Multi-user virtual power plant scheduling method integrating risk seeking preference or disgust preference

Abstract

The invention relates to the technical field of virtual power plant optimal scheduling methods, in particular to a multi-user virtual power plant scheduling method integrating risk seeking preference or disgust preference, which comprises the following steps of: establishing a virtual power plant operator model in a deterministic scene; establishing a virtual power plant operator economic dispatching model containing the optimal value and the conditional value-at-risk; establishing a heating ventilation air conditioner user economic dispatching model; based on the game theory, establishing a non-cooperative game economic dispatching model of a virtual power plant operator-multiple heating, ventilation and air conditioning users; and based on an ADMM algorithm, carrying out distributed solution on the non-cooperative game economic dispatching model of the virtual power plant operator-multiple heating ventilation air conditioner users. According to the method, a new thought is provided for the scheduling operation decision of the virtual power plant operator under the influence of multiple uncertainties, and a method is also provided for the energy transaction optimization of the virtual power plant operator and the energy purchasing user.

Description

Technical Field

[0001] This invention relates to the technical field of virtual power plant optimization scheduling methods, and is a multi-user virtual power plant scheduling method that integrates risk-based preference or aversion preference. Background Technology

[0002] Currently, the energy crisis and global warming are becoming increasingly prominent issues, attracting widespread attention. A Virtual Power Plant (VPP) is a virtual power plant that aggregates a large number of distributed resources using advanced communication technologies and control theories. Through internal optimization and control, it coordinates the operation of various types of distributed resources, such as a large number of new energy sources and large-scale electric vehicles, achieving efficient allocation of internal resources.

[0003] Existing research on optimal scheduling of virtual power plants mainly focuses on managing the risks arising from uncertainties in the source-load-price environment during scheduling operations. Generally, decision-makers employ robust optimization, distributed robust optimization, and conditional value at risk (VaR) methods to manage risks associated with uncertainty. These methods can also be applied to the optimal scheduling of virtual power plants. However, in these methods, the decision-maker's risk preference is typically set to risk-neutral or risk-averse. In actual operation of virtual power plants, operators may seek to take risks in order to obtain unexpected profits when faced with uncertainty. Therefore, it is necessary to develop a stochastic scheduling model for virtual power plants with an awareness of unexpected profits for practical decision-making by virtual power plant operators.

[0004] Huang Bin (Huang Bin, Jiang Long, Bi Ke, et al. Optimization method for integrated energy dispatch of virtual power plants considering comprehensive demand response [J]. Electrical Age, 2024, (03): 27-30.) constructs a virtual power plant integrated energy model, and the integrated energy dispatch during the operation of the integrated energy microgrid is realized through the exchange of different energy sources. An objective function is constructed to optimize the general energy microgrid. Based on comprehensive demand response, the basic process of virtual power plant integrated energy dispatch optimization is designed to achieve integrated energy dispatch optimization.

[0005] Patent application CN119093353B discloses an optimized scheduling method for virtual power plants in a new energy power system. The method determines the objective function and corresponding constraints for the optimized scheduling of virtual power plants. Based on the objective function and constraints, an optimized scheduling model for virtual power plants is constructed. Simulated annealing is used to solve the optimized scheduling model to obtain the optimized scheduling scheme. In determining the objective function for the optimized scheduling of virtual power plants, a backward reduction method is used to reduce the wind-solar power generation scenario to establish the objective function, thus fully considering the correlation between wind and solar power generation while reducing computational complexity.

[0006] Patent application CN119419799B discloses a power dispatching method and related equipment for a virtual power plant. The method involves obtaining a distributed charging network graph, including a first light pole unit and a second light pole unit; obtaining the basic implicit representation vectors of the first and second light pole units; processing the basic implicit representation vectors of the first and second light pole units based on a machine learning network to determine the inference relationship between the first and second light pole units in the distributed charging network graph; adjusting the machine learning network based on the actual and inference relationships between the first and second light pole units in the distributed charging network graph; and determining the implicit representation vector of each light pole unit in the distributed charging network graph based on the adjusted machine learning network. Summary of the Invention

[0007] This invention provides a multi-user virtual power plant scheduling method that integrates risk-based preferences or aversion preferences, overcoming the shortcomings of the prior art and offering a new approach to virtual power plant operators' scheduling and operation decisions under multiple uncertainties.

