Electric power spot clearing model construction method based on multiple agents and ADMM

Abstract

The invention provides an electric power spot clearing model construction method based on multiple agents and an ADMM, and the method comprises the steps: constructing a multi-agent frame, and carrying out the independent optimization of each agent according to a preset cost parameter and a risk quantification model; based on the probability distribution characteristics of wind and light output, a wind and light output scene is generated through Monte Carlo simulation, scene reduction is carried out through a clustering algorithm, and a high-probability typical scene is extracted as optimization input; an ADMM algorithm is used to carry out distributed iterative optimization on SCUC and SCED models of a power grid coordination agent, the power grid coordination agent calculates a power clearing price vector according to report and quotation information of each agent, and a risk quantification result is embedded into a target function of a thermal power and energy storage agent; and iteratively updating dual variables of the ADMM algorithm until the supply-demand balance deviation and the clearing electricity price deviation are both smaller than a preset threshold value, completing convergence calculation of the clearing model, and outputting an optimal quotation strategy of each market subject.

Description

Technical Field

[0001] This invention belongs to the field of power system technology, specifically relating to a method for constructing a power spot clearing model based on multi-agent and ADMM. Background Technology

[0002] Power system optimization and market clearing models, as core supporting technologies for new power systems, are widely used in market mechanism design for large-scale grid integration of new energy sources. With the advancement of "dual-carbon" goals, the penetration rate of intermittent power sources such as wind and solar power continues to increase, posing a challenge to the traditional centralized clearing model dominated by thermal power. Specifically, the existing technology system covers the entire process from load forecasting to unit combination, including key aspects such as SCUC safety-constrained unit combination and SCED safety-constrained economic dispatch. However, its core deficiency lies in its failure to effectively integrate new energy uncertainty modeling with market participant risk quantification mechanisms. Summary of the Invention

[0003] The present invention aims to at least partially solve one of the technical problems in the related art.

[0004] Therefore, the first objective of this invention is to propose a method for constructing a power spot clearing model based on multi-agent and ADMM.

[0005] The second objective of this invention is to propose a device for constructing a power spot clearing model based on multi-agent and ADMM.

[0006] The third objective of this invention is to provide a computer device.

[0007] The fourth objective of this invention is to provide a non-transitory computer-readable storage medium.

[0008] To achieve the above objectives, a first aspect of the present invention proposes a method for constructing a power spot clearing model based on multi-agent and ADMM, comprising: S1. Construct a multi-agent framework that includes wind power intelligent agents, photovoltaic intelligent agents, thermal power intelligent agents, energy storage intelligent agents, load intelligent agents and grid coordination intelligent agents. Each intelligent agent is independently optimized according to preset cost parameters and risk quantification models. S2, based on the probability distribution characteristics of wind and solar power output, uses Monte Carlo simulation to generate wind and solar power output scenarios, and uses a clustering algorithm to reduce the scenarios and extract high-probability typical scenarios as optimization inputs; S3. The ADMM algorithm is used to perform distributed iterative optimization of the SCUC and SCED models of the power grid coordination agent. The power grid coordination agent calculates the clearing price vector based on the reporting and bidding information of each agent, and embeds the risk quantification results into the objective functions of the thermal power and energy storage agents. S4 iteratively updates the dual variables of the ADMM algorithm until the supply-demand balance deviation and the clearing price deviation are both less than the preset threshold, thus completing the convergence calculation of the clearing model and outputting the optimal bidding strategy for each market participant.

[0009] In one embodiment of the present invention, S1 includes: S11, the optimization objective functions for the wind power intelligent agent and the photovoltaic power generation intelligent agent respectively include a maximization of revenue term. and minimize cost item ,in, and This refers to the electricity market price and renewable energy generation at time t. Initial construction investment costs, depreciation, financing costs, and operation and maintenance costs; S12, The optimization objective function of the thermal power intelligent agent includes maximizing the benefit term. and minimize cost item ,in, The output of a thermal power plant at time t. This refers to the clearing price of electricity for thermal power plants; , and These are the quadratic cost coefficient, the linear cost coefficient, and the no-load cost of thermal power units.

[0010] In one embodiment of the present invention, S2 includes: S21, the wind speed probability distribution adopts the Weibull distribution function. , To perform modeling, in the formula, Let be the wind speed frequency, k be the shape parameter, and c be the scale parameter. The average wind speed, It is a gamma function; S22, Scene reduction is performed by calculating Euclidean distance using the K-means clustering algorithm. ,in, for The Euclidean distance, where P is the number of attributes. Belonging to the same dataset, the wind and solar power output scenarios are clustered into K typical scenarios.

