Game Optimization Method for Electric Energy and Frequency Regulation Ancillary Services in a Power System with a High Proportion of Wind Power

Abstract

The present invention relates to a game optimization method for electric energy and frequency regulation ancillary services in a power system with a high proportion of wind power. The method comprises the following steps: 1) establishing a game model for thermal power plants considering opportunity costs and a game model for wind power plants considering opportunity costs under different wind speeds; 2) constructing a stochastic game model for the joint participation of wind power plants and thermal power plants in the electric energy and frequency regulation markets; 3) establishing a bilateral game model for thermal power and wind power to achieve the goal of wind power plants selling more and purchasing less; 4) based on the SAC algorithm, considering the constraints of each power generation company, establishing a Multi-Agent-SAC deep reinforcement learning environment to simulate the stochastic game behavior of power generation companies, and obtaining the optimal game strategies, day-ahead and intra-day output of each unit. Compared with the prior art, the present invention has the advantages of improving the revenue of wind power plants, reducing the total cost of electric energy and frequency regulation ancillary services, and increasing the wind power consumption rate.

Description

Technical Field

[0001] The present invention relates to the field of new power system control, and particularly to a game optimization method for electric energy and frequency modulation auxiliary services in a power system with a high wind power penetration rate. Background Art

[0002] The wind power penetration rate is gradually increasing, bringing major challenges to the safe operation of the power system. The prominent characteristics of future high-proportion renewable energy power systems are as follows: If the flexible regulation of coal-fired power plants and hydropower plants is mainly relied on to achieve the complementarity with new energy sources such as wind and light, there will be problems that the output of thermal power units is close to the lower limit and the downward adjustment potential is insufficient in most time periods.

[0003] In order to improve the frequency response ability of a new power system with a high proportion of wind power and maintain the system frequency stability, some scholars have studied the wind power frequency modulation control strategy, such as blade pitch angle control, rotor speed control, virtual inertia control, etc. There are also many literatures studying the problem of wind power participating in the power market game, but using traditional methods such as CPLEX to solve the game optimization model cannot reflect the complex game competition behaviors of each power generator as market participants; Solving stochastic games by reinforcement learning has been applied to game optimization in power systems, but the current related research does not consider the uncertainty of wind power output, intelligent decision-making, and the joint participation of thermal power and wind power in the electricity energy market and frequency modulation market at the same time; Existing literatures study the joint clearing of the electricity energy market and the frequency modulation market, but cannot well handle the problem of how wind power producers declare strategies to maximize their interests when the wind power is in different wind speed zones; There are many literatures studying how renewable energy power producers avoid risks and improve individual benefits, but there are still problems that wind power producers cannot balance the goals of avoiding risks, increasing the consumption rate, and increasing individual benefits; Therefore, studying the game optimization method for electricity energy and frequency modulation markets matching the new power system with a high proportion of wind power and designing a suitable market trading plan are of great significance for the future development of high-proportion renewable energy. Summary of the Invention

[0004] The purpose of the present invention is to overcome the defects of the above-mentioned existing technologies and provide a game optimization method for electric energy and frequency modulation auxiliary services in a power system with a high wind power penetration rate.

[0005] The purpose of the present invention can be achieved through the following technical solutions:

[0006] A game optimization method for electric energy and frequency modulation auxiliary services in a power system with a high wind power penetration rate, used to optimize resource allocation and improve the wind power consumption rate. The method includes the following steps:

[0007] 1) Establish a stochastic game mathematical model of thermal power producers considering opportunity cost and a stochastic game mathematical model of wind power producers considering opportunity cost under different wind speeds;

[0008] 2) Construct a stochastic game model for the joint participation of wind power producers and thermal power producers in the stochastic game of the electricity energy and frequency regulation markets;

[0009] 3) Establish a stochastic game model for the bilateral market of wind and thermal power to achieve the goal of wind power producers selling more and purchasing less;

[0010] 4) Based on the SAC algorithm, considering the constraints of each power producer, establish a Multi-Agent-SAC deep reinforcement learning environment to simulate the stochastic game behavior of power producers, and obtain the optimal game strategies, day-ahead and intra-day output of each unit.