[0008] The technical solution of this invention is achieved through the following measures: a multi-user virtual power plant scheduling method that integrates risk-based preference or aversion preference, comprising the following steps:

[0009] S01: Establish a virtual power plant operator model under deterministic scenarios;

[0010] S02: Considering the uncertainty of source-load-price scenario, based on the optimal value theory, the potential optimal situation is quantified and incorporated into the overall dispatch decision framework of the virtual power plant operator. Based on conditional risk value, extreme losses are quantified and incorporated into the overall dispatch decision framework of the virtual power plant operator. Finally, an economic dispatch model of the virtual power plant operator containing optimal value and conditional risk value is established.

[0011] S03: Establish an economic dispatch model for HVAC users;

[0012] S04: Based on game theory, establish a non-cooperative game-theoretic economic dispatch model for virtual power plant operators and multiple HVAC users;

[0013] S05: Based on the ADMM algorithm, a distributed solution is provided for the non-cooperative game-theoretic economic scheduling model of virtual power plant operators and multiple HVAC users.

[0014] The following are further optimizations and / or improvements to the above-mentioned technical solution:

[0015] Furthermore, in S01, the virtual power plant operator model in the deterministic scenario includes four types, specifically:

[0016] The first type: Virtual power plant operator and external power grid interaction model:

[0017]

[0018] μ buy +μ sell ≤1 (3)

[0019] in, Electricity purchased by operators from external power grids; Electricity sold by operators to external power grids; μ buy This is the status bit for the operator purchasing electricity from the external power grid; μ sell This is the status bit for operators selling electricity to the external power grid; This represents the maximum power available for interaction with the external power grid.

[0020] The second type: Interaction model between virtual power plant operators and other users:

[0021]

[0022] in, The electricity traded between the virtual power plant operator and user i at time t; The upper limit of the transaction volume between the virtual power plant operator and user i at time t;

[0023] The third type: an energy storage system model, which constrains the charging and discharging power of the energy storage system aggregated by the virtual power plant.

[0024]

[0025]

[0026] μ ech +μ edis ≤1 (11)

[0027] in, Real-time energy storage capacity for electrical energy storage; The charging efficiency of electrical energy storage; The discharge efficiency of electrical energy storage; The charging power for electrical energy storage; The discharge power of electrical energy storage; μ ech The state bit for charging the electrical energy storage; μ edis This is the state position for discharging stored electrical energy; This is the upper limit of the real-time storage capacity of electrical energy storage. This is the lower limit of the real-time energy storage capacity of electrical energy storage; The upper limit of the charging power for electrical energy storage; This represents the upper limit of the discharge power of the electrical energy storage. This represents the energy storage capacity at moment 1. The energy storage capacity at time 0; The charging power of the stored electrical energy at moment 1; This represents the energy storage capacity from time 2 to time 24. This represents the energy storage capacity at times 1 to 23. The charging power of the stored energy from time 2 to time 24; This represents the discharge power of the stored energy from time 2 to time 24. The energy storage capacity at time 24;

[0028] The fourth type: Gas turbine model, where the gas turbine burns natural gas to generate heat energy, and its heat output and constraints are as follows:

[0029]

[0030] in, For the power generation of gas turbines; The power generation efficiency of the gas turbine; The amount of gas consumed by the gas turbine; The upper limit for gas turbine power generation; H LHV It has a low calorific value for natural gas;

[0031] Fifth type: Power balance constraint

[0032]

[0033] in, Electricity purchased by operators from external power grids; Electricity sold by operators to external power grids; For the power generation of gas turbines; The discharge power of electrical energy storage; The charging power for electrical energy storage; Contribute to renewable energy forecasting; To predict load power; This represents the electricity traded between the virtual power plant operator and user i at time t.

