Market abuse identification method and system based on improved support vector machine, medium and processor

Abstract

The invention provides a market power abuse identification method based on an improved support vector machine. The method comprises the following steps: S1, constructing an index system for market power abuse identification; s2, screening the indexes by using an LASSO regression algorithm to obtain a final index system; s3, optimizing a parameter penalty coefficient c and a kernel parameter g of the SVM by using a particle swarm algorithm; s4, constructing an SVM recognition model by using the penalty coefficient c and the kernel parameter g; and S5, performing classification identification on the generator set market behaviors by using the SVM identification model to obtain an identification result. Real-time identification of the abused market power is realized by utilizing a more efficient solving method, the abused market power behavior of a power generation enterprise can be more scientifically and accurately judged, supervision of the power market is realized, a foundation is laid for finding and reducing the risk of the market power in real time, and development of credit supervision of a power market subject is further promoted.

Description

Technical Field

[0001] The present invention relates to the technical field of market power abuse identification, and in particular to a market power abuse identification method, system, medium and processor based on an improved support vector machine. Background Art

[0002] With the accelerated advancement of the spot market, accurate and real-time identification of market power abuse is a key task in the management of power market abuse. Traditional identification of market power abuse by power generation companies mainly relies on expert decision-making. For example, the Chinese invention patent application with application number CN202110404468.7 discloses a method for identifying market power abuse by power generation units. By establishing a set of market power abuse discrimination indicators, the indicators in the set are screened based on the random forest algorithm to determine the indicators used in the market power abuse discrimination model; a discrimination model for the risk of market power abuse by power generation units is established based on the isolation forest algorithm to identify the abuse of market power by power generation units and obtain a set of units with potential risk of market power abuse; an expert system is used to obtain a market power risk logic deduction model to track units with potential risk of market power abuse; and abnormal label samples are generated for units that are determined to be abusing market power and input into the data system. At the same time, the Chinese invention patent application with application number CN202311010550.7 discloses a method, device and computer equipment for assessing market power abuse, including obtaining the declaration data, clearing data and daily demand data of all power generation enterprises in the power market; constructing multiple dimensions of abuse of market power evaluation indicators based on the declaration data, clearing data and daily demand data of the power market of power generation enterprises in the power market; constructing a comprehensive evaluation index for identifying power generation enterprises' abuse of market power based on multiple dimensions of abuse of market power evaluation indicators; and evaluating whether power generation enterprises have abused market power based on the comprehensive evaluation indicators and outputting the evaluation results. However, there are not many effective automatic monitoring methods for the extremely unbalanced data sets and the gradually increasing amount of data in actual market transactions, resulting in the failure to timely and accurately discover market abuse, which increases the credit risk of power market transactions.

[0003] In view of this, a method, system, medium and processor for identifying market power abuse based on an improved support vector machine are needed. Summary of the invention

[0004] In view of the extremely unbalanced data sets and the gradually increasing data volume in market transactions in the prior art, there are not many effective automatic monitoring methods. The present invention provides a method, system, medium and processor for identifying market power abuse based on an improved support vector machine, which can not only realize automatic identification of extremely unbalanced data sets and the gradually increasing data volume in transactions, but also minimize the credit risk of power market transactions. The specific technical scheme is as follows:

[0005] A method for identifying market power abuse based on an improved support vector machine, comprising:

[0006] S1: Construct an indicator system for identifying market power abuse;

[0007] S2: Use LASSO regression algorithm to screen the indicators to obtain the final indicator system;

[0008] S3: Use particle swarm optimization to optimize the SVM parameter penalty coefficient c and kernel parameter g;

[0009] S4: Use the penalty coefficient c and kernel parameter g to build the SVM recognition model;

[0010] S5: Use the SVM recognition model to classify and identify the market behavior of the generator set to obtain the recognition result.

[0011] Furthermore, the indicator system includes structural indicators, behavioral indicators and impact indicators.

