Abnormal power quotation behavior identification method and device based on clustering algorithm

Abstract

The invention relates to an abnormal power quotation behavior identification method and device based on a clustering algorithm, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: carrying out a Legenna index test, a residual supply rate index test and a behavior influence test on market electricity price data to obtain a test result; screening out the target generator set which does not pass the test, and replacing the original quoted price of the target generator set with a preset reference quoted price; performing price influence test on the target generator set by using a preset scheduling model to obtain an analysis result of the electricity price persistence behavior, and processing to obtain an initial clustering result of the abnormal power quotation behavior data; and performing a similarity test on the quotation curve of each cluster generator set, and performing a retention level test on the corresponding quotation of each cluster generator set to obtain an identification result of the abnormal power quotation behavior. The method can effectively identify the abnormal power quotation behavior of the power generator.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for identifying abnormal electricity bidding behavior based on a clustering algorithm. Background Art

[0002] In a landscape where multiple market structures, including spot markets and medium- and long-term trading, coexist, some power producers may manipulate node electricity prices through strategic bidding or exploit market power to engage in non-competitive bidding. This not only distorts the market regulation function of price signals but also potentially triggers market operational risks, threatening the safe and economic operation of the power system. However, current monitoring of power producer bidding behavior relies primarily on manual judgment and statistical analysis based on fixed thresholds, making it difficult to effectively identify hidden and time-varying abnormal bidding patterns in complex market environments.

[0003] Therefore, there is an urgent need for a method, device, computer equipment, computer-readable storage medium and computer program product for identifying abnormal electricity bidding behavior based on a clustering algorithm, which can effectively identify the abnormal electricity bidding behavior of power producers to ensure the healthy operation of the electricity market. Summary of the Invention

[0004] Based on this, it is necessary to provide a method, device, computer equipment, computer-readable storage medium and computer program product based on clustering algorithm to effectively identify abnormal electricity bidding behavior of power producers in accordance with the above technical problems to ensure the healthy operation of the power market.

[0005] In a first aspect, the present application provides a method for identifying abnormal electricity bidding behavior based on a clustering algorithm, comprising:

[0006] collecting market electricity price data, performing a Lerner index test, a surplus supply rate index test, and a behavior impact test on the market electricity price data, and obtaining test results;

[0007] According to the test results, the target generators that failed the test are screened out, and the original bids of the target generators are replaced with the preset benchmark bids to obtain adjusted power bid data;

[0008] Based on the adjusted power quotation data, the preset dispatch model is used to conduct price impact tests on the target power generation units to obtain the analysis results of the power price retention behavior;

[0009] Based on the analysis results of the electricity price retention behavior, the generator groups other than the target generator group are divided into different clusters, and the target generator group that fails the test is used as the cluster center to obtain the initial clustering results of the abnormal electricity bidding behavior data;

[0010] Based on the initial clustering results, a similarity test is performed on the quotation curves of each cluster of generators, and a retention level test is performed on the corresponding quotations of each cluster of generators to obtain the identification results of abnormal power quotation behavior.

[0011] In one embodiment, the collecting of market electricity price data, performing a Lerner index test, a surplus supply rate index test, and a behavior impact test on the market electricity price data, and obtaining test results include:

[0012] Collecting market clearing price and marginal cost data of power producers and calculating the Lerner index, and conducting a Lerner index test based on the Lerner index to obtain quantitative indicators of the monopoly power level of market participants;

[0013] Obtaining capacity data of each generator set in the market and total market load demand data, and calculating a surplus supply rate index, performing a surplus supply rate index test based on the surplus supply rate index, and obtaining a quantitative indicator result of the market control power of the generator;

[0014] Obtain actual quotes and preset benchmark quotes for generator sets;

[0015] Conduct behavioral impact tests based on actual quotations and preset benchmark quotations to obtain preliminary judgment results on whether there is market power manipulation; conduct price impact tests based on actual quotations and preset benchmark quotations to obtain quantitative results on the degree of impact of actual quotations on market electricity prices.

[0016] In one embodiment, before performing the price impact test on the target generator set using the preset scheduling model, the method further includes:

[0017] Obtain the winning bid output data and cost curve data of each generator set;

[0018] Establishing a preset scheduling model based on the marked output data and the cost curve data;

[0019] Determine the constraints of the preset scheduling model, wherein the constraints include system balance constraints, system reserve constraints, transmission line transmission capacity constraints, generator set ramp constraints, generator set maximum and minimum output constraints, and line power flow constraints.

[0020] In one embodiment, the similarity test is performed on the bid curves of the generator groups in each cluster based on the initial clustering results, and the retention level test is performed on the bids of the generator groups in each cluster to obtain the identification results of abnormal power bidding behavior, including:

[0021] According to the initial clustering results, the Euclidean distance between the cluster center of each cluster and other generator groups is calculated to obtain the initial clustering distance;

[0022] According to the initial clustering distance, the clusters to which each generator group belongs are reallocated, and the cluster center of each cluster is updated;

[0023] Based on the updated cluster center, return to the step of recalculating the Euclidean distance between the cluster center of each cluster and other generator groups until the clustering algorithm converges to obtain the clustering result of abnormal power bidding behavior;

[0024] Obtain the cluster center of each cluster and the quotation curve data of other generators;

[0025] Performing a similarity test on the quotation curves of each cluster of generator sets based on the quotation curve data to obtain a similarity test result;

[0026] According to the clustering results of abnormal power bidding behaviors, the bid retention levels of the generators in each cluster are calculated, and the identification results of abnormal power bidding behaviors are obtained.

[0027] In one embodiment, the Euclidean distance is calculated as follows:

[0028] Obtain the total number of price segments of market quotations, as well as the quotes and quantities of each generator set in each price segment;

[0029] Calculate the difference between the cluster center of each cluster and the bid and quantity difference of other generators in each price range;

[0030] The square values ​​of the bid and quantity differences of all price segments are taken, and the square roots are added together to obtain the Euclidean distance between the cluster center and other generators.

[0031] In one embodiment, the price impact test on the target generator set is performed using a preset scheduling model based on the adjusted power quotation data, including:

[0032] Based on the adjusted electricity quotation data, calculate the ratio between the original market quotation and the preset benchmark quotation, and use it as the price impact test factor;

[0033] Based on the price impact test factor, a price impact test is performed on the target power generation unit using a preset scheduling model.

[0034] In a second aspect, the present application also provides a device for identifying abnormal power bidding behavior based on a clustering algorithm, comprising:

[0035] a data acquisition module, configured to collect market electricity price data, perform a Lerner index test, a surplus supply rate index test, and a behavior impact test on the market electricity price data, and obtain test results;

[0036] A data processing module is used to screen out target generator sets that fail the test based on the test results, replace the original bids of the target generator sets with preset benchmark bids, and obtain adjusted power bid data;

[0037] The price impact test module is used to conduct price impact tests on target generators based on the adjusted power quotation data using a preset dispatch model to obtain analysis results on power price retention behavior;

[0038] a bidding behavior identification module, configured to divide the generator sets other than the target generator set into different clusters based on the analysis results of the electricity price retention behavior, use the target generator set that failed the test as the cluster center, and obtain an initial clustering result of the abnormal electricity bidding behavior data;

[0039] The quotation behavior identification module is also used to perform similarity tests on the quotation curves of each cluster of generator sets based on the initial clustering results, and to perform retention level tests on the corresponding quotations of each cluster of generator sets, so as to obtain the identification results of abnormal power quotation behavior.