[0034] Furthermore, in step S02, a virtual power plant operator economic dispatch model containing optimal value and conditional risk value is finally established. The specific steps are as follows:

[0035] Step 1: The objective function of the stochastic optimization model used by risk-seeking decision-makers is:

[0036]

[0037] Where x is a vector of variables; ζ is a vector of uncertain parameters; f(x,ζ) is the expected profit; Eζ {f(x,ζ)} is the expected value of f(x,ζ); B ζ {f(x,ζ)} is a statistical measure of high profits under the optimal expected profit distribution; β rs It is the risk-seeking parameter; the optimal value, abbreviated as VAB, is a statistical measure, defined as:

[0038]

[0039] Where, α VAB It is the probability parameter of VAB, α VAB ∈(0,1); VAB(α) VAB The value of x is equal to or greater than η, guaranteeing that the result is equal to or greater than η. b The probability of profit is not less than α. VAB Maximum η b In this case, VAB(α) VAB x) can be considered as being in α VAB The lower bound of high profit under the optimal scenario of ×100% expected profit distribution allows risk-seeking decision-makers to choose α. VAB The value of is determined to achieve an ideal trade-off between total expected profit and high profit in the best scenario;

[0040] The "optimal value" (VAB) refers to the lower bound of the best expected profit for a Virtual Power Plant Operator (VPP Operator). It is a statistical measure of the lower bound of the "high profit" portion that a VPP operator can achieve in the expected profit distribution under all uncertain scenarios (source-load-price fluctuations).

[0041] Step 2: Simultaneously considering the high profit under the best-case scenario and the risk under the worst-case scenario, based on the optimal value and risk metric, we propose an objective function, expressed as:

[0042]

[0043] Among them, R ζ {f(x,ζ)} is the risk measure of f(x,ζ), β ra It is the risk aversion parameter, β ra ∈(0,1); β ra A larger β indicates a higher level of risk aversion among participants. Therefore, the sum of all weight parameters in the objective function should equal 1. ra =β rs When β = 0, a risk-neutral model is obtained; when β = 0, a risk-neutral model is obtained. ra When β = 0, the risk-seeking model is obtained; when β = 0, the risk-seeking model is obtained. rs When = 0, the risk aversion model is obtained;

[0044] The objective function of the virtual power plant is the weighted sum of expected profit, optimal value, and conditional risk value, expressed as:

[0045]

[0046] Here, Ξ is the set of decision variables consisting of all variables, and the risk seeks parameter β. rs and risk aversion parameter β ra These are the weights assigned to the optimal value and the conditional risk value, respectively; pr w Let w be the probability of scenario w; For the profit of the virtual power plant in scenario w; π CVAR Conditional Value at Risk (VaR) (an indicator of extreme losses);

[0047] In scenario w, the expected profit of the virtual power plant is expressed as:

[0048]

[0049] in, Electricity purchased for virtual power plants; Electricity sales volume of the virtual power plant; For virtual power plant electricity purchase price; The electricity price of the virtual power plant; p gas For gas price; p e This represents the operation and maintenance cost coefficient for energy storage. This refers to gas consumption. The charging power at time t in scenario w; Let be the discharge power at time t in scenario w;

[0050] The conditional value at risk model is represented by the constraints in the first three rows of the following equation, π VAB The value is considered as in α VAB The lower bound of high profit in the best-case scenario of ×100% expected profit, assuming the expected profit distribution is known, is used to calculate π. VAB The closed form is Equation (16). In the scenario-based stochastic optimization problem, the expected profit distribution is unknown and determined by the decision variables of the virtual power plant. In this case, by maximizing π in the objective function Equation (18) VAB And include the constraints in rows 5 through 7 of the following formula to calculate π. VAB ;

[0051]

