Multi-collaboration-degree fusion-based collusion identification method and system for electricity market subject pairs

Abstract

The invention relates to an electricity market subject pair collusion identification method and system based on multi-collaboration degree fusion, and the method comprises the steps: constructing a collaboration degree index system of an electricity market subject pair from the perspective of driving conditions, market behaviors and result influences for an electricity market collusion behavior pattern; power generation side-transaction sequence data and power sale side-transaction sequence data are collected, each index value in the coordination degree index system is calculated, all index values under each view angle are fused, and a coordination degree matrix of each power market subject pair under each view angle is calculated; based on a view angle dominating relation and Softmax self-adaptive weighting, the collaboration degree matrixes of all view angles are fused, and a comprehensive collaboration degree index of the electricity market subject pair is obtained; and performing collusion risk grade division on the power market subject to realize collusion identification. The method has the advantages of fully identifying the abnormal relationship between the power market subjects, realizing stable identification in a complex scene and the like.

Description

Technical Field

[0001] This invention relates to the field of power market risk identification, and in particular to a method and system for identifying collusion between power market participants based on multi-degree synergy fusion. Background Technology

[0002] To achieve effective market regulation, maintain fair market transactions, and ensure the healthy development of the electricity market, it is urgent to establish an efficient and accurate mechanism for identifying and warning of market collusion, providing scientific support for subsequent regulatory actions.

[0003] Chinese patent application publication number CN120087985A discloses a method for identifying abnormal bidding behavior of electricity market participants. Its core lies in constructing a feature set of abnormal bidding behavior and identifying the types of abnormal behavior by comparing it with market operation data. This scheme provides an effective tool for identifying overt abnormal bidding behavior by individual market participants. However, when dealing with highly concealed collusive market risks, it suffers from shortcomings such as a lack of analysis of the correlation between multiple participants and a single source of risk judgment, resulting in limited confidence in identifying complex collusive behavior.

[0004] Chinese patent application publication number CN119130598A discloses a scheme for identifying collusive bidding in the electricity spot market. Its core lies in using data mining and cluster analysis techniques to discover suspicious collusive groups from the declaration data of market participants. While this method deepens the understanding of behavioral patterns, its core analysis relies on the declaration data itself. The constructed analytical indicators only reflect the similarity of bidding behavior, and its ability to identify other collusive behaviors is limited. It struggles to reveal the underlying motivations for collusion and also suffers from weak interpretability of clustering results and sensitivity to parameter settings.

[0005] In summary, there is currently a lack of a method and system for identifying collusion among electricity market participants to solve or partially solve the aforementioned problems. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the existing technology by providing a collusion identification method and system for power market entities based on multi-degree synergy fusion, so as to solve or partially solve the problems of existing methods lacking correlation analysis between multiple entities and having a single source of risk judgment.

[0007] The objective of this invention can be achieved through the following technical solutions: One aspect of the present invention provides a method for identifying collusion between pairs of electricity market participants based on multi-degree-of-cooperation fusion, comprising the following steps: To address the collusion patterns in the electricity market, a coordination index system for electricity market participants is constructed from the perspectives of exercise conditions, market behavior, and the impact of outcomes. Collect transaction sequence data from the power generation side and the power sales side, calculate the values ​​of each indicator in the coordination index system, and fuse all indicator values ​​from each perspective based on the confidence entropy evidence theory to calculate the coordination matrix of each power market participant from each perspective. By fusing the synergy matrices of various perspectives based on perspective dominance and Softmax adaptive weighting, a comprehensive synergy index for power market participants is obtained. Based on the comprehensive coordination index, collusion risk levels are classified among electricity market participants to achieve collusion identification.

