A power system demand response optimization method based on game model

Abstract

The present invention provides a power system demand response optimization method based on a game model. By constructing a master-slave game model, the decision-making process of the independent system operator (ISO) and multiple large users is divided into an upper layer and a lower layer for optimization. The upper layer model is dominated by the ISO, and the system operating cost is minimized by dynamically adjusting the power supply, flexible ramping product configuration and incentive level; the lower layer model is led by each major user to independently optimize its response amount according to the incentive strategy provided by the ISO to maximize its own benefits. Through a multi-stage iterative solution strategy and multi-region coordinated optimization, the present invention improves the system's solution accuracy and convergence efficiency, and improves the stability of the power system.

Description

Technical Field

[0001] The present invention relates to the field of power system optimization and dispatching, and in particular to a power system demand response optimization method based on a game model. Background Art

[0002] With the rapid development of renewable energy and its large-scale access to the power system, the operation of the power system is facing more and more challenges. In particular, due to the intermittent and fluctuating nature of renewable energy such as wind and solar energy, the load balance of the system has become more complex and unstable. Traditional power dispatching methods mainly rely on the regulation capabilities of the power generation side to maintain the stability of the power system. However, with the increasing complexity of the power system and the continuous changes in electricity demand, it is difficult to meet the stable operation requirements of the system by relying solely on the regulation of the power generation side. Therefore, demand-side response has gradually become an important auxiliary service means, which helps the system achieve a balance between supply and demand of electricity by encouraging users to adjust their own electricity consumption behavior.

[0003] At present, demand response mainly achieves user response through direct load control or price incentives. However, most existing demand response strategies adopt single-layer optimization or simple control methods, which lack sufficient consideration of the differentiated response behavior of individual users and make it difficult to effectively coordinate the response characteristics of different types of users. In addition, with the gradual opening of the electricity market, the demand for coordinated optimization of multi-regional power systems is also increasing. The existing single optimization strategy is difficult to meet the operation requirements of complex power systems. Therefore, a power system demand response optimization method based on game model is proposed. Summary of the Invention

[0004] In view of this, the present invention provides a power system demand response optimization method based on a game model to solve or alleviate the technical problems existing in the prior art and at least provide a beneficial option.

[0005] The technical solution of the present invention is implemented as follows: a method for optimizing power system demand response based on a game model, comprising the following optimization steps:

[0006] Establishment of user response cost function: For multiple users, a response cost function is established that can quantify the economic cost of users under different incentive levels. The response cost function is:

[0007] in, Indicates the The response cost of each user, Represents a user Demand response amount, parameters 、 、 Represent the quadratic term, linear term, and fixed cost of user response costs, respectively. These parameters are determined by regression analysis of historical user data to accurately reflect the user's response characteristics and cost changes under different conditions, reflecting their different sensitivities to incentives.

[0008] Steps for constructing the master-slave game model: Based on the cost function, a two-layer game model is constructed between the independent operator ISO and the large user to simulate the interaction process between the two. Specifically, the steps include:

[0009] The upper model is led by ISO, whose goal is to adjust the power supply , flexible climbing products flat , to minimize the total cost of power operation, the optimization objective function is:

[0010]

[0011] in, is the total number of scheduling cycles, is the number of users participating in demand response, and Indicates time period In the upper-level optimization process, ISO makes comprehensive decisions based on market price fluctuations, user response capabilities, and system requirements to ensure power supply and demand balance and system stability while reducing operating costs.