[0012] Furthermore, the use of the LASSO regression algorithm to screen the indicators to obtain the final indicator system includes the following steps:

[0013] Use L1 norm as constraint;

[0014] Construct a penalty function to make the coefficients of some indicators 0 during the regression process, thereby achieving indicator screening;

[0015] After eliminating the indicators with a result of 0, the VIF multicollinearity test was performed;

[0016] After testing, the final indicator system is obtained.

[0017] Furthermore, the calculation formula for the estimated value of the LASSO regression algorithm variable selection is as follows:

[0018]

[0019] In the above formula, n is the number of power generation enterprises; w is the number of indicators; x ij is the index value; β j is the regression parameter corresponding to the jth indicator, if β j If it is zero, the indicator j can be omitted; β0 is the initial regression parameter; y i The result of the abuse determination.

[0020] Furthermore, the use of the particle swarm algorithm to optimize the parameter penalty coefficient c and the kernel parameter g of the SVM includes the following steps:

[0021] S31: construct SVM model;

[0022] S32: Set the population size, maximum number of iterations, inertia weight, cognitive learning factor, social learning factor and fitness function of the particle swarm algorithm, and randomly generate the initial position and speed of the particles;

[0023] S33: Select the accuracy of SVM to evaluate the fitness of each particle and update the individual extreme value of the particle;

[0024] S34: Find the extreme value of the particle group and determine whether the termination condition is met. If so, proceed to the next step. Otherwise, update the speed and position of the particles, form a new population, and repeat step S33;

[0025] S35: After reaching the maximum number of iterations, the search ends and the penalty coefficient c and the kernel parameter g corresponding to the individual extreme value of the particle with the smallest fitness are output.

[0026] Furthermore, the use of the penalty coefficient c and the kernel parameter g to construct the SVM recognition model includes the following steps:

[0027] The support vector machine SVM recognition model is constructed as follows:

[0028]

[0029] y i (w T x i +b)…1-ξ i ;

[0030] ξ i …0;

[0031] w T x i +b=0;

[0032] According to the Lagrange duality principle, by changing w, ξ i , b find the partial derivative, and find its dual form, the SVM recognition model can be transformed into:

[0033]

[0034] c…λ i …0;

[0035] Get the optimal normal vector w of the optimal hyperplane * and the optimal intercept b * for:

[0036]

[0037] K(x i ,x j ) = exp(-g||x i -x j|| 2 );

[0038]

[0039] According to the optimal normal vector w * and the optimal intercept b * , the final SVM recognition model is constructed as follows:

[0040] f(x i )=sign(w * x i +b * );

[0041] Among them, K(x i ,x j ) is the kernel function; λ i , j are the Lagrangian operators corresponding to the i-th and j-th samples respectively; x i 、x j are the index values ​​corresponding to the i-th and j-th samples respectively; y i ,y j are the label values ​​corresponding to the i-th and j-th samples respectively; ξ i is the slack variable; c is the penalty coefficient, that is, the larger the c, the greater the penalty for misclassification; w and b are the normal vector and intercept of the hyperplane, respectively, w T is the transpose of w, and n is the number of power generation companies.

[0042] Furthermore, the method further comprises the following steps:

[0043] An example analysis is performed on the recognition result data to evaluate the recognition effect of the SVM recognition model.

[0044] A market power abuse identification system based on an improved support vector machine is applied to the above-mentioned market power abuse identification method based on an improved support vector machine, comprising:

[0045] A system building module, which is used to build an indicator system for identifying market power abuse;

[0046] A screening module is used to screen the indicators using the LASSO regression algorithm to obtain the final indicator system;

[0047] An optimization module, which is used to optimize the parameter penalty coefficient c and kernel parameter g of SVM using a particle swarm algorithm;

[0048] A model building module, which is used to build an SVM recognition model using a penalty coefficient c and a kernel parameter g;

[0049] The identification module is used to classify and identify the market behavior of the generator set using the SVM identification model to obtain the identification result.

