Distributed virtual power plant game optimization scheduling method

Abstract

The invention discloses a distributed virtual power plant game optimization scheduling method, and belongs to the field of optimization scheduling. According to the method, aiming at the characteristics that the distributed power supplies are distributed dispersedly and have randomness and volatility when the large-scale distributed power supplies are accessed to the network, multi-virtual power plant coordinated scheduling under the non-cooperative game is reasonably carried out, the economic benefit of multi-virtual power plant operation is improved, and energy utilization is fully optimized; considering the influence of the renewable energy prediction error along with the time scale, and establishing a multi-virtual power plant multi-time scale optimization scheduling model based on distributed model prediction control; taking a single virtual power plant as a whole to participate in the electricity market, and establishing a multi-virtual power plant non-cooperative game model; the two-way auction mechanism is applied to the electricity market, and the two-way auction process is analyzed through a non-cooperative game model; by considering distributed virtual power plant optimization scheduling under the competition relationship, the operation economy and transaction electric quantity are optimized, and flexible interaction and reasonable electric energy distribution between virtual power plants are realized.

Description

Technical Field

[0001] The present invention belongs to the field of optimal scheduling, and particularly relates to a game-based optimal scheduling method for a distributed virtual power plant. Background Art

[0002] In the process of the energy system gradually evolving towards a more flexible and sustainable direction, the virtual power plant has become a remarkable concept. The virtual power plant plays an extremely important role in the intelligent management of distributed energy, the rational distribution of electric energy, and the optimization of market scheduling. Due to the unstable supply of distributed renewable energy, its output is intermittent, fluctuating, and random. Therefore, its particularity needs to be fully considered during system scheduling. At the same time, the transactions between virtual power plants involve the electricity buying and selling between different energy assets. Since different virtual power plants have different energy combinations and production advantages, and the operators they belong to may also be different, there is an interest competition relationship between them, and competitive advantages can be brought through the understanding and flexible application of market mechanisms. Therefore, it is necessary to study a reasonable distributed optimization scheduling method to cope with the extremely uncertain renewable energy output, improve the flexibility of multiple virtual power plants in energy interaction, and at the same time, considering the factor of carbon emission reduction, promote the gradual development of virtual power plants towards a low-carbon economy.

[0003] Currently, in order to achieve refined control, the control period needs to be studied in different time scales. In addition, the interaction variables of each VPP exist as system constraint conditions at the same time, and the system model is complex, which brings great pressure to the upper-layer calculation; and the actually operating VPPs belong to different operators, and they all hope to minimize communication as much as possible to protect the information security of their own operations, while the distributed optimization method can effectively solve these problems. With the expansion of the scale and quantity of virtual power plants, most investors participate in the power generation competition in the same region, so non-cooperative game has gradually become the game mode for the fair competition of virtual power plants. Due to the existence of an exact solution of Nash equilibrium, non-cooperative game has been widely applied in practice.

[0004] Through the Model Predictive Control (MPC) algorithm, the accurate prediction and real-time scheduling of renewable energy output are realized, enabling the virtual power plant to more flexibly respond to energy fluctuations and motivating the virtual power plant to gradually reduce carbon emissions; considering the competition relationship, a non-cooperative game model for multiple virtual power plants is established, and the inter-plant transaction electricity is determined based on two-way auction quotes. The coordination of non-cooperative game can reduce the total operating cost of virtual power plants. Each virtual power plant shows self-interest, fully optimizes energy utilization, and flexibly formulates reasonable strategies according to its own conditions. Summary of the Invention

[0005] The object of the present invention is to overcome the defects in the above-mentioned background technology, and a distributed virtual power plant game optimization scheduling method is proposed to solve the problems in the prior art that, in view of the characteristics of the output fluctuation of photovoltaic units and the difficulty in predicting the load, considering the competition relationship, rationally coordinating and optimizing the operation of multiple virtual power plants through non-cooperative games, improving the operation economy of the virtual power plants and ensuring that each virtual power plant makes full use of energy.

[0006] The present invention adopts the following technical solutions to solve the above technical problems:

[0007] A distributed virtual power plant game optimization scheduling method includes the following steps:

[0008] Step S1: First, construct a stepped carbon cost model, divide the stepped interval according to the carbon emission quotas of each virtual power plant, and set the carbon emission penalty prices for different stepped intervals; secondly, construct a multi-virtual power plant day-ahead optimization model, with the goal of minimizing the sum of the overall daily operation cost and the stepped carbon trading cost, and optimize the exchange power of the external power grid connection line, the power generation of micro gas turbines, the control of controllable loads, and the operation cost of energy storage; then, construct a multi-virtual power plant intraday rolling optimization model, with the goal of minimizing the operation cost and the stepped carbon trading cost within the rolling period, and optimize the power generation of gas turbines, the operation of energy storage, and the carbon emission cost; finally, construct a multi-virtual power plant real-time feedback correction model, with the goal of minimizing the adjustment amount of the output of adjustable resources, and perform distributed optimization using the synchronous alternating direction multiplier method, calculate the control variable increment and update the output variable;

[0009] Step S2: Take each virtual power plant as a game participant, define the strategy set to include the trading electricity volume and the quotation set; establish a multi-virtual power plant non-cooperative game model with the lowest operation cost of each virtual power plant as the utility function;

[0010] Step S3: Based on the electricity trading market mechanism, divide the virtual power plants into buyers or sellers according to their net power status, and use the multi-to-multi two-way auction form for transaction matching, and achieve supply-demand balance through quotation rule judgment, transaction matching, and cyclic transactions; define the transaction process, construct the optimization scheduling of the distributed virtual power plant under the competition relationship, dynamically update the virtual power plant operation cost function, and optimize the scheduling strategy through multiple rounds of transactions until the market closing conditions are met.