[0012] The lower-level model is led by major users, based on the incentive level provided by ISO and their respective response cost functions , choose the best response amount , so as to reduce the response cost while obtaining incentive compensation;

[0013] Optimization of the upper model: During the optimization process of the upper model, ISO adjusts the power supply according to the feedback of users. , flexible climbing products and motivation levels Dynamically adjust to minimize the total operating cost while satisfying the following constraints:

[0014] Power balance constraints: In each period , must meet ,in For the period Total load demand to ensure stable operation of the system in all scheduling cycles;

[0015] Flexibility requirement constraint: total response volume Meeting the system's flexibility needs at different times to cope with demand changes and fluctuations caused by renewable energy integration;

[0016] Incentive level non-negativity constraint: setting the incentive level , and adjust the incentive level according to the real-time needs of the system to improve the effectiveness of the incentive;

[0017] Lower-level user response optimization: major users are based on the incentive level set by the upper-level ISO and their respective response costs, independently optimizing their response volumes , the user's goal is to minimize the difference between the response cost and the incentive compensation, that is:

[0018]

[0019] By solving the KKT conditions of the lower model, the user In the period The optimal response is:

[0020]

[0021] The user's response amount will be dynamically adjusted as the incentive level changes, and the size of the response amount is also affected by the coefficient of its cost function;

[0022] Iterative solution of the master-slave game model: Through iterative optimization to solve the master-slave game model, gradually adjust the upper ISO 、 and , by calculating the optimal response amount for each user The results are fed back to the upper model. ISO uses the gradient descent optimization algorithm to dynamically adjust the power supply based on user feedback. , flexible climbing products and motivation levels , through continuous iterative optimization, until the decision variables of the upper and lower models converge to stable values.

[0023] Further preferably, the parameters of the user response cost function are 、 、 This is obtained by performing regression analysis on the user's historical data. The specific steps are as follows:

[0024] S1. Collect historical demand response data of multiple large users, including the response volume of each user under different incentive levels , corresponding incentive costs , and the actual economic costs incurred, data comes from users' electricity consumption records, historical response behaviors, and the implementation results of market incentive policies;

[0025] S2. Use regression analysis method to fit the data and determine the quadratic term coefficient in the response cost function , linear term coefficient , and fixed cost items , the fitting method uses the least square method, the goal is to minimize the fitting error, so that the response cost function Accurately describe the economic costs to users under different response amounts.

[0026] Further preferably, the regression analysis further uses a gradient boosting regression tree learning algorithm to capture nonlinear relationships in user response characteristics and adaptively adjust response curve parameters for different types of users.

[0027] Further preferably, the upper model of the master-slave game model includes the incentive level Adaptive dynamic adjustment to minimize the total operating cost of the system, the iterative optimization algorithm selects the gradient descent method to adjust the power supply according to the user's response feedback in each iteration , flexibility products and motivation levels renew.

[0028] Further preferably, the iterative solution process adopts a multi-stage convergence strategy, uses a fast iterative algorithm to roughly optimize the system and quickly approach the optimal solution; then uses a refined iterative algorithm to further optimize the decision variables. Finally, when the change in the optimization target value of the upper and lower models is less than the set convergence threshold, the iteration is stopped and the final power supply and incentive strategy are output.

[0029] Further preferably, the lower model calculates the user's response amount When the user receives the stimulus signal, the response delay of the user is modeled. The response delay is estimated by analyzing the user's historical response data to accurately simulate the user's actual response behavior.

[0030] Further preferred, it is characterized in that the power balance constraints of the upper-level model also include the coordinated optimization of multi-regional power systems. In the case of multiple power regions, the power exchange, transmission loss and response capability differences between regions are taken into account in the upper-level model through inter-regional power exchange constraints and transmission constraints.

[0031] Further preferably, it is characterized in that: the iterative solution process includes multi-level convergence judgment, which includes: judging the change range of the optimization objective function value of the upper model in multiple iterations, and if the change range is less than a set threshold, it is considered that the upper model has converged;

[0032] Determine the adjustment range of the user response of the lower model in multiple iterations. If the change range is less than the set threshold, the lower model is considered to have converged.

[0033] When both the upper and lower models meet the convergence conditions, the iteration is stopped and the final power supply, flexibility product configuration and incentive level are taken as the optimal configuration.