[0008] As a preferred technical solution, the process of calculating the coordination degree matrix of each electricity market participant from various perspectives includes the following steps: The indicator values ​​are standardized and homogenized, and fuzzification is achieved through evaluation level division. Membership functions are then established for each evaluation level. In the formula, , They represent the first n From the perspective of the first k The mean and standard deviation of each indicator at each assessment level; the reliability interval of the membership function is... The interval width Indicates the first n From the perspective of the first k The uncertainty of the reliability measurement of each indicator at each assessment level l Indicates the rating level; Considering the center and width of the confidence interval, calculate the confidence entropy of each indicator from different perspectives. : In the formula, K The total number of indicators from this perspective; Calculate the weights of each indicator from different perspectives based on confidence entropy. : ; The calculated index value belongs to the first l Confidence level of each assessment level m nk,l Construct the confidence assignment matrix M n = ( m nk,l ) K×5 ,in m nk,l satisfy ; Based on the confidence assignment matrix, an improved evidence theory algorithm using confidence entropy weights is employed to fuse all indicator values ​​from each perspective, yielding confidence fusion results for different levels of indicator values ​​across various perspectives. m n,l ; Based on the fusion results of confidence levels of indicator values ​​at different perspectives, fusion is performed at the level dimension to obtain the market participants' confidence in (…). i, j The degree of collusion and cooperation from this perspective forms a degree of cooperation matrix: .

[0009] As a preferred technical solution, the confidence fusion result is described above. m n,l The calculation process includes the following steps: Weights of each indicator from different perspectives Generate a 5-dimensional square matrix with all values ​​being 0. A With 5-dimensional column vectors Z These are used to store the initial confidence assignment matrix and the initial confidence, and to initialize the uncertainty factor, respectively. V =0; Traversal k =1,2,…, K Iterative updates A , Z and V : In the formula, M n Assign a matrix to the confidence level. M n (k,:) represents the confidence assignment matrix. M n The k The line, diag(), is a diagonal matrix. T For transpose, a 11 This represents the element in the first row and first column; Based on uncertainty factor Fusion confidence assignment matrix M n The confidence levels of the evaluation index values ​​at different levels from various perspectives were fused to obtain the results. m n,l : .

[0010] As a preferred technical solution, the process of obtaining the comprehensive coordination index of the electricity market participants includes the following steps: Calculate each pair of viewpoints n ,m ( n ≠ m The main body below ( i, j The probability of pairwise dominance ; Computational perspective n Average dominance The larger the value, the stronger the explanatory power and dominance of that perspective in the subject pair: In the formula, N Total number of viewpoints; The weights are obtained using Softmax normalization. ; Based on weight By weighting and synthesizing the synergy from the three perspectives, a comprehensive synergy index of collusion between the subjects is obtained. : .

[0011] As a preferred technical solution, the pairwise dominance probability Calculate using the following formula: in, Indicates market entities' ( i, j ) in the n Collusion and coordination degree values ​​from various perspectives The temperature coefficient used to control the smoothness of the dominance function, The standard deviation of the synergy among the three perspectives is represented by the coefficient. It is a regulating factor.

[0012] As a preferred technical solution, the weights Calculate using the following formula: in, >0 is an adjustment coefficient used to control the weight distribution and its sensitivity.

[0013] As a preferred technical solution, classifying the collusion risk level of electricity market participants includes the following steps: Based on expert experience and historical data analysis, a multi-level threshold range for the comprehensive synergy index is set, and a multi-level early warning system is constructed. The continuous synergy index is mapped to discrete early warning levels, forming a basic framework for hierarchical classification. Based on the calculated comprehensive coordination index of power market participants and the multi-level early warning system, the corresponding early warning level of each market participant is matched, and corresponding risk early warning signals are generated to classify and identify participants with different risk levels. Differentiated measures will be taken against electricity market participants based on the warning level.

[0014] As a preferred technical solution, the coordination index system includes the following indicators from the perspective of exercise conditions: total market share, cost structure similarity, adjustable capacity similarity, and historical transaction interaction; indicators from the perspective of market behavior: price curve similarity, marginal dominance alternation, price range overlap, volume reporting curve similarity, volume reporting peak-shifting complementarity, and sales volume coordination; and indicators from the perspective of outcome impact: main electricity price deviation, transaction volume deviation, main revenue deviation, and relative efficiency impact.

