Power market manipulation behavior identification method based on multi-source heterogeneous data fusion

Abstract

The invention discloses a power market manipulation behavior identification method based on multi-source heterogeneous data fusion, and relates to the technical field of market transaction risk prevention and control, and the method comprises the steps: collecting multi-source heterogeneous market operation data of all power generation enterprises in a power market; based on the multi-source heterogeneous market operation data, outputting a weighted average quotation, a retention ratio, a high quotation ratio and a dynamic market share of a corresponding unit, and performing normalization processing; and analyzing the transaction data, judging the market manipulation behavior type of the power generation enterprise corresponding to the unit with the abnormal data, and performing warning according to the determined market manipulation behavior type. According to the method, the related power generation enterprises are warned according to the determined market manipulation behavior type, and the purpose of identifying and correcting the market manipulation behavior is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of market transaction risk prevention and control, and specifically to a method for identifying power market manipulation behaviors based on multi-source heterogeneous data fusion. Background Art

[0002] With the increasing complexity of the power market and the advancement of market-oriented reforms, market manipulation behaviors have posed challenges to the fairness, efficiency, and stability of the power market. In order to effectively supervise market behaviors and maintain market order, there is an urgent need for a method and technology that can accurately identify various manipulation behaviors in the power market.

[0003] For the above-mentioned artificially created transaction risks, a relatively prominent feature is that some abnormal phenomena will occur during the bidding process, which provides a valuable starting point for identifying the improper transaction behaviors of power generation enterprises. And how to reasonably screen and process the huge amount of original transaction data in the power market, and through what calculation methods to process these data, so as to obtain more comprehensive and accurate discrimination indicators to accurately identify these market manipulation behaviors has become one of the main difficulties in the effective supervision of the power market.

[0004] However, the current research work on the supervision of power market manipulation behaviors is still relatively scarce, and there are still problems such as insufficient research depth and the urgent need to improve relevant theories and methods. At present, it is still impossible to accurately prevent, identify, and correct these artificially created transaction risks, and naturally it is also difficult to restrict the behavior of power generation enterprises behind them from obtaining improper benefits. This poses challenges to maintaining fair competition and market efficiency in the power market and ensuring the sustainable and healthy development of the power market.

[0005] In summary, the existing supervision and identification methods for the market manipulation behaviors of power generation enterprises still cannot fully solve the problem that some power generation enterprises carry out unfair competition through abnormal bidding to obtain excessive profits and disrupt the fair competition order of the power market. Summary of the Invention

[0006] In view of the above existing problems, the present invention is proposed.

[0007] Therefore, the technical problem solved by the present invention is that the existing supervision and identification methods for the market manipulation behaviors of power generation enterprises have the problem that they cannot accurately prevent, identify, and correct these artificially created transaction risks, and naturally it is also difficult to restrict the behavior of power generation enterprises behind them from obtaining improper benefits.

[0008] To solve the above technical problems, the present invention provides the following technical solutions: A method for identifying power market manipulation behaviors based on multi-source heterogeneous data fusion, including collecting multi-source heterogeneous market operation data of all power generation enterprises in the power market; based on the multi-source heterogeneous market operation data, outputting the weighted average bid price, withholding ratio, high bid price ratio, and dynamic market share of the corresponding units, and performing normalization processing; analyzing the transaction data to determine the type of market manipulation behavior of the power generation enterprise corresponding to the unit with abnormal data, and giving a warning according to the determined type of market manipulation behavior.

[0009] As a preferred solution of the method for identifying power market manipulation behaviors based on multi-source heterogeneous data fusion according to the present invention, wherein: the multi-source heterogeneous market operation data includes data such as the sub-period bid prices, sub-period bid volumes, and maximum power generation of all units of the enterprise participating in transactions currently and historically.

