fuzzy comprehensive evaluation method and a utility function to construct a comprehensive reputation evaluation model; a comprehensive reputation evaluation model is constructed through empirical analysis and verification, and suggestions of adopting power consumption limitation punishment for malicious users are proposed by comparing the reputation level predicted by the reputation evaluation model and actual performance.

Description

Technical Field

[0001] The present invention belongs to the technical field of reputation evaluation of distributed adjustable resources in power systems, and more specifically, relates to a distributed adjustable resource reputation evaluation method and system taking into account penalties for malicious users. Background Art

[0002] In the power market, the power supply and demand relationship is affected by a variety of key factors, such as seasonal changes, climate conditions, and labor hours, and presents obvious dynamic changes. This dynamic change makes it difficult to stabilize the supply and demand relationship in the power market, affecting the prediction and management of the power market. The trading behaviors of power market players are ever-changing, their subjective motivations are complex, and their credit status and risks also change dynamically. This complexity and uncertainty poses severe challenges to the existing power market credit assessment system, making it difficult to fully and accurately reflect the credit status of market players.

[0003] Most of the existing power market reputation assessment systems rely on traditional data analysis methods, which often fail to effectively cope with the complexity of the behavior of power market players and the dynamic changes in reputation risks. As a result, the assessment results are often not accurate enough, and it is impossible to fully predict and manage reputation risks. In addition, the current risk prevention and control strategies often lack pertinence and effectiveness, resulting in insufficient risk control measures in market operations. These problems have seriously affected the stability and operational efficiency of the power market. Summary of the invention

[0004] In order to solve the deficiencies in the prior art, the present invention provides a distributed adjustable resource reputation evaluation method and system taking into account the penalty of malicious users. By establishing a rating index system and an evaluation model, it has good performance in terms of sorting consistency, inverse order, sample separation, robustness, etc., so as to better improve decision-making efficiency.

[0005] The present invention adopts the following technical solution.

[0006] A first aspect of the present invention provides a distributed adjustable resource reputation evaluation method taking into account malicious user penalties, comprising the following steps:

[0007] Obtain indicators that affect the reputation of power market entities and construct a comprehensive rating index system; based on the comprehensive rating index system, with the goal of minimizing the deviation between subjective weights and objective weights, weight the index system through a combination of game theory, hierarchical weighting method and anti-entropy weight method to obtain initial comprehensive weights; based on the initial comprehensive weights, introduce the dynamic characteristics of the subject's reputation as an adaptive adjustment coefficient, adaptively adjust the subjective weights and objective weights, and then optimize the final comprehensive weights with the goal of minimizing the deviation between the subjective weights and objective weights after adaptive adjustment; based on the final comprehensive weights, combine fuzzy comprehensive evaluation method and utility function to construct a comprehensive reputation evaluation model; verify through empirical analysis, construct a comprehensive reputation evaluation model, and put forward reputation evaluation suggestions by comparing the reputation level predicted by the reputation evaluation model with the actual performance.

[0008] Preferably, the indicators that affect the reputation of the power market subject are obtained, and a comprehensive rating indicator system is constructed, including: constructing a reputation effect path transmission model of the transaction behavior of the distributed adjustable resource subject, as follows:

[0009] Obtain various key factors that affect the credit of distributed adjustable resource subjects and perform standardization, calculate the direct grey correlation between the standardized key factors, and construct a direct grey relationship matrix;

[0010] Based on the direct grey relationship matrix, the real direct influence matrix X is calculated as follows: ij ] n×n :

[0011]

[0012] In the formula, represents the influence degree of the kth key factor i on the key factor j after grey processing; K represents the reachability matrix; x ij It represents the average direct impact of all judged key factors i on key factor j;

[0013] The real-numbered direct impact matrix is ​​used as the comprehensive impact measure between key factors;

[0014] The direct impact of real numbering on matrix normalization is as follows:

[0015]

[0016] In the formula, A represents the real number after normalization, which directly affects the matrix; a ij It represents the normalized direct impact value of key factor i on key factor j; It represents the total influence of key factor i on all key factors; Indicates the maximum total influence of all key factors;

[0017] The comprehensive influence matrix T is calculated using the following relationship:

[0018]

[0019] The reputation effect path transmission model is as follows:

[0020] M=f(T)

[0021] Where I is the unit matrix; It means that it includes all direct effects and all indirect effects; for matrix A, O is a zero matrix.

[0022] Preferably, the key factors affecting the credit of the distributed adjustable resource subject are obtained and standardized, and the direct grey correlation between the standardized key factors is calculated to construct a direct grey relationship matrix, which is as follows:

[0023] The relationship between the key influencing factors was evaluated using a gray number ranging from 0.1 to 0.9;

[0024] The gray number is clarified by the following relationship and converted into a certain value:

[0025]

[0026] In the formula, and They represent the lower and upper bounds of the degree of influence of key factor i judged by the kth evaluation source on key factor j; and They represent the lower and upper bounds of the influence of key factor i on key factor j after clarification; represents the extreme value of the grey judgment matrix of the kth evaluation source, Used to represent the maximum difference in the judgment of the impact degree among all key factors;

[0027] The clarified values ​​are standardized using the following relationship to make the evaluation of different key factors on the same scale:

[0028]

[0029] Where: represents the normalized clarity value of key factor i of the kth judgment with respect to key factor j; It indicates the influence degree of the key factor i of the kth judgment on the key factor j after gray processing;

[0030] Based on the standardized clarity values, a direct grey relationship matrix is ​​constructed, and the elements in the matrix represent the degree of direct influence between the key factors; the direct grey relationship matrix is ​​as follows:

[0031]

[0032] Where: represents the direct influence of the ith key factor on the jth key factor; k = 1, 2, ..., K represents the number of different evaluation sources, and n = 13 represents the number of key factors; It represents the influence degree of key factor i on key factor j as considered by k evaluation sources, which is included in the direct grey relationship matrix and has

[0033] Preferably, in the reputation effect path conduction model, based on the obtained comprehensive influence matrix T, a deconstruction and interaction structure analysis model is established, and the reachable relationship matrix K is calculated using the following relationship:

[0034] H=T+I

[0035] H=[h ij ] n×n

[0036]

[0037] K=[k ij ] n×n ,n=13

[0038] In the formula, I represents the unit matrix; H represents the overall influence matrix; λ represents the exogenous threshold; h ij represents the key factor in the i-th row and j-th column of the overall impact matrix H; k ij Represents the key factor of the i-th row and j-th column in the reachability matrix K;

[0039] Calculate the reachable set R of each key factor based on the reachable matrix i and the predecessor set A i , by verifying whether R i =R i ∩A i To identify key factors at different levels, gradually remove the identified key factors until all key factors are layered; draw a multi-level key factor association path diagram based on the hierarchical division results and the reachability matrix, display all key factors and their interrelationships, so as to characterize the path of reputation transmission from multi-subject transaction behaviors;

[0040] The reachable set R i and the predecessor set A i as follows:

[0041] R i ={C j |C j ∈C,r ij ≠0},i=1,2,...,13; j=1,2,...,13

[0042] A i ={C j |C j ∈C,r ji ≠0},i=1,2,...,13; j=1,2,...,13

[0043] C={C1,C2,...,C 13}

[0044] Where:

[0045] R i Indicates that from the key factor C i All key factors for arrival;

[0046] A i Indicates that all key factors C i Key factors that have an impact;

[0047] r ji Indicates the key factor C j For key factor C i impact.