[0040] In a third aspect, the present application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0041] collecting market electricity price data, performing a Lerner index test, a surplus supply rate index test, and a behavior impact test on the market electricity price data, and obtaining test results;

[0042] According to the test results, the target generators that failed the test are screened out, and the original bids of the target generators are replaced with the preset benchmark bids to obtain adjusted power bid data;

[0043] Based on the adjusted power quotation data, the preset dispatch model is used to conduct price impact tests on the target power generation units to obtain the analysis results of the power price retention behavior;

[0044] Based on the analysis results of the electricity price retention behavior, the generator groups other than the target generator group are divided into different clusters, and the target generator group that fails the test is used as the cluster center to obtain the initial clustering results of the abnormal electricity bidding behavior data;

[0045] Based on the initial clustering results, a similarity test is performed on the quotation curves of each cluster of generators, and a retention level test is performed on the corresponding quotations of each cluster of generators to obtain the identification results of abnormal power quotation behavior.

[0046] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0047] collecting market electricity price data, performing a Lerner index test, a surplus supply rate index test, and a behavior impact test on the market electricity price data, and obtaining test results;

[0048] According to the test results, the target generators that failed the test are screened out, and the original bids of the target generators are replaced with the preset benchmark bids to obtain adjusted power bid data;

[0049] Based on the adjusted power quotation data, the preset dispatch model is used to conduct price impact tests on the target power generation units to obtain the analysis results of the power price retention behavior;

[0050] Based on the analysis results of the electricity price retention behavior, the generator groups other than the target generator group are divided into different clusters, and the target generator group that fails the test is used as the cluster center to obtain the initial clustering results of the abnormal electricity bidding behavior data;

[0051] Based on the initial clustering results, a similarity test is performed on the quotation curves of each cluster of generators, and a retention level test is performed on the corresponding quotations of each cluster of generators to obtain the identification results of abnormal power quotation behavior.

[0052] In a fifth aspect, the present application further provides a computer program product, comprising a computer program, which, when executed by a processor, implements the following steps:

[0053] collecting market electricity price data, performing a Lerner index test, a surplus supply rate index test, and a behavior impact test on the market electricity price data, and obtaining test results;

[0054] According to the test results, the target generators that failed the test are screened out, and the original bids of the target generators are replaced with the preset benchmark bids to obtain adjusted power bid data;

[0055] Based on the adjusted power quotation data, the preset dispatch model is used to conduct price impact tests on the target power generation units to obtain the analysis results of the power price retention behavior;

[0056] Based on the analysis results of the electricity price retention behavior, the generator groups other than the target generator group are divided into different clusters, and the target generator group that fails the test is used as the cluster center to obtain the initial clustering results of the abnormal electricity bidding behavior data;

[0057] Based on the initial clustering results, a similarity test is performed on the quotation curves of each cluster of generators, and a retention level test is performed on the corresponding quotations of each cluster of generators to obtain the identification results of abnormal power quotation behavior.

[0058] The above-mentioned abnormal power bidding behavior identification method, device, computer equipment, computer-readable storage medium and computer program product based on clustering algorithm can systematically screen out abnormal bidding behavior of generators through Lerner index test, surplus supply rate index test and behavior impact test; for target generator sets that fail the test, a preset benchmark quotation is used to replace their original quotation, and the price impact test is re-performed to further verify whether their bidding behavior is abnormal, thereby reducing the possibility of missed judgment; the generator sets are divided into different clusters through the clustering method, and the units that fail the test are used as the cluster center, which can quickly identify groups of units that may engage in collusion and improve supervision efficiency; based on the initial clustering results, the quotation curves of the generator sets in each cluster are subjected to similarity test and retention level test, which can dynamically adjust the clustering results to adapt to changes in the market environment; through price impact test and retention behavior analysis, the retention behavior of generators can be discovered in a timely manner to avoid its negative impact on market prices and system security; through cluster analysis and quotation curve similarity test, collusion between generator sets can be identified to prevent them from obtaining excess profits by jointly raising quotations.

[0059] In summary, this method can not only identify abnormal bidding behavior, but also provide market regulators with detailed abnormal behavior data and analysis results through cluster analysis and retention level testing, helping them to formulate targeted regulatory measures and maintain the normal order of the power market. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments of the present application or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying any creative work.

[0061] Figure 1 FIG2 is a diagram showing an application environment of a method for identifying abnormal power bidding behavior based on a clustering algorithm in one embodiment;

[0062] Figure 2 1 is a flow chart of a method for identifying abnormal power bidding behavior based on a clustering algorithm in one embodiment;

[0063] Figure 3 1 is a flow chart of a method for identifying abnormal power bidding behavior based on a clustering algorithm in another embodiment;

[0064] Figure 4 1 is a structural block diagram of an abnormal power bidding behavior identification device based on a clustering algorithm in one embodiment;

[0065] Figure 5FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0067] The abnormal power bidding behavior identification method based on clustering algorithm provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown, the terminal 102 communicates with the server 104 via a network. The data storage system can store data that the server 104 needs to process. The data storage system can be integrated on the server 104 or placed on the cloud or other network servers.

[0068] The server 104 collects market electricity price data through the terminal 102, and performs a Lerner index test, a surplus supply rate index test, and a behavior impact test on the market electricity price data to obtain test results; based on the test results, the target generator groups that fail the test are screened out, and the original quotations of the target generator groups are replaced with preset benchmark quotations to obtain adjusted electricity quotation data; based on the adjusted electricity quotation data, a price impact test is performed on the target generator groups using a preset scheduling model to obtain analysis results of electricity price retention behavior; based on the analysis results of electricity price retention behavior, the generator groups other than the target generator groups are divided into different clusters, and the target generator groups that fail the test are used as cluster centers to obtain initial clustering results of abnormal electricity quotation behavior data; based on the initial clustering results, a similarity test is performed on the quotation curves of the generator groups in each cluster, and a retention level test is performed on the corresponding quotations of the generator groups in each cluster to obtain identification results of abnormal electricity quotation behavior.

[0069] The terminal 102 may be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices may include smart speakers, smart TVs, smart air conditioners, smart car devices, projectors, and the like. Portable wearable devices may include smart watches, smart bracelets, head-mounted devices, and the like. Head-mounted devices may include virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, and the like. The server 104 may be an independent physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server providing cloud computing services.

[0070] In an exemplary embodiment, Figure 2As shown in the figure, a method for identifying abnormal power bidding behavior based on clustering algorithm is provided. Figure 1 The server 104 in the example is used as an example to illustrate the process, including the following steps S202 to S210.

[0071] Step S202 : collecting market electricity price data, performing a Lerner index test, a surplus supply rate index test, and a behavior impact test on the market electricity price data, and obtaining test results.

[0072] Specifically, we collect bid data from power generators in the electricity market, including information such as the bid price, output, and cost of each generator, as well as data on the market's total load demand. This data, which typically includes the generator's marginal cost, actual bid price, clearing price, and total market load demand, forms the basis for subsequent analysis.

