Power system transaction simulation method based on multi-agent proxy

Abstract

The invention provides an electric power system transaction simulation method based on multi-agent agency, and belongs to the technical field of electric power system simulation, and the method comprises the steps: collecting the historical operation data of an electric power system, constructing a multi-agent system comprising a power generation agent, a flexible load agent and an electricity selling agent, and carrying out the Monte Carlo simulation and scene generation. According to the method, a deep reinforcement learning algorithm is adopted, a revenue function and a decision model of each agent are defined, market clearing calculation is executed, and transaction simulation is performed by constructing a verification scene library. The system can evaluate agent strategies, calculate market key indexes and feed market information back to agents, intelligence and dynamic optimization of electricity transaction decision are achieved, and the operation efficiency and transaction accuracy of the electricity market are effectively improved. The method provided by the invention solves the technical problem in the prior art that the interaction and learning mechanism between market participants are not fully considered, so that the simulation result and the actual market performance have large deviation.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system simulation, and in particular, relates to a power system transaction simulation method based on multi-agent agents. Background Art

[0002] With the deepening of electricity market reform and the large-scale access of new resources such as new energy, energy storage, and demand response, the transaction form of the power system has become highly complex.

[0003] With its distributed and adaptive characteristics, the intelligent system has become an ideal tool for simulating the power market. It can simulate the behavior and decision-making of market participants such as power generation companies and large users. Intelligent agent technology has been successfully applied in power generation company quotation, demand side management and reliability management in power market transactions. Intelligent algorithms, such as Q-learning and DDPG, are used to simulate the self-learning and optimal decision-making process of market participants to adapt to the dynamic changes of the market. Power market simulation platforms, such as ElecSim, provide functions such as simulated clearing and intelligent quotation, support simulation of multiple market models, help market players understand the market system, and build an intelligent transaction decision-making support system. The development of these technologies has provided a solid theoretical and technical foundation for power market simulation and promoted the process of power market reform.

[0004] Traditional power market transaction simulation methods are mainly based on game theory and equilibrium theory. These methods usually assume that market participants are completely rational and have complete information. However, in the actual power market, participants often face constraints such as incomplete information and limited rationality, which makes it difficult for traditional methods to accurately simulate real market behavior. At present, scholars at home and abroad have conducted extensive research on power market transaction modeling methods. One type of method constructs the optimal bidding strategy of market participants based on Nash equilibrium theory, such as proposing a clearing model that considers unit ramping constraints, or establishing a market equilibrium model that considers transmission constraints; another type of method studies the dynamic game process of market participants based on evolutionary game theory, such as analyzing the evolutionary stability of power producers' bidding strategies. However, these methods generally have the technical problem of not fully considering the interaction and learning mechanism between market participants, resulting in a large deviation between the simulation results and the actual market performance. Summary of the invention

[0005] In view of this, the present invention provides a power system transaction simulation method based on multi-agent agents, which can solve the technical problem that the existing technology generally does not fully consider the interaction and learning mechanism between market participants, resulting in a large deviation between the simulation results and the actual market performance.

[0006] The present invention is implemented as follows: The present invention provides a multi-agent agent-based power system transaction simulation method, comprising the following steps: S01. Collecting historical operation data of the power system, wherein the historical operation data includes power generation output data, load demand data, electricity price data, and network topology data; S02. Build an electricity trading environment, and set the number of market participants, the number of quotation periods, the number of capacity segments, contract data, the upper and lower quotation limits based on the historical operation data; S03, performing Monte Carlo simulation to generate a wind power output scenario set, a load demand scenario set, and a market price scenario set, and verifying the scenario set through a probability density distribution test; S04. Establish a multi-agent system, wherein the multi-agent system includes a power generation agent, a flexible load agent, and a power sales agent, and defines the state space, action space, strategy space, and reward function of each agent based on a partially observable Markov decision process; S05, defining the profit function of the power generation agent, dividing the power generation agent into a thermal power generation agent, a hydropower generation agent, and a wind power generation agent according to the power generation type, and using the historical operation data to train the cost coefficient of each power generation agent; S06. Construct a demand response model of a flexible load intelligent agent, and optimize it by solving a flexible load optimization equation group, wherein the flexible load optimization equation group includes an objective function equation, a load balance equation, a comfort constraint equation, and a transfer constraint equation; S07. Calculate the input parameters of the flexible load optimization equation group through regression analysis, including reduction compensation cost coefficient, transfer scheduling cost coefficient, power purchase and sale cost coefficient, user discomfort cost coefficient, load adjustment upper limit value, load adjustment lower limit value, and user satisfaction threshold value; S08. Establishing a joint decision-making model for power purchase and sales of the power sales intelligent entity, wherein the joint decision-making model includes a spot market quotation strategy and a retail market pricing strategy, and sets a minimum power value and a maximum power value based on the historical operation data; S09. Execute market clearing calculations, perform unconstrained clearing and constrained clearing based on the principle of supply and demand balance, and obtain market clearing electricity prices and market clearing electricity quantities; S10, using a deep reinforcement learning algorithm to train the multi-agent system, using the market-clearing electricity price and the market-clearing electricity quantity as training data, and updating the strategy parameters of each agent; S11, constructing a verification scenario library, and generating test samples based on the wind power output scenario set, the load demand scenario set, and the market price scenario set; S12, executing transaction simulation in the verification scenario library, evaluating the strategy parameters of the multi-agent system, and calculating the total market transaction volume, node power value, agent income value, agent profit value, and agent market share value; S13. Publish market information, and feed back the total market transaction volume, node electricity value, agent revenue value, agent profit value, and agent market share value to the multi-agent system for the next round of transaction decision optimization.

[0007] On the basis of the above technical solution, the power system transaction simulation method based on multi-agent agent of the present invention can also be improved as follows: Among them, the objective function equation is used to minimize the total system cost, and the input includes reduction compensation cost, transfer scheduling cost, electricity purchase and sales cost, and user discomfort cost, and the output is the optimal load response plan. The load balance equation is used to ensure the balance of power supply and demand. The input includes the total power on the demand side, renewable energy output, grid input power, and energy storage system output. The output is the power balance state of each time period. The comfort constraint equation is used to ensure the user's electricity experience. The input includes the user's original electricity demand, load adjustment amount, and user satisfaction threshold. The output is the load adjustment range that meets the user's comfort requirements. The transfer constraint equation is used to control the rationality of load transfer. The input includes the upper and lower limits of the load allowed to be transferred in and out in each time period, and the original load curve. The output is a load scheduling plan that meets the transfer constraints.