[0011] In the above step 1), the upper-layer objective of the stochastic game mathematical model of thermal power producers considering opportunity cost is to minimize the total cost of electric energy and frequency regulation ancillary services, and its expression is:

[0012]

[0013] where, a i is the square term of the cost coefficient of thermal power producer i, is the declared price and declared capacity of thermal power producer i in the electricity energy market at time t, is the declared price and declared capacity of thermal power producer i in the frequency regulation ancillary service market at time t, and M is the total number of thermal power producers.

[0014] In the above step 1), the lower-layer objective of the stochastic game mathematical model of thermal power producers considering opportunity cost is to maximize the individual revenue of power producers, and its expression is:

[0015]

[0016] where, are the clearing prices of the electricity energy market and the frequency regulation market at time t respectively, and β j,t is the opportunity cost coefficient, which takes the value of 1 when the revenue obtained by thermal power producer i participating in the frequency regulation market is less than the revenue lost in the electricity energy market, and 0 otherwise, is the winning capacity of thermal power producer i in the electricity energy market without participating in the frequency regulation market.

[0017] In the above step 1), the upper-layer objective of the stochastic game mathematical model of wind power producers considering opportunity cost under different wind speeds is to minimize the total cost of electric energy and frequency regulation ancillary services, and its expression is:

[0018]

[0019] where x takes the values of 1, 2, and 3, representing low wind speed, medium wind speed, and high wind speed respectively, are the declared price and declared capacity of wind power producer j in the electricity energy market at time t under wind speed zone x respectively, Let \(p_{xjt}\) and \(q_{xjt}\) be the declared price and declared capacity of wind power producer \(j\) at time \(t\) in wind speed zone \(x\) in the electricity energy market, and \(N\) be the total number of wind power producers.

[0020] In step 1), the lower - level objective of the stochastic game mathematical model of wind power producers considering the opportunity cost at different wind speeds is to maximize the individual profit of power producers, and its expression is:

[0021]

[0022] Where, \(q_{xjt}^0\) is the winning capacity of wind power producer \(j\) at time \(t\) in wind speed zone \(x\) in the electricity energy market without participating in the frequency regulation market.

[0023] In step 2), the stochastic game model in which wind power producers and thermal power producers jointly participate in the electricity energy and frequency regulation markets consists of game players, state space, action strategies, and profit functions. The game players include multiple wind power producers and thermal power producers. The state space, that is, the constraint conditions, are specifically:

[0024]

[0025] Where, \(P_t\) D,t and \(P_{ft}\) RegD,t are the system load demand and frequency regulation demand at time \(t\) respectively, \(P_{it}^{\max}\) i,max and \(P_{it}^{\min}\) i,min are the upper and lower limits of the output of thermal power producer \(i\), \(\widetilde{P}_{xjt}\) is the predicted output value of wind power producer \(j\) in wind speed zone \(x\), \(P_{bd}^t\) bd,t , \(x_{bd}\) bd are the power flow, power flow transfer limit, and reactance between node \(b\) and node \(d\), \(\delta_b^t\) b,t and \(\delta_d^t\) d,t are the phase angles of node \(b\) and node \(d\) at time period \(t\);

[0026] The action strategies are specifically:

[0027]

[0028]

[0029] Where, \(b_i\) i and \(C_i\) pi are the first - order cost coefficient and frequency regulation capacity cost coefficient of thermal power producer \(i\) respectively; are the electricity energy market cost coefficient and frequency regulation capacity cost coefficient in wind speed zone \(x\) respectively;

[0030] The profit function is specifically:

[0031]

[0032] Among them, R i,t , R j,t are the revenues of thermal power merchant i and wind power merchant j at time t, respectively.

[0033] In step 3), in the intraday bilateral trading between wind power merchants and thermal power merchants, a game model of the thermal-wind bilateral market is constructed to achieve the goal of wind power merchants selling more and purchasing less. Then, the stochastic game model of the thermal-wind bilateral market consists of players, state space, action strategies, and revenue functions.