[0015] As a preferred technical solution, the behavioral patterns include similar pricing, alternating price increases, staggered pricing, similar volume reporting, off-peak volume reporting, and volume splitting.

[0016] Another aspect of the present invention provides a collusion identification system for power market participant pairs based on multi-degree of cooperation fusion, used to implement the aforementioned collusion identification method for power market participant pairs, the system comprising: The microservice-based data governance module, built on the Spring Cloud architecture, is used to connect to multi-source electricity market data and complete the collection, cleaning, storage and preprocessing of data such as market participant information, declaration data and clearing results, providing data services for upper-level analysis and calculation. The credibility entropy improved evidence theory fusion module is used to receive standardized indicator data provided by the data governance module, and use the improved evidence theory to fuse information on the evaluation values ​​of the three dimensions of exercise conditions, market behavior, and outcome impact, and calculate the subject's synergy index. Based on the perspective dominance relationship and Softmax adaptive weight calculation module, the perspective dominance score is calculated based on the pairwise dominance relationship between the synergy of each perspective. Then, the Softmax function with temperature coefficient is introduced to normalize the dominance score and obtain the adaptive weight of each perspective, so as to realize the dynamic adjustment of the weight according to the feature differences of different power market entities. The risk monitoring and handling application module is implemented based on the Spring Cloud framework. Based on the synergy index obtained by fusion computing, it completes risk level determination, real-time monitoring dashboard display and risk warning release, and generates corresponding market or handling suggestions to achieve closed-loop management of collusion risk throughout the entire process.

[0017] Compared with the prior art, the present invention has at least one of the following beneficial effects: (1) Fully identify abnormal relationships among electricity market participants: Based on the actual operation of the electricity market and specific collusion behavior patterns, this invention establishes a method for calculating the collusion coordination index of participants from three perspectives: conditions for collusion, participant behavior, and the impact of results. This method not only covers the entire chain of collusion strategies, but also, through hierarchical coordination calculation and integration, can characterize and identify abnormal collaborative relationships among market participants from multiple angles.

[0018] (2) Achieve stable identification in complex scenarios: This invention uses the DS evidence theory improved by confidence entropy to fuse multidimensional indicators within each perspective, effectively reducing the interference of conflicting evidence on the results and improving the discriminability of the evaluation results. This method has strong robustness to abnormal data and some missing indicators, and can stably identify potential collusion under complex market conditions.

[0019] (3) Dynamic evaluation is possible: This invention introduces perspective dominance relationship analysis and Softmax adaptive weighting method to integrate the synergy of different perspectives into the overall synergy, realizing two-layer adaptive fusion from the indicator level to the perspective level, forming a multi-layer fusion system that combines quantitative data at the lower level with difference-sensitive data at the upper level. In addition, the evaluation results can be continuously updated in conjunction with real-time market data to support dynamic risk warning and regulatory decision-making.

[0020] (4) Facilitates decision-making: This invention supports differentiated analysis of different entities, enabling refined risk management. It helps improve the accuracy and differentiation of identifying potential collusion among power market entities, and can provide scientific, quantitative, and explainable decision-making basis for market supervision and administrative handling, thereby improving regulatory efficiency and market fairness. Attached Figure Description

[0021] Figure 1 This is a flowchart of the collusion identification method for power market participants based on multi-degree fusion in the embodiment; Figure 2 This is a flowchart illustrating the collusion identification method in the embodiment; Figure 3 This is a schematic diagram of a collusion identification system for power market participants based on multi-degree fusion in an embodiment. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0023] Example 1 To address the problems of the aforementioned existing technologies, this embodiment provides a collusion identification method for power market participant pairs based on multi-cooperation degree fusion. By collecting power market transaction data, it first calculates a cooperation degree index system for participant pairs from three perspectives: exercise conditions, market behavior, and impact of outcomes. Then, based on the improved DS evidence theory of confidence entropy, it fuses the index systems of participant pairs from each perspective to generate a cooperation degree matrix for the participant pairs. Finally, it performs weighted integration of the cooperation degrees from the three perspectives to calculate the cooperation degree index of the power market participant pairs, and issues differentiated collusion warning signals based on the index levels.