[0010] As a preferred solution of the method for identifying power market manipulation behaviors based on multi-source heterogeneous data fusion according to the present invention, wherein: based on the multi-source heterogeneous market operation data, outputting the weighted average bid price, withholding ratio, high bid price ratio, and dynamic market share of the corresponding units, and performing normalization processing includes outputting the multi-source heterogeneous market operation data, and constructing a weighted average bid price model expressed as:

[0011]

[0012] wherein, I represents the total number of bid segments; t represents the bid segment to be detected; i represents the first bid segment after the contract power; P t represents the declared electricity price of the t-th segment; Q t represents the declared electricity volume of the t-th segment;

[0013] Constructing a high bid price ratio model expressed as:

[0014]

[0015] wherein, Q hb represents the high bid electricity volume of the unit, and the definition method is to select the sum of the electricity volumes corresponding to the bids greater than or equal to μ + kσ of all bids except the highest bid; wherein, μ represents the average value of all bids, σ represents the standard deviation of all bids, k represents the confidence factor, and the value of k depends on the tolerance of the power market, and k is proportional to the market's tolerance of high bids;

[0016] Constructing a unit dynamic market share model expressed as:

[0017]

[0018] wherein, S j represents the market share of the j-th unit; It represents the winning bid electricity quantity of the j-th unit; J represents the total number of units in the market.

[0019] As a preferred embodiment of the method for identifying electricity market manipulation behavior based on multi-source heterogeneous data fusion according to the present invention, wherein: the normalization process includes normalizing the calculation result using the Z-score method, which is expressed as:

[0020]

[0021] Wherein, x represents the original data value, μ represents the average value of the data group where the original data is located, σ represents the standard deviation of the data group where the original data is located, and x' represents the normalized data value.

[0022] As a preferred embodiment of the method for identifying electricity market manipulation behavior based on multi-source heterogeneous data fusion according to the present invention, wherein: analyzing the transaction data to determine the type of market manipulation behavior of the power generation enterprise corresponding to the unit with abnormal data, and giving a warning according to the determined type of market manipulation behavior includes analyzing the transaction data using a machine learning algorithm, and determining the type of market manipulation behavior of the power generation enterprise corresponding to the unit with abnormal data;

[0023] The machine learning algorithm is the K-means algorithm, including:

[0024] Using the K-means++ initialization method to initialize the data, obtaining K centroids, K≥1, depending on the machine performance of the implementation operation;

[0025] For each sample x in the dataset i , the distance to each centroid c j is expressed using the Euclidean method as:

[0026]

[0027] Wherein, x ik represents the k-th feature of the sample x i , c jk represents the k-th feature of the centroid c j , n represents the dimension of the feature, and each sample x i is assigned to the category corresponding to the centroid c j with the closest distance;

[0028] For each category j, recalculate the centroid c j of the category, and the centroid is expressed as:

[0029]

[0030] Wherein, N j represents the number of samples in category j, C jDenote all samples in class j;

[0031] Repeat the sample assignment and centroid update until the centroid no longer changes, and this state is called convergence;

[0032] Set a detection radius, and the values within the detection radius are determined to be normal, while the values outside the detection radius are determined to be abnormal.

[0033] As a preferred solution of the method for identifying power market manipulation behavior based on multi-source heterogeneous data fusion according to the present invention, wherein: analyzing the trading data to determine the market manipulation behavior type of the power generation enterprise corresponding to the abnormal data, and giving a warning according to the determined market manipulation behavior type. The implementation steps of the K-means++ initialization method include:

[0034] Randomly select a data point as the first centroid;

[0035] For each unselected data point, output the shortest distance D(x) to the selected centroid;

[0036] When selecting the next centroid, randomly select a data point from the probability distribution with D(x) 2 as the weight;

[0037] Repeat until K centroids are selected.

[0038] As a preferred solution of the method for identifying power market manipulation behavior based on multi-source heterogeneous data fusion according to the present invention, wherein: the market manipulation behavior types include capacity withholding, extreme bidding, and jump bidding;

[0039] Define the values of the withholding ratio R and the high bidding ratio HR higher than the detection radius r as possibly having capacity withholding behavior;

[0040] Define the weighted average bid and the unit market share S j higher than the detection radius r as possibly having extreme bidding behavior;

[0041] Define the total winning electricity quantity Q of the unit w and the last section of the bid higher than the detection radius r, while the weighted average bid lower than the detection radius r as possibly having jump bidding behavior.