[0034] The players of the stochastic game model of the thermal-wind bilateral market include multiple wind power merchants and thermal power merchants. The state space of the stochastic game model of the thermal-wind bilateral market is specifically:

[0035]

[0036] Among them, are the declared capacities of wind power merchant j sold and purchased at time t in the intraday bilateral market, respectively, is the total winning capacity of wind power merchant j in the dayahead electricity energy market and frequency regulation market at time t, is the output of wind power merchant j at time t in the dayahead.

[0037] The action strategies of the stochastic game model of the thermal-wind bilateral market are specifically:

[0038] If the wind power is less than the dayahead winning capacity, purchase from the thermal power merchant, then there is:

[0039]

[0040] Among them, is the declared purchase electricity price of wind power merchant j at time t in the intraday bilateral market, is the clearing price of the electricity energy market at time t, and K penalty is the penalty coefficient for the actual deviation of the wind power merchant from the dayahead winning capacity;

[0041] If the wind power is greater than the predicted value, sell electricity to the bilateral market. When the real-time load is less than the predicted load demand dayahead, at this time, the thermal power merchant is the electricity purchaser and the wind turbine is the electricity seller, then there is:

[0042]

[0043] Among them, is the declared selling electricity price of wind power merchant j at time t in the intraday bilateral market;

[0044] The revenue function of the stochastic game model of the thermal-wind bilateral market is specifically:

[0045]

[0046] Among them, are the electricity purchase and electricity sales revenue returns of wind power merchant j in the bilateral market at time t, are the trading price and trading volume when wind power merchant j has insufficient output at time t; are the trading price and trading volume when wind power merchant j has excess output at time t.

[0047]

[0048] Among them, are the electricity sales and electricity purchase revenue returns of thermal power merchant i in the bilateral market at time t, are the trading price and trading volume of thermal power merchant i when wind power merchant j has insufficient output at time t, where are the trading price and trading volume when thermal power merchant i has insufficient output at time t, Each agent adopts strategies i and j respectively under states and and Combined with the revenues R i,t and j,t calculate the revenue as a reference for the stochastic game strategy of the power generator and then adjust the strategy.

[0049] Compared with the prior art, the present invention has the following advantages:

[0050] First, improve the revenue of wind power merchants: The present invention considers establishing a bilateral thermal-wind power trading market before the intraday real-time balancing market. Wind power merchants can sell more and buy less, reducing the penalty costs of wind power merchants and enabling them to obtain higher revenues.

[0051] Second, reduce the total cost of electric energy and frequency regulation ancillary services: While considering the opportunity costs of wind power merchants in different wind speed zones, it can increase the output of wind power merchants in the market and reduce the output of thermal power merchants with higher generation costs. For a system where thermal and wind power jointly participate in stochastic games, this method can promote the optimal allocation of market resources and reduce the total cost of electric energy and frequency regulation ancillary services.

[0052] Third, improve the wind power consumption rate: Before the intraday real-time balancing market, establish a bilateral thermal-wind power trading market, enable wind power merchants to sell more and buy less, sell the excess output to thermal power merchants, and improve the wind power consumption rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 is the day-ahead wind power and load forecast.

[0054] Figure 2 is the intraday wind power output and load demand.

[0055] Figure 3 Comparison of power generation company revenues for different scenarios in the embodiments.

[0056] Figure 4 Comparison of wind power consumption rates for each scenario in the embodiments.

[0057] Figure 5 Comparison of the total costs of electric energy and frequency regulation ancillary services for each scenario at different times of the day in the embodiments.

[0058] Figure 6 Schematic diagram of the method flow of the present invention. Detailed implementation manners

[0059] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0060] Embodiment

[0061] The present invention proposes a joint game optimization method for an electric energy market and a frequency regulation market with a high proportion of wind power. First, the present invention constructs a stochastic game mathematical model of a thermal power generation company considering opportunity costs and a stochastic game mathematical model of a wind power generation company considering opportunity costs under different wind speeds, and establishes an objective function considering minimizing the total cost of purchasing electric energy and frequency regulation ancillary services by the operator and maximizing the individual revenue of the power generation company; and constructs a corresponding stochastic game model based on the stochastic game models of the thermal power generation company and the wind power generation company, and elaborates on each part such as the game player, the game state, the game strategy, and the game revenue function in detail;