[0008] Furthermore, the time period is set as follows: one day is set to 96 time periods, corresponding to one time period every 15 minutes. The number of capacity segments is 3 to 5 segments, and each capacity segment corresponds to a different quotation strategy. For the quotation upper limit and quotation lower limit, the quotation upper and lower limits are set according to the historical transaction price, the upper limit is 1.2 times the historical highest price, and the lower limit is 0.8 times the historical lowest price. The user satisfaction threshold is set to 0.8. The upper and lower limits of the load adjustment are set to plus or minus 30% of the rated load.

[0009] Furthermore, in step S03, considering the correlation between wind power, load and electricity price, the scenario generation process is specifically expressed as follows: Wind power output scenario generation equation: ; In the formula, For the scene In the period Wind power output value (MW); For the period The predicted wind power output value (MW); For the period The standard deviation of the prediction error (MW); is a random variable that follows a standard normal distribution; is the correlation coefficient between wind power output and load; is the correlation coefficient between wind power output and electricity price; is the load demand; For electricity price.

[0010] Probability constraints: ; In the formula, is the allowable prediction error range (MW); is the confidence level, which is generally taken as 0.95.

[0011] In step S05, the cost function of the power generation agent takes into account factors such as the start and stop status of the unit and the climbing constraint: Cost function of thermal power generation agent: ; In the formula, The start / stop status of the unit (0 / 1); is the startup cost (yuan); is the continuous running time (h); is the efficiency reduction coefficient; is the output change (MW); is the climbing cost coefficient.

[0012] Constraints: ; ; In the formula, They are upper and lower ramp rate limits (MW / h) respectively.

[0013] In step S06, the flexible load optimization equations consider the mutual influence between loads: Objective function equation: ; In the formula, For load The interaction coefficient of ; For load In the period Adjustment amount (MW); is the discomfort sensitivity coefficient.

[0014] Load balance equation: ; In the formula, are the efficiency functions of renewable energy, power grid, and energy storage, respectively; It is the charge state of the energy storage system.

[0015] In step S08, the decision model of the power selling agent considers market competition and user response: Quotation strategy: ; In the formula, For competitors Impact factor; For competitors market share; For adjustable load capacity; is the total load capacity.

[0016] Retail market pricing strategy: ; In the formula, is the current electricity demand density; is the reference demand density.

[0017] Principal component constraints: ; ; In the formula, is the orthogonal transformation matrix; Score the principal components.

[0018] In step S09, the market clearing process takes into account network constraints: Objective function: ; In the formula, is the quote variance.

[0019] Power flow constraints: ; ; In the formula, For Node Power flow between is the node admittance; is the node phase angle; are the node power generation and load respectively.

[0020] Probability constraints: ; In the formula, is the line transmission capacity; is the confidence level.

[0021] Variable coupling constraints: ; In the formula, is a coupling matrix that describes the mutual influence of power generation, electricity price and load.

[0022] The variables or functions involved in each equation are explained in detail below.

[0023] 1. In the wind power output scenario generation equation: Correlation coefficient How to obtain: ; In the formula, is the historical data sample point; is the average value of historical data; is the sample size.

[0024] Confidence Level Determination of: Calculate the probability distribution function of forecast error based on historical data statistical analysis , so that .

[0025] 2. In the cost function of thermal power generation agent: Efficiency reduction coefficient How to obtain: ; In the formula, for The power generation efficiency at all times; is the rated power generation efficiency; The statistical period.

[0026] Climbing cost coefficient Determination of: ; In the formula, Additional costs incurred for ramping; is the output change; For rated output.

[0027] 3. In the flexible load optimization equation: Interaction coefficient How to obtain: ; In the formula, is the cross elasticity of the load response; For rated load.

[0028] Efficiency function The expression: ; ; ; In the formula, is the rated efficiency; is the efficiency reduction coefficient; is the rated power; is the reference state of charge.

[0029] 4. In the decision-making model of power sales agent: Competitor influence factor How to obtain: ; In the formula, is the sensitivity of market share to electricity prices; Market share of this retailer; Electricity prices for competitors.

[0030] Orthogonal transformation matrix Build: ; ; In the formula, is the covariance between the quote price and the retail price; are the variances of the quoted price and the retail price respectively.

[0031] 5. During the market clearing process: Quote Variance Calculation: ; In the formula, For historical quote data; It is the average of historical quotes.

[0032] Coupling Matrix Build: ; The matrix elements are obtained through regression analysis: ; In the formula, is the output variable (electricity generation); are input variables (electricity price, load); For the working point.

[0033] Node admittance Calculation: ; In the formula, is the line reactance, obtained through actual measurement or equipment parameters.

[0034] Compared with the prior art, the beneficial effects of the power system transaction simulation method based on multi-agent agent provided by the present invention are as follows: First, by constructing a scenario generation model that considers the correlation between wind power output, load demand and market price, the coupling relationship of multiple uncertain factors is accurately characterized. Second, by introducing dynamic characteristics such as unit efficiency decay and ramp cost, a more realistic power generation agent cost function is established. Third, by designing an optimization equation group that considers the interaction between loads, accurate modeling of flexible load agents is achieved. Fourth, by constructing a power sales agent decision-making model based on market competition and user response, the rationality of the retail market pricing strategy is improved. Fifth, by introducing probability constraints and principal component constraints, the reliability of market clearing results is guaranteed. It effectively solves the technical problem that the prior art generally does not fully consider the interaction and learning mechanism between market participants, resulting in a large deviation between the simulation results and the actual market performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A flow chart of the method provided by the present invention; Figure 2 Schematic diagram of the composition of a multi-intelligent agent system in an embodiment of the present invention. DETAILED DESCRIPTION

[0036] like Figure 1 As shown, it is a flow chart of a power system transaction simulation method based on multi-agent agent provided by the present invention. The specific implementation methods of the steps of the present invention are described in detail below: The specific implementation method of step S01 is to collect the power system operation data in real time through the data acquisition system. First, the distributed data collector is used to collect the real-time output data of each power plant. The sampling period is 5 minutes. The data includes active power, reactive power, generator set operation status and other information; secondly, the smart meter is arranged to collect user load data. The sampling period is 15 minutes, and the user's power consumption, power factor, voltage and current and other parameters are recorded; again, the real-time electricity price data of each time period is obtained from the power trading center, including the day-ahead market price, real-time market price, frequency modulation market price, etc.; finally, the topological structure data of the transmission network is collected, including bus node data, line impedance parameters, transformer ratio data, etc. The purpose of this step is to establish a historical database to provide data support for subsequent scenario generation and model training.