[0011] Further, the construction of the stepped carbon cost model in step S1 is specifically as follows:

[0012] According to the different limited carbon emission quotas of each virtual power plant, use it as the stepped division standard and make corresponding interval divisions; the stepped carbon emission cost model is as follows:

[0013]

[0014] In the formula, qc0 is the low - price carbon emission penalty price when the carbon emission value of the virtual power plant meets the quota value, q c1 is the carbon emission penalty price of the first tier, q c2 is the carbon emission penalty price of the second tier, q c3 is the carbon emission penalty price of the third tier, g is the length of the carbon emission tier interval, A trad is the trading share of the system actually participating in the carbon trading market within a carbon trading cycle.

[0015] Furthermore, the specific process of constructing the day - ahead optimization model of the multi - virtual power plant in step S1 is as follows:

[0016] Comprehensively considering the randomness of load and renewable energy output and operation economy, a day - ahead optimization model is constructed with the goal of minimizing the sum of the overall daily operation cost of the multi - virtual power plant and the stepped carbon trading cost:

[0017]

[0018] In the formula, is the cost of the exchange power of the overall virtual power plant and the external power grid, is the power generation cost of the micro - gas turbine j in the virtual power plant at time t, is the cost generated by the regulation of the controllable load at time t, is the operation cost of the energy storage j at time t, γ is the carbon emission coefficient per unit power limit of the virtual power plant, C RQ carbon,j,t is the carbon emission calculation cost.

[0019] Furthermore, the specific process of constructing the intra - day rolling optimization model of the multi - virtual power plant in step S1 is as follows:

[0020] In the intra - day rolling optimization stage, a model is established with the goal of minimizing the operation cost and the stepped carbon trading cost of the multi - virtual power plant in the time period [t0, t0 + M△t]:

[0021]

[0022] In the formula, T RN is [t0, t0 + M△t], are respectively the exchange power cost of the overall virtual power plant and the external power grid, the gas turbine power generation cost, the carbon emission cost, and the energy storage operation cost in the intra - day rolling optimization stage.

[0023] Furthermore, the specific process of constructing the real - time feedback correction model of the multi - virtual power plant in step S1 is as follows:

[0024] This stage is divided into 288 sampling periods for real - time feedback correction at 5 - minute intervals, and the synchronous alternating direction multiplier method is used for distributed optimization within each period;

[0025] The real-time feedback correction stage establishes a model with the goal of minimizing the adjustable resource output adjustment amount at the current moment:

[0026]

[0027] In the formula, P i,MT,ref is the reference value of the output of the micro gas turbine in the i-th subsystem, and P i,ESS,ref is the reference value of the charging and discharging power of the energy storage device in the i-th subsystem, and P i,MT is the predicted output value of the output of the micro gas turbine in the i-th subsystem, and P i,ESS is the predicted output value of the charging and discharging power of the energy storage device in the i-th subsystem, and △u i is the increment of the control variable in the i-th subsystem;

[0028] Set the intraday rolling optimization index as the minimum weighted sum of the variance between the actual output and the reference trajectory and the variance of the control variable. Solve the control variable of the system according to the alternating direction multiplier method, and then obtain the system output variable from the control variable.

[0029] Furthermore, the establishment of the non-cooperative game model of the multi-virtual power plant in step S2 is specifically as follows:

[0030] Step S2.1: Regard each virtual power plant as a whole as a participant in the power market, and there is a non-cooperative game between them. The set N of the participating subjects can be re-expressed as N = {VPP1, VPP2,..., VPPn}, where VPP is a distributed virtual power plant;

[0031] Step S2.2: Considering that the traded electricity volume between virtual power plants is related to the quotation, this paper takes the traded electricity volume and the quotation together as the strategy set of the game, that is, S = {S1, S2,..., S n}, S n = {P jh,n , q bid,n}, where P jh,n and q bid,n are the sets of the traded electricity volume and the traded electricity price of virtual power plant n with other virtual power plants respectively;

[0032] Step S2.3: Construct a non-cooperative game model of the multi-virtual power plant with the minimum operating cost of each virtual power plant at time t as the utility function.