[0048] Preferably, based on the constructed reputation effect path conduction model, the influence of each key factor on the reputation of the power market subject is analyzed, and the key indicators that have a significant impact on the subject's reputation are identified; through investigation, the importance of each identified key indicator is evaluated within the range of [0,10], and the conservative value and optimistic value of the importance of each indicator are obtained;

[0049] Based on the survey results, the conservative triangular fuzzy number of each key indicator is calculated. and optimistic triangular fuzzy numbers

[0050] Calculate the significant consistency value G for each key indicator i To check the consistency of the evaluation sources:

[0051] like It means that the evaluation sources have completely consistent judgments on indicator i, and the significant consistency value G of indicator i i as follows:

[0052]

[0053] like Then calculate the grayscale interval value of the indicator and If M i ≥Z i , the significant consistency value G of index i i as follows:

[0054]

[0055] If M i <Z i , indicating that the evaluation sources have inconsistent judgment results on indicator i, then repeat the above steps until the significant consistency value G of all indicators is calculated. i ;

[0056] The indicators are screened and the comprehensive rating indicator system is constructed according to the significant consistency value as follows:

[0057] Setting G i The threshold of is the average of the conservative and optimistic values;

[0058] If G i If it is greater than the threshold, the indicator is retained;

[0059] If G i If it is less than or equal to the threshold, the indicator is deleted.

[0060] Preferably, the conservative triangular fuzzy number of each key indicator is and optimistic triangular fuzzy numbers In the above example, the lower and upper limits of the conservative value of index i are and It represents the geometric mean of the conservative values ​​of index i except the lower limit and upper limit;

[0061] The lower and upper limits of the optimistic value for indicator i are and It represents the geometric mean of the optimistic values ​​of indicator i except the lower limit and upper limit.

[0062] Based on the subjective weighting of linear weighted fuzzy assessment LBWA, the evaluation source opinions are selected from the indicator set C = {C1, C2, ..., C n} select the most important indicator and define it as the optimal indicator C B , and stratify the remaining indicators; calculate the influence function of the indicators in each layer, use the elasticity coefficient r0 to determine the weight, and obtain the weight distribution of subjective weighting;

[0063] Using the entropy concept in information theory, the anti-entropy value of each indicator is calculated and normalized to obtain the objective weight of each indicator;

[0064] The indicator weight w is obtained by the anti-entropy weight AEW method using the following relationship1i :

[0065]

[0066] r ij =x ij / ∑ j x ij

[0067] w 1i =h i / ∑ i h i

[0068] Where:

[0069] h i Represents the anti-entropy value of the i-th indicator;

[0070] r ji Indicates the key factor C in the overall impact matrix R j For key factor C i The impact of

[0071] Weight w 1i is the anti-entropy value h i The normalized result of is used to indicate the relative importance of the i-th indicator among all indicators;

[0072] The subjective and objective weights are combined through game theory methods to define a basic weight set, and the combined weights are optimized to minimize the deviation between the subjective and objective weights to obtain the initial comprehensive weights;

[0073] The initial comprehensive weights based on game theory are as follows:

[0074]

[0075] Where:

[0076] represents the initial comprehensive weight obtained based on game theory optimization;

[0077] represents the optimal allocation coefficient of the lth basic weight;

[0078] W l Represents the initial weight vector obtained by the lth weighting method.

[0079] Preferably, based on the subjective weighting of the linear weighted fuzzy assessment LBWA, the index set C = {C1, C2, ..., C n} select the most important indicator and define it as the optimal indicator C B, and divide the remaining indicators into layers; calculate the influence function of the indicators in each layer, use the elasticity coefficient r0 to determine the weight, and obtain the weight distribution of subjective weighting, as follows:

[0080] The specific stratification rules for indicators are as follows:

[0081] S1: The difference between the importance and the optimal indicator is within the set range, and the importance is no more than 2 times lower than the optimal indicator; S2: The importance is more than 2 times but no more than 3 times lower than the optimal indicator; S k Layer: indicators whose importance is more than 3 times lower than the optimal indicator and increase in multiples;

[0082] The importance of determining indicators is as follows:

[0083] Set for any indicator C i , the importance is expressed as S(C i ), all indicators are divided into different levels according to their importance S k (k=1,2,...,K), the set of each level is represented by S=S1∪S2∪…∪S K ; For any p,q∈{1,2,…,K}, if p≠q, then S p ∩S q =φ, indicating that each level does not overlap; p and q are the indexes of the level, used to indicate different levels;

[0084] The importance within the comparison layer is as follows:

[0085] At all levels In the above example, we determine the optimal index for the level Compare the importance of other indicators in the layer with the optimal indicator and record the results Indicates that in the k-level, the i-th k The importance of each indicator compared with the optimal indicator of this layer; The value of is defined as an integer in the interval [0, r]; when k = 1, and The constant r of the indicator importance determined by the stratification results is calculated using the following relationship:

[0086] r=max{|S1|,|S2|,…,|S K |}

[0087] In the formula, |S K | represents the set S k The number of indicators in ;

[0088] The elasticity coefficient r0 is determined based on the hierarchical weighted assessment LBWA as follows:

[0089] The constant r represents the maximum difference in the importance judgment of the indicator in the layer. According to the value of r, the elasticity coefficient r0 is defined to satisfy r0>r, and r0=r+1;

[0090] The influence function of each indicator is calculated using the following relationship:

[0091]

[0092] In the formula, Indicates the i-th level in the k-th level k indicators;

[0093] according to The indicator weight is calculated using the following relationship:

[0094]

[0095] In the formula, k represents the number of layers; N k Indicates the number of indicators at the kth level;

[0096] The weights of the remaining indicators are as follows:

[0097]

[0098] In the formula, Indicates the i-th level in the k-th level k The weight of each indicator;

[0099] The subjective weights of all levels and indicators are sorted into a set w 2i .

[0100] Preferably, based on the initial comprehensive weight, the dynamic characteristics of the subject's reputation are introduced as an adaptive adjustment coefficient, and the subjective weight and the objective weight are adaptively adjusted. Then, the final comprehensive weight is optimized with the goal of minimizing the deviation between the subjective weight and the objective weight after the adaptive adjustment, as follows:

[0101] A weighted normalized decision matrix based on the multi-attribute ranking MARCOS method of complex ratio comparison is established to reflect the dynamic reputation characteristics, and the number of alternatives is set to m, the number of indicators is n, and the indicator value is x. ij (i=1,2,…,n,j=1,2,…,m) multiple attribute decision making problem, the original decision matrix is ​​X=[x ij ] n×m ; Normalized decision matrix X * as follows:

[0102] Standardize benefit-based indicators and cost-based indicators:

[0103]

[0104] In the formula, x ij Represents the indicator value in the original decision matrix; Indicates the normalized indicator value; It represents the maximum value of the i-th indicator among all the schemes; represents the minimum value of the ith indicator among all solutions; for interval indicators, determine the optimal interval of the indicator [x ij,BL ,x ij,UP ] and the theoretical interval [x ij,min ,x ij,max ], if the indicator value is within the optimal interval, the normalized value is 1; if it is at the boundary of the theoretical interval, it is 0; if it is between the two, the normalized value is calculated by linear interpolation; the standardized decision matrix is ​​obtained

[0105] The normalized decision matrix X * Multiply it by the initial indicator weight and calculate the weighted normalized decision matrix R using the following relationship:

[0106]

[0107] In the formula, represents the final weight.