[0037] The specific implementation method of step S02 is to initialize the configuration of the power trading environment based on the collected historical data. First, the number of market participants is set according to the number of power generation enterprises, power sales companies, and power users who have participated in the transaction in history, generally ranging from 50 to 200; secondly, the number of quotation periods is set according to the actual trading rules, usually 96 periods, corresponding to one period every 15 minutes; again, the number of capacity segments is set, generally 3 to 5 segments, and each capacity segment corresponds to a different quotation strategy; then, the contract data is set according to the long-term agreement and bilateral contract, including the electricity volume and electricity price information of the monthly contract and annual contract; finally, the upper and lower limits of the quotation are set according to the historical transaction price, generally the upper limit is 1.2 times the historical highest price, and the lower limit is 0.8 times the historical lowest price. The purpose of this step is to build a trading environment that conforms to the actual situation.

[0038] The specific implementation method of step S03 is to use the Latin hypercube sampling method to generate random scenarios. First, the probability distribution space of wind power, user load, and market electricity price is divided into equal probability subspaces, and the number of samples is generally 1000; secondly, considering the correlation between variables, the Nataf transformation is used to convert the related random variables into independent random variables; again, the inverse transformation method is used to generate samples that conform to the marginal distribution; then the chi-square test method is used to test the probability density distribution of the generated scenarios, and the confidence level is set to 0.95; finally, the valid scenarios are screened according to the test results to form the final scenario set. The purpose of this step is to generate a statistically representative scenario sample.

[0039] The specific implementation method of step S04 is to construct a multi-agent system based on a partially observable Markov decision process. First, the state space of each agent is defined, including its own state and observable environmental state. The state vector dimension is generally 10 to 20 dimensions; secondly, the action space is defined, including continuous variables and discrete variables, such as power generation adjustment, load response, electricity price quotation, etc.; thirdly, the strategy space is defined, and the probability distribution of state to action is mapped using a random strategy network; then the reward function is designed, taking into account economic benefits, technical constraints and market rules, and the weight coefficient of the reward function is determined by expert experience. The purpose of this step is to establish a decision model framework for the agent.

[0040] The specific implementation method of step S05 is to establish a benefit function model for different types of power generation agents. First, the fuel cost coefficient of the thermal power generation unit is trained by the polynomial fitting method, taking into account the start-stop cost, ramp cost and efficiency decay; secondly, the opportunity cost model of hydropower generation is established, taking into account reservoir scheduling and cascade optimization; thirdly, the prediction error cost model of wind power generation is constructed, taking into account the penalty cost caused by output uncertainty; and then the least squares method is used to perform regression analysis on the historical operation data of various types of generators to obtain the coefficient value of the cost function. The purpose of this step is to accurately describe the economic characteristics of the power generation agent.

[0041] The specific implementation method of step S06 is to build a flexible load demand response model that takes into account user experience. First, establish the objective function, including the reduction compensation cost, transfer scheduling cost, power purchase and sales cost, and user discomfort cost; secondly, establish power balance constraints to ensure the balance of power generation and consumption in each period; thirdly, establish comfort constraints to set the allowable range of load adjustment according to user types; and then establish load transfer constraints to ensure the conservation of total power before and after load transfer. The purpose of this step is to achieve optimal scheduling of flexible loads.

[0042] The specific implementation method of step S07 is to determine the key parameters of the flexible load optimization equation group through data analysis. First, the reduction compensation cost coefficient is calculated by multivariate regression analysis method, and the general value range is 0.5 to 2 yuan per kilowatt-hour; secondly, the transfer scheduling cost coefficient is calculated based on historical response data, and the general value range is 0.2 to 1 yuan per kilowatt-hour; thirdly, the purchase and sale cost coefficient is determined according to market transaction data, and the real-time electricity price is determined; then the user discomfort cost coefficient is obtained through questionnaire survey, and the general value range is 1 to 5 yuan per kilowatt-hour; finally, the upper and lower limits of the load adjustment are set according to the equipment parameters, which are generally plus or minus 30 percentage points of the rated load, and the user satisfaction threshold is set to 0.8. The purpose of this step is to ensure the practicality of the optimization model.

[0043] The specific implementation method of step S08 is to establish a market transaction decision model for the power sales agent. First, a spot market quotation strategy considering market competition is constructed, and the opponent modeling method is used to predict the quotation behavior of competitors; secondly, a retail market pricing strategy considering user response is established, and a nonlinear response function is used to describe the user's sensitivity to electricity prices; thirdly, the guaranteed electricity value and the capped electricity value are set based on historical data, which are generally 0.8 times and 1.2 times the average electricity price respectively; and then the principal component analysis method is used to establish the coupling relationship between the quotation strategy and the pricing strategy. The purpose of this step is to optimize the transaction decision of the power seller.

[0044] The specific implementation method of step S09 is to perform market clearing calculation based on DC power flow. First, unconstrained clearing calculation is performed, and the initial clearing result is determined according to the price priority principle; secondly, constrained clearing calculation is performed considering network constraints, and the DC power flow optimization problem is solved by the interior point method; again, the line power flow distribution is calculated according to the node injection power to check whether the line transmission capacity constraint is met; then the final clearing electricity price and clearing electricity quantity are determined based on the supply and demand balance principle. The purpose of this step is to obtain a market clearing result that meets physical constraints.

[0045] The specific implementation of step S10 is to train the agent using a deep reinforcement learning algorithm based on a dominant strategy. First, the value network and the policy network are constructed, and a multilayer perceptron is used as a function approximator; secondly, the experience replay mechanism is used to store transaction data, and the replay pool capacity is set to 10,000; thirdly, the policy gradient method is used to update the policy network parameters, and the learning rate is set to 0.001; then the temporal difference algorithm is used to update the value network parameters, and the discount factor is set to 0.95. The purpose of this step is to improve the decision-making ability of the agent.

[0046] The specific implementation method of step S11 is to build a verification scenario library for testing. First, cluster analysis is performed on the wind power output scenario set, and the scenarios are divided into 3 to 5 categories using a Gaussian mixture model; secondly, feature extraction is performed on the load demand scenario set, and the dimension is reduced using the principal component analysis method; thirdly, probability distribution analysis is performed on the market price scenario set, and the price distribution is fitted using the kernel density estimation method; and then test samples are selected based on the typicality and representativeness of the scenario features. The purpose of this step is to build an effective verification data set.