[0024] In this embodiment, an improved DS evidence theory based on confidence entropy is used to identify multi-degree collusion among pairs of electricity market participants. First, taking pairs of electricity market participants as the evaluation object, and combining their transaction forms and technical characteristics, six typical collusion behavior patterns are summarized from two aspects: bidding behavior and quantity reporting behavior. Then, focusing on the complete chain of collusion behavior from its formation to its impact, an indicator system suitable for identifying the risk of collusion between participants is constructed, including 3 primary indicators and 14 secondary indicators.

[0025] Among them, the indicator set from the perspective of exercise conditions focuses on external conditions such as market structure, unit characteristics, and supply and demand patterns to assess whether there is an objective basis for collusion among the entities; the indicator set from the perspective of market behavior focuses on the consistency of bidding and volume reporting strategies to identify whether there are obvious collusion or synchronization characteristics; and the indicator set from the perspective of outcome impact assesses whether collusion actually affects the market price mechanism or resource allocation from the dimensions of clearing price, transaction volume, and revenue distribution, providing verification support for the identification conclusions.

[0026] Secondly, based on the aforementioned three-perspective indicator system, an improved DS evidence theory based on confidence entropy is used to fuse multidimensional indicators within each perspective, thereby generating a corresponding perspective synergy index. This method, by introducing a confidence entropy mechanism, effectively reduces the interference of conflicting evidence on the fusion results, improving the model's ability to identify abnormal behavior and the discriminative power of the evaluation results.

[0027] Subsequently, a first-level synergy fusion method based on perspective dominance and Softmax adaptive weighting was adopted to integrate the synergy indices from three perspectives—exercise conditions, market behavior, and outcome impact—to obtain the comprehensive collusion synergy degree of the subject pair. When this value exceeds the threshold set by experts or the historical statistical threshold, the subject pair is judged to have a high suspicion of collusion, thus achieving quantitative identification and interpretive judgment of potential collusion behavior.

[0028] Ultimately, based on the comprehensive coordination assessment results, corresponding risk warnings will be issued, and corresponding market disposal or administrative management measures will be implemented to achieve intelligent identification, early warning, and intervention of collusion by new types of electricity market participants.

[0029] See Figure 1 and Figure 2 This method includes steps S1-S4. Step S1: Based on different collusion behavior patterns in the electricity market, construct a coordination index system for the subject pair from three perspectives: exercise conditions, market behavior, and impact of results. Collect data from the generation and sales sides of a certain transaction sequence, calculate the values ​​of each index, and standardize them.

[0030] For details, see Figure 2 Step S1 may include steps S11-S14: Step S11: Define the assessment object and construct the collusion pattern library. Using the electricity market entities (such as two power generation companies, or a power generation company and a power sales company) as the basic assessment unit, systematically analyze their transaction forms and technical characteristics. From the two aspects of bidding behavior and quantity reporting behavior, summarize and extract six typical collusion behavior patterns to form a collusion pattern library, as shown in Table 1, which serves as the behavioral basis for the construction of the indicator system.

[0031] Table 1. Typical Collusion Patterns Among Electricity Market Entities Step S12 establishes the top-level framework of the three-tiered indicator system. Based on the complete chain of collusion from its formation to its impact, a top-level framework is established, consisting of three primary indicators: "conditions for exercise," "market behavior," and "impact of the outcome." This framework corresponds to the "probability," "consistency," and "effectiveness" of collusion, aiming to comprehensively assess the risks of collusion.