[0050] A computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the above-mentioned market power abuse identification method based on improved support vector machine.

[0051] A processor is used to run a program, wherein the program executes the above-mentioned market power abuse identification method based on improved support vector machine when running.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] 1. This method combines a large amount of indicator data and uses a more efficient solution method to achieve real-time identification of abuse of market power. It can more scientifically and accurately determine the abuse of market power by power generation companies, realize the supervision of the power market, and lay the foundation for real-time discovery and reduction of market power risks, thereby promoting the development of credit supervision of power market players. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the description of the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.

[0055] Figure 1 A flowchart of a method for identifying market power abuse based on an improved support vector machine is shown below;

[0056] Figure 2 This is a schematic diagram of the structure of a market power abuse identification system based on an improved support vector machine. DETAILED DESCRIPTION

[0057] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0059] It should also be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0060] It should be further understood that the term "and / or" used in the present description and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0061] Embodiment 1

[0062] like Figure 1 The figure shows a flow chart of a method for identifying market power abuse based on an improved support vector machine, including:

[0063] S1: Construct an indicator system for identifying market power abuse.

[0064] First, combined with the actual situation of the power market, a market power abuse identification index system is constructed. Taking into account the internal factors of the trading behavior of power generators and the external factors of the market environment, this method constructs a generator market power abuse identification index set X from three aspects: structural indicators, behavioral indicators, and impact indicators. That is, the index system includes structural indicators, behavioral indicators, and impact indicators.

[0065] S2: Use the LASSO regression algorithm to screen the indicators to obtain the final indicator system.

[0066] In the specific implementation, the indicator data is reduced in dimension based on the characteristics of high-dimensional data in the power market. The LASSO regression algorithm is used to screen the indicators, which is a method to solve the problem of multicollinearity of indicators. The LASSO regression algorithm is used to screen the indicators to obtain the final indicator system, which includes the following steps:

[0067] Use L1 norm as constraint;

[0068] Construct a penalty function to make the coefficients of some indicators 0 during the regression process, thereby achieving indicator screening;

[0069] After eliminating the indicators with a result of 0, the VIF multicollinearity test was performed;

[0070] After testing, the final indicator system is obtained.

[0071] In a specific implementation, the calculation formula for the estimated value of the LASSO regression algorithm variable selection is as follows:

[0072]

[0073] In the above formula, n is the number of power generation enterprises; w is the number of indicators; x ij is the jth index value of the given i-th sample point; β j is the regression parameter corresponding to the jth indicator, if β j If it is zero, the indicator j can be omitted; β0 is the initial regression parameter; y i is the abuse determination result; i=1,2,...,n,j=1,2,...,w.

[0074] S3: Use particle swarm optimization to optimize the SVM parameter penalty coefficient c and kernel parameter g.

[0075] With the continuous expansion of market players and training sets, and the increase in the number of centralized bidding quotation segments, the increase in bidding data is inevitable, but SVM training is very time-consuming when the amount of data is large, and the time expenditure for parameter optimization is even greater. This method proposes an improved support vector machine SVM and uses the particle swarm optimization (PSO) algorithm to optimize the parameter penalty coefficient c and kernel parameter g of SVM.

[0076] In a specific implementation, the particle swarm algorithm is used to optimize the parameter penalty coefficient c and the kernel parameter g of the SVM, including the following steps:

[0077] S31: Data preparation and building SVM model;

[0078] S32: PSO parameter setting and particle initialization, setting the population size, maximum number of iterations, inertia weight, cognitive learning factor, social learning factor and fitness function of the particle swarm PSO algorithm, and randomly generating the initial position and speed of the particles;

[0079] S33: Fitness evaluation, select the accuracy of SVM to evaluate the fitness of each particle and update the individual extreme value of the particle;

[0080] S34: Find the extreme value of the particle group and determine whether the termination condition is met. If so, proceed to the next step. Otherwise, update the speed and position of the particles, form a new population, and repeat step S33;

[0081] S35: After reaching the maximum number of iterations, the search ends and the penalty coefficient c and the kernel parameter g corresponding to the individual extreme value of the particle with the smallest fitness are output.