[0047] The specific implementation method of step S12 is to conduct a transaction simulation test in the verification scenario. First, the total market transaction volume is calculated, and the supply and demand matching situation in each period is counted; secondly, the electricity value of each node is calculated, and the spatial distribution characteristics of the price are analyzed; thirdly, the income value of each intelligent agent is calculated, including power generation income, power sales income and demand response income; then, the profit value and market share value of each intelligent agent are calculated to evaluate its competitiveness. The purpose of this step is to verify the effectiveness of the intelligent agent strategy.

[0048] The specific implementation method of step S13 is to establish a market information feedback mechanism. First, the total market transaction volume information is released to all market players for the overall supply and demand situation judgment; secondly, the node power value information is released to the market players in the relevant areas for price signal analysis; thirdly, the agent income value and profit value information are fed back to each agent for strategy evaluation and adjustment; and then the market share value information is released to the regulatory agency for market power monitoring. The purpose of this step is to promote the healthy development of the market.

[0049] In a specific implementation of step S01, data collection involves the following mathematical representation: 1) Mathematical expression of active power collection: , in represents the three-phase active power at time t (MW), Indicates the effective value of line voltage (kV), Indicates the effective value of line current (kA), Indicates the power factor.

[0050] 2) Mathematical expression of load data collection: , in represents the total load at time t (MW), represents the load power of the i-th user (MW), represents the load participation factor, Indicates the total number of users.

[0051] 3) Statistical processing of electricity price data: , in represents the weighted average electricity price at time t (yuan / MWh), represents the electricity price in the ith market (yuan / MWh), Indicates the corresponding transaction volume (MWh), Indicates the number of market types.

[0052] In a specific implementation of step S02, the environment configuration involves the following mathematical representation: 1) Market participant capacity calculation: , in represents the total transaction capacity (MW), represents the capacity of the i-th generator (MW), represents the capacity of the jth load quotient (MW), represents the capacity of the kth electricity retailer (MW).

[0053] 2) Setting of upper and lower limits of quotation: , , in Respectively represent the upper and lower limits of the quotation (yuan / MWh), represents the historical price series, is the adjustment coefficient, Represents the standard deviation function.

[0054] The specific implementation of step S03 also involves the following mathematical model: 1) Latin Hypercube Sampling: , in represents the jth sample point of the i-th variable, represents the inverse function of the marginal distribution, represents the permutation of the jth interval, represents a uniformly distributed random number between [0,1], Indicates the sample size.

[0055] 2) Nataf transformation: , in represents the transformed standard normal variable, represents the inverse function of the standard normal distribution, Represents the marginal distribution function of the original variable.

[0056] 3) Scene probability density test: , in represents the observed frequency, represents the theoretical frequency, Indicates the number of intervals.

[0057] The specific implementation of step S04 also involves the following decision model: 1) State transition probability: , in Indicates the state transition count, Indicates the number of occurrences of state-action pairs.

[0058] 2) Strategy function: , in represents the state-action value function, Represents policy network parameters.

[0059] 3) Reward calculation: , in represents the current period income, Indicates the technical breach penalty, Indicates compliance with market rules.

[0060] The specific implementation of step S05 also involves the following cost model: 1) Calculation of efficiency attenuation coefficient: , in represents the power generation efficiency at time t, Indicates the rated power generation efficiency, Indicates the statistical period.

[0061] 2) Determination of climbing cost coefficient: , in represents the additional cost caused by ramping, Indicates the change in output force, Indicates rated output.

[0062] The specific implementation of step S06 also involves the following optimization model: 1) Load interaction coefficient: , in represents the cross elasticity of the load response, Indicates rated load.

[0063] 2) Equipment efficiency function: , , , in Indicates the rated efficiency, represents the efficiency reduction coefficient, Indicates the reference state of charge.

[0064] The specific implementation of step S07 also involves the following parameter model: 1) Calculation of reduction compensation cost coefficient: , in represents the reduction compensation cost coefficient (yuan / MWh), represents the compensation coefficient, represents the electricity price in time period t (yuan / MWh), Indicates the load reduction (MW).

[0065] 2) Transfer scheduling cost coefficient: , in represents the transfer dispatch cost coefficient (yuan / MWh), represents the adjustment coefficient, Indicates the transition from period t to The load in MW during the period.

[0066] 3) Discomfort cost coefficient: , in represents the discomfort cost coefficient (yuan / MW²), represents the sensitivity coefficient, Represents the user importance weight.

[0067] The specific implementation of step S08 also involves the following decision model: 1) Competitor influence factors: , in represents the sensitivity of market share to electricity prices, Indicates the market share of this retailer. Indicates competitor electricity prices.

[0068] 2) Orthogonal transformation matrix: , , in represents the covariance between the quote and the retail price, Represent the variance of the quoted price and the retail price respectively.

[0069] The specific implementation of step S09 also involves the following clearing model: 1) Quote variance calculation: , in Represents historical quote data, Indicates the average value of historical quotes.

[0070] 2) Coupling matrix construction: , The matrix elements are obtained through regression analysis: , in represents the output variable (electricity generation), represents input variables (electricity price, load), Indicates the working point.

[0071] The specific implementation of step S10 also involves the following learning algorithm: 1) Policy gradient update: , in represents a parameterized strategy, represents the action-value function.

[0072] 2) Value network loss function: , in represents the parameterized value function, Represents the discount factor.

[0073] 3) Experience replay update: , in represents the learning rate, Indicates the batch size, Represents the timing difference error.

[0074] The specific implementation of step S11 sets the following scene model: 1) Gaussian mixture model: , in represents the weight of the kth Gaussian component, denote the mean vector and covariance matrix respectively.

[0075] 2) Principal Component Analysis: , in represents the original data matrix, represents the left and right singular vectors, represents the singular value matrix.

[0076] Steps S12 and S13 mainly involve result evaluation and information release, and also include the following indicator calculations: 1) Market concentration index: , in represents the Herfindahl-Hirschman index, Represents the market share of market player i.

[0077] 2) Price discovery efficiency: , in represents the price discovery efficiency, represents the clearing price, Indicates the reference price, Represents the mean reference price.