[0073] Among them, a) Calculation of the Lerner Index: Regarding retention behavior analysis, the most commonly used measurement indicator is the Lerner Index (L), which measures the level of monopoly power of market participants by the degree of deviation between the market clearing price and marginal cost. The specific definition is:

[0074]

[0075] Where P is the current market clearing price; MC is the unit's marginal cost. In an ideal market, the unit's LI is 0; in a typical market, it is slightly greater than 0. However, if the unit's LI is too high, it indicates that the unit may be holding back. A reasonable range for selecting the LI is (0, 0.25).

[0076] b) Residual supply index test: The residual supply index (RSI) of a power company refers to the sum of the market shares of all units except the unit in question during a certain period of time. It can be expressed as:

[0077]

[0078] Where: m is the number of all groups in the market; q j is the capacity of the jth unit; q i is the capacity of the i-th unit; D is the total market load demand. The smaller the RSI number of the unit, the stronger its ability to control the market electricity price, and the more likely it is to collude with other power generators to bid in order to raise the market clearing price and obtain excess profits. i <1, indicating that without the supply of unit i, the market demand will not be met. At this point, the unit has complete control over the market electricity price.

[0079] c) Behavior Impact Test: The behavior impact test is to compare the actual quotation of the unit with the reference quotation. If the actual quotation of the unit deviates too much from the reference quotation, it indicates that the behavior impact test has failed. The behavior impact test factor α is defined as:

[0080]

[0081] Where: P bid is the actual quotation of the unit; P ref The benchmark price for the market. In the New York State electricity market, the behavioral impact test fails when α > 2. In the New England and Midwestern power markets, the behavioral impact test fails when α > 3. my country's power market has a higher degree of concentration among market participants, necessitating stricter standards. Therefore, the behavioral impact test fails when the behavioral impact factor α > 1.5.

[0082] Step S204 , based on the test results, the target generator sets that fail the test are screened out, and the original bids of the target generator sets are replaced with preset benchmark bids to obtain adjusted power bid data.

[0083] Specifically, the Lerner Index, Excess Supply Index, and Behavioral Impact Tests were performed on market electricity price data to obtain test results for each generator. Generators failing the tests were determined based on set thresholds (e.g., a Lerner Index exceeding 0.25, a Behavioral Impact Factor exceeding 1.5, a Price Impact Factor exceeding 2, etc.). Generators failing the tests were considered "target generators," indicating potential issues with their bidding behavior, such as withholding or collusion.

[0084] A preset benchmark quote is a reasonable, competitive quote, typically based on the market average or some other fair pricing mechanism. It replaces the target generator's original quote to eliminate the potential for abnormal quotes to interfere with market analysis and ensure the accuracy and fairness of subsequent analysis.

[0085] For target generators that fail the test, their original bids are replaced with preset benchmark bids. This substitution simulates a market environment free of abnormal bidding behavior, allowing for more accurate analysis of the behavior of other generators and assessment of the overall market status. The resulting power bid data is called "adjusted power bid data." This adjusted data is closer to an idealized market state and eliminates the distortion of market price formation mechanisms caused by abnormal bids.

[0086] Step S206 , based on the adjusted power quotation data, a price impact test is performed on the target power generating set using a preset scheduling model to obtain an analysis result of the power price retention behavior.

[0087] Specifically, pre-defined dispatch models (such as the SCED model) are tools used for electricity market clearing calculations. Their goal is to minimize the system's total operating costs while satisfying system security constraints. These dispatch models can simulate the operating state of the electricity market and calculate the market-clearing price and winning bid output of each unit under different bidding conditions.

[0088] The dispatch model is run through the adjusted electricity bidding data to analyze the specific impact of the bidding behavior of the target generator set on the market price, and to determine whether there is any retention behavior. The dispatch model is run using the adjusted electricity bidding data (i.e., replacing the original bidding of the target generator set) to obtain a new market clearing result, including the new market clearing price and the winning bid output of each unit.

[0089] The security constrained economic dispatch (SCED) model is an algorithm used in the electricity market to calculate the time-of-use electricity price at each node. The SCED model aims to optimize system cost, namely:

[0090]

[0091] Where: p i (t) is the winning bid output of the i-th unit at time t; C(p i (t)) is usually the cost curve of the unit, and in the power market bidding environment, it is the bidding curve.

[0092] If the unit fails the behavior impact test, it will be subjected to the price impact test. The difference between the two is that the behavior impact test measures the degree to which the unit's actual bid deviates from the reference bid, while the price impact test measures the degree to which the bid affects the marginal node electricity price. The price impact test factor β is defined as:

[0093]

[0094] Where: P ori The market price when the dispatch center uses the original quotation of the unit for clearing is called the original market quotation; P att The market price at which the dispatch center uses a preset benchmark quote instead of the original quote for clearing is called the alternative market price. Based on the operating experience of foreign power markets, the price impact test fails when β > 2.

[0095] If the price impact test factor β exceeds the set threshold (such as 2), it is considered that the bidding behavior of the target generator group has a significant impact on the market price and there may be retention behavior.

[0096] Step S208: Based on the analysis results of the electricity price retention behavior, the generator groups other than the target generator group are divided into different clusters, and the target generator group that fails the test is used as the cluster center to obtain the initial clustering results of the abnormal electricity bidding behavior data.

[0097] Specifically, through cluster analysis, other generators are divided into different clusters in order to identify groups of generators that may engage in collusion.

[0098] Cluster analysis is a data mining technique that divides data points (in this case, generators) into clusters, ensuring that data points within the same cluster have high similarity, while data points between different clusters have low similarity. Similarity between generators is typically measured using metrics such as bid curve similarity and bid level similarity. The target generators that failed the test have been identified as potentially exhibiting abnormal bidding behavior. Using these generators as cluster centers allows for more effective identification of other generators with similar behavior. Using the target generator as the cluster center, the similarity between other generators and these centers is calculated (e.g., bid curve similarity, bid level similarity, etc.), and similar generators are grouped into the same cluster. The initial clustering results obtained through these steps can preliminarily identify groups of generators that may be engaging in collusion. Generators within these clusters may be using coordinated bidding to inflate market prices and generate excess profits. Generators within each cluster exhibit high similarity in bidding behavior, such as similar bid curves and similar bid levels. These characteristics indicate the potential for collusion among the generators within the cluster.

[0099] In step S210 , based on the initial clustering results, similarity tests are performed on the bid curves of the generator groups in each cluster, and retention level tests are performed on the bids corresponding to the generator groups in each cluster to obtain identification results of abnormal power bid behaviors.

[0100] Specifically, through the similarity test of the bidding curves, we can further confirm whether the bidding behaviors of each unit in the cluster are highly consistent, so as to determine whether there is collusion.

[0101] The similarity of the bidding curves of the generator set pairs is weighted averaged to obtain the similarity of the bidding curves of the generator set cluster. The similarity of the bidding curves of the generator set pairs is measured by the area between the bidding curves of the two units. The similarity of the bidding curves of the generator set pair formed by units A and B is ρ AB Expressed as:

[0102]

[0103] Where: p A (ε) and p B (ε) are the functional expressions of the bid and quotation ratio curves of generator sets A and B respectively; ε is the quotation ratio of the generator sets.