[0011] In one embodiment of the present invention, S3 further includes: S31, the SCUC model includes system load balance constraints. ,in, This represents the output of generator unit i within the province during time period t. This represents the planned power of tie line j in time period t, where NT is the total number of tie lines. The system load for time period t; S32, the SCED model includes power output constraints for renewable energy power plants. Where E represents the collection of new energy power stations. The predicted output of the new energy power station i during time period t.

[0012] In one embodiment of the present invention, S4 includes: S41, the dual variable iterative formula adopts... , To update, in the formula, Let t represent the Lagrange multiplier vector in the ADMM optimization algorithm for agent i in the (k+1)th iteration, where t is the 96-point time interval. As a supply-demand deviation penalty factor; S42, the formula for calculating the supply-demand deviation is: , ,in and Let these represent the power generation vector and load power vector of node i in the k-th iteration, respectively. Let represent the supply-demand deviation vector of node i in the k-th iteration, and t be the 96-point time period.

[0013] To achieve the above objectives, a second aspect of the present invention provides an apparatus for constructing a power spot clearing model based on multi-agent and ADMM, comprising: The multi-agent framework construction module is used to construct a multi-agent framework that includes wind power agents, photovoltaic agents, thermal power agents, energy storage agents, load agents, and grid coordination agents. Each agent is independently optimized according to preset cost parameters and risk quantification models. The wind and solar power output scene generation and reduction module is used to generate wind and solar power output scenes based on the probability distribution characteristics of wind and solar power output using Monte Carlo simulation, and to reduce the scenes using a clustering algorithm, extracting high-probability typical scenes as optimization inputs; The ADMM distributed iterative optimization module is used to perform distributed iterative optimization of the SCUC and SCED models of the grid coordination agent using the ADMM algorithm. The grid coordination agent calculates the cleared electricity price vector based on the reported quantity and price information of each agent and embeds the risk quantification results into the objective functions of the thermal power and energy storage agents. The dual variable iterative convergence module is used to iteratively update the dual variables of the ADMM algorithm until the supply-demand balance deviation and the clearing price deviation are both less than the preset threshold, thus completing the convergence calculation of the clearing model and outputting the optimal bidding strategy for each market participant.

[0014] This invention discloses a method and apparatus for constructing a power spot market clearing model based on multi-agent and ADMM algorithms. The power spot market clearing optimization method, based on the multi-agent framework and ADMM algorithm, considers the uncertainties and risk of wind and solar power generation. It can effectively handle the uncertainties of wind and solar power generation, achieve a quantitative balance between returns and risks for market participants, and improve the economic efficiency of power spot market clearing and the safety of grid operation.

[0015] To achieve the above objectives, a third aspect of this application provides a computer device, including a processor and a memory; wherein the processor runs a program corresponding to the executable program code stored in the memory, for implementing the electricity spot clearing model construction method based on multi-agent and ADMM as described in the first aspect embodiment.

[0016] To achieve the above objectives, a fourth aspect of this application provides a non-transitory computer-readable storage medium storing a computer program that, when executed by a processor, implements the method for constructing a power spot clearing model based on multi-agent and ADMM as described in the first aspect embodiment.

[0017] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0018] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart of a method for constructing a power spot clearing model based on multi-agent and ADMM according to an embodiment of the present invention; Figure 2 This is an architecture diagram of an electricity spot clearing model based on a multi-agent framework and ADMM algorithm, considering wind and solar uncertainties and revenue risks according to an embodiment of the present invention. Figure 3 This is a structural diagram of a power spot clearing model construction device based on multi-agent and ADMM according to an embodiment of the present invention; Figure 4 It is a computer device according to an embodiment of the present invention. Detailed Implementation

[0019] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0021] The following description, with reference to the accompanying drawings, describes a method and apparatus for constructing a power spot clearing model based on multi-agent and ADMM, according to an embodiment of the present invention.