[0033] Furthermore, the power trading market mechanism in step S3 is specifically as follows:

[0034] Within a virtual power plant, the surplus degree of the renewable energy output to the load can be represented by the net power:

[0035] PNet,i P(t) = P RES,i P(t) - P Load,i P(t)(5)

[0036] Wherein, P RES,i P(t) is the power generation of renewable energy, and P Load,i is the electricity consumption of the load; when P Net,i P(t) < 0, it indicates a power shortage, and more controllable power sources need to be called, or electricity needs to be purchased from other virtual power plants, which is called the buyer; when P Net,i P(t) > 0, it indicates a power surplus, and additional energy needs to be stored, the surplus needs to be absorbed, or electricity needs to be sold to other virtual power plants, which is called the seller; due to the cost of power transmission, the power generation of DG units and the power consumption of the load must be balanced first; in the two-way auction market, the buyers and sellers participating in the transaction in a many-to-many form can submit quotation information at any time within the specified quotation time.

[0037] Furthermore, the specific trading steps of the power trading market are as follows:

[0038] Step S3.1: Submission of quotation and product quantity information;

[0039] Step S3.2: Quotation rule judgment: Judge whether the quotation is reasonable and whether to accept it;

[0040] Step S3.3: Matching of trading buyers and sellers: Match according to the trading rules, and confirm parameters such as price and quantity;

[0041] Step S3.4: Information disclosure: Disclose the quotations, order, historical trading results and other relevant information of both parties to maintain the transparency of the market;

[0042] Step S3.5: Circular trading: According to the trading rules, judge whether a new round of trading is required. If so, return to Step S3.1; if the electricity quantity of the seller cannot meet the buyer's demand, purchase electric energy from the distribution network to achieve the balance of supply and demand;

[0043] By establishing a non - cooperative game model to analyze the two - way auction process, under the psychology of maximizing the expected benefits of non - cooperative virtual power plant operators, the two participants at the edge of the intersection set will reach a deal, and their quotations will be used as trading references; when the bids of more than one buyer or seller are the same, they should be regarded as a whole first to calculate the trading volume, and then the trading volume will be specifically allocated according to their respective net power bidding ratios.

[0044] Furthermore, the optimization scheduling of distributed virtual power plants considering the competition relationship in Step S3 is specifically as follows:

[0045] Based on the established DMPC virtual power plant model, an optimal dispatching strategy for the virtual power plant considering the competition relationship is constructed, where DMPC is distributed model predictive control; define the sum of power shortages of all buyer virtual power plants as:

[0046]

[0047] The one-time transaction volume for concluding a deal is:

[0048]

[0049] In the formula, and are the total net powers of the seller and the buyer in the concluded transaction respectively; σ is the transaction status, with a value of 1 representing the buyer and a value of 0 representing the seller; when ω ∈ {0, 1}, there is only one buyer at this time; when There are other buyers in the market waiting for transactions at this time;

[0050] Considering the information security and privacy, the information transmission is restricted, and only necessary information is exchanged between virtual power plants:

[0051] m i (t) = [i, |P Net (t)|, price, ID] (8)

[0052] In the formula, T is the length of the optimization layer, t ∈ T, i is the code of the virtual power plant, price is the bidding price, ID is the identity identifier, ID = [D sale , D buy , D sale is the seller, D buy is the buyer, both are binary integer variables;

[0053] Then the information set shared with other virtual power plants is:

[0054] MES i = [m i (1), m i (2), …, m i (T)] (9)

[0055] After receiving the shared information of other virtual power plants, it enters the electricity trading market; next, after the trading process, the MPC optimization strategy of the virtual power plant itself will also change accordingly, and the operating cost function becomes:

[0056]

[0057] C trad,i,t = ∑ pricet (l)·P trad,i,t (l) (11)

[0058]

[0059] Wherein, l is the trading round, and price t (l) is the price of this round of trading.

[0060] Furthermore, the trading process is specifically as follows: When the respective operators determine a round of trading, they will recalculate the operating costs. If both parties are satisfied with the trading, they can confirm and broadcast it. Once any one of the two parties is not satisfied with the result, the trading will be cancelled, and the dissatisfied party will withdraw and wait for the next bid; Confirming the trading means the end of this round of trading, the identity identification value returns to zero, and the virtual power plant will re-enter the next round of trading according to its updated status; The above process loops continuously until the conditions for opening the market are not met;

[0061] After the trading is completed, each virtual power plant conducts power transmission; then it enters the independent operation state, obtains the optimal scheduling plan, inputs the control quantity u = [u1, u2,..., u MΔt , and then performs real-time feedback correction on a smaller time scale until the loop ends when t = t + MΔt.

[0062] Compared with the prior art, the present invention adopts the above technical solutions and has the following beneficial effects:

[0063] (1) The present invention can effectively cope with the uncertainties brought by different scenarios, simplify the model on the basis of ensuring stable operation, and improve the calculation speed.

[0064] (2) Through the coordination of non-cooperative games, the present invention significantly reduces the total operating cost of virtual power plants. Each virtual power plant demonstrates self-interest, can fully optimize energy utilization, and can flexibly formulate reasonable strategies according to its own conditions. Description of the Drawings

[0065] In order to more clearly illustrate the technical solutions of the present application, the accompanying drawings required for use in the embodiments will be briefly introduced below. Obviously, for those of ordinary skill in the art, other accompanying drawings can also be obtained based on these drawings without creative efforts.