[0032] Step S13: The system refines the secondary indicators under each dimension. Based on the aforementioned three-dimensional framework, the system constructs a specific evaluation set containing 14 secondary indicators: For the primary indicator of "exercise conditions perspective," the focus is on assessing whether the entities possess the objective basis for implementing coordinated behavior; its secondary indicators mainly originate from external data such as market structure, unit characteristics, and supply and demand patterns. For the primary indicator of "market behavior perspective," the focus is on detecting whether the entities exhibit coordinated or synchronized characteristics in their bidding and quantity reporting strategies; its secondary indicators are obtained through statistical analysis of historical and real-time reporting data. For the primary indicator of "outcome impact perspective," the focus is on assessing whether suspicious coordinated behavior has actually changed the market clearing outcome and resource allocation efficiency. Its secondary indicators are extracted from market clearing data. Thus, a complete and clearly hierarchical 3-level, 14-indicator system is formed, as shown in Table 2, providing a comprehensive basis for subsequent evidence fusion and coordination degree calculation.

[0033] Table 2. Multi-perspective indicator library for collusion identification by electricity market participants. Step S14: Collect transaction data for a certain transaction sequence between the power generation side and the power sales side, and calculate the indicators of each indicator set in a03). Let the first... n ( n The index values ​​under the perspectives of (1, 2, 3) are respectively c nk After standardizing and homogenizing the indicators, we obtain .

[0034] Step S2: Using the confidence entropy evidence theory, all synergy index values ​​under each perspective are fused to calculate the synergy matrix of all subject pairs under the three perspectives.

[0035] For details, see Figure 2 Step S2 may include steps S21-S26: Step S21, the index value from step S14 Fuzzy processing is performed to divide the index calculation results into five levels: I, II, III, IV, and V, and a level is established. l The membership function for each evaluation level is: The confidence interval of the membership function is ,in, , They represent the first n From the perspective of the first k The mean and standard deviation of each indicator at each evaluation level, and the interval width. This indicates the uncertainty in the reliability measurement of different indicators at each assessment level; and They represent the indicators respectively c nk In the l The upper and lower limits of trust level estimation under each rating level.

[0036] Step S22: Considering the center and width of the confidence interval, calculate the indicators from different perspectives. Improved reliability entropy: In the formula, K The total number of indicators from this perspective. H nk,l It reflects the uncertainty and reliability of the evidence. The smaller the value, the lower the uncertainty, and the higher the weight can be assigned when assigning weights.

[0037] Step S23: Use the improved confidence entropy to obtain the weights of each indicator from different perspectives: Step S24, calculate Belongs to the l Confidence level of each assessment level m nk,l Construct the confidence assignment matrix M n = ( m nk,l ) K×5 ,in m nk,l satisfy .

[0038] Step S25: Using an improved evidence theory algorithm based on confidence entropy weights, the evidence is analyzed... Evidence fusion is performed. Specifically, step S25 may include S251-S253: Step S251: Generate a 5-dimensional square matrix with all values ​​being 0. A With 5-dimensional column vectors Z These are used to store the initial confidence assignment matrix and the initial confidence, respectively. Z=w n1 M (1,:) T Simultaneously initialize uncertainty factors. V =0.

[0039] Step S252: Traverse k =1,2,…, K Iterative updates A , Z and V : Step S253: Based on the uncertainty factor in step S252 Fusion confidence assignment matrix M n The confidence levels of the evaluation index values ​​at different levels from various perspectives were fused to obtain the results. m n,l : Step S26, set the scores for the five evaluation levels as follows: R =[0.2, 0.4, 0.6, 0.8, 1.0], then the market entities' ( i, j The degree of collusion and cooperation from this perspective is: Step S3: Using a method based on perspective dominance and Softmax adaptive weighting, the synergy matrix of the three perspectives is fused to obtain the comprehensive synergy index of the subject pair.