[0022] Example 1 Figure 1 This is a flowchart of a method for constructing a power spot clearing model based on multi-agent and ADMM according to an embodiment of the present invention, as follows: Figure 1 As shown, it includes: S1. Construct a multi-agent framework that includes wind power intelligent agents, photovoltaic intelligent agents, thermal power intelligent agents, energy storage intelligent agents, load intelligent agents and grid coordination intelligent agents. Each intelligent agent is independently optimized according to preset cost parameters and risk quantification models. S2, based on the probability distribution characteristics of wind and solar power output, uses Monte Carlo simulation to generate wind and solar power output scenarios, and uses a clustering algorithm to reduce the scenarios and extract high-probability typical scenarios as optimization inputs; S3. The ADMM algorithm is used to perform distributed iterative optimization of the SCUC and SCED models of the power grid coordination agent. The power grid coordination agent calculates the clearing price vector based on the reporting and bidding information of each agent, and embeds the risk quantification results into the objective functions of the thermal power and energy storage agents. S4 iteratively updates the dual variables of the ADMM algorithm until the supply-demand balance deviation and the clearing price deviation are both less than the preset threshold, thus completing the convergence calculation of the clearing model and outputting the optimal bidding strategy for each market participant.

[0023] The present invention provides a method for constructing a power spot market clearing model based on multi-agent and ADMM, which can effectively handle the uncertainty of wind and solar power generation, realize the quantitative balance between returns and risks for market participants, and improve the economic efficiency of power spot market clearing and the safety of grid operation.

[0024] Example 2 The following describes in detail, with reference to the accompanying drawings, a method for constructing a power spot clearing model based on multi-agent and ADMM according to an embodiment of the present invention.

[0025] This invention proposes a spot market clearing model for electricity based on a multi-agent framework and the ADMM algorithm, considering the uncertainties and revenue risks of wind and solar power. Its improvements lie in: A spot market clearing model for electricity is proposed, based on a multi-agent framework and the ADMM algorithm, considering the uncertainties and revenue risks of wind and solar power. The process is shown in the attached figure. Figure 2 As shown, it includes the following steps: I. This patent adopts a multi-agent framework design, which is mainly divided into wind power agents, photovoltaic power generation agents, thermal power agents, energy storage agents, load agents, and grid coordination agents. Power generation and load agents can adjust cost and benefit risk parameters to form multiple different agents to participate in spot market optimization and clearing. The distributed optimization algorithm ADMM algorithm is used for optimization and iterative clearing. After completion, it enters step II.

[0026] II. Collect parameters for each agent. For wind power agents, photovoltaic power agents, thermal power agents, energy storage agents, and load agents, collect information related to cost, revenue, and risk. For grid agents, collect information such as grid topology, power flow constraints, and clearing parameters. Grid agents can generally be simulated by referring to publicly available electrical network data from IEEE. Wind power and photovoltaic power are modeled and scenario reduced. Proceed to step III.

[0027] III. Apply the ADMM algorithm for overall distributed optimization, and proceed to step IV.

[0028] IV. In the ADMM algorithm, the dual variable (multiplier) needs to be updated according to the iterative formula. After iterative optimization, the algorithm proceeds to step V.

[0029] V. When the supply-demand balance deviation and the clearing price deviation are less than the threshold, stop iterative optimization, the optimization clearing converges, and the supply-demand deviation penalty factor is temporarily set to 0.1, then proceed to step VI.

[0030] VI. Finally, each power generator and energy storage operator obtains the optimal pricing strategy at this moment, taking into account the uncertainties and risks of wind and solar power.

[0031] Furthermore, in step I, The optimization objective function of the power grid coordination agent is: SCUC model objective function: , In the formula, N represents the total number of generating units, U represents the number of users, and T represents the total number of time periods considered, with 96 time periods considered per day. This represents the load of unit i during time period t. This represents the output of unit i during time period t. These are the operating cost, startup cost, and no-load cost of unit i in time period t, respectively. The unit operating cost is a multi-segment linear function related to the output range declared by the unit and the corresponding electricity price. For user i, the electricity purchase cost during time period t, users who report quantity but do not quote a price will be given priority for clearing out the highest price by default; , ES represents the charging and discharging power of the independent energy storage power station and the virtual power plant unit es during time period t, respectively, and ES represents the total number of independent energy storage power stations and virtual power plants; These represent the charging and discharging costs of an independent energy storage power station and a virtual power plant unit es, respectively, during time period t. The charging cost is a multi-segment linear function related to the output range of each segment of the charging section of the charging and discharging quotation curve declared by the energy storage power station and the corresponding electricity price. The discharging cost is a multi-segment linear function related to the output range of each segment of the discharging section of the charging and discharging quotation curve declared by the energy storage power station and the corresponding electricity price. M is the network flow constraint relaxation penalty factor; , respectively, are the forward and reverse power flow relaxation variables for branch l; NL is the total number of lines; , respectively, represent the forward and reverse current relaxation variables for section s; NS represents the total number of sections; When unit i switches to the start-up state during time period t, the start-up cost needs to be included. When the unit is started, the no-load cost needs to be included. The unit operating cost is determined by multiplying the energy price corresponding to the bid output range declared by unit i by the bid electricity price.