[0062] Next, based on the day-ahead winning capacity of each unit, the real-time wind power output, and the load demand, a stochastic game model for the day-ahead bilateral trading market of thermal and wind power is established, considering maximizing the individual revenues of the wind power generation company and the thermal power generation company and minimizing the total cost of purchasing electric energy and frequency regulation ancillary services by the operator, so as to achieve more sales and less purchases by the wind power generation company, and at the same time satisfy the lower limit constraint of the wind power generation company's quotation when the wind power output is insufficient, the upper limit constraint of the thermal power generation company's quotation, and the selling electricity quotation constraint of the wind power generation company when the wind power output is more than the day-ahead winning bid and the load demand is less than the day-ahead load forecast; a corresponding stochastic game model is established based on the above constraints and the target revenue function;

[0063] Finally, through the Multi-Agent Soft Actor-Critic (MA-SAC) deep reinforcement learning environment, the stochastic game behaviors of each power generation company are simulated, and the optimal game strategies, day-ahead clearing, and the power output of each power generation company at different times of the day are obtained.

[0064] As Figure 6 shown, a joint game optimization method for an electric energy market and a frequency regulation market with a high proportion of wind power according to the present invention specifically includes the following steps:

[0065] 1) Establish a stochastic game mathematical model for thermal power generators considering opportunity cost in detail, and a stochastic game mathematical model for wind power generators considering opportunity cost under different wind speeds;

[0066] 2) Construct a stochastic game model in which wind power generators and thermal power generators jointly participate in the electric energy and frequency regulation markets;

[0067] 3) Establish an intraday bilateral trading market for wind power generators and thermal power generators, and a stochastic game model for bilateral trading between wind and thermal power, to achieve the goal of wind power generators selling more and purchasing less;

[0068] 4) Based on the Soft Actor-Critic (SAC) algorithm, considering the constraints of each power generator, establish a Multi-Agent-SAC deep reinforcement learning environment to simulate the stochastic game behavior of power generators, and obtain the optimal game strategies, day-ahead and intraday power outputs of each unit.

[0069] In step 1), the stochastic game mathematical model of power generators considering opportunity cost should satisfy the upper and lower layer objectives as much as possible. The upper layer objective is specifically to minimize the total cost of the operator's purchase of electric energy and frequency regulation auxiliary services, and the lower layer objective is specifically to maximize the individual revenue of power generators. Respectively construct a stochastic game model for thermal power generators considering opportunity cost and a stochastic game mathematical model for wind power generators considering opportunity cost under different wind speeds. The specific model expressions are as follows:

[0070] (1) Stochastic game model for thermal power generators considering opportunity cost

[0071] The upper layer objective is to minimize the total cost of electric energy and frequency regulation auxiliary services, and the expression is as follows:

[0072]

[0073] where, a i is the square term of the cost coefficient of thermal power generator i; is the declared price and declared capacity of thermal power generator i in the electric energy market at time t; is the declared price and declared capacity of thermal power generator i in the frequency regulation auxiliary service market at time t, and M is the total number of thermal power generators;

[0074] The lower layer objective is to maximize the individual revenue of power generators, and the specific expression is as follows:

[0075]

[0076] where, are the clearing prices of the electric energy market and the frequency regulation market at time t respectively; β j,t is the opportunity cost coefficient, which takes the value of 1 when the revenue obtained by thermal power generator i participating in the frequency regulation market is less than the revenue lost in the electric energy market, otherwise it is 0; The winning capacity of thermal power plant i in the electricity energy market without participating in the frequency regulation market.

[0077] (2) Stochastic game mathematical model of wind power producers considering opportunity cost under different wind speeds

[0078] The upper - layer objective is to minimize the total cost of electric energy and frequency regulation ancillary services, and the expression is as follows:

[0079]

[0080] Among them, x takes 1, 2, 3, representing low wind speed, medium wind speed, and high wind speed respectively; are the declared price and declared capacity of wind power producer j at time t in the electricity energy market under wind speed zone x respectively; are the declared price and declared capacity of wind power producer j at time t in the electricity energy market under wind speed zone x respectively, and N is the total number of wind power producers;

[0081] The lower - layer objective is to maximize the individual revenue of power generators, and the specific expression is as follows:

[0082]

[0083] Among them, is the winning capacity of wind power producer j at time t in the electricity energy market without participating in the frequency regulation market under wind speed zone x.