[0104] The generator cluster bid curve similarity index overcomes the limitation of being limited to analyzing bid behavior between two generators. It can now analyze bid behavior from more than two generators. The greater the similarity index, the more similar the bid curves of several generators within the cluster, and the greater the likelihood of collusion among the generators. A reasonable standard for setting this index is (0, 0.8).

[0105] Power generators form a collusive alliance to jointly raise market prices through negotiation. The bids of alliance members vary, but to achieve excess profits, the alliance's overall bid level is always high. Therefore, this embodiment measures the overall bid level of a cluster by constructing a cluster retention level indicator, calculated as follows:

[0106]

[0107] Where: I m N is the quotation retention level index of the unit cluster; l is the number of units in the unit cluster; is the average bid of the i-th unit in the market; N n is the number of all units in the market. If the unit cluster retention level index is high, it indicates that the overall bidding level of the unit cluster is higher than that of other market members, indicating a collusive alliance. The reasonable range of this index is (0, 1.2), which can be dynamically adjusted by market managers based on market operations.

[0108] In the above-mentioned abnormal power bidding behavior identification method based on clustering algorithm, the Lerner index test, the surplus supply rate index test and the behavior impact test can systematically screen out the abnormal bidding behavior of generators; for the target generators that fail the test, the preset benchmark quotation is used to replace their original quotation, and the price impact test is re-performed to further verify whether their bidding behavior is abnormal, thereby reducing the possibility of missed judgment; the generators are divided into different clusters by the clustering method, and the generators that fail the test are used as the cluster center, which can quickly identify the group of generators that may have collusion behavior and improve the supervision efficiency; according to the initial clustering results, the quotation curves of the generators in each cluster are subjected to similarity tests and retention level tests, which can dynamically adjust the clustering results to adapt to changes in the market environment; through price impact tests and retention behavior analysis, the retention behavior of generators can be discovered in a timely manner to avoid its negative impact on market prices and system safety; through cluster analysis and quotation curve similarity tests, collusion between generators can be identified to prevent them from obtaining excess profits by jointly raising quotations.

[0109] In an exemplary embodiment, market electricity price data is collected, and a Lerner index test, a surplus supply rate index test, and a behavior impact test are performed on the market electricity price data to obtain test results, including:

[0110] Collect market clearing price and marginal cost data of power generators, calculate the Lerner index, and conduct Lerner index test based on the Lerner index to obtain quantitative indicators of the monopoly power level of market participants;

[0111] Obtain the capacity data of each generator set in the market and the total market load demand data, calculate the surplus supply rate index, conduct a surplus supply rate index test based on the surplus supply rate index, and obtain the quantitative indicator results of the market control power of the generator;

[0112] Obtain actual quotes and preset benchmark quotes for generator sets;

[0113] Conduct behavioral impact tests based on actual quotations and preset benchmark quotations to obtain preliminary judgment results on whether there is market power manipulation; conduct price impact tests based on actual quotations and preset benchmark quotations to obtain quantitative results on the degree of impact of actual quotations on market electricity prices.

[0114] Specifically, the Lerner Index (LI) quantifies the level of monopoly power of power generators. This index reflects the degree of deviation between the market-clearing price and marginal cost. A LI close to 0 indicates a near-perfectly competitive market, with generators' bids close to marginal cost. A LI significantly greater than 0 indicates that generators may possess market power, with bids deviating from marginal cost. The acceptable range is typically set between 0 and 0.25; values ​​exceeding this range may indicate market manipulation.

[0115] The residual supply index (RSI) quantifies a generator's control over market electricity prices. A smaller RSI indicates a greater control over market electricity prices. An RSI close to 1 indicates that the combined capacity of other generators in the market is close to the total market load demand, and the generator's control over market electricity prices is weak. An RSI significantly less than 1 indicates that the combined capacity of other generators in the market is insufficient to meet the total market load demand, and the generator has a strong control over market electricity prices. Reasonable range: Typically set to (0, 1); values ​​less than 1 indicate potential market control.

[0116] By comparing actual bids with benchmark bids, we determine whether the generator's bid deviates significantly from reasonable levels, providing a preliminary assessment of market power manipulation. An α close to 1 indicates that the actual bid is close to the benchmark and the generator's bid is reasonable. An α significantly greater than 1 indicates that the actual bid is significantly higher than the benchmark, suggesting market power manipulation. The reasonable range is typically set between (1 and 1.5). Exceeding this range indicates a failure to pass the behavioral impact test.

[0117] In this example, by collecting market electricity price data and conducting Lerner Index, Excess Supply Index, and Behavior Impact Tests, we quantify the market behavior of power producers and identify any market power manipulation. These tests measure power producers' market power and bidding behavior from different perspectives, providing a scientific basis for market regulation and helping maintain fairness and efficiency in the electricity market.

[0118] In an exemplary embodiment, before performing a price impact test on a target generator set using a preset scheduling model, the process further includes:

[0119] Obtain the winning bid output data and cost curve data of each generator set;

[0120] Establish a preset scheduling model based on the marked force data and cost curve data;

[0121] Determine the constraints of the preset scheduling model, which include system balance constraints, system reserve constraints, transmission line transmission capacity constraints, generator set ramp constraints, generator set maximum and minimum output constraints, and line flow constraints.

[0122] Specifically, the winning bid output refers to the power output accepted by the dispatch center for each generator during the electricity market clearing process. This data reflects the actual operating status of each generator at the time of market clearing and serves as a fundamental component of the dispatch model. The cost curve describes the operating costs of a generator at different output levels. This typically includes both fixed and variable costs. Cost curve data is used to calculate the operating costs of a generator and is a crucial component of the dispatch model's optimization objective function.

[0123] The winning bid output data and cost curve data are used as training sets to train a preset dispatch model (such as the SCED model). The preset dispatch model is used to simulate the clearing process of the electricity market and determine the output allocation of each generator unit to minimize the total operating cost of the system while meeting the system's safety constraints. In order to ensure that the operation results of the dispatch model meet the actual operating requirements of the power system, a series of constraints need to be set. These constraints include:

[0124] The system equilibrium constraint is:

[0125]

[0126] Where: p i (t) is the winning bid output of the i-th unit at time t; p d (t) is the total active power demand of the system at time t.

[0127] The system standby constraints are:

[0128]

[0129] Where: p r,i (t) is the spare capacity of the i-th unit that won the bid at time t; r i (t) is the spare capacity required by the system at time t.

[0130] The transmission capacity constraint of the transmission line is:

[0131]

[0132] Where: p ij (t), and are the transmission power of the line at time t and the upper and lower limits of the transmission power respectively.

[0133] The generator set climbing constraint is:

[0134] -Δp i ≤p i (t)-p i (t-1)≤Δp i ;

[0135] Where Δp i is the climbing constraint of the first unit.

[0136] The maximum and minimum output constraints of the generator set are:

[0137]

[0138] Where, and are the lower and upper output limits of the i-th unit respectively.