[0034] The embodiment of the present invention adopts the above technical solution, which has the following advantages:

[0035] The present invention constructs a master-slave game model. The independent system operator (ISO) takes the lead at the upper level to optimize the power supply, flexible ramping product configuration and incentive level to minimize the system operating costs. At the lower level, each major user independently optimizes its response according to the incentive strategy, thereby better reflecting and utilizing the individual response characteristics of the user. It adopts iterative optimization to solve the master-slave game model, gradually adjusts the system parameters, and makes the upper and lower layers reach game equilibrium. At the same time, through cross-regional power exchange and transmission constraints, the balance of power supply and demand between multi-regional power systems is achieved. On the one hand, the present invention can accurately determine the differentiated response of users to incentives and improve the efficiency of demand response. On the other hand, through multi-stage iterative optimization and multi-level convergence judgment, the convergence speed and solution accuracy are improved, and the operational stability of the power system is improved.

[0036] The above summary is for illustrative purposes only and is not intended to be limiting in any way. In addition to the illustrative aspects, embodiments and features described above, further aspects, embodiments and features of the present invention will be readily apparent by reference to the accompanying drawings and the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0038] Figure 1 It is a flow chart of the optimization steps of the present invention;

[0039] Figure 2 A diagram showing the steps of various strategies and judgments of the present invention;

[0040] Figure 3 This is a multi-stage convergence strategy judgment flow chart of the present invention;

[0041] Figure 4 This is the iterative solution flow chart of the present invention. DETAILED DESCRIPTION

[0042] Hereinafter, only certain exemplary embodiments are briefly described. As will be appreciated by those skilled in the art, the described embodiments may be modified in various ways without departing from the spirit or scope of the present invention. Therefore, the drawings and description are to be considered as illustrative in nature and not restrictive.

[0043] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0044] Example 1

[0045] like Figure 1-4 As shown, this embodiment discloses a power system demand response optimization method based on a game model, including the following optimization steps:

[0046] The purpose of establishing the user response cost function is to construct a cost model that can accurately reflect the user's response behavior under different incentive levels. The specific implementation method is as follows:

[0047] It is necessary to collect historical demand response data of users, including the response amount of each user under different incentive levels (reduced electricity consumption or transferred load), the corresponding incentive cost (the amount of compensation the user receives when participating in demand response), and the actual economic cost borne by the user (production adjustment cost or increase in equipment operating cost). The data comes from the electricity consumption records and historical response behavior of multiple large users, obtained through historical records provided by power companies, smart meter monitoring data, or relevant records of users participating in demand response programs; regression analysis methods are used to fit the parameters of the user response cost function, including the quadratic term coefficient , linear term coefficient and fixed cost items The regression analysis method used is the least squares method, the goal is to minimize the fitting error and make the established response cost function It can accurately describe the economic cost of users under different response amounts. The specific fitting process is to transform the user's response amount and the corresponding actual cost data as input, and the least squares method is used to solve the parameters 、 and , so as to minimize the sum of squares of the residuals of the regression equation and effectively find the parameter values that best fit the model with the user's historical response data.

[0048] The master-slave game model simulates the interaction between the independent system operator (ISO) and large users to find the optimal strategy to minimize the total system operating cost and achieve a balance between power supply and demand. The specific implementation steps are as follows:

[0049] Based on the establishment of the user response cost function, a two-layer master-slave game model is constructed, which includes an upper model and a lower model. The upper model is led by ISO and is responsible for deciding the power supply in the power system. , flexible climbing products and motivation levels The lower-level model is led by major users and is responsible for optimizing their respective response volumes. ,The master-slave game model allows the upper and lower layers to influence and adjust each other through an ,iterative approach to gradually approach the optimal solution and ,achieve the optimal configuration of the entire power system;

[0050] In the construction of the upper-level model, the goal of ISO is to minimize the total operating cost of the system. The optimization objective function can be expressed as:

[0051]

[0052] in, is the total number of scheduling cycles, is the number of users participating in demand response, and Indicates time period In the upper-level optimization process, ISO makes comprehensive decisions based on market price fluctuations, user response capabilities, and system requirements to ensure power supply and demand balance and system stability while reducing operating costs.