[0078] Specifically, the principle of the present invention is as follows: First, in the scene generation link, conditional probability distribution is used to characterize the correlation between wind power, load and electricity price. By introducing the correlation coefficient matrix, the mapping relationship between variables is established, which overcomes the defect of the traditional independent sampling method that ignores the correlation of variables. At the same time, the prediction error range is limited by probability constraints to ensure the rationality of the generated scene. Secondly, in terms of power generation intelligent agent modeling, a cost function considering efficiency decay is constructed based on thermodynamic principles and equipment characteristics. The efficiency decay coefficient is obtained by regression of historical operation data, and the ramp cost coefficient is calculated based on incremental cost. This method better reflects the dynamic characteristics of the unit. Third, in terms of flexible load modeling, a multi-objective optimization equation group is designed based on user utility theory. By introducing the load interaction influence coefficient, the correlation effect between different user groups is described, and the coefficient is calculated by the cross elasticity of the load response. The efficiency function adopts a piecewise nonlinear form to accurately reflect the operating characteristics of the equipment under different working conditions. Fourth, in terms of power sales intelligent agent decision-making, a competition model is constructed based on market microstructure theory. By calculating the sensitivity of market share to electricity price, the competitor influence factor is obtained, and a quantitative description of the market competition situation is achieved. The Logistic function is used to characterize the nonlinear relationship between electricity price and demand, and the orthogonality of decision variables is ensured by principal component analysis. Fifth, in the market clearing link, an optimization model considering network constraints is constructed based on the node admittance matrix. By introducing the coupling matrix, the mapping relationship between power generation, electricity price and load is established. The matrix is ​​obtained through local linearization and regression analysis. Finally, the deep reinforcement learning algorithm is used to train the intelligent agent, and the online learning and policy optimization of the intelligent agent are realized through value function approximation and policy gradient method. This method overcomes the problem of low computational efficiency of traditional optimization algorithms and can adapt to the dynamic changes of the market environment.

[0079] A specific embodiment 1 of the present invention is provided below. The specific implementation of each step in this embodiment 1 is described in detail as follows: The specific implementation of step S01 is to collect power system operation data in real time through a distributed data acquisition system, and use a SCADA system (data acquisition and monitoring control system) to collect power generation unit data in a cycle of every 5 minutes, wherein the active power collection adopts a three-phase power calculation formula , in the parameters It represents the three-phase active power at time t, in megawatts, is the effective value of the line voltage, in kilovolts, is the effective value of the line current, in kiloamperes, The power factor is calculated by collecting load data through smart meters with a sampling period of 15 minutes and using the load aggregation calculation formula Data processing, including represents the total load at time t, in megawatts, represents the load power of the i-th user, represents the load participation factor, Indicates the total number of users. The system obtains real-time electricity price information by subscribing to the data interface of the power trading center and adopts the weighted average calculation formula Handling of multiple market electricity prices, where represents the weighted average electricity price at time t, in RMB per MWh, represents the electricity price in the ith market, Indicates the corresponding transaction volume. It represents the number of market types and collects network topology data, including bus node number, branch impedance parameters, transformer ratio and other information. It uses a relational database to store and manage the data. It performs quality checks on the collected data, including data integrity, validity and consistency checks, removes outliers and completes the data. It uses linear interpolation to process missing data. The purpose of this step is to build a complete historical database to provide data support for subsequent modeling analysis and scenario generation.

[0080] The specific implementation method of step S02 is to initialize the configuration of the power trading environment based on historical data, using the total balance calculation formula Determine the number of market participants, where represents the total transaction capacity, represents the capacity of the i-th generator, represents the capacity of the jth load quotient, It represents the capacity of the kth electricity seller. The number of participants is generally between 50 and 200. According to the trading rules, a day is divided into 96 time periods for quotation. Each period is 15 minutes. The power generation capacity is divided into 3 to 5 quotation segments. The historical price statistics method is used to set the upper and lower limits of the quotation. The upper limit calculation formula is: The lower limit calculation formula is ,in Respectively represent the upper and lower limits of the quotation, represents the historical price series, is the adjustment coefficient, and its general values ​​are 1.2, 0.2, 0.8, 0.2, Represents the standard deviation function. A fixed trading volume is set according to bilateral agreements and long-term contracts. The contract coverage is generally 60 to 80 percent of the total trading volume. The purpose of this step is to build a standardized and orderly trading environment.

[0081] The specific implementation method of step S03 is to use the Monte Carlo method to generate transaction scenarios. First, the Latin hypercube sampling algorithm is used to generate basic scenarios. The scenario generation formula is: ,in represents the jth sample point of the i-th variable, represents the inverse function of the marginal distribution, represents the permutation of the jth interval, Represents a random number uniformly distributed from 0 to 1. Represents the number of samples, and uses Nataf transformation to process variable correlation. The transformation formula is: ,in represents the transformed standard normal variable, represents the inverse function of the standard normal distribution, Represents the marginal distribution function of the original variable. When generating wind power output scenarios, the correlation with load and electricity price is considered. The scenario generation formula is: , where the meaning of each parameter is the same as above. The chi-square test method is used to verify the probability distribution of the scenario. The test statistic calculation formula is: ,in represents the observed frequency, represents the theoretical frequency, represents the number of intervals, and the significance level is set to 0.05. The purpose of this step is to generate representative trading scenarios.

[0082] The specific implementation method of step S04 is to construct a multi-agent system based on a partially observable Markov decision process, define the state space of each agent to include the local observable state and the environment state, and use the state transition probability calculation formula Describes the state evolution, where Indicates the state transition count, Represents the number of occurrences of state-action pairs, defines a continuous action space to describe power generation regulation, load response and electricity price quotation, and adopts a random strategy based on a strategy network. The strategy function is ,in represents the state-action-value function, Represent the policy network parameters and design the comprehensive reward function ,in represents the current period income, represents the technical breach penalty, Indicates market rule compliance, weight coefficient Take 0.5, 0.3, and 0.2 respectively. The purpose of this step is to establish the decision-making model framework of the intelligent agent.