[0108] Preferably, based on the final comprehensive weight, a comprehensive reputation evaluation model is constructed by combining the fuzzy comprehensive evaluation method and the utility function, as follows:

[0109] Based on the weighted normalized decision matrix, the positive and negative ideal solutions of MARCOS are calculated as follows:

[0110]

[0111] In the formula, S + represents the set of positive ideal solutions, the optimal value of each indicator; S - represents the set of negative ideal solutions, the worst value of each indicator; It represents the maximum value of the i-th indicator among all the schemes; It represents the minimum value of the i-th indicator among all the schemes;

[0112] The difference and quotient of the alternative solutions relative to the positive and negative ideal solutions are calculated using the following relationship:

[0113]

[0114] Where: It represents the difference between the value of the i-th indicator in the j-th solution and the positive ideal solution; It represents the difference between the value of the i-th indicator in the j-th solution and the negative ideal solution; represents the quotient of the value of the i-th indicator in the j-th solution and the positive ideal solution; represents the quotient of the value of the i-th indicator in the j-th solution and the negative ideal solution;

[0115] The geometric and numerical similarity grey relational degree is calculated using the following relationship:

[0116]

[0117] Where: Indicates the jth solution relative to the positive ideal solution The geometric similarity grey relational degree of Indicates the jth solution relative to the negative ideal solution The geometric similarity grey relational degree of

[0118] The numerical similarity grey relational degree is calculated using the following relationship:

[0119]

[0120]

[0121] Where: Indicates the jth solution relative to the positive ideal solution The numerical similarity grey relational degree; Indicates the jth solution relative to the negative ideal solution The numerical similarity grey relational degree;

[0122] The comprehensive grey relational degree is calculated using the following relationship:

[0123]

[0124] Where: Indicates the jth solution relative to the positive ideal solution The comprehensive grey relational degree of Indicates the jth solution relative to the negative ideal solution The comprehensive grey relational degree of

[0125] The utility function is calculated using the following relationship:

[0126]

[0127] Where: Indicates the jth solution relative to the positive ideal solution The utility function of Indicates the jth solution relative to the negative ideal solution The utility function of

[0128] The utility function is summarized as follows:

[0129]

[0130] Where: f * (γ j ) represents the comprehensive utility function value of the jth solution, which is used to comprehensively evaluate the pros and cons of the solutions.

[0131] Preferably, a comprehensive reputation evaluation model is constructed through empirical analysis and verification, and by comparing the reputation level predicted by the reputation evaluation model with the actual performance, suggestions for electricity restriction penalties for malicious users are proposed, as follows:

[0132] Use the constructed comprehensive reputation evaluation model to predict the reputation level of each market entity, and calculate the expected reputation value of each entity based on the input of the comprehensive reputation evaluation model;

[0133] The reputation level predicted by the model is compared with the actual performance of market entities collected; market entities below the set threshold are defined as malicious users; the formula for the power demand of malicious users is as follows:

[0134]

[0135] Where:

[0136] represents the actual power consumption of malicious user e in period t;

[0137] represents the minimum power consumption of all users in period t;

[0138] represents the maximum power consumption of all users in period t;

[0139] rand represents a random number between 0 and 1;

[0140] N mal represents the set of malicious users;

[0141] Penalize malicious users for electricity demand, including:

[0142] Set the power consumption of malicious users to the minimum power consumption of all users;

[0143] Replace false data provided by malicious users and increase penalties;

[0144] The power consumption of malicious users in period t is set as follows:

[0145]

[0146] Where:

[0147] is the power consumption after being identified as a malicious user, indicating the minimum power consumption of all users in period t.

[0148] The second aspect of the present invention provides a distributed adjustable resource reputation evaluation system taking into account malicious user penalties, including: an indicator acquisition and rating system construction module, an initial comprehensive weight assignment module, a dynamic weight adjustment module, a comprehensive reputation evaluation model construction module, and an empirical analysis and evaluation suggestion module;

[0149] The indicator acquisition and rating system construction module is used to obtain key indicators that affect the reputation of power market entities. By constructing a path transmission model of the effect of distributed adjustable resource entity transaction behavior on reputation, it describes how different key factors affect the reputation of market entities through transaction behavior, collects and identifies key indicators that affect the reputation of power market entities, and constructs a comprehensive rating indicator system.

[0150] The initial comprehensive weighting module is used to assign weights to each indicator in the comprehensive rating index system using a game theory combined hierarchical weighting method and an anti-entropy weighting method, with the goal of minimizing the deviation between subjective weights and objective weights, and calculating the initial comprehensive weights;

[0151] The dynamic weight adjustment module is used to introduce the dynamic characteristics of the subject's reputation as an adaptive adjustment coefficient, dynamically adjust the initial comprehensive weight, and optimize the final comprehensive weight;

[0152] The comprehensive reputation evaluation model construction module is used to combine the fuzzy comprehensive evaluation method and the utility function, build a comprehensive reputation evaluation model based on the comprehensive rating index system and comprehensive weights, and evaluate the reputation level of power market entities;

[0153] The empirical analysis and evaluation suggestion module is used to verify based on empirical analysis, compare the reputation level predicted by the reputation evaluation model with the actual performance, and make reputation evaluation suggestions.

[0154] The third aspect of the present invention provides a terminal, including a processor and a storage medium; the storage medium is used to store instructions; the processor is used to operate according to the instructions to execute the steps of a distributed adjustable resource reputation evaluation method taking into account malicious user penalties.

[0155] A fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a distributed adjustable resource reputation evaluation method that takes into account malicious user penalties.

[0156] Compared with the prior art, the beneficial effects of the present invention include at least:

[0157] (1) The present invention introduces the dynamic characteristics of subject reputation as an adaptive adjustment coefficient, so that the weight allocation can respond to the changes in the reputation of power market subjects in real time. The dynamic adjustment capability overcomes the shortcomings of traditional static evaluation methods and is more in line with the actual situation of the power market.

[0158] (2) The present invention combines the hierarchical weighting method and the anti-entropy weight method through game theory, comprehensively considers the deviation of subjective and objective weights, and optimizes them. Compared with the traditional single weighting method, the present invention is more scientific and reasonable in weight allocation;

[0159] (3) The present invention provides a more comprehensive reputation evaluation result by combining the fuzzy comprehensive evaluation method and the utility function. Compared with the traditional evaluation method, the comprehensive evaluation model of the present invention can better handle complex and uncertain market information and reduce errors;

[0160] (4) The present invention verifies the reputation evaluation model through empirical analysis and proposes power consumption restriction and penalty recommendations for malicious users, so that the evaluation results have practical application value and improve the stability and efficiency of market operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0161] Figure 1 It is a flow chart of the analysis of the transmission path of the distributed adjustable resource subject transaction behavior on the reputation effect provided in accordance with the embodiment of the present invention;

[0162] Figure 2 It is a framework diagram of a hybrid multi-attribute decision-making reputation rating model provided in accordance with an embodiment of the present invention;

[0163] Figure 3 The present invention provides a general flow chart of a distributed adjustable resource reputation evaluation method taking into account the penalty for malicious users. DETAILED DESCRIPTION

[0164] In order to make the purpose, technical scheme and advantages of the present invention clearer, the technical scheme of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. The embodiments described in this application are only part of the embodiments of the present invention, not all of them. Based on the spirit of the present invention, other embodiments obtained by ordinary technicians in this field without creative work are all within the scope of protection of the present invention.

[0165] The following is a further detailed description of the specific implementation of the distributed adjustable resource reputation evaluation method considering the penalty of malicious users in conjunction with the accompanying drawings and embodiments. The following embodiments are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0166] like Figure 1The flowchart of the analysis of the transmission path of the trading behavior on the reputation shown in the figure uses the grey theory and the analysis method of the decision-making test and evaluation laboratory to explore the influence of the attributes and importance of the trading behavior of multiple entities in the power market on each key influencing factor in the reputation path transmission model, and divides the hierarchical structure by explaining the structural model method, and studies the transmission path of the trading behavior of multiple entities in the power market on the reputation, which is helpful to establish a reference basis for the multiple entity reputation rating model system.