[0032] SCED model objective function: .

[0033] SCUC constraints of the power grid coordination agent: 1) System load balance constraints: , in, This represents the output of generator unit i within the province during time period t. This represents the planned power of tie line j in time period t (input is positive, output is negative), where NT is the total number of tie lines. Let t be the system load during time period t.

[0034] 2) Upper and lower limits of unit output constraints: , In the formula, and These represent the maximum and minimum output of unit g, respectively.

[0035] 3) Unit ramp-up constraints:

[0036] , In the formula, and These represent the maximum uphill and downhill rates of unit g, respectively.

[0037] 4) System positive and standby capacity constraints: To ensure system power balance and prevent supply-demand imbalances caused by load forecasting errors and various operational accidents, the entire system generally needs to maintain a certain reserve capacity. The total daily operating capacity must meet the system's minimum reserve capacity. The system's positive reserve capacity constraint can be described as follows: , in, This indicates the start / stop status of unit i during time period t. =0 indicates that the unit is shut down. =1 indicates that the unit is started; This represents the maximum output of unit i during time period t. The system's positive standby capacity requirement for time period t.

[0038] 5) System negative reserve capacity constraint: The system negative reserve capacity constraint can be described as: , in, Let i be the minimum output of unit i during time period t; The system's negative backup capacity requirement for time period t.

[0039] 6) System spinning reserve capacity constraints: The total upward and downward adjustment capacity of the unit output at each time period must meet the actual upward and downward adjustment requirements for rotating standby.

[0040] , , in, The maximum ramp rate of unit i. The maximum downhill / climb rate of unit i; , These are the maximum and minimum output of unit i during time period t, respectively; , The requirements for rotating backup are adjusted upwards and downwards for time period t, respectively.

[0041] 7) Minimum continuous start-up and shutdown time constraints for the unit: Due to the physical properties and actual operational requirements of thermal power units, they are required to meet a minimum continuous start-up / shutdown time. This minimum continuous start-up / shutdown time constraint can be described as follows: , , in, This represents the start-up and shutdown status of unit i during time period t; , These are the minimum continuous operating time and minimum continuous downtime of the unit. , The duration of continuous operation and continuous shutdown of unit i during time period t can be represented by state variables. To indicate: , , 8) Maximum number of start-stop cycles for the unit: First, define the variable for switching between starting and stopping.

[0042] definition To indicate whether unit i switches to shutdown state during time period t, the following condition must be met: , definition Whether unit i switches to the start-up state during time period t depends on the following condition: , The start-stop limit for unit i can be expressed as follows: , .

[0043] 9) Branch flow constraints: Branch flow constraints can be described as: , in, , These are the power flow transmission limits of branch l, respectively; The generator output power transfer distribution factor of the node where unit i is located to branch l; is the generator output power transfer distribution factor of the node where tie line j is located to branch l; K is the number of nodes in the system; Let be the generator output power transfer distribution factor of node k to branch l; Let be the bus load value of node k in time period t. These are the forward and reverse power flow relaxation variables for branch l, respectively.

[0044] 10) Cross-sectional power flow constraints: Considering the power flow constraints at the critical section, these constraints can be described as follows: , in, , These represent the power flow transmission limits at section s, respectively. The generator output power transfer distribution factor of the node where unit i is located to section s; The generator output power transfer distribution factor of the node where the tie line j is located to section s; Let be the generator output power transfer distribution factor at node k to section s. These are the forward and reverse kinetic flow relaxation variables for section s, respectively.

[0045] 11) Output constraints of new energy power plants: , Where E represents a collection of new energy power stations. The predicted power output of renewable energy power station i in time period t. That is, the market output of renewable energy power station before the day should be less than the predicted output value of renewable energy power station.

[0046] The SCUC model also needs to consider the charging and discharging power constraints of independent energy storage power stations and virtual power plant units, the state of charge (SOC) constraints of independent energy storage power stations, the SOC constraints at the start and end of the operating day of independent energy storage power stations, the number of charge and discharge cycles of independent energy storage power stations, and the constraints of simultaneous charging and discharging of independent energy storage power stations within the same hour. These constraints need to be combined with the application information or uniformly set by the dispatching agency, and basically cover the constraints during the internal optimization of independent energy storage.