[0042] Another object of the present invention is to provide a power market manipulation behavior recognition system based on multi-source heterogeneous data fusion, which can output the weighted average bid price, withholding ratio, high bid ratio and dynamic market share of the corresponding unit by using multi-source heterogeneous market operation data, and perform normalization processing, solving the problem that the current supervision and recognition methods for the market manipulation behavior of power generation enterprises cannot accurately prevent, identify and correct these artificially created trading risks.

[0043] As a preferred embodiment of the power market manipulation behavior recognition system based on multi-source heterogeneous data fusion according to the present invention, it includes a data acquisition module, a data processing module, an analysis and judgment module, and a warning and prompt module; the data acquisition module is used to collect the market operation data of power generation enterprises in the power market and collect the multi-source heterogeneous market operation data of all power generation enterprises in the power market; the data processing module is used to use the data collected by the data acquisition module to output the weighted average bid price, withholding ratio, high bid ratio and dynamic market share of the unit, and perform data cleaning on the results to achieve data normalization; the smoothing processing module is used to further identify the reasons for the market manipulation behavior of the power generation enterprise to be detected, identify abnormal data by using advanced machine learning algorithms, and judge the market manipulation behavior type of the power generation enterprise corresponding to the unit with abnormal data; the smoothing processing module is used to analyze the judgment result of the analysis and judgment module and warn the relevant power generation enterprises.

[0044] A computer device includes a memory and a processor, and the memory stores a computer program. When the processor executes the computer program, the steps of the power market manipulation behavior recognition method based on multi-source heterogeneous data fusion are realized.

[0045] A computer-readable storage medium stores a computer program thereon. When the computer program is executed by a processor, the steps of the power market manipulation behavior recognition method based on multi-source heterogeneous data fusion are realized.

[0046] Advantages of the present invention: The method for identifying power market manipulation behaviors based on multi-source heterogeneous data fusion provided by the present invention collects multi-source heterogeneous market operation data of all power generation enterprises in the power market, including at least data such as the time-of-use quotes, time-of-use quantities reported, and maximum power generation of all units participating in transactions of these enterprises at present and in history; according to the collected data, calculate the weighted average quote, withholding ratio, high quote ratio, and dynamic market share of the corresponding units respectively, and normalize the calculation results; use advanced machine learning algorithms to analyze transaction data, and determine the type of market manipulation behavior of the power generation enterprise corresponding to the unit with abnormal data; according to the type of market manipulation behavior determined in the third step, give warnings to relevant power generation enterprises to achieve the purpose of identifying and correcting market manipulation behaviors; the proposed identification system is a practical product of the identification method, which strengthens the operability of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0048] Figure 1 It is the overall flowchart of a method for identifying power market manipulation behaviors based on multi-source heterogeneous data fusion provided by the first embodiment of the present invention.

[0049] Figure 2 It is the structural block diagram of a system for identifying power market manipulation behaviors based on multi-source heterogeneous data fusion provided by the third embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0050] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the following will make a detailed description of the specific embodiments of the present invention in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0051] Example 1, referring to Figure 1 , which is an embodiment of the present invention, provides a method for identifying power market manipulation behaviors based on multi-source heterogeneous data fusion, including:

[0052] S1: Collect multi-source heterogeneous market operation data of all power generation enterprises in the power market.

[0053] Further, collect the multi-source heterogeneous market operation data of all power generation enterprises in the power market, including at least the data of the segmented quotes, segmented quantities reported, and maximum power generation of all units participating in transactions by these enterprises at present and historically.

[0054] S2: Based on the multi-source heterogeneous market operation data, output the weighted average quotes, retention ratios, high quote ratios, and dynamic market shares of the corresponding units, and perform normalization processing.