[0084] In step 2), the stochastic game model in which wind power producers and thermal power plants jointly participate in the stochastic game of electricity energy and frequency regulation market consists of game players, state space, action strategies, and revenue functions. The specific model expression is:

[0085] (1) Game players (Agent)

[0086] Wind power producer 1, wind power producer 2, wind power producer 3, thermal power plant 1, thermal power plant 2, thermal power plant 3;

[0087] (2) State space X i 、X j (State): That is, the constraint conditions, including the upper and lower limits of unit output, node load demand, etc., then there are:

[0088]

[0089] Among them, P D,t 、P RegD,t are the system load demand and frequency regulation demand at time t respectively, P i,max 、P i,min are the upper and lower limits of the output of thermal power plant i, The predicted output value of wind power producer j under wind speed zone x, P bd,t 、 x bdFor the line power flow, power flow transfer limit, reactance between node b and node d, δ b,t , δ d,t are the phase angles of node b and node d at time period t;

[0090] (3) Game strategy (Action)

[0091]

[0092]

[0093] Among them, b i , C pi are the primary cost coefficient and frequency regulation capacity cost coefficient of thermal power plant i respectively; are the electricity energy market cost coefficient and frequency regulation capacity cost coefficient under wind speed zone x respectively;

[0094] (4) Reward function (Reward)

[0095]

[0096] Among them, R i,t , R j,t are the rewards of thermal power plant i and wind power plant j at time t respectively;

[0097] In step 3), establish an intraday bilateral trading stochastic game market for wind power plants and thermal power plants, construct a stochastic game model for the wind-thermal bilateral market, and achieve the goal of wind power plants selling more and purchasing less. The specific stochastic game model is as follows:

[0098] (1) Players (Agent)

[0099] Wind power plant 1, wind power plant 2, wind power plant 3, thermal power plant 1, thermal power plant 2, thermal power plant 3;

[0100] (2) State space (State)

[0101]

[0102] Among them, are the declared selling and purchasing capacities of wind power plant j in the intraday bilateral market at time t respectively; is the total winning capacity of wind power plant j in the electricity energy market and frequency regulation market at time t in the day-ahead; is the output of wind power plant j at time t in the day-ahead;

[0103] (3) Game strategy (Action)

[0104] If the wind power is less than the day-ahead winning capacity and it can purchase from the thermal power plant, then there is:

[0105]

[0106] Among them, is the declared price of the electricity purchased by wind power merchant j at time t in the intraday bilateral market; is the clearing price of the electricity energy market at time t; K penalty is the penalty coefficient for the deviation of the actual capacity of the wind power merchant from the capacity won in the day-ahead market.

[0107] If the wind power is greater than the predicted value, electricity is sold to the bilateral market. When the real-time load is less than the load demand predicted in the day-ahead, the thermal power merchant acts as the electricity purchaser and the wind turbine generator acts as the electricity seller. Then there is:

[0108]

[0109] Among them, is the declared price of the electricity sold by wind power merchant j at time t in the intraday bilateral market.

[0110] (4) Reward function (Reward)

[0111]

[0112] Among them, are the electricity purchase and sale revenue returns of wind power merchant j in the bilateral market at time t, respectively, is the transaction price and transaction capacity when wind power merchant j has insufficient output at time t; is the transaction price and transaction capacity when wind power merchant j has excess output at time t.

[0113]

[0114] Among them, are the electricity sale and purchase revenue returns of thermal power merchant i in the bilateral market at time t, respectively, is the transaction price and transaction capacity of thermal power merchant i when wind power merchant j has insufficient output at time t, where is the transaction price and transaction capacity when thermal power merchant i has insufficient output at time t,

[0115] Each Agent adopts strategies i 、X j under states and and Combined with R i,t 、R j,t The calculated revenue is used as a reference for the game strategy of the power generator to adjust the strategy.