[0139] The line power flow constraint is:

[0140]

[0141] Where: is the transmission power flow limit of line l; g l-i is the generator output transfer distribution factor of the node where unit i is located on line l; g l-j is the generator output transfer distribution factor of the node where the tie line j is located on line l; p j,t is the output power of tie line j, is the forward and reverse power flow relaxation factor of line l; N is the total number of units participating in the bidding; N T is the total number of contact lines.

[0142] In this embodiment, before using a preset dispatch model to conduct price impact testing on target generators, by obtaining the winning bid output data and cost curve data for each generator, and establishing a preset dispatch model that incorporates constraints such as system balance, reserve, transmission line transmission capacity, ramping, maximum and minimum output, and line power flow, it provides an accurate and reliable foundation for subsequent testing. This process not only improves the accuracy of analysis and enhances the scientific nature of supervision, but also ensures the fairness and efficiency of the power market and enhances the reliability and dynamic adaptability of regulatory decisions, thereby effectively identifying abnormal bidding behavior by generators and maintaining the normal operation of the power market.

[0143] In an exemplary embodiment, Figure 3 As shown in the figure, based on the initial clustering results, similarity tests are performed on the bid curves of each cluster of generators, and retention level tests are performed on the corresponding bids of each cluster of generators. The identification results of abnormal power bid behavior are obtained, including:

[0144] Step S302: Calculate the Euclidean distance between the cluster center of each cluster and other generator groups based on the initial clustering result to obtain the initial clustering distance;

[0145] Step S304: reallocate the clusters to which each generator set belongs based on the initial cluster distance, and update the cluster center of each cluster;

[0146] Step S306: Based on the updated cluster centers, return to the step of recalculating the Euclidean distances between the cluster centers of each cluster and other generators, until the clustering algorithm converges to obtain a clustering result of abnormal power bidding behavior;

[0147] Step S308, obtaining the cluster center of each cluster and the quotation curve data of other power generation groups;

[0148] Step S310: performing a similarity test on the quotation curves of each cluster of generator sets based on the quotation curve data to obtain a similarity test result;

[0149] Step S312 , based on the clustering results of abnormal power bidding behaviors, the bid retention levels of the power generators in each cluster are calculated to obtain the identification results of the abnormal power bidding behaviors.

[0150] Specifically, k units that fail the behavior impact test are selected and used as the initial cluster centers {λ1,λ2,…,λ k}; the remaining j units are the objects to be clustered, denoted as {e1,e2,…,e j}.

[0151] The Euclidean distance between the cluster center of each cluster and other generators is calculated to evaluate the similarity between each generator and the cluster center. The clusters to which each generator belongs are reallocated based on the Euclidean distance from each generator to the cluster center, and the cluster center of each cluster is updated. The new cluster center calculation method is:

[0152]

[0153] Where: is the new cluster center; W is the number of samples in the cluster; e w is the wth sample in this cluster. Return to the step of recalculating the Euclidean distance between each cluster center and the other generators until the clustering algorithm converges to obtain the clustering results of abnormal power bidding behavior. By reallocating and updating the cluster centers, the clustering results are optimized to better match the actual data distribution.

[0154] The similarity of the unit pair quotation curves is weighted averaged to obtain the similarity of the unit cluster quotation curves. The similarity of the unit pair quotation curves is measured by the area between the two unit quotation curves. The similarity of the quotation curves of the unit pair formed by units A and B is ρ AB Expressed as:

[0155]

[0156] Where: p A (ε) and p B (ε) are the functional expressions of the quotation and quotation ratio curves of units A and B respectively; ε is the quotation ratio of the units.

[0157] The similarity of bidding curves in a cluster overcomes the limitation of analyzing bidding behavior only between two generators. It can now analyze the bidding behavior of more than two generators. The greater the similarity index of a cluster's bidding curves, the more similar the bidding curves of several generators within the cluster are, and the greater the likelihood of collusion among the generators. A reasonable standard for setting this index is (0, 0.8).

[0158] Power producers form a collusion alliance to jointly raise market prices through negotiation. The bids of alliance members vary, but in order to obtain excess profits, the overall bid level of the alliance is always high. Therefore, this paper constructs a unit cluster retention level index to measure the overall bid of the unit cluster. Its calculation formula is:

[0159]

[0160] Where: I m N is the quotation retention level index of the unit cluster; l is the number of units in the unit cluster; is the average bid of the i-th unit in the market; N n is the number of all units in the market. If the unit cluster retention level index is high, it indicates that the overall bidding level of the unit cluster is higher than that of other market members, indicating a collusive alliance. The reasonable range of this index is (0, 1.2), which can be dynamically adjusted by market managers based on market operations.

[0161] In this embodiment, the aforementioned clustering algorithm-based analysis process can systematically identify abnormal power bidding behavior by power generators. Starting from the initial clustering results, this method iteratively optimizes cluster centers and reallocates clusters to ensure the stability and accuracy of the clustering results. Combined with a bid curve similarity test and a bid retention level test, the consistency and rationality of bids by power generators within a cluster can be quantified. Ultimately, this method can effectively identify power generators that may be engaging in collusion or retention, providing a scientific basis for power market regulation, enhancing the scientific nature and reliability of market regulation, and helping to maintain the fairness and efficiency of the power market.

[0162] In an exemplary embodiment, the Euclidean distance is calculated as follows:

[0163] Obtain the total number of price segments of market quotations, as well as the quotes and quantities of each generator set in each price segment;

[0164] Calculate the difference between the cluster center of each cluster and the bid and quantity difference of other generators in each price range;

[0165] The square values ​​of the bid and quantity differences of all price segments are taken, and the square roots are added together to obtain the Euclidean distance between the cluster center and other generators.

[0166] Specifically, calculate the centers of each initial cluster With other units The Euclidean distance D(j,k) between them is:

[0167]

[0168] Where: Y is the total number of price segments quoted in the market; is the quotation of unit j in the yth price segment; is the quotation of unit k in the yth price segment; is the reported quantity of unit j in the yth price segment; is the reported quantity of unit k in the yth price segment.

[0169] In an exemplary embodiment, a price impact test is performed on a target generator set using a preset dispatch model based on the adjusted power bid data, including:

[0170] Based on the adjusted electricity quotation data, calculate the ratio between the original market quotation and the preset benchmark quotation, and use it as the price impact test factor;

[0171] Based on the price impact test factors, the preset scheduling model is used to conduct price impact tests on the target power generation units.

[0172] Specifically, the adjusted electricity bid data is input into the dispatch model. The dispatch model is run to generate new market-clearing results, including the new market-clearing price and the winning bid output of each unit. A price impact test factor is calculated based on the ratio of the original market bid to the preset benchmark bid. By comparing the price impact test factor with a set threshold (e.g., 2), it is determined whether the target generator's bid has had a significant impact on the market price.

[0173] If β>2, it is considered that the target generator's bidding behavior has a significant impact on the market price, and there may be abnormal bidding behavior. If β≤2, it is considered that the target generator's bidding behavior has a small impact on the market price and the bidding behavior is relatively reasonable.