[0066] Figure 1 is a flowchart of a distributed virtual power plant game optimization scheduling method;

[0067] Figure 2 is the reference value of renewable energy in the embodiment;

[0068] Figure 3 is the reference value of load in the embodiment;

[0069] Figure 4 is the power market trading situation in the embodiment;

[0070] Figure 5 is the comparison of the operating costs of virtual power plant 1 before and after the game in the embodiment;

[0071] Figure 6 is the comparison of the operating costs of virtual power plant 2 before and after the game in the embodiment;

[0072] Figure 7 is the comparison of the operating costs of virtual power plant 3 before and after the game in the embodiment. Detailed implementation manners

[0073] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0074] As Figure 1 shown, a distributed virtual power plant game optimization scheduling method includes the following steps:

[0075] Step S1: First, construct a stepped carbon cost model, divide the stepped intervals according to the carbon emission quotas of each virtual power plant, and set the carbon emission penalty prices for different stepped intervals; secondly, construct a multi-virtual power plant day-ahead optimization model, with the goal of minimizing the sum of the overall daily operating cost and the stepped carbon trading cost, and optimize the exchange power of the external power grid connection line, the power generation of the micro gas turbine, the controllable load regulation, and the energy storage operating cost; then, construct a multi-virtual power plant intraday rolling optimization model, with the goal of minimizing the operating cost and the stepped carbon trading cost within the rolling period, and optimize the power generation of the gas turbine, the energy storage operation, and the carbon emission cost; finally, construct a multi-virtual power plant real-time feedback correction model, with the goal of minimizing the adjustment amount of the adjustable resource output, and perform distributed optimization using the synchronous alternating direction multiplier method, calculate the control variable increment, and update the output variable;

[0076] Step S2: Take each virtual power plant as a game participant, define the strategy set to include the trading electricity quantity and the quotation set; establish a multi-virtual power plant non-cooperative game model with the lowest operating cost of each virtual power plant as the utility function;

[0077] Step S3: Based on the electricity trading market mechanism, divide into buyers or sellers according to the net power status of the virtual power plant, and use the multi-to-multi two-way auction form for trading matching. Achieve supply-demand balance through quotation rule judgment, trading matching, and cyclic trading; define the trading process, construct an optimized dispatching of distributed virtual power plants under competitive relationships, dynamically update the operating cost function of the virtual power plant, and optimize the dispatching strategy through multiple rounds of trading until the market closing conditions are met.

[0078] Further, the construction of the stepped carbon cost model in step S1 is specifically as follows:

[0079] According to the different limited carbon emission quotas of each virtual power plant, use it as the standard for ladder division and conduct corresponding interval division; the stepped carbon emission cost model is as follows:

[0080]

[0081] In the formula, q c0 is the low-carbon emission penalty price when the carbon emission value of the virtual power plant meets the quota value, q c1 is the first-step carbon emission penalty price, q c2 is the second-step carbon emission penalty price, q c3 is the third-step carbon emission penalty price, g is the length of the carbon emission ladder interval, A trad is the trading share of the system actually participating in the carbon trading market within a carbon trading cycle.

[0082] Further, the construction of the day-ahead optimization model for multiple virtual power plants in step S1 is specifically as follows:

[0083] Comprehensively considering the randomness of load and renewable energy output and operation economy, construct a day-ahead optimization model with the goal of minimizing the sum of the overall daily operating cost of multiple virtual power plants and the stepped carbon trading cost:

[0084]

[0085] In the formula, is the cost of the exchange power of the overall virtual power plant and the external power grid connection line, is the power generation cost of the micro gas turbine j in the virtual power plant at time t, is the cost generated by the regulation of the controllable load at time t, is the operating cost of the energy storage j at time t, γ is the carbon emission coefficient per unit power limit of the virtual power plant, C RQ carbon,j,t is the carbon emission calculation cost.

[0086] Further, the construction of the intra-day rolling optimization model for multiple virtual power plants in step S1 is specifically as follows:

[0087] In the intraday rolling optimization stage, a model is established with the goal of minimizing the operating cost and the stepped carbon trading cost of the multi-virtual power plant during the time period [t0, t0+MΔt]:

[0088]

[0089] Where T RN is [t0, t0+MΔt], are respectively the power exchange cost of the overall virtual power plant and the external power grid, the gas turbine power generation cost, the carbon emission cost, and the energy storage operation cost during the intraday rolling optimization stage.

[0090] Furthermore, the specific construction of the multi-virtual power plant real-time feedback correction model in step S1 is as follows:

[0091] This stage takes 5 minutes as the interval and is divided into 288 sampling periods for real-time feedback correction. The synchronous alternating direction multiplier method is used for distributed optimization within each period;

[0092] A model is established with the goal of minimizing the adjustment amount of the adjustable resource output at the current moment during the real-time feedback correction stage:

[0093]

[0094] Where P i,MT,ref is the reference value of the micro gas turbine output in the i-th subsystem, P i,ESS,ref is the reference value of the charge and discharge power of the energy storage device in the i-th subsystem, P i,MT is the predicted output value of the micro gas turbine output in the i-th subsystem, P i,ESS is the predicted output value of the charge and discharge power of the energy storage device in the i-th subsystem, Δu i is the control variable increment of the i-th subsystem;

[0095] The optimization index of the intraday rolling is set as the minimum weighted sum of the variance between the actual output and the reference trajectory and the variance of the control variable. The control variable of the system is solved according to the alternating direction multiplier method, and then the system output variable is obtained from the control variable.