[0167] like Figure 2 The framework of the hybrid multi-attribute decision-making reputation rating model shown in the figure includes two steps: indicator weighting and attribute integration. An indicator weighting technique using game theory combined anti-entropy weight (AEW) and hierarchical weighting (LBWA) is innovatively proposed, and an attribute integration method suitable for multiple alternatives is proposed to evaluate the reputation level of alternatives through a comprehensive utility function.

[0168] like Figure 3 As shown, embodiment 1 of the present invention provides a distributed adjustable resource reputation evaluation method taking into account malicious user penalties, comprising the following steps:

[0169] Obtain indicators that affect the reputation of power market players and build a comprehensive rating indicator system;

[0170] Obtaining indicators that affect the reputation of power market entities includes: constructing a reputation effect path transmission model for the transaction behavior of distributed adjustable resource entities, as follows:

[0171] Determine the key influencing factor system and the direct grey relationship matrix. Invite different evaluation sources in related fields to construct the direct grey relationship matrix based on the questionnaire. (k=1,2,...,K represents the number of evaluation sources, n=13 represents the number of key factors), thus obtaining the judgment of the correlation between the influencing factors from each evaluation source. It represents the influence degree of key factor i on key factor j as considered by k evaluation sources, which is included in the direct grey coefficient matrix and has

[0172] The elements in the relationship matrix are gray numbers, which are obtained from the language scale of the evaluation source. Formulas (1)-(4) are the process of clarifying the gray numbers in the matrix, and then obtaining the real number direct influence matrix.

[0173]

[0174] In the formula, the gray number The lower and upper bounds of and Gray Number The clarified lower and upper bounds are expressed as and is the influence degree of key factor i on key factor j after grey system processing as considered by the kth evaluation source; is the normalized clarity value; is the range of the grey judgment matrix of the evaluation source k, and the calculation formula is:

[0175]

[0176] Calculate the real direct impact matrix and the clear direct grey relationship matrix of all evaluation sources Further directly affects the matrix X = [x ij ] n×n The calculation method is:

[0177]

[0178] Determine the comprehensive impact matrix of DEMATEL. Based on the real number direct impact matrix A = [a ij ] n×n :

[0179]

[0180] For matrix A, we have (O is a zero matrix), the comprehensive influence matrix T is calculated according to the following formula:

[0181]

[0182] Among them, I is the unit matrix.

[0183] The centrality M of each key factor is calculated according to the following formula i And the cause degree N i :

[0184] M i =f i +e i (i=1,2,...,n) (9)

[0185] N i =f i -e i (i=1,2,...,n) (10)

[0186] In the formula, f i and e i The key factors are C i The influence and influence of are calculated by the following formula:

[0187]

[0188] The larger the centrality, the more importance the system attaches to the indicator; if the causal degree is greater than 0, it means that the total impact of the key factor on it is greater than other key factors in the system, and it can be defined as a cause key factor; if the causal degree is less than 0, it means that the total impact of other key factors in the system on the key factor is greater than that of the key factor, and it can be defined as a result key factor.

[0189] Determine the reachability matrix of ISM. Using the comprehensive influence matrix T, the overall influence matrix H = [h ij ] n×n ,n=13 can be calculated as follows:

[0190] H=T+I (13)

[0191] The reachability matrix K = [k ij ] n×n , n = 13 is based on the overall influence matrix H and is determined according to the following formula:

[0192]

[0193] In the formula, λ is an exogenously given threshold. The larger λ is, the simpler the final system will be; conversely, the more complex the system will be. The node degree is defined as the key factor C in K i The sum of the values ​​of the elements in the row and column can be further used to determine the value of λ according to the node degree. The value of λ needs to make the node degree of each key factor moderate, so as to ensure that the relationship between each factor in the system is not too complex or simple.

[0194] Determine the level of each key factor. Let the set of influencing key factors be denoted as C = {C1, C2, ..., C 13}, then the reachable set R of each key factor can be i and the predecessor set A i It is expressed as:

[0195] R i ={C j |C j ∈C,r ij ≠0},i=1,2,...,13 (15)

[0196] A i ={C j |C j ∈C,r ji ≠0},i=1,2,...,13 (16)

[0197] For the key factors affecting C i Verify that the following holds true:

[0198] R i =Ri ∩A i (17)

[0199] If formula (16) holds true, then the key factor C i The priority is defined as the highest. Next, delete the i-th row and i-th column of the matrix K, and determine again whether there is a key factor in the matrix so that equation (17) holds. If so, the priority of the key factor is defined as the second level. Continue to remove the row and column where the key factor is located in the matrix K, and make the next judgment until all key factors are judged. Finally, determine the priority level of each key factor according to the order in which each key factor is removed.

[0200] Draw multi-level key factor association paths. According to the results of the key factor stratification and the relationship between the key factors reflected by the reachable matrix elements, make a multi-level key factor association path diagram, which includes all the key factors. Then study the relationship between the subject's transaction behavior and the path of reputation transmission and other factors.

[0201] Based on the constructed transmission model, identify the key indicators that affect the credibility of power market players and build a comprehensive rating indicator system;

[0202] A questionnaire survey was conducted to invite a group of evaluation sources to determine the conservative and optimistic values ​​of the importance of each indicator in the interval [0,10].

[0203] Conservative triangular fuzzy numbers of each index and optimistic triangular fuzzy numbers It can be calculated from the questionnaire results of the evaluation source. Among them, the lower limit and upper limit of the conservative value of indicator i for the evaluation source are and is the geometric mean of the conservative values ​​of indicator i from all evaluation sources except the lower limit and upper limit; similarly, the lower limit and upper limit of the optimistic value of indicator i from the evaluation sources are and It is the geometric mean of the optimistic values ​​of indicator i from all evaluation sources except the lower limit and upper limit.

[0204] Check the consistency of the evaluation source opinions and calculate the significant consistency value G of each indicator i .

[0205] like It means that the evaluation sources give completely consistent judgments on indicator i. At this time, the significant consistency value G of indicator i is i for:

[0206]

[0207] like Then it is necessary to further calculate the grayscale interval value of the indicator and If M i ≥Z i , at this time, the significant consistency value G of indicator i i for:

[0208]

[0209] If M i <Z i , which means that the evaluation sources give inconsistent judgment results for indicator i, and the significant consistency value G cannot be calculated. i Therefore, repeat the above steps until the significant consistency value G of all indicators is calculated. i .

[0210] Indicator screening according to significant consistency value G i This value not only shows the degree of recognition of the evaluation sources on the evaluation of indicator i, but also reflects the importance the evaluation sources attach to the evaluation of this index. i The more the indicator increases, the more important it is. i Based on the threshold, the indicators with important influence are screened. In this method, G i The threshold of is set to the average of the conservative value and the optimistic value, that is, (0+10) / 2=5. i If it is greater than the threshold, the indicator is retained; otherwise, it is deleted.

[0211] Based on the comprehensive rating index system, with the goal of minimizing the deviation between subjective weights and objective weights, the index system is weighted by combining the hierarchical weighting method and the anti-entropy weight method based on game theory to obtain the initial comprehensive weights.

[0212] Subjective weighting based on LBWA. The optimal indicator is determined based on the opinions of the evaluation sources, and the indicator set C = {C1,C1,...,C n The most important indicator among them is defined as the optimal indicator C B .