[0047] SCED constraints of the power grid coordination agent: 1) System load balance constraints: , in, This represents the output of generator unit i within the province during time period t. This represents the planned power of tie line j in time period t (input is positive, output is negative), where NT is the total number of tie lines. Let t be the system load during time period t.

[0048] 2) Upper and lower limits of unit output constraints: , In the formula, and Let represent the maximum and minimum output of unit g, respectively. For units that are shut down in the SCUC optimization results, both the upper and lower limits of the generator power are set to zero.

[0049] 3) Unit ramp-up constraints: , , In the formula, and These represent the maximum uphill and downhill rates of unit g, respectively.

[0050] 4) System spinning reserve capacity constraints: The total upward and downward adjustment capacity of the unit output at each time period must meet the actual upward and downward adjustment requirements for rotating standby.

[0051] , , in, The maximum ramp rate of unit i. The maximum downhill / climb rate of unit i; , These are the maximum and minimum output of unit i during time period t, respectively; , The requirements for rotating backup are adjusted upwards and downwards for time period t, respectively.

[0052] 5) Branch flow constraints: Branch flow constraints can be described as: , in, , These are the power flow transmission limits of branch l, respectively; The generator output power transfer distribution factor of the node where unit i is located to branch l; is the generator output power transfer distribution factor of the node where tie line j is located to branch l; K is the number of nodes in the system; Let be the generator output power transfer distribution factor of node k to branch l; Let be the bus load value of node k in time period t. These are the forward and reverse power flow relaxation variables for branch l, respectively.

[0053] 6) Cross-sectional power flow constraints: Considering the power flow constraints at the critical section, these constraints can be described as follows: , in, , These represent the power flow transmission limits at section s, respectively. The generator output power transfer distribution factor of the node where unit i is located to section s; The generator output power transfer distribution factor of the node where the tie line j is located to section s; Let be the generator output power transfer distribution factor at node k to section s. These are the forward and reverse kinetic flow relaxation variables for section s, respectively.

[0054] 7) Output constraints of new energy power plants: , Where E represents a collection of new energy power stations. The predicted power output of renewable energy power station i in time period t. That is, the market output of renewable energy power station before the day should be less than the predicted output value of renewable energy power station.

[0055] 8) The relevant constraints of independent energy storage power stations, pumped storage power stations, and virtual power plant units are consistent with the day-ahead safety constraint unit combination (SCUC) model.

[0056] The objective function for optimizing a thermal power plant intelligent agent is: , , In the formula, The output of a thermal power plant at time t; This refers to the clearing price of electricity for thermal power plants; , and These are the quadratic cost coefficient, the linear cost coefficient, and the no-load cost of thermal power units.

[0057] The constraints on thermal power intelligent agents mainly consider bidding rules, maximum generating capacity, and fastest adjustment power, which are consistent with those in the clearing model.

[0058] The optimization objective functions for photovoltaic power generation intelligent agents and wind power intelligent agents are: , , In the formula, and This refers to the electricity market price and renewable energy generation at time t. Initial construction investment costs, depreciation, financial costs, and operation and maintenance costs.

[0059] Constraints on photovoltaic power generation intelligent agents and wind power intelligent agents: , In the formula, These refer to the maximum output of wind power and photovoltaic units, respectively. These are the actual grid-connected power outputs of wind power and photovoltaic units, respectively. These are the abandoned amounts of wind power and photovoltaic power units, respectively.

[0060] The optimization objective function of the energy storage agent is: , , In the formula, For energy storage profit function; The energy storage discharge price vector; The energy storage charging price vector; , These are the charging and discharging power vectors of energy storage in the i-th time period; C inv It is the investment and construction cost, C ope It's the cost of operation and maintenance.

[0061] Constraints of energy storage agents: , In the formula, The step size is 1h. This is the minimum discharge coefficient for energy storage, with a value of 0.1; Let be the amount of electricity stored in energy storage unit j at time t; Let t be the energy state of energy storage system i at time t; Let be the charging and discharging power of energy storage system i at time t; This is the upper limit of the energy storage charging and discharging power.

[0062] The objective function for optimizing the load agent is: , In the formula, Electricity cost for load; and This refers to the load at time t and the corresponding electricity price.

[0063] The main constraint of the load agent is that critical loads need to be guaranteed a reliable power supply.