[0082] Step S33 will evaluate the fitness, and step S34 is to determine the iterative termination condition. When step S34 determines that the termination condition is met, the penalty coefficient c and the kernel parameter g corresponding to the individual extreme value with the minimum fitness saved in step S33 will be used as the optimization result.

[0083] S4: Use the penalty coefficient c and kernel parameter g to build the SVM recognition model.

[0084] For linearly separable data, the principle of support vector machine in market power abuse identification is as follows:

[0085] Assume the given training set is:

[0086] {(x1,y1),(x2,y2),...,(x n ,y n )};

[0087] Where n is the number of power generation enterprises; y n ∈{1, -1} is the recognition result of the definition algorithm, 1 means no abuse, -1 means abuse; x n Provide quotation data for power generation companies.

[0088] Assume the optimal classification interface is:

[0089] w T x i +b=0;

[0090] Therefore, the use of penalty coefficient c and kernel parameter g to construct a linearly separable support vector machine (SVM) SVM recognition model includes the following steps:

[0091] The support vector machine SVM recognition model is constructed as follows:

[0092]

[0093] y i (w T x i +b)…1-ξ i ;

[0094] ξ i …0;

[0095] w T x i +b=0;

[0096] According to the Lagrange duality principle, by changing w, ξ i , b find the partial derivative, and find its dual form, the SVM recognition model can be transformed into:

[0097]

[0098] c…λ i …0;

[0099] Get the optimal normal vector w of the optimal hyperplane * and the optimal intercept b * for:

[0100]

[0101] K(x i ,x j ) = exp(-g||x i -x j || 2 );

[0102] This quadratic programming problem can be solved if the solution is Its optimal hyperplane is

[0103] According to the optimal normal vector w * and the optimal intercept b * , the final SVM recognition model is constructed as follows:

[0104] f(x i )=sign(w * x i +b * );

[0105] Among them, K(x i ,x j ) is the kernel function; λ i , j are the Lagrangian operators corresponding to the i-th and j-th samples respectively; x i 、x j are the index values ​​corresponding to the i-th and j-th samples, i=1,...,n; y i ,y j are the label values ​​corresponding to the i-th and j-th samples respectively; ξ i is the slack variable; c is the penalty coefficient, that is, the larger the c, the greater the penalty for misclassification; w and b are the normal vector and intercept of the hyperplane, respectively, w T is the transpose of w, and n is the number of power generation companies.

[0106] The SVM recognition model requires the input of the penalty coefficient c and the kernel parameter g. When using the SVM recognition model for classification, a suitable kernel function K(x i ,x j ) = exp(-g||x i -xj || 2 ) is crucial, and the kernel parameter g is an important parameter in the kernel function.

[0107] S5: Use the SVM recognition model to classify and identify the market behavior of the generator set to obtain the recognition result.

[0108] Furthermore, the method further comprises the following steps:

[0109] An example analysis is performed on the recognition result data to evaluate the recognition effect of the SVM recognition model.

[0110] The results of market power abuse identification can be divided into abuse and non-abuse. The raw data of abuse identification include abuse data n v and non-abuse of data n The results of the analysis can be divided into four categories: Abuse data is identified as abuse data vv , abused data is identified as non-abused data n vn , non-abused data is identified as non-abused data n nn , non-abused data is identified as abusive data n nv The algorithm is evaluated by accuracy and recall, and the calculation formula is as follows:

[0111]

[0112] The correct rate Z q It can reflect the overall accuracy level of the SVM recognition model recognition and classification; Z h The recall rate can reflect the recognition rate of abused data, while ignoring the accuracy rate of non-abused data. Abuse identification should find an algorithm with high accuracy rate under the premise that the abused data is fully identified.