[0213] Indicator stratification. According to the importance of the indicator, the remaining indicators except the optimal indicator are divided into different levels, and the stratification is based on the following:

[0214] S1 layer: Among the remaining indicators, indicators with the same importance as the optimal indicator are allocated to this layer, or indicators with less than twice (not including twice) the importance of the optimal indicator are allocated to this layer.

[0215] S2 layer: Among the remaining indicators, the indicators that are two to three times (not including three times) less important than the optimal indicator are classified into this layer.

[0216] S K Layer: Among the remaining indicators, the indicators that are K times to K+1 times (excluding K+1 times) less important than the optimal indicator are classified into this layer.

[0217] The importance of the indicator can be preliminarily determined by the above indicator stratification. Assume that for any indicator C i , its importance can be expressed by S(C i ), then for each level S k For (k=1,2,...,K), we have:

[0218] S=S1∪S2∪…∪S K (20)

[0219] And for any p,q∈{1,2,…,K}, if p≠q, then S p ∩S q =φ.

[0220] The importance of indicators is judged in different levels. In the above example, we determine the optimal index of this layer. Then compare the importance of other indicators in the layer with this optimal indicator and record the results Indicates that in the k-level, the i-th k The importance of an indicator compared with the optimal indicator of this layer, that is, the greater the importance of the indicator, The smaller the value of is, the smaller the is defined as an integer. In the special case of k = 1, we have and r is a constant of the importance of the indicator determined by the stratification results, which is:

[0221] r=max{|S1|,|S2|,…,|S K |} (21)

[0222] In the formula, |S K | represents the set S K The number of indicators in .

[0223] Determine the LBWA elasticity coefficient r0. The constant r represents the maximum difference in the importance judgment of the indicator in the layer. According to the r value, the elasticity coefficient r0 is defined to satisfy r0>r, and usually r0=r+1.

[0224] Calculate the influence function of each indicator The formula is as follows:

[0225]

[0226] In the formula, Indicates the i-th level in the k-th level k indicators.

[0227] Calculate the indicator weights according to The weight of the optimal indicator can be expressed as:

[0228]

[0229] Furthermore, the weights of the remaining indicators are:

[0230]

[0231] At this point, the subjective weights of all indicators are obtained, denoted as w 2i .

[0232] Based on the objective weighting of AEW, entropy is a concept that measures the degree of disorder in the thermodynamic system that was later introduced into information theory. Assume that there are possible states in the system with a probability of m for each state, and the probability of each possible state appearing is p. j (j=1,2,...,m), then it can be expressed by entropy:

[0233]

[0234] Where 0≤p j ≤1 and there is ∑ j p j =1.

[0235] According to the AEW method, assuming a multi-attribute decision-making problem with m objects to be evaluated and n indicators, x ij (i=1,2,...,n,j=1,2,...m) to represent the index value, and X=[x ij ] n×m To represent the decision matrix. Therefore, the anti-entropy value of each indicator can be expressed as:

[0236]

[0237] In the formula, represents the dimensionless value of the indicator value. Furthermore, the AEW weight of the indicator w 1i It can be obtained by normalizing the anti-entropy value of the indicator:

[0238] w 1i =h i / ∑ i h i (27)

[0239] Assume that we get the weight vector W of n indicatorsl (l=1,2,…,L) can be calculated using L weighting methods, so the basic weight set W={W1,W2,…,W L} can be obtained, then define the combined weight W inte is a linear combination of L basic weights, namely:

[0240]

[0241] In the formula, is the distribution coefficient of the lth basic weight. It can be deduced from this that in order to find the best combination weight In this section, the idea of ​​game theory is used to optimize the basic weight redistribution coefficient in equation (28). The linear combination composed of L basic weights has infinite possibilities. The optimization goal is to minimize the deviation between the optimal combination weight and all basic weights, which can be expressed as:

[0242]

[0243] In the formula, is the decision variable, ||V||2 represents the second norm of vector V, and there is

[0244] Using the mature MATLAB commercial parser, the optimal value can be obtained The combined weights based on game theory can be expressed as follows:

[0245]

[0246] Specifically in this section, there are two basic weights to be combined: W1 is the objective weight based on AEW, and W2 is the subjective weight based on LBWA. The combined weight can effectively balance the importance of the indicators represented by the subjective and objective weights, and minimize the difference between the combined weight and the subjective and objective weights, ensuring that the combined weight accurately reflects the characteristics of the indicator itself and that the original data information of the indicator can be fully utilized.

[0247] Based on the initial comprehensive weight, the dynamic characteristics of the subject's reputation are introduced as the adaptive adjustment coefficient to adaptively adjust the subjective weight and the objective weight. Then, the final comprehensive weight is optimized with the goal of minimizing the deviation between the subjective weight and the objective weight after adaptive adjustment.

[0248] Establish a MARCOS weighted normalized decision matrix, assuming that the number of equipment options is m, the number of indicators is n, and the indicator value is x ij (i=1,2,…,n,j=1,2,…,m) multiple attribute decision making problem, the original decision matrix is ​​X=[x ij ] n×mConsidering that there may be negative indicators as benefit indicators (such as operating income growth rate, operating profit margin, etc.), this paper standardizes the original decision matrix in the following way:

[0249]

[0250] In the formula, if the benefit indicator x ij <0, then set x ij is a very small positive number, thus ensuring All are positive numbers.

[0251] For interval indicators, the optimal interval of the indicator [x ij,BL ,x ij,UP ] and the theoretical interval [x ij,min ,x ij,max ] should be determined first. Further, when x ij ∈[x ij,BL ,x ij,UP ], there is When x ij =x ij,min or x ij =x ij,max Sometimes, there are When x ij ∈[x ij,min ,x ij,BL ]∪[x ij,UP ,x ij,max ], calculated by linear interpolation Then, the original corresponding matrix can be standardized, thus obtaining a standardized decision matrix

[0252] Multiply the normalized decision matrix and the indicator weights to obtain the weighted normalized matrix R = [r ij ] n×m ,in, is the final weight.

[0253] Based on the final comprehensive weight, a comprehensive reputation evaluation model is constructed by combining fuzzy comprehensive evaluation method and utility function;

[0254] Based on the above weighted normalized decision matrix, the positive and negative ideal solutions of MARCOS are calculated:

[0255]

[0256] Compute the differences and quotients of alternatives with respect to positive and negative ideal solutions:

[0257]

[0258] In the formula, the difference between the index and the positive ideal solution and the negative ideal solution are expressed as and to represent; the quotients are expressed by and To express.

[0259] Geometric similarity grey relational degree:

[0260]

[0261] In the formula, the geometric similarity grey relational degree of alternative solution j relative to the positive and negative ideal solutions is expressed as and To express.

[0262] Grey relational degree of numerical similarity:

[0263]

[0264] In the formula, the numerical similarity grey relational degree of alternative solution j relative to the positive and negative ideal solutions is expressed as and To express.

[0265] Comprehensive grey relational degree:

[0266]

[0267] In the formula, the utility of alternative j relative to the positive and negative ideal solutions is expressed as and To express.

[0268] Utility function:

[0269]

[0270] In the formula, the utility function of alternative j relative to the positive and negative ideal solutions is expressed as and To express.

[0271] Comprehensive utility function:

[0272]

[0273] In the formula, f(γ j ) is the comprehensive utility function of alternative j. f(γ j ) is larger, indicating that the multi-attribute decision-making result of alternative j is better. * (γ j ) is the modified utility function value, which is in the interval [1 / e,1]. Based on this, the credit rating of multiple entities in the electricity market can be intuitively judged.