[0064] The dual variable (multiplier) in the ADMM algorithm, and the iterative formula for the clearing price: , , In the formula, This represents the Lagrange multiplier vector in the ADMM optimization algorithm of agent i during the (k+1)th iteration; t represents the 96-point time interval. The supply-demand deviation penalty factor is tentatively set at 0.1.

[0065] , , In the formula, Let represent the supply-demand deviation vector of node i in the k-th iteration; t represents the 96-point time period.

[0066] The power generation and load agents, including wind power agents, photovoltaic power generation agents, thermal power agents, energy storage agents, and load agents, perform internal optimization based only on the electricity price at the previous moment (the initial electricity price set during the initial optimization) and their own revenue and risk quantification model, while simultaneously transmitting their optimized quantity and price quotes.

[0067] The grid coordination agent, acting as the recipient of quantity and price quotes from other agents, performs SCUC and SCED clearing to calculate the electricity price at that moment. In cases of non-convergence of power flow, it is necessary to readjust the parameters of each agent and the initial electricity price settings.

[0068] The revenue-risk quantification model for energy storage and thermal power intelligent agents primarily employs Value-Based Arbitration (VAR) for risk assessment. Using Monte Carlo simulation methods (mainly targeting wind and solar uncertainties), the maximum possible loss at a given moment is estimated under different confidence levels, i.e., the VAR value at the corresponding confidence level. A revenue-risk quantification term is added to the objective function of the energy storage and thermal power intelligent agents.

[0069] Furthermore, in step II, the uncertainty modeling of wind and solar energy is performed as follows: The probability distribution function of wind speed follows a Weibull distribution, which can be simplified as follows: , , In the formula, is the wind speed frequency; k is the shape parameter; c is the scale parameter; Average wind speed; This is a gamma function.

[0070] The probability curve of light intensity follows a beta distribution, which can be simplified as follows: , In the formula, and r is the shape parameter; r is the light radiation intensity; r max This represents the maximum light intensity.

[0071] Monte Carlo simulation combined with K-means clustering is used to reduce the number of scenarios and obtain high-probability typical wind and solar power output scenarios. The k-means clustering algorithm is then used to extract wind or solar power output. Euclidean distance is used to measure correlation during k-means clustering. , in, for The Euclidean distance, where P is the number of attributes. They belong to the same dataset.

[0072] The objective function of the k-means clustering algorithm is... : , ), in, Indicates when the sample The value is 1 if the cluster is classified as cluster k, and 0 otherwise. The mean vector of cluster k.

[0073] from Randomly selected from the sample data The centroids are used as the initial cluster centers. The process begins by looping through the data, calculating the distance from each sample point to a specific centroid, and assigning the sample to the centroid closest to it, resulting in K clusters. For each cluster, the average distance of all sample points assigned to that cluster is calculated and used as the new centroid.

[0074] Example 3 To achieve the above embodiments, such as Figure 3 As shown, this embodiment also provides a power spot clearing model construction device 10 based on multi-agent and ADMM. The device 10 includes a multi-agent framework construction module 100, a wind and solar power output scenario generation and reduction module 200, an ADMM distributed iterative optimization module 300, and a dual variable iterative convergence module 400.

[0075] The multi-agent framework construction module 100 is used to construct a multi-agent framework that includes wind power intelligent agents, photovoltaic intelligent agents, thermal power intelligent agents, energy storage intelligent agents, load intelligent agents and grid coordination intelligent agents. Each intelligent agent is independently optimized according to preset cost parameters and risk quantification models. The wind and solar power output scene generation and reduction module 200 is used to generate wind and solar power output scenes based on the probability distribution characteristics of wind and solar power output using Monte Carlo simulation, and to reduce the scenes using a clustering algorithm, extracting high-probability typical scenes as optimization inputs. ADMM Distributed Iterative Optimization Module 300 is used to perform distributed iterative optimization of the SCUC and SCED models of the power grid coordination agent using the ADMM algorithm. The power grid coordination agent calculates the cleared electricity price vector based on the reported quantity and price information of each agent and embeds the risk quantification results into the objective functions of the thermal power and energy storage agents. The dual variable iterative convergence module 400 is used to iteratively update the dual variables of the ADMM algorithm until the supply-demand balance deviation and the clearing price deviation are both less than the preset threshold, thus completing the convergence calculation of the clearing model and outputting the optimal bidding strategy of each market participant.