[0096] Furthermore, the specific establishment of the multi-virtual power plant non-cooperative game model in step S2 is as follows:

[0097] Step S2.1: Each virtual power plant is regarded as a whole as a participant in the power market, and there is a non-cooperative game between them. The set of participating entities N can be re-expressed as N = {VPP1, VPP2,..., VPPn}, where VPP is a distributed virtual power plant;

[0098] Step S2.2: Considering that the transaction electricity volume between virtual power plants is related to the quoted price, in this paper, the transaction electricity volume and the quoted price are jointly used as the strategy set of the game, that is, S = {S1, S2, …, S n}, S n = {P jh,n , q bid,n}, where P jh,n and q bid,n are the sets of the transaction electricity volume and the transaction electricity price of virtual power plant n and other virtual power plants respectively;

[0099] Step S2.3: Construct a non - cooperative game model of multiple virtual power plants with the minimum operating cost of each virtual power plant at time t as the utility function.

[0100] Furthermore, the specific electricity trading market mechanism in step S3 is as follows:

[0101] Within a virtual power plant, the surplus degree of renewable energy output to the load can be represented by the net power:

[0102] P Net,i (t) = P RES,i (t) - P Load,i (t) (17)

[0103] In the formula, P RES,i (t) is the renewable energy power generation, and P Load,i is the load power consumption; when P Net,i (t) < 0, it indicates a power shortage, and more controllable power sources need to be called or electricity needs to be purchased from other virtual power plants, which is called a buyer; when P Net,i (t) > 0, it indicates a power surplus, and additional energy needs to be stored, the surplus needs to be absorbed, or electricity needs to be sold to other virtual power plants, which is called a seller; due to the cost of power transmission, the power generation of DG units and the power consumption of the load must be balanced first; in the two - way auction market, the buyers and sellers participating in the transaction in a many - to - many form can submit quotation information at any time within the specified quotation time.

[0104] Furthermore, the specific trading steps of the electricity trading market are:

[0105] Step S3.1: Submission of quotation and product quantity information;

[0106] Step S3.2: Quotation rule judgment: Judge whether the quotation is reasonable and whether to accept it;

[0107] Step S3.3: Matching of trading buyers and sellers: Match according to the trading rules and confirm parameters such as price and quantity;

[0108] Step S3.4: Information disclosure: Disclose the quotations, orders, historical trading results and other relevant information of both parties to maintain market transparency;

[0109] Step S3.5, Circular trading: According to the trading rules, determine whether a new round of trading is required. If so, return to Step S3.1; if the seller's electricity quantity cannot meet the buyer's demand, purchase electric energy from the distribution network to achieve supply-demand balance.

[0110] By establishing a non-cooperative game model to analyze the two-way auction process, under the psychology of maximizing the expected benefits of non-cooperative virtual power plant operators, the two participants at the edge of the intersection set will reach a deal and use their quotes as a trading reference; when the bids of more than one buyer or seller are the same, they should be regarded as a whole to calculate the trading volume first, and then allocate the trading volume specifically according to their respective net power bidding ratios.

[0111] Furthermore, the specific optimization scheduling of the distributed virtual power plant considering the competition relationship in Step S3 is as follows:

[0112] Based on the established DMPC virtual power plant model, construct an optimization scheduling strategy for the virtual power plant considering the competition relationship, where DMPC is distributed model predictive control; define the sum of the power shortages of all buyer virtual power plants as:

[0113]

[0114] The trading volume of a single deal reached is:

[0115]

[0116] In the formula, and are the total net powers of the seller and the buyer of the concluded transaction respectively; σ is the trading state, with a value of 1 representing the buyer and a value of 0 representing the seller; when ω ∈ {0, 1}, there is only one buyer at this time; when at this time, there are other buyers in the market waiting for transactions;

[0117] Considering information security and privacy, restrict information transmission, and only exchange necessary information between virtual power plants:

[0118] m i (t) = [i, |P Net (t)|, price, ID] (20)

[0119] In the formula, T is the length of the optimization layer, t ∈ T, i is the code number of the virtual power plant, price is the bidding price, ID is the identity identifier, ID = [D sale , D buy , D saleFor the seller, D buy For the buyer, both are binary integer variables;

[0120] Then the set of information shared with other virtual power plants is:

[0121] MES i = [m i (1), m i (2), …, m i (T)] (21)

[0122] After receiving the shared information from other virtual power plants, it enters the electricity trading market; next, after the trading process, the MPC optimization strategy of the virtual power plant itself will also change accordingly, and the operating cost function becomes:

[0123]

[0124] C trad,i,t = ∑price t (l)·P trad,i,t (l) (23)

[0125]

[0126] In the formula, l is the trading round, and price t (l) is the price of this round of trading.