[0174] In this embodiment, by calculating the price impact test factor and using a preset scheduling model to conduct price impact testing on the target generator, it is possible to quantitatively assess the actual impact of the target generator's bidding behavior on market prices. This process not only improves the accuracy of identifying abnormal bidding behavior but also enhances the scientific nature and reliability of market supervision. This method can effectively identify generators that may be engaging in market power manipulation, providing a strong basis for market supervision and maintaining the fairness and efficiency of the electricity market.

[0175] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0176] Based on the same inventive concept, embodiments of the present application also provide a device for identifying abnormal power bidding behavior based on a clustering algorithm, which is used to implement the aforementioned method for identifying abnormal power bidding behavior based on a clustering algorithm. The solution provided by this device is similar to the solution described in the aforementioned method. Therefore, the specific limitations of the embodiments of one or more devices for identifying abnormal power bidding behavior based on a clustering algorithm provided below can be found in the limitations of the method for identifying abnormal power bidding behavior based on a clustering algorithm above, and will not be repeated here.

[0177] In an exemplary embodiment, Figure 4 As shown, a device for identifying abnormal power bidding behavior based on a clustering algorithm is provided, comprising:

[0178] The data collection module 402 is used to collect market electricity price data, perform Lerner index test, excess supply rate index test and behavior impact test on the market electricity price data, and obtain test results;

[0179] The data processing module 404 is used to screen out target generators that fail the test based on the test results, and replace the original bids of the target generators with preset benchmark bids to obtain adjusted power bid data;

[0180] The price impact test module 406 is used to perform a price impact test on the target generator set based on the adjusted power quotation data using a preset scheduling model to obtain an analysis result of the power price retention behavior;

[0181] The bidding behavior identification module 408 is used to divide the generator groups other than the target generator group into different clusters based on the analysis results of the electricity price retention behavior, use the target generator group that failed the test as the cluster center, and obtain the initial clustering results of the abnormal electricity bidding behavior data;

[0182] The bidding behavior identification module 410 is further used to perform similarity tests on the bidding curves of each cluster of generator groups based on the initial clustering results, and to perform retention level tests on the corresponding bids of each cluster of generator groups, to obtain identification results of abnormal power bidding behavior.

[0183] In an exemplary embodiment, the data acquisition module 402 is specifically used to collect the market clearing price and marginal cost data of power producers, calculate the Lerner index, perform the Lerner index test based on the Lerner index, and obtain the quantitative index results of the monopoly power level of market participants; obtain the capacity data of each power generation unit in the market and the total market load demand data, and calculate the surplus supply rate index, perform the surplus supply rate index test based on the surplus supply rate index, and obtain the quantitative index results of the market control power of the power generator; obtain the actual quotation and preset benchmark quotation of the power generation unit; perform the behavior impact test based on the actual quotation and the preset benchmark quotation to obtain the preliminary judgment result of whether there is market power manipulation behavior; perform the price impact test based on the actual quotation and the preset benchmark quotation to obtain the quantitative result of the degree of influence of the actual quotation on the market electricity price.

[0184] In an exemplary embodiment, a scheduling model establishment module is used to obtain the winning output data and cost curve data of each generator set; establish a preset scheduling model based on the bid output data and cost curve data; and determine the constraints of the preset scheduling model, which include system balance constraints, system backup constraints, transmission line transmission capacity constraints, generator set climbing constraints, generator set maximum and minimum output constraints, and line flow constraints.

[0185] In an exemplary embodiment, the bidding behavior identification module 410 is also used to calculate the Euclidean distance between the cluster center of each cluster and other generator groups based on the initial clustering results to obtain the initial clustering distance; based on the initial clustering distance, the cluster to which each generator group belongs is reallocated, and the cluster center of each cluster is updated; based on the updated cluster center, return to the step of recalculating the Euclidean distance between the cluster center of each cluster and other generator groups until the clustering result of abnormal power bidding behavior is obtained by convergence using the clustering algorithm; obtain the bidding curve data of the cluster center of each cluster and other generator groups; based on the bidding curve data, perform a similarity test on the bidding curves of each cluster generator group to obtain a similarity test result; based on the clustering result of abnormal power bidding behavior, calculate the bidding retention level of each cluster generator group to obtain the identification result of abnormal power bidding behavior.

[0186] In an exemplary embodiment, the Euclidean distance is calculated as follows:

[0187] Obtain the total number of price segments of market quotations, as well as the quotes and quantities of each generator set in each price segment;

[0188] Calculate the difference between the cluster center of each cluster and the bid and quantity difference of other generators in each price range;

[0189] The square values ​​of the bid and quantity differences of all price segments are taken, and the square roots are added together to obtain the Euclidean distance between the cluster center and other generators.

[0190] In an exemplary embodiment, the price impact test module 406 is used to calculate the ratio between the original market quotation and the preset benchmark quotation based on the adjusted electricity quotation data, and use it as a price impact test factor; based on the price impact test factor, a price impact test is performed on the target power generation unit using a preset scheduling model.

[0191] Each module in the aforementioned clustering algorithm-based abnormal electricity bidding behavior identification device can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in hardware form, or can be stored in a computer device memory in software form, so that the processor can call and execute the corresponding operations of each module.

[0192] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Figure 5 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O) and a communication interface. The processor, memory and input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store market electricity price data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for identifying abnormal electricity quotation behavior based on a clustering algorithm is implemented.

[0193] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0194] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0195] Collect market electricity price data, conduct Lerner index test, surplus supply rate index test and behavior impact test on the market electricity price data, and obtain test results;

[0196] According to the test results, the target generators that failed the test are screened out, and the original bids of the target generators are replaced with the preset benchmark bids to obtain adjusted power bid data;

[0197] Based on the adjusted power quotation data, the preset dispatch model is used to conduct price impact tests on the target power generation units to obtain the analysis results of the power price retention behavior;

[0198] Based on the analysis results of electricity price retention behavior, the generators other than the target generator are divided into different clusters, and the target generator that fails the test is used as the cluster center to obtain the initial clustering results of abnormal electricity bidding behavior data;

[0199] Based on the initial clustering results, a similarity test is performed on the quotation curves of each cluster of generators, and a retention level test is performed on the corresponding quotations of each cluster of generators to obtain the identification results of abnormal power quotation behavior.

[0200] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0201] Collect market clearing price and marginal cost data of power generators, calculate the Lerner index, and conduct Lerner index test based on the Lerner index to obtain quantitative indicators of the monopoly power level of market participants;

[0202] Obtain the capacity data of each generator set in the market and the total market load demand data, calculate the surplus supply rate index, conduct a surplus supply rate index test based on the surplus supply rate index, and obtain the quantitative indicator results of the market control power of the generator;

[0203] Obtain actual quotes and preset benchmark quotes for generator sets;

[0204] Conduct behavioral impact tests based on actual quotations and preset benchmark quotations to obtain preliminary judgment results on whether there is market power manipulation; conduct price impact tests based on actual quotations and preset benchmark quotations to obtain quantitative results on the degree of impact of actual quotations on market electricity prices.

[0205] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0206] Obtain the winning bid output data and cost curve data of each generator set;

[0207] Establish a preset scheduling model based on the marked force data and cost curve data;

[0208] Determine the constraints of the preset scheduling model, which include system balance constraints, system reserve constraints, transmission line transmission capacity constraints, generator set ramp constraints, generator set maximum and minimum output constraints, and line flow constraints.