[0113] The example analysis of this step uses the data set constructed by the power market simulation experiment for example analysis. The recognition results of other recognition methods and the PSO-SVM recognition method of this scheme are shown in Table 1. As can be seen from Table 1, PSO-SVM has the highest accuracy and accuracy.

[0114] Table 1 Support vector machine test accuracy and recall

[0115] Evaluation indicators PSO-SVM KNN Complex Tree SVM Accuracy 100％ 80％ 85％ 88％ Recall 100％ 76％ 80％ 80％

[0116] The identification method of this application combines a large amount of indicator data and uses a more efficient solution method to achieve real-time identification of abuse of market power. It can more scientifically and accurately determine the abuse of market power by power generation companies, achieve supervision of the power market, and lay the foundation for real-time discovery and reduction of market power risks, thereby promoting the development of credit supervision of power market entities.

[0117] Embodiment 2

[0118] like Figure 2 As shown, a market power abuse identification system based on an improved support vector machine is applied to the market power abuse identification method based on an improved support vector machine described above, comprising:

[0119] A system building module, which is used to build an indicator system for identifying market power abuse;

[0120] A screening module is used to screen the indicators using the LASSO regression algorithm to obtain the final indicator system;

[0121] An optimization module, which is used to optimize the parameter penalty coefficient c and kernel parameter g of SVM using a particle swarm algorithm;

[0122] A model building module, which is used to build an SVM recognition model using a penalty coefficient c and a kernel parameter g;

[0123] The identification module is used to classify and identify the market behavior of the generator set using the SVM identification model to obtain the identification result.

[0124] Embodiment 3

[0125] A computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the above-mentioned market power abuse identification method based on improved support vector machine.

[0126] Embodiment 4

[0127] A processor is used to run a program, wherein the program executes the above-mentioned market power abuse identification method based on improved support vector machine when running.

[0128] The present application provides a method for identifying market power abuse based on an improved support vector machine, including: S1: constructing an index system for identifying market power abuse; S2: using the LASSO regression algorithm to screen the indexes to obtain the final index system; S3: using the particle swarm algorithm to optimize the SVM parameter penalty coefficient c and kernel parameter g; S4: using the penalty coefficient c and kernel parameter g to construct an SVM identification model; S5: using the SVM identification model to classify and identify the market behavior of the generator set to obtain the identification result. Using a more efficient solution method to achieve real-time identification of abuse of market power can more scientifically and accurately determine the abuse of market power by power generation enterprises, realize the supervision of the power market, lay the foundation for real-time discovery and reduction of market power risks, and promote the development of credit supervision of power market entities.

[0129] Those of ordinary skill in the art will appreciate that the units of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition of each example has been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0130] In the embodiments provided by the present invention, it should be understood that the division of units is only a logical function division, and there may be other division methods in actual implementation, for example, multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored, etc.

[0131] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0132] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-0nlyMemory), random access memory (RAM, RandomAccessMemory), mobile hard disk, magnetic disk or optical disk, etc., which can store program code.

[0133] 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 aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. A method for identifying market power abuse based on an improved support vector machine, characterized in that: include: S1: Construct an indicator system for identifying market power abuse; S2: Use LASSO regression algorithm to screen the indicators to obtain the final indicator system; S3: Use particle swarm optimization to optimize the SVM parameter penalty coefficient c and kernel parameter g; S4: Use the penalty coefficient c and kernel parameter g to build the SVM recognition model; S5: Use the SVM recognition model to classify and identify the market behavior of the generator set to obtain the recognition result.

2. The method for identifying market power abuse based on improved support vector machine according to claim 1, characterized in that: The indicator system includes structural indicators, behavioral indicators and impact indicators.