[0127] Furthermore, the specific trading process is as follows: when the respective operators determine a round of trading between the two parties, they will recalculate the operating cost. If both parties are satisfied with the transaction, they can confirm and broadcast it. Once any one of the two parties is not satisfied with the result, the transaction will be cancelled, and the dissatisfied party will withdraw and wait for the next bid; confirming the transaction means the end of this round of trading, the identity identification value returns to zero, and the virtual power plant re-enters the next round of trading according to its updated status; the above process loops continuously until the conditions for opening the market are not met;

[0128] After the transaction is completed, each virtual power plant conducts power transmission; then it enters the independent operation state, obtains the optimal scheduling plan, inputs the control quantity u = [u1, u2, …, u MΔt , and then conducts real-time feedback correction on a smaller time scale until the loop ends when t = t + MΔt.

[0129] To illustrate the effectiveness of this invention patent, a system containing three virtual power plants is established for simulation analysis. Based on the distributed coordinated scheduling model of multiple virtual power plants under non-cooperative game established in this invention patent, the DMPC scheduling results and the feasibility of the DMPC strategy considering the competition relationship are analyzed respectively, and the corresponding results are obtained.

[0130] The parameter situation is as follows: The internal unit parameters of each virtual power plant are shown in Table 1, and the time-of-use electricity price is shown in Table 2. According to the scenario analysis method, the output of photovoltaic units in a certain actual area is clustered into 4 scenarios, and a typical scenario with the highest occurrence probability is selected as the reference trajectory of renewable energy output, such as Figure 2 shown. The load power in each virtual power plant is as Figure 3 shown. The operating loss cost of energy storage is taken as 0.05 yuan / kWh, the power generation cost of a micro gas turbine is taken as 0.66 yuan / kWh, and the unit compensation cost of interruptible load is taken as 0.8 yuan / kWh. In the ADMM algorithm, considering that the number of convergence times is affected by the step size ρ, the step size is taken as 1.5, the maximum number of iterations is 50 times, and the convergence accuracy is 10 -3 .

[0131] Table 1 Internal capacity parameters of each virtual power plant

[0132]

[0133] Table 2 Time-of-use electricity price

[0134]

[0135] The calorific value of natural gas is taken as 35564 kJ / m3, the carbon dioxide generation rate is 1.885 kg / m3, the carbon emission coefficient of the reference standard coal for external power purchase is taken as 0.7476 kgCO2 / kg standard coal, and the stepped carbon trading price is shown in Table 3. The proposed DMPC strategy based on non-cooperative game is applied to the multi-virtual power plant system constructed above.

[0136] Table 3 Stepped carbon trading price

[0137]

[0138] The electricity market trading situation within a day is as Figure 4 shown. Positive power represents purchase, and negative represents sale. For virtual power plants, the price of buying electricity from adjacent virtual power plants is lower than that of buying electricity from the large power grid, and the price of selling surplus electricity to adjacent virtual power plants is higher than that of selling it to the large power grid. Since virtual power plant 3 has a serious power shortage and can accept buying more energy at a higher price, it basically assumes the role of the buyer. And virtual power plant 2 basically assumes the role of the seller. Moreover, under the established electricity trading strategy considering competitive relationships, the resource-advantaged party cannot monopolize, and each virtual power plant has the opportunity to participate in each round of trading, enabling each participant to obtain a satisfactory result in the final transaction settlement stage. The non-cooperative game algorithm can encourage virtual power plants to quote competitively and independently adjust their quotes according to their own interests to meet their own maximum interests.

[0139] Such as Figures 5 - 7As shown, the operating cost of the virtual power plant has been reduced to a certain extent, and the most obvious changes come from virtual power plant 1 and virtual power plant 3. The operating cost curve of virtual power plant 2 has decreased significantly during the period from 6:00 to 16:00. This is because under the non - cooperative game algorithm, the virtual power plant makes independent decisions and flexibly adjusts its operating strategy according to factors such as market demand, price information, and its own resource status. By introducing competitive pressure, it promotes the virtual power plant to allocate and utilize resources more effectively.

[0140] Table 4 Statistical results of operating costs

[0141]

[0142] The total daily costs of the three virtual power plants are shown in Table 4. After adopting the game strategy considering the competitive relationship, the cost of virtual power plant 1 has decreased by 11.43% compared with that before the game, and the costs of virtual power plant 2 and virtual power plant 3 have decreased by 6.63% and 7.75% respectively. In each round of game iteration, each virtual power plant weighs the internal power generation cost and the electricity sales revenue in the non - cooperative game scenario of equal competition, and adjusts its strategy according to the game results of the previous moment. The independence and autonomy of individual virtual power plants are emphasized, and there is no relationship between leaders and followers. Through the competitive decisions of participants, it promotes the market to improve efficiency, and the market price and resource allocation can reach equilibrium more quickly.