[0116] Embodiment

[0117] Take the output prediction and load prediction of the Belgian wind farm as the day-ahead data, and the actual output and load demand of each wind power merchant as the intra-day data, as shown respectively in Figure 1 、 Figure 2 ; The cost parameters of the thermal power merchant are shown in Table 1, and the strategy range of the thermal power merchant is The parameters of the wind power merchant are shown in Table 2, and the strategy range of the wind power merchant is

[0118] Table 1 Cost parameters of the thermal power merchant

[0119]

[0120] Table 2 Parameters of the wind power merchant

[0121]

[0122] To compare the advantages of applying MA-SAC proposed in the present invention to the game optimization of wind power merchants and thermal power merchants participating in electric energy and frequency regulation ancillary services with other methods for solving the model established in the present invention, a comparative analysis is carried out on MA-SAC and MA-DDPG (Multi-Agent Deep Deterministic Policy Gradient), PER-MA-DDPG (Prioritized Experience Replay MA-DDPG) for solving the game model proposed in the present invention. The profit curves of the power generation merchant individuals under the above algorithms are as shown in Figure 3 ; As can be seen from Figure 3 , when using MA-SAC to solve, the power generation merchant individual can obtain higher profits; the average consumption time of obtaining the optimal game strategy under the above algorithms is shown in Table 3. Figure 3 Combined with the convergence steps and time in Table 3 of the convergence curve, it can be seen that the time for the method of the present invention to obtain the optimal strategy is the least, having the advantage of faster convergence.

[0123] Table 3 Average consumption time of obtaining the optimal strategy by different deep reinforcement learning methods

[0124]

[0125] Based on the above analysis, the present invention applies MA-SAC to the game optimization of the electric energy and frequency regulation ancillary service market with a high proportion of wind power. To compare the impacts of the opportunity cost under different wind speeds and the establishment of a bilateral thermal-wind power trading market proposed in the present invention on the profits of wind power merchants, the total cost of electric energy and frequency regulation ancillary services, and the wind power consumption rate, the following four schemes are set in the embodiment:

[0126] Scheme 1: The wind power merchant random game model in different wind speed zones considers the same opportunity cost

[0127] Scenario 2: A stochastic game model for power producers in different wind speed zones considering different opportunity costs

[0128] Scenario 3: A stochastic game model for power producers in different wind speed zones considering the same opportunity cost, and establishing a bilateral thermal-wind power trading market within a day

[0129] Scenario 4: A stochastic game model for power producers in different wind speed zones considering different opportunity costs, and establishing a bilateral thermal-wind power trading market within a day

[0130] As can be seen from Table 4, the revenues of Generator 1, Generator 2, and Generator 3 under Scenario 4 decreased by 0.44%, 15.665%, and 15.196% respectively compared with Scenario 3; the revenues of Wind Power Producer 1, Wind Power Producer 2, and Wind Power Producer 3 under Scenario 4 increased by 1.78%, 3.694%, and 2.5691% respectively compared with Scenario 3; as can be seen from Table 1, the power generation cost of Generator 1 is lower than that of Generator 2 and Generator 3, and the revenue loss of Generator 1 is smaller; therefore, the opportunity cost at different wind speeds also affects the revenue of thermal power producers, which can give priority to power generation for power producers with lower power generation costs, increase the revenue of wind power producers, and at the same time reduce the total cost of electric energy and frequency regulation ancillary services, so as to maximize social benefits.

[0131] Table 4 Revenue changes of each power producer under different scenarios

[0132]

[0133] Figure 4 is the wind power consumption rate in each period under different scenarios. The wind power consumption under Scenario 4 is significantly higher than that under Scenario 1 and Scenario 2; compared with Scenario 3, the wind power consumption rate under Scenario 4 increases more significantly; therefore, in the bilateral thermal-wind power trading market established by the present invention, wind power producers sell their excess output to thermal power producers and reduce wind curtailment, which can effectively improve the consumption of wind power producers. Therefore, Scenario 4 can promote the role of wind power producers in the optimal allocation of market resources in a high-proportion wind power system and reduce the role of power producers with higher power generation costs.