[0040] For details, see Figure 2 Step S3 may include steps S31-S34: Step S31: To characterize the differences in the coordination strength of each viewpoint and quantify the relative dominance of each viewpoint, calculate the relative dominance of each pair of viewpoints. n , m ( n ≠ m Paired dominance probability : In the formula Indicates perspective n right m The strength of dominance, The temperature coefficient used to control the smoothness of the dominance function can be adaptively adjusted. in The standard deviation of the synergy among the three perspectives is used to characterize its overall dispersion; the coefficients As an adjustment factor, when the difference between viewpoints is large (std value increases). Follow Increasing the weight of the dominant perspective can enhance its sensitivity to differences in cooperability, thus giving it higher weight; when the difference in perspectives is small, Choosing a low to medium value can avoid excessive weight bias towards a single perspective, thereby improving the stability and robustness of the fusion result. To balance perspective discrimination and weight stability, a value of [value missing] can be selected. =0.6.

[0041] Step S32, calculate the viewpoint n Average dominance The larger the value, the stronger the explanatory power and dominance of that perspective in the subject pair. N =3 represents the total number of viewpoints: Step S33: Obtain the weights using Softmax normalization. ,in, > 0 is an adjustment coefficient used to control the weight distribution and its sensitivity. Larger values ​​can enhance Softmax's sensitivity to viewpoint differences, giving significantly higher weights to views with higher coherence. When the value is small, weight differentiation can be suppressed, allowing information from multiple perspectives to participate in the fusion in a balanced manner. =6 can achieve a better balance between differentiation and stability, and is suitable for scenarios involving the identification of collaborative behavior in the power market: Step S34: Based on the weights determined in step S33, the synergy of the three perspectives is weighted and synthesized to further obtain the collusion synergy index of the subject pair. for: Step S4: Based on the level of the comprehensive coordination index, issue corresponding risk warnings to market entities and take matching market-based disposal and administrative management measures.

[0042] For details, see Figure 2 Step S4 may include steps S41-S43: Step S41: Based on expert experience and historical data analysis, set multi-level threshold ranges for the comprehensive synergy index, construct a multi-level early warning system of "low risk - medium risk - high risk", map the continuous synergy index into discrete early warning levels, and form a basic framework for hierarchical classification. Step S42: Based on the comprehensive synergy index calculated in step S3, automatically determine the corresponding early warning level of the market entity and generate the corresponding risk early warning signal to realize the classification and identification of entities with different risk levels. Step S43: Activate differentiated handling mechanisms based on the warning level: For low- and medium-risk entities, implement market-based measures, such as restricting contract transfers and suspending some trading qualifications; for high-risk entities, strengthen supervision and administrative intervention, including mandatory delisting, initiating guaranteed power supply services, regulatory interviews, and administrative fines, in order to maintain market order.

[0043] Example 2 Based on Example 1, this example provides a collusion identification system for power market participant pairs based on multi-degree of cooperation fusion, used to implement the collusion identification method for power market participant pairs in Example 1. See [link to example]. Figure 3 The system is based on the Spring Cloud microservice architecture and includes modules such as data governance, confidence entropy evidence theory, and Softmax adaptive weighted calculation to develop a collusion analysis and evaluation system for new types of power market participants, enabling risk monitoring and handling of collusion risks among new types of power market participants.

[0044] Specifically, the system includes: (1) Microservice-based data governance module.

[0045] Based on the Spring Cloud architecture, a data governance microservice was designed and implemented. This module is responsible for connecting to multi-source electricity market data, collecting, cleaning, storing, and preprocessing data such as market participant information, declaration data, and clearing results, providing high-quality, standardized data services for upper-level analysis and computation. (2) Credibility entropy improves the evidence theory fusion module.

[0046] Implementing a Confidence Entropy-Improved Evidence Theory Fusion Module: This module develops a core computational microservice based on the improved DS evidence theory using confidence entropy. It receives standardized indicator data from the data governance module and uses the improved evidence theory to fuse information from the evaluation values ​​of the three dimensions of "exercise conditions, market behavior, and outcome impact," calculating a precise subject-to-stake synergy index. (3) Based on the viewpoint dominance relationship and Softmax adaptive weight calculation module.