[0053] In the game-based power system demand response optimization method of this invention, the upper and lower optimization processes are led by the Independent System Operator (ISO) and multiple large users, respectively. The ISO adjusts the overall system operation during the upper-level optimization process, while the large users independently determine their response behavior based on the ISO's incentive strategy during the lower-level optimization process. The following are specific implementations of these two optimization processes:

[0054] ISO is responsible for the overall resource allocation and cost minimization of the system in the upper model. ISO needs to dynamically adjust the power supply according to market conditions, system requirements and user feedback. , flexible climbing products and motivation levels , the specific implementation steps are as follows:

[0055] The power balance constraint requires that in each period , the power supply and demand in the system reaches a balance, and the power balance constraint formula is: ,in, For the period The amount of electricity supply, For users The amount of demand response, For the period The total load demand, Indicates the number of users participating in demand response. Power balance constraints ensure that the supply and demand of electricity are equal in each scheduling cycle to avoid power shortages or power surpluses.

[0056] ISO can be adjusted based on user response and For example, when the user's demand response is large, the ISO reduces the amount of power supplied , thereby reducing electricity procurement costs; when user response is insufficient, ISO needs to increase or motivation level To encourage more users to respond.

[0057] Flexibility demand constraints deal with the load fluctuation problem caused by the access of renewable energy in the power system. Renewable energy has uncertainty and intermittency, which can easily cause rapid changes in the load of the power system. Therefore, flexibility demand constraints require a total response amount Satisfy the system during the period Flexibility needs to cope with demand-side volatility: , ISO dynamically adjusts according to real-time flexibility requirements The configuration amount ensures that the system has sufficient adjustment capacity to cope with load changes. For example, in the case of a sudden increase in wind power or photovoltaic output, the user's response amount or power supply can be appropriately reduced through ISO to avoid system instability.

[0058] To ensure the rationality and effectiveness of incentives, the incentive level Must be non-negative, that is: , ISO adjusts the incentive level based on user feedback and the actual needs of the system. Dynamic adjustment and adjustment of incentive levels are achieved through the following methods:

[0059] When the user's response volume is insufficient, increase To attract more users to respond;

[0060] When the user's response volume is large, reduce To reduce costs.

[0061] The user's optimization problem is solved by solving the KKT condition of the lower model to obtain the optimal response. The KKT condition is used to determine the extreme point of the optimization problem under the constraint conditions. By solving the following equation, the user's In the period The optimal response is:

[0062]

[0063] From the formula, we can see that the user's response volume With the motivation level The higher the incentive level, the greater the user's response; conversely, the response decreases; in addition, the size of the response is also affected by the cost coefficient. and The impact, for example, when the user's quadratic cost coefficient When is larger, the increase in the response volume will be smaller, reflecting that the response cost increases faster.

[0064] The optimization of the upper-level ISO and the optimization of the lower-level user response are alternating iterative processes. The ISO adjusts the system's decision variables based on user feedback, and the user optimizes the response based on the ISO's decision. This process continues until the upper and lower-level decision variables converge. Through iterative optimization, the system operating cost is minimized while meeting the constraints of power balance and flexibility requirements.

[0065] Solve the master-slave game model through iterative optimization to gradually adjust the power supply of the upper ISO , flexible climbing products and motivation levels , to achieve the optimal configuration of the system, the iterative process adopts a multi-stage convergence strategy. In the initial stage, the gradient descent method is used to roughly optimize the system. By calculating the gradient information of the system, it quickly approaches the optimal solution. The advantage of the gradient descent method is its high computational efficiency and the ability to quickly converge to a region close to the optimal solution. On this basis, the quasi-Newton method is used to approximate the Hessian matrix using the second-order derivative information to accelerate the convergence speed and improve the optimization accuracy, effectively avoiding the dilemma of local optimality, while reducing the amount of calculation and improving the overall solution efficiency. During the iteration process, the upper ISO will dynamically adjust the decision variables according to the user's feedback 、 and , while the major users in the lower layer optimize their own responses according to the incentive level set by ISO, forming a feedback loop of interaction between the upper and lower layers. After each iteration, the system will determine whether the objective functions of the upper and lower layers have changed significantly. If the change is less than the set threshold, it is considered that it has converged to a stable solution and the iteration process can be stopped.