[0134] Figure 5 is the cost of electric energy and frequency regulation ancillary services in each period within a day under different scenarios. By Figure 5 it can be seen that the total costs of Scenario 3 and Scenario 4 are lower than those of Scenario 1 and Scenario 2; the total costs of electric energy and frequency regulation ancillary services in each scenario within a day are 3176066.001$, 3095239.67$, 3183770.489$, and 3091530.224$ respectively, and the cost of Scenario 4 is reduced by 2.66% compared with Scenario 1. Therefore, considering the opportunity cost at different wind speeds and establishing a bilateral thermal-wind power trading market within a day for wind power producers to sell more and buy less can reduce the total cost of electric energy and frequency regulation ancillary services in the power market.

[0135] It can be seen that the solution 4 for establishing a bilateral thermal-wind power trading market considering the opportunity cost under different wind speeds can take into account the goals of improving the revenue of wind power producers, reducing the total cost of electric energy and ancillary services, and increasing the wind power consumption rate. From this, it can be concluded that the solution method, stochastic game model, and market trading solution proposed in the present invention can achieve the consumption of a high proportion of wind power and the optimal allocation of market resources.

[0136] In summary, the present method first establishes a stochastic game model for wind power producers considering the opportunity cost under different wind speeds and a stochastic game model for thermal power producers considering the opportunity cost, and models the stochastic game behavior of power producers. Based on the day-ahead winning capacity of each unit, the real-time wind power output, and the load demand, a stochastic game model for the bilateral thermal-wind power trading market considering the maximization of the individual revenue of wind power producers and thermal power producers and the minimization of the total cost of the operator's purchase of electric energy and frequency regulation ancillary services is established. Therefore, the method proposed in the present invention has the advantages of low cost of electric energy and ancillary services and high individual revenue of power producers. In addition, in the bilateral market established in the present invention, when the load demand is less than the day-ahead load forecast and the actual output of the wind power producer is greater than the day-ahead winning bid, the wind power producer sells the excess power to the thermal power producer, which can not only increase the consumption volume of the wind power producer, but also reduce the power generation cost of the thermal power producer to a certain extent through the conversion of the power generation role. Therefore, the present invention can effectively improve the revenue of wind power producers, reduce the total cost of electric energy and ancillary services, and increase the consumption rate of wind power producers.