[0076] Furthermore, the aforementioned multi-agent framework building module 100 is also used for: The optimization objective functions for wind power intelligent agents and photovoltaic power generation intelligent agents respectively include maximizing the revenue term. and minimize cost item ,in, and This refers to the electricity market price and renewable energy generation at time t. Initial construction investment costs, depreciation, financing costs, and operation and maintenance costs; The optimization objective function of the thermal power intelligent agent includes maximizing the revenue term. and minimize cost item ,in, The output of a thermal power plant at time t. This refers to the clearing price of electricity for thermal power plants; , and These are the quadratic cost coefficient, the linear cost coefficient, and the no-load cost of thermal power units.

[0077] Furthermore, the aforementioned landscape output scene generation and reduction module 200 is also used for: The wind speed probability distribution adopts the Weibull distribution function. , To perform modeling, in the formula, Let be the wind speed frequency, k be the shape parameter, and c be the scale parameter. The average wind speed, It is a gamma function; Scene reduction was performed by calculating the Euclidean distance using the K-means clustering algorithm. ,in, for The Euclidean distance, where P is the number of attributes. Belonging to the same dataset, the wind and solar power output scenarios are clustered into K typical scenarios.

[0078] Furthermore, the ADMM distributed iterative optimization module 300 described above is also used for: The SCUC model includes system load balance constraints. ,in, This represents the output of generator unit i within the province during time period t. This represents the planned power of tie line j in time period t, where NT is the total number of tie lines. The system load for time period t; The SCED model includes power output constraints for renewable energy power plants. Where E represents the collection of new energy power stations. The predicted output of the new energy power station i during time period t.

[0079] Furthermore, the dual variable iterative convergence module 400 described above is also used for: The dual variable iterative formula adopts , To update, in the formula, Let t represent the Lagrange multiplier vector in the ADMM optimization algorithm for agent i in the (k+1)th iteration, where t is the 96-point time interval. As a supply-demand deviation penalty factor; The formula for calculating the supply-demand deviation is: , ,in and Let these represent the power generation vector and load power vector of node i in the k-th iteration, respectively. Let represent the supply-demand deviation vector of node i in the k-th iteration, and t be the 96-point time period.

[0080] An embodiment of the present invention provides a power spot market clearing model construction device based on multi-agent and ADMM, which can effectively handle the uncertainty of wind and solar power generation, realize the quantitative balance between returns and risks for market participants, and improve the economic efficiency of power spot market clearing and the safety of grid operation.

[0081] To implement the methods of the above embodiments, the present invention also provides a computer device, such as... Figure 4 As shown, the computer device 600 includes a memory 601 and a processor 602; wherein, the processor 602 reads executable program code stored in the memory 601 to run a program corresponding to the executable program code, so as to implement the various steps of the method described above.

[0082] To implement the above embodiments, this application also proposes a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the method described in the foregoing embodiments.

[0083] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0084] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

Claims

1. A method for constructing an electricity spot clearing model based on multi-agent and ADMM, characterized in that, include: S1. Construct a multi-agent framework that includes wind power intelligent agents, photovoltaic intelligent agents, thermal power intelligent agents, energy storage intelligent agents, load intelligent agents and grid coordination intelligent agents. Each intelligent agent is independently optimized according to preset cost parameters and risk quantification models. S2, based on the probability distribution characteristics of wind and solar power output, uses Monte Carlo simulation to generate wind and solar power output scenarios, and uses a clustering algorithm to reduce the scenarios and extract high-probability typical scenarios as optimization inputs; S3. The ADMM algorithm is used to perform distributed iterative optimization of the SCUC and SCED models of the power grid coordination agent. The power grid coordination agent calculates the clearing price vector based on the reporting and bidding information of each agent, and embeds the risk quantification results into the objective functions of the thermal power and energy storage agents. S4 iteratively updates the dual variables of the ADMM algorithm until the supply-demand balance deviation and the clearing price deviation are both less than the preset threshold, thus completing the convergence calculation of the clearing model and outputting the optimal bidding strategy for each market participant.

2. The method as described in claim 1, characterized in that, S1 includes: S11, the optimization objective functions for the wind power intelligent agent and the photovoltaic power generation intelligent agent respectively include a maximization of revenue term. and minimize cost item ,in, and This refers to the electricity market price and renewable energy generation at time t. Initial construction investment costs, depreciation, financing costs, and operation and maintenance costs; S12, The optimization objective function of the thermal power intelligent agent includes maximizing the benefit term. and minimize cost item ,in, The output of a thermal power plant at time t. This refers to the clearing price of electricity for thermal power plants; , and These are the quadratic cost coefficient, the linear cost coefficient, and the no-load cost of thermal power units.