[0143] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A distributed virtual power plant game optimization scheduling method, characterized in that: The method comprises the following steps: Step S1, first, construct a stepped carbon cost model, divide the step intervals according to the carbon emission quota of each virtual power plant, and set the carbon emission penalty price in different step intervals; secondly, construct a day-ahead optimization model for multiple virtual power plants, with the goal of minimizing the sum of the overall daily operating cost and the step-type carbon trading cost, and optimize the external grid interconnection line exchange power, micro gas turbine power generation, controllable load regulation and energy storage operation costs; then, construct a daily rolling optimization model for multiple virtual power plants, with the goal of minimizing the operating cost and the step-type carbon trading cost within the rolling period, and optimize the gas turbine power generation, energy storage operation and carbon emission costs; finally, construct a real-time feedback correction model for multiple virtual power plants, with the goal of minimizing the adjustable resource output adjustment, and use the synchronous alternating direction multiplier method for distributed optimization, calculate the control variable increment and update the output variable; Step S2: taking each virtual power plant as a game participant, defining a strategy set including a transaction power and a quotation set; Taking the lowest operating cost of each virtual power plant as the utility function, a non-cooperative game model of multiple virtual power plants is established. Step S3: Based on the electricity trading market mechanism, virtual power plants are divided into buyers or sellers according to their net power status, and transactions are matched in the form of many-to-many two-way auctions. Supply and demand balance is achieved through quotation rule judgment, transaction matching and circular transactions. The transaction process is defined, and the optimal scheduling of distributed virtual power plants under competitive relations is constructed. The operating cost function of virtual power plants is dynamically updated, and the scheduling strategy is optimized through multiple rounds of transactions until the market closing conditions are met.

2. A distributed virtual power plant game optimization scheduling method according to claim 1, characterized in that: The step S1 in constructing the stepwise carbon cost model is specifically as follows: According to the different carbon emission limits of each virtual power plant, it is used as the ladder division standard and the corresponding interval division is carried out; the ladder carbon emission cost model is as follows: In the formula, q c0 is the low carbon emission penalty price when the carbon emission value of the virtual power plant meets the quota value, q c1 is the first-tier carbon emission penalty price, q c2 is the second-tier carbon emission penalty price, q c3 is the third-tier carbon emission penalty price, g is the length of the carbon emission step interval, A trad It refers to the trading share of the system that actually participates in the carbon trading market during a carbon trading cycle.

3. A distributed virtual power plant game optimization scheduling method according to claim 1, characterized in that: The step S1 in which the multi-virtual power plant day-ahead optimization model is constructed is specifically as follows: Taking into account the randomness of load and renewable energy output, as well as the economic efficiency of operation, a day-ahead optimization model is constructed with the goal of minimizing the sum of the overall daily operation cost of multiple virtual power plants and the ladder-type carbon trading cost: In the formula, is the power exchange cost between the virtual power plant as a whole and the external power grid interconnection line, is the power generation cost of micro gas turbine j in the virtual power plant at time t, is the cost of controllable load regulation at time t, is the operating cost of energy storage j at time t, γ is the carbon emission coefficient of the unit power limit of the virtual power plant, C RQcarbon,j,t Putting a cost on carbon emissions.

4. A distributed virtual power plant game optimization scheduling method according to claim 1, characterized in that: The construction of the intraday rolling optimization model of multiple virtual power plants in step S1 is specifically as follows: In the intraday rolling optimization stage, the model is established with the goal of minimizing the operating costs and ladder-type carbon trading costs of multiple virtual power plants within the period [t0, t0+M△t]: Where, T RN is [t0, t0+M△t], They are respectively the power exchange cost between the virtual power plant as a whole and the external grid interconnection line, the gas turbine power generation cost, the carbon emission cost and the energy storage operation cost during the intraday rolling optimization phase.

5. A distributed virtual power plant game optimization scheduling method according to claim 1, characterized in that: The real-time feedback correction model of multiple virtual power plants is constructed in step S1 as follows: This stage is divided into 288 sampling periods with an interval of 5 minutes for real-time feedback correction, and the synchronous alternating direction multiplier method is used for distributed optimization in each period; In the real-time feedback correction stage, the model is established with the goal of minimizing the output adjustment of adjustable resources at the current moment: Where P i,MT,ref is the reference value of the micro gas turbine output in the ith subsystem, P i,ESS,ref is the reference value of the charging and discharging power of the energy storage device in the i-th subsystem, P i,MT is the predicted output value of the micro gas turbine in the ith subsystem, P i,ESS is the predicted output value of the charging and discharging power of the energy storage device in the i-th subsystem, △u i is the control variable increment of the ith subsystem; The optimization index of intraday rolling is set to the minimum weighted sum of the variance of the actual output and the reference trajectory and the variance of the control variable. The control variable of the system is solved according to the alternating direction multiplier method, and then the system output variable is calculated from the control variable.