[0083] The specific implementation method of step S05 is to establish a profit function model for different types of power generation agents, taking into account the dynamic characteristics and constraints of the power generation unit. For thermal power generation units, a cost function that considers the start-stop state and climbing constraints is adopted. ,in represents the power output at time t, Indicates the start and stop status of the unit at time t, represents the fuel cost coefficient, obtained by least squares fitting, represents the startup cost, Indicates the efficiency attenuation coefficient, using the efficiency statistical formula Calculate, where represents the power generation efficiency at time t, Indicates rated power generation efficiency, climbing cost coefficient Using the cost analysis formula Determine that for hydropower generating units, establish an opportunity cost model considering reservoir operation ,in Indicates hydropower output. Represents the water storage capacity of the reservoir. represents the maximum storage capacity, Represents the opportunity cost coefficient. For wind turbines, a prediction error cost model is constructed. ,in Indicates the actual wind power output, represents the wind power forecast value, represents the penalty cost coefficient. The operating constraints of each generator set include upper and lower limits of output, ramp rate, and minimum start and stop time. The purpose of this step is to establish an accurate power generation cost model.

[0084] The specific implementation method of step S06 is to construct a flexible load demand response model that takes into account user experience, using a multi-objective optimization method, and the objective function is ,in Indicates the amount of load reduction, represents the load transfer amount, Indicates the amount of electricity purchased. Represent the reduction, transfer and electricity purchase cost coefficients respectively, represents the discomfort sensitivity coefficient, and the load balance constraint is ,in Indicates the actual load, represents the base load, They represent the incoming and outgoing loads respectively. The equipment efficiency constraint is described by a piecewise function. The renewable energy efficiency function is: , the grid efficiency function is , the energy storage efficiency function is , where the meaning of each parameter is the same as above. The purpose of this step is to achieve optimal scheduling of load.

[0085] The specific implementation method of step S07 is to determine the key parameters of the flexible load optimization equation group through data analysis, and use multivariate regression analysis to calculate the reduction compensation cost coefficient. The calculation formula is: ,in It represents the compensation coefficient, which is generally between 1.2 and 1.5. The formula for calculating the transfer scheduling cost coefficient is: ,in It represents the adjustment coefficient, which is generally between 0.8 and 1.2. The calculation formula of the discomfort cost coefficient is: ,in It represents the sensitivity coefficient, which is generally set at 1.5 to 2.5. The upper and lower limits of the load adjustment are set to plus or minus 30 percentage points of the rated load, and the user satisfaction threshold is set to 0.8. The purpose of this step is to determine the parameter values ​​of the optimization model.

[0086] The specific implementation method of step S08 is to establish a market transaction decision model for the power sales agent. The quotation strategy adopts a dynamic pricing method that takes into account market competition. The quotation function is ,in Indicates the quoted value. represents the base electricity price, Indicates the quotation adjustment coefficient and competitor influence factor The elastic analysis method is used for calculation, and the calculation formula is: ,in Represents the sensitivity of market share to electricity price. The retail market adopts a dynamic pricing strategy based on user demand. The pricing function is ,in represents the current electricity demand density, represents the reference demand density, It represents the price response coefficient, which is generally between 2 and 5. The principal component analysis method is used to establish the coupling relationship between the quotation strategy and the pricing strategy. The coupling matrix is: ,in , represents the covariance between the bid price and the retail price. The purpose of this step is to optimize the transaction strategy of the retailer.

[0087] The specific implementation method of step S09 is to perform market clearing calculations taking into account network constraints, using a social welfare maximization method based on DC power flow, with the objective function being ,in It represents the quote variance, and the calculation formula is: , the power flow constraints include the power balance equation and the node power injection equation ,in represents the node admittance, represents the node phase angle, They represent the node power generation and load respectively. The interior point method is used to solve the optimization problem. The convergence accuracy is set to 0.001 and the maximum number of iterations is 1000. The purpose of this step is to obtain the market clearing result that meets the physical constraints.

[0088] The specific implementation method of step S10 is to use a deep reinforcement learning algorithm to train the agent and use a policy gradient method to update the policy network parameters. The gradient calculation formula is: ,in represents a parameterized strategy, represents the action value function, the value network is updated using the temporal difference algorithm, and the loss function is ,in represents the parameterized value function, Represents the discount factor, with a value of 0.95. The experience replay mechanism is used to store transaction data. The update formula is: ,in represents the learning rate, with a value of 0.001. Represents the batch size, which is set to 32. The purpose of this step is to improve the decision-making ability of the agent.

[0089] The specific implementation method of step S11 is to build a verification scenario library, cluster the wind power output scenarios using a Gaussian mixture model, and the probability density function is ,in represents the weight of the kth Gaussian component, Represent the mean vector and covariance matrix respectively. The maximum expectation algorithm is used for parameter estimation. The principal component analysis is used to reduce the dimension of the load demand scenario. The matrix decomposition form is: ,in represents the original data matrix, Represents the left and right singular vectors, selects the principal components with cumulative variance contribution greater than 85%, and uses kernel density estimation to fit the probability distribution of the market price scenario. The purpose of this step is to construct a valid validation data set.

[0090] The specific implementation method of step S12 is to conduct a transaction simulation test in a verification scenario, calculate the market concentration index, and use the Herfindahl Hirschman Index. The calculation formula is: ,in It represents the market share of market player i and evaluates the price discovery efficiency. The calculation formula is: ,in represents the price discovery efficiency, represents the clearing price, Represents the reference price, calculates the revenue value, profit value and market share value of the agent, and evaluates its competitiveness. The purpose of this step is to verify the effectiveness of the agent's strategy.

[0091] The specific implementation method of step S13 is to establish a market information feedback mechanism, classify and organize information such as total market transaction volume, node electricity price, intelligent body income and market share, use real-time database for storage and release, set the information update cycle to 5 minutes, ensure data security through encrypted transmission, set the information access level according to the authority of the market subject, and realize the hierarchical management of market information. The purpose of this step is to promote the healthy development of the market. Through the organic combination of the above steps, a complete power system transaction simulation method is constructed, which realizes the intelligent decision-making of market subjects and the effective operation of the market.

[0092] The following is an example 2 of a specific application scenario of the present invention: This example takes a provincial power market as an example to illustrate the specific application process of the multi-agent-based power system transaction simulation method. The provincial power market includes 75 power generators, 120 power sales companies and 3,000 large users, with a total installed capacity of 85,000 MW, including 45,000 MW of thermal power units, 25,000 MW of hydropower units, and 15,000 MW of wind power units, with an annual electricity demand of 380 billion kWh.

[0093] First, the data acquisition system is used to collect the operation data of the power system. The data collected from the power generation units show that the average utilization hours of the units are 4,500 hours, the equivalent availability factor of the thermal power units is 0.92, the equivalent availability factor of the hydropower units is 0.98, and the average capacity factor of the wind power units is 0.25. The dynamic characteristics data of the generator units are shown in Table 1.