[0274] A comprehensive reputation evaluation model is constructed through empirical analysis and verification. By comparing the reputation level predicted by the reputation evaluation model with the actual performance, suggestions for electricity restriction penalties for malicious users are put forward.

[0275] Preferably, malicious users are defined as market entities whose reputation value is lower than the set threshold. They often randomly select data to report, which damages the power consumption range and destroys the stability of the model, resulting in the failure of electricity prices to converge and users to maximize their own utility, thus causing social welfare losses. The calculation formula for the power demand of malicious users is:

[0276]

[0277] represents the actual power consumption of malicious user e in period t, rand represents a random number between 0 and 1, and N mal Represents a collection of malicious users.

[0278] This method proposes to distinguish malicious users from the received electricity demand data and punish them with the minimum electricity consumption of all users, replace the false data provided by malicious users, and increase the punishment. Out of personal will and to avoid damage to their own utility, malicious users are restricted in their electricity consumption, and users are more inclined to report real electricity consumption data. This is conducive to preventing non-malicious users from turning into malicious users and maintaining the stability of the power system, because the minimum electricity consumption at this time can only meet the basic electricity needs of users, and cannot meet the normal or higher quality electricity needs of users. In time period t, the minimum electricity consumption of all users is set to the actual maximum electricity value that malicious users can consume as a penalty, that is:

[0279]

[0280] In the formula, The amount of electricity used after being identified as a malicious user.

[0281] Example 2 of the present invention provides a distributed adjustable resource reputation evaluation system taking into account the penalty of malicious users, including: an indicator acquisition and rating system construction module, an initial comprehensive weight assignment module, a dynamic weight adjustment module, a comprehensive reputation evaluation model construction module, and an empirical analysis and evaluation suggestion module;

[0282] The indicator acquisition and rating system construction module is used to obtain key indicators that affect the reputation of power market entities. By constructing a path transmission model of the effect of distributed adjustable resource entity transaction behavior on reputation, it describes how different key factors affect the reputation of market entities through transaction behavior, collects and identifies key indicators that affect the reputation of power market entities, and constructs a comprehensive rating indicator system.

[0283] The initial comprehensive weighting module is used to assign weights to each indicator in the comprehensive rating index system using a game theory combined hierarchical weighting method and an anti-entropy weighting method, with the goal of minimizing the deviation between subjective weights and objective weights, and calculating the initial comprehensive weights;

[0284] The dynamic weight adjustment module is used to introduce the dynamic characteristics of the subject's reputation as an adaptive adjustment coefficient, dynamically adjust the initial comprehensive weight, and optimize the final comprehensive weight;

[0285] The comprehensive reputation evaluation model construction module is used to combine the fuzzy comprehensive evaluation method and the utility function, build a comprehensive reputation evaluation model based on the comprehensive rating index system and comprehensive weights, and evaluate the reputation level of power market entities;

[0286] The empirical analysis and evaluation suggestion module is used to verify based on empirical analysis, compare the reputation level predicted by the reputation evaluation model with the actual performance, and make reputation evaluation suggestions.

[0287] Example 3 of the present invention provides a terminal, including a processor and a storage medium; the storage medium is used to store instructions; the processor is used to operate according to the instructions to execute the steps of a distributed adjustable resource reputation evaluation method that takes into account malicious user penalties.

[0288] Example 4 of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a distributed adjustable resource reputation evaluation method that takes into account penalties for malicious users.

[0289] The beneficial effects of the present invention include at least:

[0290] (1) The present invention introduces the dynamic characteristics of subject reputation as an adaptive adjustment coefficient, so that the weight allocation can respond to the changes in the reputation of power market subjects in real time. The dynamic adjustment capability overcomes the shortcomings of traditional static evaluation methods and is more in line with the actual situation of the power market.

[0291] (2) The present invention combines the hierarchical weighting method and the anti-entropy weight method through game theory, comprehensively considers the deviation of subjective and objective weights, and optimizes them. Compared with the traditional single weighting method, the present invention is more scientific and reasonable in weight allocation;

[0292] (3) The present invention provides a more comprehensive reputation evaluation result by combining the fuzzy comprehensive evaluation method and the utility function. Compared with the traditional evaluation method, the comprehensive evaluation model of the present invention can better handle complex and uncertain market information and reduce errors;

[0293] (4) The present invention verifies the reputation evaluation model through empirical analysis and proposes power consumption restriction and penalty recommendations for malicious users, so that the evaluation results have practical application value and improve the stability and efficiency of market operation.

[0294] The present invention may be a system, a method and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for causing a processor to implement various aspects of the present invention.

[0295] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A distributed adjustable resource reputation evaluation method taking into account malicious user penalties, characterized in that: The following steps are involved: Obtain indicators that affect the reputation of power market players and build a comprehensive rating indicator system; Based on the comprehensive rating index system, with the goal of minimizing the deviation between subjective weights and objective weights, the index system is weighted by combining the hierarchical weighting method and the anti-entropy weight method based on game theory to obtain the initial comprehensive weights. Based on the initial comprehensive weight, the dynamic characteristics of the subject's reputation are introduced as the adaptive adjustment coefficient to adaptively adjust the subjective weight and the objective weight. Then, the final comprehensive weight is optimized with the goal of minimizing the deviation between the subjective weight and the objective weight after adaptive adjustment. Based on the final comprehensive weight, combined with fuzzy comprehensive evaluation method and utility function, a comprehensive reputation evaluation model is constructed; verified through empirical analysis, a comprehensive reputation evaluation model is constructed, and reputation evaluation recommendations are put forward by comparing the reputation level predicted by the reputation evaluation model with the actual performance.

2. A distributed adjustable resource reputation evaluation method taking into account malicious user penalties according to claim 1, characterized in that: Obtain indicators that affect the reputation of power market entities and build a comprehensive rating indicator system, including: building a reputation path transmission model for the transaction behavior of distributed adjustable resource entities, as follows: Obtain various key factors that affect the credit of distributed adjustable resource subjects and perform standardization, calculate the direct grey correlation between the standardized key factors, and construct a direct grey relationship matrix; Based on the direct grey relationship matrix, the real direct influence matrix X is calculated as follows: ij ] n×n : In the formula, represents the influence degree of the kth key factor i on the key factor j after grey processing; K represents the reachability matrix; x ij It represents the average direct impact of all judged key factors i on key factor j; The real-numbered direct impact matrix is ​​used as the comprehensive impact measure between key factors; The direct impact of real numbering on matrix normalization is as follows: A=[a ij ] n×n In the formula, A represents the real number after normalization, which directly affects the matrix; a ij It represents the normalized direct impact value of key factor i on key factor j; It represents the total influence of key factor i on all key factors; Indicates the maximum total influence of all key factors; The comprehensive influence matrix T is calculated using the following relationship: The reputation effect path transmission model is as follows: M=f(T) Where I is the unit matrix; It means that it includes all direct effects and all indirect effects; for matrix A, O is a zero matrix.