Claims

1. A game optimization method for electric energy and frequency modulation auxiliary services in a power system with a high proportion of wind power, which is used to optimize resource allocation and improve the wind power accommodation rate, and is characterized in that, The method includes the following steps: 1) Establish a stochastic game mathematical model for thermal power plants considering opportunity cost and a stochastic game mathematical model for wind power plants considering opportunity cost under different wind speeds; 2) Construct a stochastic game model in which wind power plants and thermal power plants jointly participate in the electric energy and frequency regulation markets; 3) Establish a stochastic game model for the bilateral market of wind and fire to achieve the goal of wind power plants selling more and purchasing less; 4) Based on the SAC algorithm, considering the constraints of each power generator, establish a Multi-Agent-SAC deep reinforcement learning environment to simulate the game behavior of power generators, and obtain the optimal game strategies, day-ahead and intra-day power outputs of each unit; In step 1) above, the upper-layer objective of the stochastic game mathematical model for thermal power plants considering opportunity cost is to minimize the total cost of electric energy and frequency regulation ancillary services, and its expression is: Among them, a i is the square term of the cost coefficient of thermal power plant i, is the declared price and declared capacity of thermal power plant i in the electricity energy market at time t, is the declared price and declared capacity of thermal power plant i in the frequency regulation ancillary service market at time t, and M is the total number of thermal power plants; In step 1) above, the lower-layer objective of the stochastic game mathematical model for thermal power plants considering opportunity cost is to maximize the individual revenue of power generators, and its expression is: Among them, λ t E and λ t R are the clearing prices of the electricity energy market and the frequency regulation market at time t respectively. β i,t is the opportunity cost coefficient, which takes the value of 1 when the profit obtained by thermal power plant i participating in the frequency regulation market is less than the profit lost in the electricity energy market, and 0 otherwise. is the winning bid capacity of thermal power plant i in the electricity energy market without participating in the frequency regulation market. In step 1) above, the upper-layer objective of the stochastic game mathematical model for wind power plants considering opportunity cost under different wind speeds is to minimize the total cost of electric energy and frequency regulation ancillary services, and its expression is: Among them, x takes values of 1, 2, and 3, representing low wind speed, medium wind speed, and high wind speed respectively. They are respectively the declared price and declared capacity of the wind power merchant j at time t in the electricity energy market under the wind speed zone x. They are respectively the declared price and declared capacity of the wind power merchant j at time t in the frequency modulation ancillary service market under the wind speed zone x. N is the total number of wind power merchants. In step 1) above, the lower-layer objective of the stochastic game mathematical model for wind power plants considering opportunity cost under different wind speeds is to maximize the individual revenue of power generators, and its expression is: Among them, is the winning bid capacity of wind power merchant j at time t in the electricity energy market without participating in the frequency regulation market under wind speed zone x; In step 2) above, the stochastic game model in which wind power plants and thermal power plants jointly participate in the electric energy and frequency regulation markets consists of players, state space, action strategies, and revenue functions. The players include multiple wind power plants and thermal power plants. The state space, that is, the constraint conditions, are specifically: Among them, P D,t , P RegD,t are the system load demand and frequency regulation demand at time t respectively, P i,max , P i,min are the upper and lower limits of the output of thermal power plant i, is the predicted output value of wind power plant j under wind speed zone x, P bd,t , x bd are the line power flow, power flow transfer limit, and reactance between node b and node d, δ b,t , δ d,t are the phase angles of node b and node d at time period t; The action strategies are specifically: Among them, b i , C pi are respectively the first-term cost coefficient and the frequency regulation capacity cost coefficient of power generation company i; are respectively the electricity energy market cost coefficient and the frequency regulation capacity cost coefficient under wind speed zone x; The revenue functions are specifically: Among them, R i,t and R j,t are the revenues of thermal power merchant i and wind power merchant j at time t, respectively. In step 3) above, in the intra-day bilateral trading stochastic game market of wind power plants and thermal power plants, construct a stochastic game model for the bilateral market of wind and fire to achieve the goal of wind power plants selling more and purchasing less. Then the stochastic game model for the bilateral market of wind and fire consists of players, state space, action strategies, and revenue functions; The players of the stochastic game model for the bilateral market of wind and fire include multiple wind power plants and thermal power plants. The state space of the stochastic game model for the bilateral market of wind and fire is specifically: Among them, are the sold and purchased declared capacities of the intraday bilateral market wind power merchant j at time t, is the total winning bid capacity of the day-ahead wind power merchant j in the electricity energy market and the frequency regulation market at time t, is the output of the wind power merchant j at time t of the day-ahead; The game strategies of the stochastic game model for the bilateral market of wind and fire are specifically: If the wind power is less than the day-ahead winning capacity, purchase from the thermal power plant, then there is: Among them, is the declared price of the electricity purchased by the wind power trader j at time t in the intraday bilateral market, is the clearing price of the electricity energy market at time t in the intraday, and K penalty is the penalty coefficient for the deviation of the actual capacity of the wind power trader from the day-ahead winning bid; If the wind power is greater than the predicted value, sell electricity to the bilateral market. When the real-time load is less than the day-ahead predicted load demand, at this time the thermal power plant is the purchasing power plant and the wind turbine is the selling power plant, then there is: Among them, is the declared price of the electricity sold by the wind power merchant j at time t in the intraday bilateral market; The revenue functions of the stochastic game model for the bilateral market of wind and fire are specifically: Among them, are the electricity purchase and sale revenue returns of wind power merchant j in the bilateral market at time t, are the transaction price and transaction volume when wind power merchant j has insufficient output at time t; are the transaction price and transaction volume when wind power merchant j has excess output at time t; Among them, are the electricity sales and electricity purchase revenue returns of thermal power merchant i in the bilateral market at time t, are the transaction price and transaction volume of thermal power merchant i when wind power merchant j has insufficient output at time t, where are the transaction price and transaction volume of thermal power merchant i when it has insufficient output at time t, Each agent adopts strategies i and j respectively under states X and and Combined with the revenues R i,t and j,t The revenue is calculated and used as a reference for the power generation company's game strategy to further adjust the next strategy.

Images