[0055] Further, according to the data collected in S1, calculate the weighted average quotes, retention ratios, high quote ratios, and dynamic market shares of the corresponding units respectively, and perform normalization processing on the calculation results.

[0056] It should be noted that the following methods are used to calculate the weighted average quotes, retention ratios, high quote ratios, and dynamic market shares of the corresponding units, and perform normalization processing on the calculation results:

[0057] (1) Weighted average quote (price) The calculation formula is:

[0058]

[0059] It should be noted that I is the total number of quote segments; t is the quote segment to be detected, and multiple segments can be selected; i is the first quote segment after the contract electricity quantity; P t is the declared electricity price for the t-th segment; Q t is the declared electricity quantity for the t-th segment;

[0060] (2) The calculation formula for the retention ratio R (retention rate) is:

[0061]

[0062] It should be noted that Q max is the maximum power generation of the unit; Q b is the total declared electricity quantity of the unit

[0063] (3) The calculation formula for the high quote ratio HP (high price rate) is:

[0064]

[0065] It should be noted that Q hbThe high - quoted electricity quantity for this unit is defined as follows: Select the sum of the electricity quantities corresponding to the quotations that are greater than or equal to μ + kσ of all quotations except the highest quotation, where μ is the average value of all quotations, σ is the standard deviation of all quotations, and k is the confidence factor. The value of k depends on the tolerance of this electricity market. The larger k is, the higher the tolerance of the market for high quotations.

[0066] (4) The dynamic market share S j (share) is calculated as:

[0067]

[0068] It should be noted that S j is the market share of the j - th unit; is the winning bid electricity quantity of the j - th unit; J is the total number of units in the market.

[0069] (5) Use the Z - score method to normalize the calculation results:

[0070]

[0071] It should be noted that x is the original data value, μ is the average value of the data group where the original data is located, σ is the standard deviation of the data group where the original data is located, and x′ is the normalized data value. The significance of this step is to unify the numerical value range of multi - source heterogeneous data, facilitating the calculation of subsequent machine learning algorithms.

[0072] S3: Analyze the transaction data to determine the type of market manipulation behavior of the power generation enterprise corresponding to the unit with abnormal data, and give a warning according to the determined type of market manipulation behavior.

[0073] Furthermore, use advanced machine learning algorithms to analyze the transaction data and determine the type of market manipulation behavior of the power generation enterprise corresponding to the unit with abnormal data;

[0074] In a specific embodiment, the following method is used to determine the type of market manipulation behavior of the subject to be detected:

[0075] (1) Initialization: Use the K - means++ initialization method to initialize the data and obtain K centroids, where K≥1 and depends on the machine performance for the implementation operation;

[0076] (2) Assign samples: For each sample x i in the data set, calculate its distance to each centroid c j using the Euclidean method:

[0077]

[0078] where xik is the k-th feature of sample x i , and c jk is the k-th feature of the centroid c j , where n is the dimension of the feature. Then, each sample x i is assigned to the category corresponding to the nearest centroid c j ;

[0079] (3) Update the centroid: For each category j, recalculate the centroid c j of this category. The calculation formula for the centroid is the mean value of all samples in this category:

[0080]

[0081] where N j is the number of samples in category j, and C j are all samples in category j;

[0082] (4) Iteration: Repeat steps (2) and (3) until the centroid no longer changes significantly, and this state is called convergence;

[0083] (5) Detect outliers: Set a detection radius. Values within the detection radius are determined to be normal, and values outside the detection radius are determined to be abnormal. Combining with the Z-score theory, the value r of the detection radius should be between 2 and 3. This setting method can improve the detection accuracy of the system and prevent misdetection as much as possible.