[0094] Table 1: Dynamic characteristic parameters of generator sets

[0095] Data collected by smart meters show that industrial users account for 65 percent of the total electricity consumption, commercial users account for 25 percent, and residential users account for 10 percent. The user load characteristic data is shown in Table 2.

[0096] Table 2: User load characteristic data table

[0097] Historical price statistics show that the current average market price is 485 yuan per MWh, the real-time average market price is 520 yuan per MWh, and the frequency regulation market average price is 580 yuan per MWh. The price fluctuation characteristics are shown in Table 3.

[0098] Table 3: Market price fluctuation characteristics

[0099] Next, we will build the power trading environment. The total number of market participants is set to 200, including 75 power generation companies, 120 power sales companies, and 5 large user participants. The day is divided into 96 time periods for quotation, and the power generation capacity is divided into 3 quotation segments, namely base load segment, intermediate load segment and peak load segment, with capacity accounting for 60%, 30% and 10% respectively. The upper limit of the quotation is set to 850 yuan per megawatt-hour, and the lower limit is set to 280 yuan per megawatt-hour. The contract power coverage rate is 70%.

[0100] The Monte Carlo method was used to generate trading scenarios. Latin hypercube sampling was used to generate 1,000 basic scenarios, considering that the correlation coefficient between wind power output and load demand was 0.65, and the correlation coefficient with market electricity prices was 0.45. Through the probability density distribution test, 300 valid scenarios were selected for simulation analysis.

[0101] Construct a multi-agent system. The state space dimension of the power generation agent is 15, including information such as unit output, start and stop status, and operating time. The action space dimension is 4, including output adjustment and quotation adjustment. The state space dimension of the flexible load agent is 10, including information such as load power, electricity price signal, and comfort. The action space dimension is 3, including load reduction, transfer, and response time. The state space dimension of the power sales agent is 12, including information such as contract power, market share, and competitor prices. The action space dimension is 2, including wholesale market quotations and retail market pricing. In this embodiment, it is set that all transactions are implemented in the form of centralized bidding and clearing. In the centralized bidding market, each agent selects a bidding strategy with the goal of maximizing its own interests, and submits the selected bidding strategy to the trading center. The trading center processes the bidding and returns the clearing results to the agent. Then the agent determines the next trading strategy based on the clearing results. The constructed multi-agent system is as follows: Figure 2As shown in the figure, the system master control module is the back-end support module, which can simulate different network topologies and set the market environment according to the needs of simulation training. The environment settings include: the number of quotation periods and capacity segments of the power generation party, the number of quotation periods and capacity segments of the power purchase party, the number of market participants, contract data, and the upper and lower limits of the quotation. It can simulate the unilateral quotation or bilateral quotation market, and provide default quotations. The market environment is diverse, taking into account the contract as needed. The master control module is also the interface for the interaction of various sub-modules on the instructor's desk. The quotation processing sub-module controls the quotation process of the power market, including three stages: starting quotation, quotation in progress, and ending quotation. After the quotation market is opened on the back end, multiple agents declare data and submit the data to the database of the designated server through the communication interface. The market transaction sub-module performs unconstrained clearing and constrained clearing of power market transactions. Unconstrained clearing is based on the market mechanism of unified clearing and the principle of supply and demand balance. Generators queue up to sell electricity according to the quotation from low to high, and power purchasing enterprises queue up to purchase electricity according to the quotation from high to low. The unconstrained clearing data obtained by calculation will be used to draw an unconstrained clearing curve, marking the unconstrained market clearing price (MCP) and market clearing quantity (MCQ), visually showing the algorithm principle of unconstrained transaction clearing. Constrained clearing is based on unconstrained calculations and takes into account line transmission constraints. After the calculation is completed, the system will mark the market data and system parameters such as flow direction, node electricity price, generator capacity utilization rate, line utilization rate, transformer utilization rate, etc. on the simulated system clearing network diagram; at the same time, the network diagram will mark different background colors at each node according to the high and low range of node electricity prices, so as to intuitively mark the difference between system node electricity prices. The settlement and evaluation submodule performs settlement processing after market transactions, including the processing of contracts. The public settlement data of the system includes the total market transaction volume, social welfare, and the transaction price of the balancing node; the private settlement data includes the individual income, individual profit, individual market share, individual surplus and its ranking of market participants. Distinguishing between unconstrained and constrained market transactions will result in two sets of settlement results for evaluation. The information release submodule releases the transaction, settlement and other information that are allowed to be made public by the backend to the frontend for the multi-agent model to conduct reinforcement learning training.

[0102] The revenue function model is established for different types of power generation agents. The fuel cost coefficient a of the thermal power unit is 0.00325 yuan per kilowatt-hour square, b is 0.285 yuan per kilowatt-hour, c is 21.5 yuan per kilowatt-hour, and the startup cost is 45 yuan per kilowatt of the rated capacity of the unit. The opportunity cost coefficient of the hydropower unit is is 0.15, The wind turbine prediction error penalty coefficient is 0.08. is 0.35, is 0.25.

[0103] A flexible load demand response model is constructed. The reduction compensation cost coefficient is 1.2 yuan per kilowatt-hour, the transfer scheduling cost coefficient is 0.8 yuan per kilowatt-hour, and the discomfort cost coefficient is 2.5 yuan per kilowatt-hour. The upper limit of the load adjustment is 30% of the rated load, the lower limit is minus 30% of the rated load, and the user satisfaction threshold is set to 0.8. Among the equipment efficiency coefficients, the renewable energy efficiency attenuation coefficient is 0.15, the power grid efficiency attenuation coefficient is 0.12, and the energy storage efficiency attenuation coefficient is 0.18.

[0104] A market transaction decision model for the power sales agent is established. The base electricity price is set at 500 yuan per megawatt-hour, and the quotation adjustment coefficient range is plus or minus 15 percent. The competitor influence factor is obtained through market share sensitivity analysis, with an average value of 0.35. The retail market price response coefficient is set at 3.5, the floor electricity price is 380 yuan per megawatt-hour, and the capped electricity price is 680 yuan per megawatt-hour.

[0105] Market clearing calculations are performed. The interior point method is used to solve the DC power flow optimization problem, with a convergence accuracy of 0.001 and a maximum number of iterations of 1000. The 62 main transmission lines of the transmission network are considered, and the capacity is limited to within 90 percent of the rated capacity. The balancing error of the node injection power is controlled within 0.1 MW.