3. The method as described in claim 1, characterized in that, The S2 includes: S21, the wind speed probability distribution adopts the Weibull distribution function. , Modeling is performed, where, Let be the wind speed frequency, k be the shape parameter, and c be the scale parameter. The average wind speed, It is a gamma function; S22, Scene reduction is performed by calculating Euclidean distance using the K-means clustering algorithm. ,in, for The Euclidean distance, where P is the number of attributes. Belonging to the same dataset, the wind and solar power output scenarios are clustered into K typical scenarios.

4. The method as described in claim 1, characterized in that, The S3 further includes: S31, the SCUC model includes system load balance constraints. ,in, This represents the output of generator unit i within the province during time period t. This represents the planned power of tie line j in time period t, where NT is the total number of tie lines. The system load for time period t; S32, the SCED model includes power output constraints for renewable energy power plants. Where E represents the collection of new energy power stations. The predicted output of the new energy power station i during time period t.

5. The method as described in claim 1, characterized in that, The S4 includes: S41, the dual variable iterative formula adopts... , To update, in the formula, Let t represent the Lagrange multiplier vector in the ADMM optimization algorithm for agent i in the (k+1)th iteration, where t is the 96-point time interval. As a supply-demand deviation penalty factor; S42, the formula for calculating the supply-demand deviation is: , ,in and Let these represent the power generation vector and load power vector of node i in the k-th iteration, respectively. Let represent the supply-demand deviation vector of node i in the k-th iteration, and t be the 96-point time period.

6. A device for constructing a spot market clearing model for electricity based on multi-agent and ADMM, characterized in that, include: The multi-agent framework construction module is used to construct a multi-agent framework that includes wind power agents, photovoltaic agents, thermal power agents, energy storage agents, load agents, and grid coordination agents. Each agent is independently optimized according to preset cost parameters and risk quantification models. The wind and solar power output scene generation and reduction module is used to generate wind and solar power output scenes based on the probability distribution characteristics of wind and solar power output using Monte Carlo simulation, and to reduce the scenes using a clustering algorithm, extracting high-probability typical scenes as optimization inputs. The ADMM distributed iterative optimization module is used to perform distributed iterative optimization of the SCUC and SCED models of the grid coordination agent using the ADMM algorithm. The grid coordination agent calculates the cleared electricity price vector based on the reported quantity and price information of each agent and embeds the risk quantification results into the objective functions of the thermal power and energy storage agents. The dual variable iterative convergence module is used to iteratively update the dual variables of the ADMM algorithm until the supply-demand balance deviation and the clearing price deviation are both less than the preset threshold, thus completing the convergence calculation of the clearing model and outputting the optimal bidding strategy for each market participant.

7. The apparatus as claimed in claim 6, characterized in that, The multi-agent framework building module is also used for: The optimization objective functions for wind power intelligent agents and photovoltaic power generation intelligent agents respectively include maximizing the revenue term. and minimize cost item ,in, and This refers to the electricity market price and renewable energy generation at time t. Initial construction investment costs, depreciation, financing costs, and operation and maintenance costs; The optimization objective function of the thermal power intelligent agent includes maximizing the revenue term. and minimize cost item ,in, The output of a thermal power plant at time t. This refers to the clearing price of electricity for thermal power plants; , and These are the quadratic cost coefficient, the linear cost coefficient, and the no-load cost of thermal power units.

8. The apparatus as claimed in claim 6, characterized in that, The wind and solar power output scene generation and reduction module is also used for: The wind speed probability distribution adopts the Weibull distribution function. , Modeling is performed, where, Let be the wind speed frequency, k be the shape parameter, and c be the scale parameter. The average wind speed, It is a gamma function; Scene reduction was performed by calculating the Euclidean distance using the K-means clustering algorithm. ,in, for The Euclidean distance, where P is the number of attributes. Belonging to the same dataset, the wind and solar power output scenarios are clustered into K typical scenarios.

9. A computer device, characterized in that, Including processor and memory; The processor reads executable program code stored in the memory to run a program corresponding to the executable program code, so as to implement the method for constructing a power spot clearing model based on multi-agent and ADMM as described in any one of claims 1-5.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method for constructing a power spot clearing model based on multi-agent and ADMM as described in any one of claims 1-5.

Images