[0084] It should be noted that the implementation steps of the K-means++ initialization method are as follows:

[0085] (1) Randomly select a data point as the first centroid;

[0086] (2) For each unselected data point, calculate its shortest distance D(x) to the selected centroids;

[0087] (3) When selecting the next centroid, randomly select a data point from the probability distribution with D(x) 2 as the weight;

[0088] (4) Repeat steps 2 and 3 until K centroids are selected;

[0089] It should be noted that the types of market manipulation include capacity withholding, extreme bidding, and jump-up behavior. Values of the withholding ratio R and the high-bidding ratio HR higher than the detection radius r are defined as possible capacity withholding behaviors; the weighted average bid and the unit market share S j higher than the detection radius r are defined as possible extreme bidding behaviors; the total winning electricity quantity Q wand the last paragraph's quoted price is higher than the detection radius r, while the weighted average quoted price The value lower than the detection radius r is defined as a possible high-jump behavior.

[0090] Furthermore, according to the determined types of market manipulation behaviors, relevant power generation enterprises are warned to achieve the purpose of identifying and correcting market manipulation behaviors.

[0091] Embodiment 2, an embodiment of the present invention, provides a method for identifying power market manipulation behaviors based on multi-source heterogeneous data fusion. To verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0092] First, data collection is carried out. In the first step, data collection is required. In this embodiment, the market transaction data of a provincial power trading center from January to December 2024 is used, and 10 typical units (including thermal power, wind power, and photovoltaic) are selected. The specific data types include:

[0093] (1) Quoted price by time period (selecting a quoted price segment every 15 minutes, and the quoted price result is reserved to 1 decimal place, unit: yuan / MWh);

[0094] (2) Declared electricity volume (unit: MWh, taking integers);

[0095] (3) Maximum power generation capacity of the unit (unit: MWh);

[0096] (4) Successful bid electricity volume (unit: MWh);

[0097]

[0098]

[0099] The collection results are shown in the following table:

[0100] Then data processing is carried out. In the first step, data preprocessing is required. Missing values and outliers (such as negative quotes, over-capacity declarations, etc.) are cleaned, and the following indicators are calculated according to the corresponding formulas:

[0101] (1) Weighted average quoted price

[0102]

[0103] (2) Retention ratio (R):

[0104]

[0105] (3) High quoted price ratio (HP):

[0106]

[0107] (4) Dynamic market share (S j ):

[0108]

[0109] The calculation results are shown in the following table:

[0110]

[0111] In the second step, parameter presetting is required. According to the calculation results in the first step and the characteristics of the collected data, the following parameter settings are made in this embodiment:

[0112] (1) Confidence factor k = 1.8 (to reduce market volatility interference);

[0113] (2) Detection radius r = 2.2 (Z-score threshold);

[0114] (3) Number of K-means clusters K = 3 (verified by silhouette coefficient);

[0115] In the third step, normalization processing is required. Z-score normalization is performed on the above indicators, and the formula is:

[0116]

[0117] The normalized distances are shown in the following table:

[0118]

[0119] Then, analysis and judgment are carried out. In the first step, a calculation model needs to be constructed, including the following steps:

[0120] (1) Initialize the centroids using the K-means++ algorithm (3, numbered C1 - C3);

[0121] (2) Calculate the distance between the sample and the centroid using Euclidean distance;

[0122] (3) Set the maximum number of iterations to 100 times, and the convergence threshold is set to 1e-4;

[0123] The distribution status of the abnormal result judgment is shown in the following table:

[0124]

[0125] In the second step, determine whether it conforms to the characteristics of market manipulation;

[0126] (1) Class C3 (high risk): 3 units (G03 / G05 / G09), all detected with high jump / extreme quotes;

[0127] (2) Class C2 (suspicious): 3 units (G02 / G07 / G10), with capacity retention but relatively low abnormal offer values;

[0128] (3) False alarm verification: The retention ratio of G06 (Class C1) at 25.3% is within the normal maintenance range;

[0129] The advantages of this method are as follows:

[0130] (1) Successfully identified the "false compliance" jump behavior of G03 units. This type of offer behavior usually does not reach the threshold of traditional algorithms and is thus difficult to identify;

[0131] (2) Achieved aggregated detection of the "scattered retention" behavior of G10 units (i.e., the behavior of evading power trading market supervision through multi-day small-scale retention behavior), improving the recognition of this type of manipulation behavior.