[0209] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0210] According to the initial clustering results, the Euclidean distance between the cluster center of each cluster and other generator groups is calculated to obtain the initial clustering distance;

[0211] According to the initial clustering distance, the clusters to which each generator group belongs are reallocated, and the cluster center of each cluster is updated;

[0212] Based on the updated cluster center, return to the step of recalculating the Euclidean distance between the cluster center of each cluster and other generator groups until the clustering algorithm converges to obtain the clustering result of abnormal power bidding behavior;

[0213] Obtain the cluster center of each cluster and the quotation curve data of other generators;

[0214] Based on the quotation curve data, a similarity test is performed on the quotation curves of each cluster of generator sets to obtain a similarity test result;

[0215] According to the clustering results of abnormal power bidding behaviors, the bid retention levels of the generators in each cluster are calculated, and the identification results of abnormal power bidding behaviors are obtained.

[0216] In one embodiment, the Euclidean distance is calculated as follows:

[0217] Obtain the total number of price segments of market quotations, as well as the quotes and quantities of each generator set in each price segment;

[0218] Calculate the difference between the cluster center of each cluster and the bid and quantity difference of other generators in each price range;

[0219] The square values ​​of the bid and quantity differences of all price segments are taken, and the square roots are added together to obtain the Euclidean distance between the cluster center and other generators.

[0220] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0221] Based on the adjusted electricity quotation data, calculate the ratio between the original market quotation and the preset benchmark quotation, and use it as the price impact test factor;

[0222] Based on the price impact test factors, the preset scheduling model is used to conduct price impact tests on the target power generation units.

[0223] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0224] Collect market electricity price data, conduct Lerner index test, surplus supply rate index test and behavior impact test on the market electricity price data, and obtain test results;

[0225] According to the test results, the target generators that failed the test are screened out, and the original bids of the target generators are replaced with the preset benchmark bids to obtain adjusted power bid data;

[0226] Based on the adjusted power quotation data, the preset dispatch model is used to conduct price impact tests on the target power generation units to obtain the analysis results of the power price retention behavior;

[0227] Based on the analysis results of electricity price retention behavior, the generators other than the target generator are divided into different clusters, and the target generator that fails the test is used as the cluster center to obtain the initial clustering results of abnormal electricity bidding behavior data;

[0228] Based on the initial clustering results, a similarity test is performed on the quotation curves of each cluster of generators, and a retention level test is performed on the corresponding quotations of each cluster of generators to obtain the identification results of abnormal power quotation behavior.

[0229] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0230] Collect market clearing price and marginal cost data of power generators, calculate the Lerner index, and conduct Lerner index test based on the Lerner index to obtain quantitative indicators of the monopoly power level of market participants;

[0231] Obtain the capacity data of each generator set in the market and the total market load demand data, calculate the surplus supply rate index, conduct a surplus supply rate index test based on the surplus supply rate index, and obtain the quantitative indicator results of the market control power of the generator;

[0232] Obtain actual quotes and preset benchmark quotes for generator sets;

[0233] Conduct behavioral impact tests based on actual quotations and preset benchmark quotations to obtain preliminary judgment results on whether there is market power manipulation; conduct price impact tests based on actual quotations and preset benchmark quotations to obtain quantitative results on the degree of impact of actual quotations on market electricity prices.

[0234] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0235] Obtain the winning bid output data and cost curve data of each generator set;

[0236] Establish a preset scheduling model based on the marked force data and cost curve data;

[0237] Determine the constraints of the preset scheduling model, which include system balance constraints, system reserve constraints, transmission line transmission capacity constraints, generator set ramp constraints, generator set maximum and minimum output constraints, and line flow constraints.

[0238] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0239] According to the initial clustering results, the Euclidean distance between the cluster center of each cluster and other generator groups is calculated to obtain the initial clustering distance;

[0240] According to the initial clustering distance, the clusters to which each generator group belongs are reallocated, and the cluster center of each cluster is updated;

[0241] Based on the updated cluster center, return to the step of recalculating the Euclidean distance between the cluster center of each cluster and other generator groups until the clustering algorithm converges to obtain the clustering result of abnormal power bidding behavior;

[0242] Obtain the cluster center of each cluster and the quotation curve data of other generators;

[0243] Based on the quotation curve data, a similarity test is performed on the quotation curves of each cluster of generator sets to obtain a similarity test result;

[0244] According to the clustering results of abnormal power bidding behaviors, the bid retention levels of the generators in each cluster are calculated, and the identification results of abnormal power bidding behaviors are obtained.

[0245] In one embodiment, the Euclidean distance is calculated as follows:

[0246] Obtain the total number of price segments of market quotations, as well as the quotes and quantities of each generator set in each price segment;

[0247] Calculate the difference between the cluster center of each cluster and the bid and quantity difference of other generators in each price range;

[0248] The square values ​​of the bid and quantity differences of all price segments are taken, and the square roots are added together to obtain the Euclidean distance between the cluster center and other generators.

[0249] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0250] Based on the adjusted electricity quotation data, calculate the ratio between the original market quotation and the preset benchmark quotation, and use it as the price impact test factor;

[0251] Based on the price impact test factors, the preset scheduling model is used to conduct price impact tests on the target power generation units.

[0252] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0253] Collect market electricity price data, conduct Lerner index test, surplus supply rate index test and behavior impact test on the market electricity price data, and obtain test results;

[0254] According to the test results, the target generators that failed the test are screened out, and the original bids of the target generators are replaced with the preset benchmark bids to obtain adjusted power bid data;

[0255] Based on the adjusted power quotation data, the preset dispatch model is used to conduct price impact tests on the target power generation units to obtain the analysis results of the power price retention behavior;

[0256] Based on the analysis results of electricity price retention behavior, the generators other than the target generator are divided into different clusters, and the target generator that fails the test is used as the cluster center to obtain the initial clustering results of abnormal electricity bidding behavior data;

[0257] Based on the initial clustering results, a similarity test is performed on the quotation curves of each cluster of generators, and a retention level test is performed on the corresponding quotations of each cluster of generators to obtain the identification results of abnormal power quotation behavior.

[0258] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0259] Collect market clearing price and marginal cost data of power generators, calculate the Lerner index, and conduct Lerner index test based on the Lerner index to obtain quantitative indicators of the monopoly power level of market participants;

[0260] Obtain the capacity data of each generator set in the market and the total market load demand data, calculate the surplus supply rate index, conduct a surplus supply rate index test based on the surplus supply rate index, and obtain the quantitative indicator results of the market control power of the generator;

[0261] Obtain actual quotes and preset benchmark quotes for generator sets;

[0262] Conduct behavioral impact tests based on actual quotations and preset benchmark quotations to obtain preliminary judgment results on whether there is market power manipulation; conduct price impact tests based on actual quotations and preset benchmark quotations to obtain quantitative results on the degree of impact of actual quotations on market electricity prices.

[0263] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0264] Obtain the winning bid output data and cost curve data of each generator set;

[0265] Establish a preset scheduling model based on the marked force data and cost curve data;

[0266] Determine the constraints of the preset scheduling model, which include system balance constraints, system reserve constraints, transmission line transmission capacity constraints, generator set ramp constraints, generator set maximum and minimum output constraints, and line flow constraints.