3. The method for identifying market power abuse based on improved support vector machine according to claim 1, characterized in that: The method of using the LASSO regression algorithm to screen the indicators to obtain the final indicator system includes the following steps: Use L1 norm as constraint; Construct a penalty function to make the coefficients of some indicators 0 during the regression process, thereby achieving indicator screening; After eliminating the indicators with a result of 0, the VIF multicollinearity test was performed; After testing, the final indicator system is obtained.

4. The method for identifying market power abuse based on improved support vector machine according to claim 3, characterized in that: The calculation formula for the estimated value of the LASSO regression algorithm variable selection is as follows: In the above formula, n is the number of power generation enterprises; w is the number of indicators; x ij is the index value; β j is the regression parameter corresponding to the jth indicator, if β j If it is zero, the indicator j can be omitted; β0 is the initial regression parameter; y i The result of the abuse determination.

5. The method for identifying market power abuse based on improved support vector machine according to claim 1, characterized in that: The particle swarm algorithm is used to optimize the parameter penalty coefficient c and the kernel parameter g of the SVM, comprising the following steps: S31: construct SVM model; S32: Set the population size, maximum number of iterations, inertia weight, cognitive learning factor, social learning factor and fitness function of the particle swarm algorithm, and randomly generate the initial position and speed of the particles; S33: Select the accuracy of SVM to evaluate the fitness of each particle and update the individual extreme value of the particle; S34: Find the extreme value of the particle group and determine whether the termination condition is met. If so, proceed to the next step. Otherwise, update the speed and position of the particles, form a new population, and repeat step S33; S35: After reaching the maximum number of iterations, the search ends and the penalty coefficient c and the kernel parameter g corresponding to the individual extreme value of the particle with the smallest fitness are output.

6. The method for identifying market power abuse based on improved support vector machine according to claim 5, characterized in that: The method of using the penalty coefficient c and the kernel parameter g to construct the SVM recognition model includes the following steps: The support vector machine SVM recognition model is constructed as follows: According to the Lagrange duality principle, by changing w, ξ i , b find the partial derivative, and find its dual form, the SVM recognition model can be transformed into: Get the optimal normal vector w of the optimal hyperplane v and the optimal intercept b * for: According to the optimal normal vector w * and the optimal intercept b * , the final SVM recognition model is constructed as follows: f(x i )=sign(w * x i +b * ); Among them, K(x i ,x j ) is the kernel function; λ i , j are the Lagrangian operators corresponding to the i-th and j-th samples respectively; x i 、x j are the index values ​​corresponding to the i-th and j-th samples respectively; y i ,y j are the label values ​​corresponding to the i-th and j-th samples respectively; ξ i is the slack variable; c is the penalty coefficient, that is, the larger the c, the greater the penalty for misclassification; w and b are the normal vector and intercept of the hyperplane, respectively, w T is the transpose of w, and n is the number of power generation companies.

7. The method for identifying market power abuse based on improved support vector machine according to claim 1, characterized in that: The following steps are also included: An example analysis is performed on the recognition result data to evaluate the recognition effect of the SVM recognition model.

8. A market power abuse identification system based on improved support vector machine, characterized in that: The method for identifying market power abuse based on an improved support vector machine as applied to any one of claims 1 to 7 comprises: A system building module, which is used to build an indicator system for identifying market power abuse; A screening module is used to screen the indicators using the LASSO regression algorithm to obtain the final indicator system; An optimization module, which is used to optimize the parameter penalty coefficient c and kernel parameter g of SVM using a particle swarm algorithm; A model building module, which is used to build an SVM recognition model using a penalty coefficient c and a kernel parameter g; The identification module is used to classify and identify the market behavior of the generator set using the SVM identification model to obtain the identification result.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the market power abuse identification method based on the improved support vector machine as described in any one of claims 1 to 7.

10. A processor, characterized in that: The processor is used to run a program, wherein the program, when running, executes the market power abuse identification method based on an improved support vector machine as described in any one of claims 1 to 7.

Images