[0066] In actual power system demand response, there is usually a certain time lag in the user's response behavior, that is, the user does not respond immediately after receiving the incentive signal, but there is a certain delay. The present invention performs time lag modeling on the calculation of the user's response amount in the iterative solution process. When optimizing the user's response amount, a response delay factor is introduced to represent the user's time lag characteristics. The size of the response delay is estimated through historical data analysis. For example, the response data of the user in previous demand response events is used to calculate the average value or distribution range of the response delay, and it is used as a parameter of the model. In this way, the realism and accuracy of the optimization results are improved, and the actual response behavior of the user is better reflected.

[0067] When the power system involves multiple regions, the present invention adds the power balance equation of each region to the upper-level model, connects the power demand and response of each region through the transmission network, and forms a global coordination optimization problem. By optimizing power exchange, the complementarity and sharing of power resources between regions are achieved, the differential resources between regions are effectively utilized, and the operating efficiency of the entire power system is improved.

[0068] During the iterative solution process, by monitoring the change amplitude of the optimization objective function of the upper-level model, when the change amplitude of the objective function is less than the set convergence threshold in multiple iterations, the upper-level model is considered to have converged. On this basis, it is also necessary to judge the change amplitude of the lower-level user response. If the adjustment amplitude of the response amount of each user is less than the set threshold in multiple iterations, the lower-level model is considered to have converged. Finally, when both the upper and lower layers meet the convergence conditions, the iteration is stopped and the final optimization result is output.

[0069] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various modifications and substitutions within the technical scope disclosed in the present invention, and such modifications and substitutions are intended to be within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A power system demand response optimization method based on a game model, characterized in that: The optimization steps include: Establishment of user response cost function: For multiple users, a response cost function is established that can quantify the economic cost of users under different incentive levels. The response cost function is: Among them, C i (R i ) represents the response cost of the i-th user, R i represents the demand response of user i, that is, the load to be adjusted; parameter a i 、b i 、c i Represent the quadratic term coefficient, linear term coefficient and fixed cost term of user response cost respectively, Quadratic term coefficient a i Reflects the nonlinear change of response cost as the response amount increases, the linear term coefficient b i The fixed cost term c represents the cost part that is linearly related to the response amount. i represents the fixed cost in the response activity; parameter a i 、b i 、c i Determined through regression analysis of historical user data, it reflects the user's response characteristics and cost changes under different conditions, as well as their different sensitivities to incentives; Steps for constructing the master-slave game model: Based on the response cost function, a two-layer game model is constructed between the independent operator ISO and the large user to simulate the interaction process between the two, including: The upper-level model is led by the ISO. The ISO's goal is to minimize the total cost of power operation by adjusting the power supply P, the flexible ramping product F, and the incentive level β. The optimization objective function is: Where T is the total number of dispatch cycles, N is the number of users participating in demand response, and λ t represents the electricity price at time period t. In the upper-level optimization, the ISO makes comprehensive decisions based on market price fluctuations, user response capabilities, and system requirements to ensure power supply and demand balance and system stability while reducing operating costs; The lower model is dominated by major users, based on the incentive level β provided by ISO and their respective response cost function C i (R i ), select the best response quantity R i , so as to reduce the response cost while obtaining incentive compensation; Optimization of the upper-level model: During the upper-level optimization process, the ISO dynamically adjusts the power supply P, the flexible ramping product F, and the incentive level β based on user response feedback to minimize the total operating cost while meeting the following constraints: Power balance constraint: In each time period t, it must satisfy Among them D t is the total load demand in period t, P t is the power supply in period t, is the response amount of user i in time period t, ensuring the stable operation of the system in all scheduling cycles; Flexibility requirement constraint: total response volume Meeting the system's flexibility needs at different times to cope with demand changes and fluctuations caused by the integration of renewable energy; Incentive level non-negativity constraint: setting the incentive level And adjust the incentive level according to the real-time needs of the system to improve the effectiveness of the incentive; Lower-level user response optimization: major users are based on the incentive level set by the upper-level ISO and their respective response costs, independently optimizing their response quantities R i , the user's goal is to minimize the difference between the response cost and the incentive compensation, that is: The formula represents the response optimization goal of user i. The user's goal is to select the optimal response amount To minimize response costs and incentive compensation The difference, where: is the incentive level of ISO to user i in period t; By solving the KKT condition of the lower model, the optimal response of user i in time period t is obtained as: The user's optimal response amount will vary with the incentive level. With the change of cost parameters, the higher the incentive level, the greater the user response; the larger the cost parameter, the higher the difficulty and cost of responding; Iterative solution of the master-slave game model: solve the master-slave game model through iterative optimization, gradually adjust the P, F and β of the upper ISO, and calculate the optimal response of each user The results are fed back to the upper-level model. Based on user feedback, ISO uses the gradient descent optimization algorithm to dynamically adjust the power supply P, flexible ramp product F, and incentive level β. Through continuous iterative optimization, the decision variables of the upper and lower-level models converge to stable values.