6. A distributed virtual power plant game optimization scheduling method according to claim 1, characterized in that: The non-cooperative game model of multiple virtual power plants established in step S2 is specifically as follows: Step S2.1, each virtual power plant is regarded as a whole as a participant in the power market, and there is a non-cooperative game between them. The set of participants N can be re-expressed as N = {VPP1, VPP2, ..., VPPn}, where VPP is a distributed virtual power plant; Step S2.2: Considering that the transaction power between virtual power plants is related to the quotation, this paper takes the transaction power and the quotation as the strategy set of the game, that is, S = {S1, S2, …, S n },S n = {P jh,n ,q bid,n }, where P jh,n and q bid,n are the transaction power and transaction power prices of virtual power plant n and other virtual power plants respectively; Step S2.3: Taking the lowest operating cost of each virtual power plant at time t as the utility function, a non-cooperative game model of multiple virtual power plants is constructed.

7. A distributed virtual power plant game optimization scheduling method according to claim 1, characterized in that: The specific mechanism of the power trading market in step S3 is as follows: In a virtual power plant, the surplus of renewable energy output to load can be expressed as net power: P Net,i (t)=P RES,i (t)-P Load,i (t) (5) Where P RES,i (t) is the electricity generated by renewable energy, P Load,i is the load power consumption; when P Net,i When (t) < 0, it indicates a power shortage and requires more controllable power sources or buying power from other virtual power plants, which is called a buyer. Net,i When (t)>0, it indicates a power surplus, and the buyer needs to store extra energy, consume the surplus, or sell electricity to other virtual power plants, which is called the seller; due to the cost of power transportation, the power generation of the DG unit and the power consumption of the load must be balanced first; the two-way auction market participates in the transaction in a many-to-many form, and the buyers and sellers can submit quotation information at any time within the specified quotation time.

8. A distributed virtual power plant game optimization scheduling method according to claim 7, characterized in that: The specific transaction steps of the power trading market are as follows: Step S3.1, submission of quotation and product quantity information; Step S3.2, quotation rule judgment: judging whether the quotation is reasonable and whether to accept it; Step S3.3, transaction buyer and seller matching: matching according to transaction rules and confirming parameters such as price and quantity; Step S3.4, information disclosure: Publish the quotations, orders, historical transaction results and other relevant information of both parties to maintain market transparency; Step S3.5, cyclic trading: according to the trading rules, determine whether a new round of trading is needed. If so, return to step S3.1; if the seller's electricity cannot meet the buyer's demand, the supply and demand balance is achieved by purchasing electricity from the distribution network; By establishing a non-cooperative game model to analyze the two-way auction process, under the psychology of non-cooperative virtual power plant operators expecting to maximize their profits, two participants at the edge of the intersection set will reach a deal and use their quotations as a reference for the transaction; when there is more than one buyer or seller with the same bid, they should first be regarded as a whole to calculate the transaction volume, and then the transaction volume should be specifically allocated according to their respective net power bid ratios.

9. A distributed virtual power plant game optimization scheduling method according to claim 1, characterized in that: The optimization scheduling of distributed virtual power plants considering competition in step S3 is specifically as follows: Based on the established DMPC virtual power plant model, an optimal scheduling strategy for virtual power plants considering competition is constructed. The DMPC is a distributed model predictive control. The sum of the power shortages of all buyers’ virtual power plants is defined as: The transaction volume for a transaction is: In the formula, and are the total net power of sellers and buyers in completed transactions respectively; σ is the transaction status, with a value of 1 for buyers and a value of 0 for sellers; When ω∈{0,1}, there is only one buyer; when P short (t), There are other buyers in the market waiting for deals at this time; Considering information security and privacy, information transmission is restricted, and only necessary information is exchanged between virtual power plants: m i (t)=[i,|P Net (t)|,price,ID] (8) Where T is the length of the optimization layer, t∈T, i is the code of the virtual power plant, price is the bid price, ID is the identity, ID=[D sale , D buy ], D sale For sellers, D buy are buyers, all are binary integer variables; The information set shared with other virtual power plants is: MES i =[m i (1),m i (2),…,m i (T)] (9) After receiving the shared information from other virtual power plants, it enters the power trading market. After the transaction process, the MPC optimization strategy of the virtual power plant itself will also change, and the operating cost function becomes: C trad,i,t =∑price t (l)·P trad,i,t (l) (11) In the formula, l is the transaction round, price t (l) is the price of the transaction.

10. A distributed virtual power plant game optimization scheduling method according to claim 9, characterized in that: The transaction process is as follows: each operator will recalculate the operating cost when both parties confirm a round of transactions. If both parties are satisfied with the transaction, they can confirm and broadcast it. Once either party is dissatisfied with the result, the transaction will be cancelled, and the dissatisfied party will withdraw and wait for the next bid. Confirming the transaction means the end of this round of transactions, the identity value will be reset to zero, and the virtual power plant will re-enter the next round of transactions according to its own updated status. The above process is repeated until the conditions for opening the market are no longer met. When the transaction is completed, each virtual power plant transmits electricity; then it enters the independent operation state, obtains the optimal dispatch plan, and inputs the control quantity u=[u1,u2,…,u MΔt ], and then feedback correction is performed in real time on a smaller time scale until the cycle ends at t = t + MΔt.

Images