[0267] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0268] According to the initial clustering results, the Euclidean distance between the cluster center of each cluster and other generator groups is calculated to obtain the initial clustering distance;

[0269] According to the initial clustering distance, the clusters to which each generator group belongs are reallocated, and the cluster center of each cluster is updated;

[0270] Based on the updated cluster center, return to the step of recalculating the Euclidean distance between the cluster center of each cluster and other generator groups until the clustering algorithm converges to obtain the clustering result of abnormal power bidding behavior;

[0271] Obtain the cluster center of each cluster and the quotation curve data of other generators;

[0272] Based on the quotation curve data, a similarity test is performed on the quotation curves of each cluster of generator sets to obtain a similarity test result;

[0273] According to the clustering results of abnormal power bidding behaviors, the bid retention levels of the generators in each cluster are calculated, and the identification results of abnormal power bidding behaviors are obtained.

[0274] In one embodiment, the Euclidean distance is calculated as follows:

[0275] Obtain the total number of price segments of market quotations, as well as the quotes and quantities of each generator set in each price segment;

[0276] Calculate the difference between the cluster center of each cluster and the bid and quantity difference of other generators in each price range;

[0277] The square values ​​of the bid and quantity differences of all price segments are taken, and the square roots are added together to obtain the Euclidean distance between the cluster center and other generators.

[0278] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0279] Based on the adjusted electricity quotation data, calculate the ratio between the original market quotation and the preset benchmark quotation, and use it as the price impact test factor;

[0280] Based on the price impact test factors, the preset scheduling model is used to conduct price impact tests on the target power generation units.

[0281] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0282] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile memory and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processor involved in the various embodiments provided herein may be, but are not limited to, a general-purpose processor, a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processor (DSP), a programmable logic unit (PLC), a data processing logic unit based on quantum computing, an artificial intelligence (AI) processor, and the like.

[0283] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0284] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A method for identifying abnormal power bidding behavior based on clustering algorithm, characterized in that: The method comprises: collecting market electricity price data, performing a Lerner index test, a surplus supply rate index test, and a behavior impact test on the market electricity price data, and obtaining test results; According to the test results, the target generators that failed the test are screened out, and the original bids of the target generators are replaced with the preset benchmark bids to obtain adjusted power bid data; Based on the adjusted power quotation data, the preset dispatch model is used to conduct price impact tests on the target power generation units to obtain the analysis results of the power price retention behavior; Based on the analysis results of the electricity price retention behavior, the generator groups other than the target generator group are divided into different clusters, and the target generator group that fails the test is used as the cluster center to obtain the initial clustering results of the abnormal electricity bidding behavior data; Based on the initial clustering results, a similarity test is performed on the quotation curves of each cluster of generators, and a retention level test is performed on the corresponding quotations of each cluster of generators to obtain the identification results of abnormal power quotation behavior.

2. The method according to claim 1, characterized in that The market electricity price data is collected, and the Lerner index test, the excess supply rate index test, and the behavior impact test are performed on the market electricity price data to obtain test results, including: Collecting market clearing price and marginal cost data of power producers and calculating the Lerner index, and conducting a Lerner index test based on the Lerner index to obtain quantitative indicators of the monopoly power level of market participants; Obtaining capacity data of each generator set in the market and total market load demand data, and calculating a surplus supply rate index, performing a surplus supply rate index test based on the surplus supply rate index, and obtaining a quantitative indicator result of the market control power of the generator; Obtain actual quotes and preset benchmark quotes for generator sets; Conduct behavioral impact tests based on actual quotations and preset benchmark quotations to obtain preliminary judgment results on whether there is market power manipulation; conduct price impact tests based on actual quotations and preset benchmark quotations to obtain quantitative results on the degree of impact of actual quotations on market electricity prices.

3. The method according to claim 1, characterized in that Before performing the price impact test on the target generator set using the preset scheduling model, the method further includes: Obtain the winning bid output data and cost curve data of each generator set; Establishing a preset scheduling model based on the marked output data and the cost curve data; Determine the constraints of the preset scheduling model, wherein the constraints include system balance constraints, system reserve constraints, transmission line transmission capacity constraints, generator set ramp constraints, generator set maximum and minimum output constraints, and line power flow constraints.

4. The method according to claim 1, wherein Based on the initial clustering results, similarity tests are performed on the bid curves of the generator groups in each cluster, and retention level tests are performed on the bids of the generator groups in each cluster to obtain the identification results of abnormal power bid behavior, including: According to the initial clustering results, the Euclidean distance between the cluster center of each cluster and other generator groups is calculated to obtain the initial clustering distance; According to the initial clustering distance, the clusters to which each generator group belongs are reallocated, and the cluster center of each cluster is updated; Based on the updated cluster center, return to the step of recalculating the Euclidean distance between the cluster center of each cluster and other generator groups until the clustering algorithm converges to obtain the clustering result of abnormal power bidding behavior; Obtain the cluster center of each cluster and the quotation curve data of other generators; Performing a similarity test on the quotation curves of each cluster of generator sets based on the quotation curve data to obtain a similarity test result; According to the clustering results of abnormal power bidding behaviors, the bid retention levels of the generators in each cluster are calculated, and the identification results of abnormal power bidding behaviors are obtained.

5. The method according to claim 4, characterized in that The calculation process of the Euclidean distance is: Obtain the total number of price segments of market quotations, as well as the quotes and quantities of each generator set in each price segment; Calculate the difference between the cluster center of each cluster and the bid and quantity difference of other generators in each price range; The square values ​​of the bid and quantity differences of all price segments are taken, and the square roots are added together to obtain the Euclidean distance between the cluster center and other generators.

6. The method according to claim 2, characterized in that The price impact test of the target generator set is performed using a preset dispatch model based on the adjusted power quotation data, including: Based on the adjusted electricity quotation data, calculate the ratio between the original market quotation and the preset benchmark quotation, and use it as the price impact test factor; Based on the price impact test factor, a price impact test is performed on the target power generation unit using a preset scheduling model.

7. A device for identifying abnormal power bidding behavior based on clustering algorithm, characterized in that: The device comprises: a data acquisition module, configured to collect market electricity price data, perform a Lerner index test, a surplus supply rate index test, and a behavior impact test on the market electricity price data, and obtain test results; A data processing module is used to screen out target generator sets that fail the test based on the test results, replace the original bids of the target generator sets with preset benchmark bids, and obtain adjusted power bid data; The price impact test module is used to conduct price impact tests on target generators based on the adjusted power quotation data using a preset dispatch model to obtain analysis results on power price retention behavior; a bidding behavior identification module, configured to divide the generator sets other than the target generator set into different clusters based on the analysis results of the electricity price retention behavior, use the target generator set that failed the test as the cluster center, and obtain an initial clustering result of the abnormal electricity bidding behavior data; The quotation behavior identification module is also used to perform similarity tests on the quotation curves of each cluster of generator sets based on the initial clustering results, and to perform retention level tests on the corresponding quotations of each cluster of generator sets, so as to obtain the identification results of abnormal power quotation behavior.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images