2. The method for optimizing power system demand response based on a game model according to claim 1, characterized in that: The parameter a of the user response cost function i 、b i 、c i This is obtained by performing regression analysis on the user's historical data. The specific steps are as follows: S1. Collect historical demand response data of multiple large users, including the response amount R of each user at different incentive levels i , the corresponding incentive level β, and the actual economic costs incurred. The data comes from users' electricity consumption records, historical response behaviors, and the implementation results of market incentive policies; S2. Use regression analysis method to fit the data and determine the quadratic term coefficient a in the response cost function i , linear term coefficient b i , and fixed cost item c i , the fitting method uses the least square method, the goal is to minimize the fitting error, so that the response cost function C i (R i ) accurately describes the economic cost to users under different response amounts.

3. The power system demand response optimization method based on a game model according to claim 2, characterized in that: The regression analysis also uses a gradient boosting regression tree learning algorithm to capture nonlinear relationships in user response characteristics and adaptively adjust response curve parameters for different types of users.

4. The method for optimizing power system demand response based on a game model according to claim 1, characterized in that: The upper model of the master-slave game model includes adaptive dynamic adjustment of the incentive level β to minimize the total operating cost of the system. The iterative optimization algorithm selects the gradient descent method to update the power supply P, flexible climbing product F and incentive level β according to the user's response feedback in each iteration.

5. The method for optimizing power system demand response based on a game model according to claim 1, characterized in that: The iterative solution process adopts a multi-stage convergence strategy, uses a fast iterative algorithm to roughly optimize the system and quickly approach the optimal solution; then uses a refined iterative algorithm to further optimize the decision variables. Finally, when the change in the optimization target value of the upper and lower models is less than the set convergence threshold, the iteration is stopped and the final power supply and incentive level are output.

6. The method for optimizing power system demand response based on a game model according to claim 1, characterized in that: The lower model calculates the user's response R i When modeling the user's response delay after receiving the stimulus signal, the response delay is estimated by analyzing the user's historical response data to accurately simulate the user's actual response behavior.

7. The power system demand response optimization method based on a game model according to claim 1, characterized in that: The power balance constraints of the upper model also include the coordinated optimization of multi-regional power systems. When there are multiple power regions, the power exchange, transmission loss and response capability differences between regions are considered in the upper model through inter-regional power exchange constraints and transmission constraints.

8. The method for optimizing power system demand response based on a game model according to claim 1, characterized in that: The iterative solution process includes multi-level convergence judgment, which includes: judging the change range of the optimization objective function value of the upper model in multiple iterations, and if the change range is less than a set threshold, it is considered that the upper model has converged; Determine the adjustment range of the user response of the lower model in multiple iterations. If the change range is less than the set threshold, the lower model is considered to have converged. When both the upper and lower models meet the convergence conditions, the iteration is stopped and the final power supply, flexible ramping product configuration, and incentive level are taken as the optimal configuration.

Images