[0106] The agent is trained using a deep reinforcement learning algorithm. Both the strategy network and the value network use a three-layer neural network structure, with 128, 64, and 32 hidden neurons, respectively, and the activation function uses the ReLU function. The learning rate is set to 0.001, the discount factor is 0.95, the experience replay pool capacity is 10,000, and the batch size is 32. The number of training rounds is 1,000, and each round contains 960 trading sessions.

[0107] Construct a verification scenario library. Use a mixture model of four Gaussian components to cluster wind power output scenarios, extract five principal components for feature expression of load demand scenarios, and use the kernel density estimation method to fit the probability distribution of market price scenarios. Select 30 typical scenarios to form a test set.

[0108] The transaction simulation test was carried out under the verification scenario. The total market transaction volume reached an average of 120,000 MWh per day, of which 75,000 MWh was through contractual electricity and 45,000 MWh was through the spot market. The average electricity price of each node was 510 yuan per MWh, and the maximum node price difference was 75 yuan per MWh. The average rate of return of power generation intelligent entities was 12 percent, the average rate of return of power sales intelligent entities was 8 percent, and the average rate of return of flexible loads through demand response was 5 percent. The market concentration index was 0.15, and the price discovery efficiency was 0.92.

[0109] Establish a market information feedback mechanism. Update market transaction data every 5 minutes, including node electricity prices, transaction volume, unit output and other information. Divide into 3 levels according to the confidentiality level, and grant corresponding access rights to market entities. It can be seen from this embodiment that the power system transaction simulation method proposed in the present invention can effectively simulate the transaction behavior of market entities and realize the scientific evaluation of the market mechanism.

[0110] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A multi-agent agent-based power system transaction simulation method, characterized in that: The following steps are involved: S01. Collecting historical operation data of the power system, wherein the historical operation data includes power generation output data, load demand data, electricity price data, and network topology data; S02. Build an electricity trading environment, and set the number of market participants, the number of quotation periods, the number of capacity segments, contract data, the upper and lower quotation limits based on the historical operation data; S03, performing Monte Carlo simulation to generate a wind power output scenario set, a load demand scenario set, and a market price scenario set, and verifying the scenario set through a probability density distribution test; S04. Establish a multi-agent system, wherein the multi-agent system includes a power generation agent, a flexible load agent, and a power sales agent, and defines the state space, action space, strategy space, and reward function of each agent based on a partially observable Markov decision process; S05, defining the profit function of the power generation agent, dividing the power generation agent into a thermal power generation agent, a hydropower generation agent, and a wind power generation agent according to the power generation type, and using the historical operation data to train the cost coefficient of each power generation agent; S06. Construct a demand response model of a flexible load intelligent agent, and optimize it by solving a flexible load optimization equation group, wherein the flexible load optimization equation group includes an objective function equation, a load balance equation, a comfort constraint equation, and a transfer constraint equation; S07. Calculate the input parameters of the flexible load optimization equation group through regression analysis, including reduction compensation cost coefficient, transfer scheduling cost coefficient, power purchase and sale cost coefficient, user discomfort cost coefficient, load adjustment upper limit value, load adjustment lower limit value, and user satisfaction threshold value; S08. Establishing a joint decision-making model for power purchase and sales of the power sales intelligent entity, wherein the joint decision-making model includes a spot market quotation strategy and a retail market pricing strategy, and sets a minimum power value and a maximum power value based on the historical operation data; S09. Execute market clearing calculations, perform unconstrained clearing and constrained clearing based on the principle of supply and demand balance, and obtain market clearing electricity prices and market clearing electricity quantities; S10, using a deep reinforcement learning algorithm to train the multi-agent system, using the market-clearing electricity price and the market-clearing electricity quantity as training data, and updating the strategy parameters of each agent; S11, constructing a verification scenario library, and generating test samples based on the wind power output scenario set, the load demand scenario set, and the market price scenario set; S12, executing transaction simulation in the verification scenario library, evaluating the strategy parameters of the multi-agent system, and calculating the total market transaction volume, node power value, agent income value, agent profit value, and agent market share value; S13. Publish market information, and feed back the total market transaction volume, node electricity value, agent revenue value, agent profit value, and agent market share value to the multi-agent system for the next round of transaction decision optimization.

2. A method for simulating power system transactions based on multi-agent agents according to claim 1, characterized in that: The objective function equation is used to minimize the total system cost. The input includes reduction compensation cost, transfer scheduling cost, electricity purchase and sales cost, and user discomfort cost. The output is the optimal load response plan.

3. A method for simulating power system transactions based on multi-agent agents according to claim 1, characterized in that: The load balance equation is used to ensure the balance between electricity supply and demand. The input includes the total power on the demand side, the output of renewable energy, the grid input power, and the output of the energy storage system. The output is the power balance state in each time period.

4. The method for simulating power system transactions based on multi-agent agents according to claim 1, characterized in that: The comfort constraint equation is used to ensure the user's electricity consumption experience. The input includes the user's original electricity demand, the load adjustment amount, and the user satisfaction threshold. The output is the load adjustment range that meets the user's comfort requirements.

5. The method for simulating power system transactions based on multi-agent agents according to claim 1, characterized in that: The transfer constraint equation is used to control the rationality of load transfer. The input includes the upper and lower limits of the load allowed to be transferred in and out in each time period and the original load curve. The output is a load scheduling plan that meets the transfer constraints.

6. A method for simulating power system transactions based on multi-agent agents according to claim 1, characterized in that: The time periods are set as follows: one day is set to 96 time periods, corresponding to one time period every 15 minutes.

7. A method for simulating power system transactions based on multi-agent agents according to claim 1, characterized in that: The number of capacity segments is 3 to 5, and each capacity segment corresponds to a different quotation strategy.

8. The method for simulating power system transactions based on multi-agent agents according to claim 1, characterized in that: The upper limit of the quotation is 1.2 times the highest historical transaction price, and the lower limit of the quotation is 0.8 times the lowest historical transaction price.

9. The method for simulating power system transactions based on multi-agent agents according to claim 1, characterized in that: The user satisfaction threshold is set to 0.

8.

10. The method for simulating power system transactions based on multi-agent agents according to claim 1, characterized in that: The upper and lower limits of the load adjustment are set to plus or minus 30% of the rated load.

Images