[0132] Example 3, referring to Figure 2 , which is an embodiment of the present invention, provides a power market manipulation behavior recognition system based on multi-source heterogeneous data fusion, including a data acquisition module, a data processing module, an analysis and judgment module, and an early warning and prompt module.

[0133] Among them, the data acquisition module is used to collect the market operation data of power generation enterprises in the power market and collect the multi-source heterogeneous market operation data of all power generation enterprises in the power market; the data processing module is used to utilize the data collected in the data acquisition module to output the weighted average offer, retention ratio, high offer ratio, and dynamic market share of the unit, and perform data cleaning on the results to achieve data normalization; the smoothing processing module is used to further identify the reasons for the market manipulation behavior of the power generation enterprises to be detected, identify abnormal data using advanced machine learning algorithms, and judge the type of market manipulation behavior of the power generation enterprises corresponding to the units with abnormal data; the smoothing processing module is used to analyze the judgment results of the analysis and judgment module and warn the relevant power generation enterprises.

[0134] If a function 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 a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in various embodiments of the present invention. The aforementioned storage medium includes: USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs, etc., all kinds of media that can store program codes.

[0135] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a predefined sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch instructions from the instruction execution system, apparatus, or device and execute the instructions), or used in combination with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in combination with an instruction execution system, apparatus, or device.

[0136] More specific examples (non-exhaustive list) of computer-readable media include the following: electrical connection parts with one or more wirings (electronic devices), portable computer disk cartridges (magnetic devices), random access memories (RAMs), read-only memories (ROMs), erasable programmable read-only memories (EPROMs or flash memories), optical fiber devices, and portable compact disc read-only memories (CDROMs). Additionally, a computer-readable medium can even be paper or other suitable media on which a program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or other appropriate processing when necessary, and then stored in a computer memory.

[0137] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc. It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

[0138] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A method for identifying electricity market manipulation behaviors based on multi-source heterogeneous data fusion, characterized in that including: Collecting multi-source heterogeneous market operation data of all power generation enterprises in the power market; Based on the multi-source heterogeneous market operation data, outputting the weighted average quotation, withholding ratio, high quotation ratio and dynamic market share of the corresponding units, and performing normalization processing; Analyzing transaction data to judge the market manipulation behavior types of power generation enterprises corresponding to units with abnormal data, and giving warnings according to the determined market manipulation behavior types.

2. The method for identifying electricity market manipulation behaviors based on multi-source heterogeneous data fusion according to claim 1, characterized in that: The multi-source heterogeneous market operation data includes data such as the sub-period quotations, sub-period reported quantities, and maximum power generation of all units of the enterprise participating in transactions currently and historically.

3. The method for identifying electricity market manipulation behavior based on multi-source heterogeneous data fusion according to claim 2, wherein: The outputting the weighted average quotation, withholding ratio, high quotation ratio and dynamic market share of the corresponding units, and performing normalization processing based on the multi-source heterogeneous market operation data includes outputting the multi-source heterogeneous market operation data and constructing a weighted average quotation model expressed as: Where, I represents the total number of quotation segments; t represents the quotation segment to be detected; i represents the first quotation segment after the contract electricity quantity; P t represents the declared electricity price of the t-th segment; Q t represents the declared electricity quantity of the t-th segment; Constructing a high quotation ratio model expressed as: Among them, Q hb represents the high-bid electricity quantity of the unit. The definition method is to select the sum of the electricity quantities corresponding to the bids that are greater than or equal to μ + kσ of all bids except the highest bid; where μ represents the average value of all bids, σ represents the standard deviation of all bids, k represents the confidence factor, and the value of k depends on the tolerance of the power market. k is proportional to the market's tolerance for high bids. Constructing a unit dynamic market share model expressed as: Among them, S j represents the market share of the j-th unit; represents the winning bid electricity of the j-th unit; J represents the total number of units in the market.