[0047] This module utilizes Python to encapsulate a weight calculation microservice, integrating perspective dominance analysis and a Softmax adaptive weighting mechanism. First, it calculates perspective dominance scores based on the pairwise dominance relationships between different perspectives. Then, it introduces a Softmax function with a temperature coefficient to normalize the dominance scores, obtaining adaptive weights for each perspective. This allows the weights to be dynamically adjusted according to the differences in characteristics among different electricity market participants, providing a basis for synergy fusion calculation.

[0048] (4) Risk monitoring and handling application module.

[0049] Within the Spring Cloud framework, the aforementioned microservices are integrated into a unified analysis and evaluation system. Based on the synergy index obtained through fusion computing, this module automatically completes risk level determination, real-time monitoring dashboard display, risk warning issuance, and generates corresponding market or administrative action suggestions, achieving closed-loop management of collusion risks throughout the entire process.

[0050] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for identifying collusion among pairs of electricity market participants based on multi-degree-of-cooperation fusion, characterized in that, Includes the following steps: To address the collusion patterns in the electricity market, a coordination index system for electricity market participants is constructed from the perspectives of exercise conditions, market behavior, and the impact of outcomes. Collect transaction sequence data from the power generation side and the power sales side, calculate the values ​​of each indicator in the coordination index system, and fuse all indicator values ​​from each perspective based on the confidence entropy evidence theory to calculate the coordination matrix of each power market participant from each perspective. By fusing the synergy matrices of various perspectives based on perspective dominance and Softmax adaptive weighting, a comprehensive synergy index for power market participants is obtained. Based on the comprehensive coordination index, collusion risk levels are classified among electricity market participants to achieve collusion identification.

2. The collusion identification method for power market participants based on multi-degree fusion according to claim 1, characterized in that, The process of calculating the coordination matrix for each electricity market participant from various perspectives includes the following steps: The indicator values ​​are standardized and homogenized, and fuzzification is achieved through evaluation level division. Membership functions are then established for each evaluation level. In the formula, , They represent the first n From the perspective of the first k The mean and standard deviation of each indicator at each assessment level; the reliability interval of the membership function is... The interval width Indicates the first n From the perspective of the first k The uncertainty of the reliability measurement of each indicator at each assessment level l Indicates the rating level; Considering the center and width of the confidence interval, calculate the confidence entropy of each indicator from different perspectives. : In the formula, K The total number of indicators from this perspective; Calculate the weights of each indicator from different perspectives based on confidence entropy. : ； The calculated index value belongs to the first l Confidence level of each assessment level m nk,l Construct the confidence assignment matrix M n = ( m nk,l ) K×5 ,in m nk,l satisfy ; Based on the confidence assignment matrix, an improved evidence theory algorithm using confidence entropy weights is employed to fuse all indicator values ​​from each perspective, yielding confidence fusion results for different levels of indicator values ​​across various perspectives. m n,l ; Based on the fusion results of confidence levels of indicator values ​​at different perspectives, fusion is performed at the level dimension to obtain the market participants' confidence in (…). i,j The degree of collusion and cooperation from this perspective forms a degree of cooperation matrix: 。 3. The collusion identification method for power market participants based on multi-degree fusion according to claim 2, characterized in that, The confidence fusion result m n,l The calculation process includes the following steps: Weights of each indicator from different perspectives Generate a 5-dimensional square matrix with all values ​​being 0. A With 5-dimensional column vectors Z These are used to store the initial confidence assignment matrix and the initial confidence, and to initialize the uncertainty factor, respectively. V =0; Traversal k =1,2,…, K Iterative updates A , Z and V : In the formula, M n Assign a matrix to the confidence level. M n (k,:) represents the confidence assignment matrix. M n The k The line, diag(), is a diagonal matrix. T For transpose, a 11 This represents the element in the first row and first column; Based on uncertainty factor Fusion confidence assignment matrix M n The confidence levels of the evaluation index values ​​at different levels from various perspectives were fused to obtain the results. m n,l : 。 4. The collusion identification method for power market participants based on multi-degree fusion according to claim 1, characterized in that, The process of obtaining the comprehensive coordination index of the electricity market participants includes the following steps: Calculate each pair of viewpoints n , m ( n ≠ m The main body below ( i,j The probability of pairwise dominance ; Computational perspective n Average dominance The larger the value, the stronger the explanatory power and dominance of that perspective in the subject pair: In the formula, N Total number of viewpoints; The weights are obtained using Softmax normalization. ; Based on weight By weighting and synthesizing the synergy from the three perspectives, a comprehensive synergy index of collusion between the subjects is obtained. : 。 5. The collusion identification method for power market participants based on multi-degree fusion according to claim 4, characterized in that, The pairwise dominance probability Calculate using the following formula: in, Indicates market entities' ( i,j ) in the n Collusion and coordination degree values ​​from various perspectives The temperature coefficient used to control the smoothness of the dominance function, The standard deviation of the synergy among the three perspectives is represented by the coefficient. It is a regulating factor.