3. A distributed adjustable resource reputation evaluation method taking into account malicious user penalties according to claim 2, characterized in that: The key factors that affect the reputation of the distributed adjustable resource subject are obtained and standardized, and the direct grey correlation between the standardized key factors is calculated to construct a direct grey relationship matrix, as follows: The relationship between the key influencing factors was evaluated using a gray number ranging from 0.1 to 0.9; The gray number is clarified by the following relationship and converted into a certain value: In the formula, and They represent the lower and upper bounds of the degree of influence of key factor i judged by the kth evaluation source on key factor j; and They represent the lower and upper bounds of the influence of key factor i on key factor j after clarification; represents the extreme value of the grey judgment matrix of the kth evaluation source, Used to represent the maximum difference in the judgment of the impact degree among all key factors; The clarified values ​​are standardized using the following relationship to make the evaluation of different key factors on the same scale: Where: represents the normalized clarity value of key factor i of the kth judgment with respect to key factor j; It indicates the influence degree of the key factor i of the kth judgment on the key factor j after gray processing; Based on the standardized clarity values, a direct grey relationship matrix is ​​constructed, and the elements in the matrix represent the degree of direct influence between the key factors; the direct grey relationship matrix is ​​as follows: Where: represents the direct influence of the ith key factor on the jth key factor; k = 1, 2, ..., K represents the number of different evaluation sources, and n = 13 represents the number of key factors; It represents the influence degree of key factor i on key factor j as considered by k evaluation sources, which is included in the direct grey relationship matrix and has 4. A distributed adjustable resource reputation evaluation method taking into account malicious user penalties according to claim 2, characterized in that: In the reputation effect path transmission model, based on the comprehensive influence matrix T, a deconstruction and interaction structure analysis model is established, and the reachable relationship matrix K is calculated using the following relationship: H=T+I H=[h ij ] n×n K=[k ij ] n×n ,n=13 In the formula, I represents the unit matrix; H represents the overall influence matrix; λ represents the exogenous threshold; h ij represents the key factor in the i-th row and j-th column of the overall impact matrix H; k ij Represents the key factor of the i-th row and j-th column in the reachability matrix K; Calculate the reachable set R of each key factor based on the reachable matrix i and the predecessor set A i , by verifying whether R i =R i ∩A i To identify key factors at different levels, gradually remove the identified key factors until all key factors are layered; Draw a multi-level key factor association path diagram based on the hierarchical division results and the reachability matrix to display all key factors and their interrelationships, so as to characterize the path of reputation transmission from multi-subject transaction behaviors; The reachable set R i and the predecessor set A i as follows: R i ={C j |C j ∈C,r ij ≠0},i=1,2,...,13;j=1,2,...,13 A i ={C j |C j ∈C,r ji ≠0},i=1,2,...,13;j=1,2,...,13 C={C1,C2,...,C 13 } Where: R i Indicates that from the key factor C i All the key factors to reach; A i Indicates that all key factors C i Key factors that have an impact; r ji Indicates the key factor C j For key factor C i impact.

5. A distributed adjustable resource reputation evaluation method taking into account malicious user penalties according to claim 1, characterized in that: Based on the constructed reputation effect path transmission model, the influence of each key factor on the reputation of the power market players is analyzed, and the key indicators that have a significant impact on the reputation of the players are identified; through investigation, the importance of each identified key indicator is evaluated within the range of [0,10], and the conservative and optimistic values ​​of the importance of each indicator are obtained; Based on the survey results, the conservative triangular fuzzy number of each key indicator is calculated. and optimistic triangular fuzzy numbers Calculate the significant consistency value C for each key indicator i To check the consistency of the evaluation sources: like It means that the evaluation sources have completely consistent judgments on indicator i, and the significant consistency value G of indicator i i as follows: like Then calculate the grayscale interval value of the indicator and If M i ≥Z i , the significant consistency value G of index i i as follows: If M i <Z i , indicating that the evaluation sources have inconsistent judgment results on indicator i, then repeat the above steps until the significant consistency value G of all indicators is calculated. i ; The indicators are screened and the comprehensive rating indicator system is constructed according to the significant consistency value as follows: Setting G i The threshold of is the average of the conservative and optimistic values; If G i If it is greater than the threshold, the indicator is retained; If G i If it is less than or equal to the threshold, the indicator is deleted.

6. A distributed adjustable resource reputation evaluation method taking into account malicious user penalties according to claim 5, characterized in that: The conservative triangular fuzzy number of each key indicator and optimistic triangular fuzzy numbers In the above example, the lower and upper limits of the conservative value of index i are and It represents the geometric mean of the conservative values ​​of index i except the lower limit and upper limit; The lower and upper limits of the optimistic value for indicator i are and It represents the geometric mean of the optimistic values ​​of indicator i except the lower limit and upper limit.

7. A distributed adjustable resource reputation evaluation method taking into account malicious user penalties according to claim 1, characterized in that: Based on the comprehensive rating index system, with the goal of minimizing the deviation between subjective weights and objective weights, the index system is weighted by combining the hierarchical weighting method and the anti-entropy weight method based on game theory to obtain the initial comprehensive weights, as follows: Based on the subjective weighting of linear weighted fuzzy assessment LBWA, the evaluation source opinions are selected from the indicator set C = {C1, C2, ..., C n } select the most important indicator and define it as the optimal indicator C B , and stratify the remaining indicators; Calculate the influence function of the indicators in each layer, use the elasticity coefficient r0 to determine the weight, and obtain the subjective weight distribution; Using the entropy concept in information theory, the anti-entropy value of each indicator is calculated and normalized to obtain the objective weight of each indicator; The indicator weight w is obtained by the anti-entropy weight AEW method using the following relationship 1i : r ij =x ij / ∑ j x ij w 1i =h i / ∑ i h i Where: h i Represents the anti-entropy value of the i-th indicator; r ji Indicates the key factor C in the overall impact matrix R j For key factor C i The impact of Weight w 1i is the anti-entropy value h i The normalized result of is used to indicate the relative importance of the i-th indicator among all indicators; The subjective and objective weights are combined through game theory methods to define a basic weight set, and the combined weights are optimized to minimize the deviation between the subjective and objective weights to obtain the initial comprehensive weights; The initial comprehensive weights based on game theory are as follows: Where: represents the initial comprehensive weight obtained based on game theory optimization; represents the optimal allocation coefficient of the lth basic weight; W l Represents the initial weight vector obtained by the lth weighting method.

8. A distributed adjustable resource reputation evaluation method taking into account malicious user penalties according to claim 7, characterized in that: Based on the subjective weighting of linear weighted fuzzy assessment LBWA, the evaluation source opinions are selected from the indicator set C = {C1, C2, ..., C n } select the most important indicator and define it as the optimal indicator C B , and divide the remaining indicators into layers; calculate the influence function of the indicators in each layer, use the elasticity coefficient r0 to determine the weight, and obtain the weight distribution of subjective weighting, as follows: The specific stratification rules for indicators are as follows: S1: The difference between the importance and the optimal indicator is within the set range, and the importance is no more than 2 times lower than the optimal indicator; S2: The importance is more than 2 times but no more than 3 times lower than the optimal indicator; S k Layer: indicators whose importance is more than 3 times lower than the optimal indicator and increase in multiples; The importance of determining indicators is as follows: Set for any indicator C i , the importance is expressed as S(C i ), all indicators are divided into different levels according to their importance S k (k=1,2,...,K), the set of each level is represented by S=S1∪S2∪…∪S K ; For any p,q∈{1,2,…,K}, if p≠q, then S p ∩S q =φ, indicating that each level does not overlap; p and q are the indexes of the level, used to indicate different levels; The importance within the comparison layer is as follows: At all levels In the above example, we determine the optimal index for the level Compare the importance of other indicators in the layer with the optimal indicator and record the results Indicates that in the k-level, the i-th k The importance of each indicator compared with the optimal indicator of this layer; The value of is defined as an integer in the interval [0, r]; when k = 1, and The constant r of the indicator importance determined by the stratification results is calculated using the following relationship: r=max{|S1|,|S2|,…,|S K |} In the formula, |S K | represents the set S k The number of indicators in ; The elasticity coefficient r0 is determined based on the hierarchical weighted assessment LBWA as follows: The constant r represents the maximum difference in the importance judgment of the indicator in the layer. According to the value of r, the elasticity coefficient r0 is defined to satisfy r0>r, and r0=r+1; The influence function of each indicator is calculated using the following relationship: In the formula, Indicates the i-th level in the k-th level k indicators; according to The indicator weight is calculated using the following relationship: In the formula, k represents the number of layers; N k Indicates the number of indicators at the kth level; The weights of the remaining indicators are as follows: In the formula, Indicates the i-th level in the k-th level k The weight of each indicator; The subjective weights of all levels and indicators are sorted into a set w 2i .