4. The method for identifying power market manipulation behavior based on multi-source heterogeneous data fusion according to claim 3, wherein: The performing normalization processing includes normalizing the calculation results using the Z-score method, expressed as: where x represents the original data value, μ represents the average value of the data group where the original data is located, σ represents the standard deviation of the data group where the original data is located, and x′ represents the normalized data value.

5. The method for identifying power market manipulation behaviors based on multi-source heterogeneous data fusion according to claim 4, wherein: The analyzing transaction data to judge the market manipulation behavior types of power generation enterprises corresponding to units with abnormal data, and giving warnings according to the determined market manipulation behavior types includes using machine learning algorithms to analyze transaction data and judging the market manipulation behavior types of power generation enterprises corresponding to units with abnormal data; The machine learning algorithm is the K-means algorithm, including: Using the K-means++ initialization method to initialize the data to obtain K centroids, K≥1, depending on the machine performance of the implementation operation; For each sample x in the dataset i , the distance to each centroid c j is represented by using the Euclidean method as follows: Among them, x ik represents the k-th feature of the sample x i , c jk represents the k-th feature of the centroid c j , n represents the dimension of the feature. Each sample x i is assigned to the category corresponding to the centroid c j that is the closest; For each category j, recalculate the centroid c of the category j , which is represented as: Among them, N j represents the number of samples in class j, and C j represents all samples in class j; Repeating to assign samples and update the centroids until the centroids no longer change and stop, and this state is called convergence; Setting a detection radius, and determining the values within the detection radius as normal and the values outside the detection radius as abnormal.

6. The method for identifying electricity market manipulation behavior based on multi-source heterogeneous data fusion according to claim 5, wherein: The analyzing transaction data to judge the market manipulation behavior types of power generation enterprises corresponding to units with abnormal data, and giving warnings according to the determined market manipulation behavior types includes the implementation steps of the K-means++ initialization method including: Randomly selecting a data point as the first centroid; For each unselected data point, outputting the shortest distance D(x) to the selected centroids; When selecting the next centroid, randomly select a data point from the probability distribution weighted by D(x) 2 ; Repeating until K centroids are selected.

7. The method for identifying power market manipulation behaviors based on multi-source heterogeneous data fusion according to claim 6, wherein: The market manipulation behavior types include capacity withholding, extreme quotation and price jump behavior; Defining the values of the withholding ratio R and the high quotation ratio HR higher than the detection radius r as possible capacity withholding behaviors; The weighted average offer and the unit market share S j Values higher than the detection radius r are defined as potentially extreme offer behaviors; The total winning bid power Q of the unit w and the last paragraph where the bid price is higher than the detection radius r, while the weighted average bid price lower than the detection radius r is defined as a possible high jump behavior.

8. A system adopting the method for identifying electricity market manipulation behaviors based on multi-source heterogeneous data fusion according to any one of claims 1 to 7, characterized in that: Including a data collection module, a data processing module, an analysis and judgment module, and a warning and prompt module; The data collection module is used to collect the market operation data of power generation enterprises in the power market and collect the multi-source heterogeneous market operation data of all power generation enterprises in the power market; The data processing module is used to utilize the data collected in the data acquisition module, output the weighted average quotation, retention ratio, high quotation ratio of the generating unit, and the dynamic market share of the generating unit, and perform data cleaning on the results to achieve data normalization; The smoothing processing module is used to further identify the reasons for the market manipulation behavior of the power generation enterprise to be detected, identify abnormal data using advanced machine learning algorithms, and determine the type of market manipulation behavior of the power generation enterprise corresponding to the generating unit with abnormal data; The smoothing processing module is used to analyze the judgment results of the judgment module and give warnings to relevant power generation enterprises.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method for identifying power market manipulation behavior based on multi-source heterogeneous data fusion according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for identifying power market manipulation behavior based on multi-source heterogeneous data fusion according to any one of claims 1 to 7.

Images