6. The collusion identification method for power market participants based on multi-degree fusion according to claim 4, characterized in that, The weights mentioned Calculate using the following formula: in, >0 is an adjustment coefficient used to control the weight distribution and its sensitivity.

7. The collusion identification method for power market participants based on multi-degree fusion according to claim 1, characterized in that, The steps for classifying the collusion risk level of electricity market participants are as follows: Based on expert experience and historical data analysis, a multi-level threshold range for the comprehensive synergy index is set, and a multi-level early warning system is constructed. The continuous synergy index is mapped to discrete early warning levels, forming a basic framework for hierarchical classification. Based on the calculated comprehensive coordination index of power market participants and the multi-level early warning system, the corresponding early warning level of each market participant is matched, and corresponding risk early warning signals are generated to classify and identify participants with different risk levels. Differentiated measures will be taken against electricity market participants based on the warning level.

8. The collusion identification method for power market participants based on multi-degree fusion according to claim 1, characterized in that, In the aforementioned coordination index system, the indicators from the perspective of exercise conditions include total market share, cost structure similarity, adjustable capacity similarity, and historical transaction interaction; the indicators from the perspective of market behavior include price curve similarity, marginal dominance alternation, price range overlap, volume reporting curve similarity, volume reporting peak-shifting complementarity, and sales volume coordination; and the indicators from the perspective of outcome impact include main electricity price deviation, transaction volume deviation, main revenue deviation, and relative efficiency impact.

9. The collusion identification method for power market participants based on multi-degree fusion according to claim 1, characterized in that, The behavioral patterns mentioned include similar bids, alternating price increases, staggered bids, similar volume reporting, off-peak volume reporting, and split volume reporting.

10. A collusion identification system for electricity market participants based on multi-degree-of-cooperation fusion, characterized in that, For implementing the collusion identification method for electricity market participants as described in any one of claims 1-9, the system comprises: The microservice-based data governance module, built on the Spring Cloud architecture, is used to connect to multi-source electricity market data and complete the collection, cleaning, storage and preprocessing of data such as market participant information, declaration data and clearing results, providing data services for upper-level analysis and calculation. The credibility entropy improved evidence theory fusion module is used to receive standardized indicator data provided by the data governance module, and use the improved evidence theory to fuse information on the evaluation values ​​of the three dimensions of exercise conditions, market behavior, and outcome impact, and calculate the subject's synergy index. Based on the perspective dominance relationship and Softmax adaptive weight calculation module, the perspective dominance score is calculated based on the pairwise dominance relationship between the synergy of each perspective. Then, the Softmax function with temperature coefficient is introduced to normalize the dominance score and obtain the adaptive weight of each perspective, so as to realize the dynamic adjustment of the weight according to the feature differences of different power market entities. The risk monitoring and handling application module is implemented based on the Spring Cloud framework. Based on the synergy index obtained by fusion computing, it completes risk level determination, real-time monitoring dashboard display and risk warning release, and generates corresponding market or handling suggestions to achieve closed-loop management of collusion risk throughout the entire process.

Images