9. A distributed adjustable resource reputation evaluation method taking into account malicious user penalties according to claim 1, characterized in that: Based on the initial comprehensive weight, the dynamic characteristics of the subject's reputation are introduced as the adaptive adjustment coefficient to adaptively adjust the subjective weight and the objective weight. Then, the final comprehensive weight is optimized with the goal of minimizing the deviation between the subjective weight and the objective weight after adaptive adjustment. The details are as follows: A weighted normalized decision matrix based on the multi-attribute ranking MARCOS method of complex ratio comparison is established to reflect the dynamic reputation characteristics, and the number of alternatives is set to m, the number of indicators is n, and the indicator value is x. ij (i=1,2,…,n,j=1,2,…,m) multiple attribute decision making problem, the original decision matrix is ​​X=[x ij ] n×m ; Normalized decision matrix X * as follows: Standardize benefit-based indicators and cost-based indicators: In the formula, x ij Represents the indicator value in the original decision matrix; Indicates the normalized indicator value; It represents the maximum value of the i-th indicator among all the schemes; represents the minimum value of the ith indicator among all solutions; for interval indicators, determine the optimal interval of the indicator [x ij,BL ,x ij,UP ] and the theoretical interval [x ij,min ,x ij,max ], if the indicator value is within the optimal interval, the normalized value is 1; If it is at the boundary of the theoretical interval, it is 0; if it is between the two, the normalized value is calculated by linear interpolation; the standardized decision matrix is ​​obtained The normalized decision matrix X * Multiply it by the initial indicator weight and calculate the weighted normalized decision matrix R using the following relationship: R＝[r ij ] n×m In the formula, represents the final weight.

10. A distributed adjustable resource reputation evaluation method taking into account malicious user penalties according to claim 1 or 9, characterized in that: Based on the final comprehensive weight, combined with the fuzzy comprehensive evaluation method and utility function, a comprehensive reputation evaluation model is constructed as follows: Based on the weighted normalized decision matrix, the positive and negative ideal solutions of MARCOS are calculated as follows: In the formula, S + represents the set of positive ideal solutions, the optimal value of each indicator; S - represents the set of negative ideal solutions, the worst value of each indicator; It represents the maximum value of the i-th indicator among all the schemes; It represents the minimum value of the i-th indicator among all the schemes; The difference and quotient of the alternative solutions relative to the positive and negative ideal solutions are calculated using the following relationship: Where: It represents the difference between the value of the i-th indicator in the j-th solution and the positive ideal solution; It represents the difference between the value of the i-th indicator in the j-th solution and the negative ideal solution; represents the quotient of the value of the i-th indicator in the j-th solution and the positive ideal solution; represents the quotient of the value of the i-th indicator in the j-th solution and the negative ideal solution; The geometric and numerical similarity grey relational degree is calculated using the following relationship: Where: Indicates the jth solution relative to the positive ideal solution The geometric similarity grey relational degree of Indicates the jth solution relative to the negative ideal solution The geometric similarity grey relational degree of The numerical similarity grey relational degree is calculated using the following relationship: Where v Indicates the jth solution relative to the positive ideal solution The numerical similarity grey relational degree; Indicates the jth solution relative to the negative ideal solution The numerical similarity grey relational degree; The comprehensive grey relational degree is calculated using the following relationship: Where: Indicates the jth solution relative to the positive ideal solution The comprehensive grey relational degree of Indicates the jth solution relative to the negative ideal solution The comprehensive grey relational degree of The utility function is calculated using the following relationship: Where: Indicates the jth solution relative to the positive ideal solution The utility function of Indicates the jth solution relative to the negative ideal solution The utility function of The utility function is summarized as follows: Where: f * (γ j ) represents the comprehensive utility function value of the jth solution, which is used to comprehensively evaluate the pros and cons of the solutions.

11. A distributed adjustable resource reputation evaluation method taking into account malicious user penalties according to claim 1, characterized in that: Through empirical analysis and verification, a comprehensive reputation evaluation model is constructed. By comparing the reputation level predicted by the reputation evaluation model with the actual performance, suggestions for restricting power consumption and punishing malicious users are proposed, as follows: Use the constructed comprehensive reputation evaluation model to predict the reputation level of each market entity, and calculate the expected reputation value of each entity based on the input of the comprehensive reputation evaluation model; The reputation level predicted by the model is compared with the actual performance of market entities collected; market entities below the set threshold are defined as malicious users; the formula for the power demand of malicious users is as follows: Where: represents the actual power consumption of malicious user e in period t; represents the minimum power consumption of all users in period t; represents the maximum power consumption of all users in period t; rand represents a random number between 0 and 1; N mal represents the set of malicious users; Penalize malicious users for electricity demand, including: Set the power consumption of malicious users to the minimum power consumption of all users; Replace false data provided by malicious users and increase penalties; The power consumption of malicious users in period t is set as follows: Where: is the power consumption after being identified as a malicious user, indicating the minimum power consumption of all users in period t.

12. A distributed adjustable resource reputation evaluation system taking into account malicious user penalties, comprising: An indicator acquisition and rating system construction module, an initial comprehensive weight assignment module, a dynamic weight adjustment module, a comprehensive reputation evaluation model construction module, and an empirical analysis and evaluation suggestion module; running a distributed adjustable resource reputation evaluation method taking into account malicious user penalties according to any one of claims 1 to 11, characterized in that: The indicator acquisition and rating system construction module is used to obtain key indicators that affect the reputation of power market entities. By constructing a path transmission model of the effect of distributed adjustable resource entity transaction behavior on reputation, it describes how different key factors affect the reputation of market entities through transaction behavior, collects and identifies key indicators that affect the reputation of power market entities, and constructs a comprehensive rating indicator system. The initial comprehensive weighting module is used to assign weights to each indicator in the comprehensive rating index system using a game theory combined hierarchical weighting method and an anti-entropy weighting method, with the goal of minimizing the deviation between subjective weights and objective weights, and calculating the initial comprehensive weights; The dynamic weight adjustment module is used to introduce the dynamic characteristics of the subject's reputation as an adaptive adjustment coefficient, dynamically adjust the initial comprehensive weight, and optimize the final comprehensive weight; The comprehensive reputation evaluation model construction module is used to combine the fuzzy comprehensive evaluation method and the utility function, build a comprehensive reputation evaluation model based on the comprehensive rating index system and comprehensive weights, and evaluate the reputation level of power market entities; The empirical analysis and evaluation suggestion module is used to verify based on empirical analysis, compare the reputation level predicted by the reputation evaluation model with the actual performance, and make reputation evaluation suggestions.

13. A terminal comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is used to operate according to the instructions to execute the steps of a distributed adjustable resource reputation evaluation method taking into account malicious user penalties according to any one of claims 1-11.

14. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of a distributed adjustable resource reputation evaluation method taking into account malicious user penalties as claimed in any one of claims 1 to 11 are implemented.

Images