[0052] Among them, pr w Let g be the probability of scenario w; w ζ represents the loss level below ζ in scenario w; M is a sufficiently large constant used for logical constraints; Scenario selection variables, if This scenario then participates in the calculation of π. VAB w represents the scene;

[0053] The Conditional Value at Risk (CVaR) model refers to the expected value of extreme losses that a virtual power plant may suffer under uncertain scenarios, used to measure the operator's risk aversion tendency. The mathematical form of this Conditional Value at Risk model in this paper is the constraint in the first to third rows of formula (20), which introduces an auxiliary variable to represent the VaR threshold and a loss variable g. w The expected loss below VaR is measured, achieving linear equivalence modeling of CVaR. In the objective function (18), CVaR is expressed with weight parameter β. ra The addition of these elements forms part of the objective function of the virtual power plant, enabling a risk-controlled, profit-maximizing scheduling strategy.

[0054] Step 3: The virtual power plant economic dispatch model considering conditional risk value and optimal value is as follows:

[0055]

[0056] Where st is a constraint, and equations (1)-(14) and (19) are constraints of the virtual power plant economic dispatch model that considers conditional risk value and optimal value.

[0057] Furthermore, in step S03, the step of establishing the HVAC user economic dispatch model is as follows:

[0058] Step 1: Determine the objective function for user i

[0059]

[0060] in: π represents the operating cost of user i; TC Indicates a penalty price for discomfort due to heat; The power exchange between user i and the virtual power plant operator; The electricity trading price between user i and virtual power plant operator; PMV i,t For thermal comfort index;

[0061] Step 2: Constraints

[0062] Constraint 1: Thermal Comfort Constraint

[0063]

[0064] Among them: PMV t For user thermal comfort indicators; θ human The average temperature of human skin in a comfortable state; θi,t Indoor temperature i represents the user's skin temperature at which comfort is achieved; M represents the human body's energy metabolism rate; R represents the indoor temperature at which the user's skin is in a comfortable state. cloth Thermal resistance of clothing; PMV min The lower limit of the user's thermal comfort index; PMV max This represents the upper limit of the user's thermal comfort index.

[0065] Constraint 2: Constraints on Electric Heating Systems

[0066]

[0067] Where: ΔT represents the time interval; C represents the heat capacity; R represents the thermal resistance; This represents the ambient temperature of the HVAC system for user i at time t. This represents the power consumption of the HVAC system for user i at time t. This represents the electrical heating power consumption of the HVAC system for user i at time t. This represents the set temperature for user i during time period t; λ i This indicates the user i's tolerance to temperature. A Boolean variable (0 or 1) representing the HVAC cooling status of user i at time t; A Boolean variable (0 or 1) representing the HVAC thermal state of user i at time t; This indicates the lower limit of the power consumption for heating, ventilation, and air conditioning (HVAC) cooling systems. This indicates the upper limit of power consumption for HVAC (Heating, Ventilation, and Air Conditioning) electric cooling. Indicates the lower limit of the heat consumption power of electric heating; Indicates the upper limit of the heat consumption of electric heating; θ i,t+1 Let θ be the indoor temperature of user i at time t+1; t The indoor temperature during the current period is equivalent to θ. i,t+1 , used in state recursion formulas; The temperature rise of user i at time t caused by cooling / heating; Heating, ventilation and air conditioning system thermal efficiency;

[0068] Constraint 3: Constraints on electricity trading with virtual power plant operators

[0069]

[0070] In the formula, Let be the electrical power exchanged between user i and user j at time t; The upper limit of the power exchanged between user i and user j; and This is the lower limit of the power consumption exchanged between user i and user j;

[0071] Constraint 4: Power Balance Constraint

[0072]

[0073] in, Let be the electrical power exchanged between user i and user j at time t; This represents the power consumption of the HVAC system for user i at time t. These represent the electrical power consumption of the HVAC system for user i at time t.

[0074] Furthermore, in step S04, the specific steps are as follows:

[0075] Step 1: Considering the clearing price between user i and virtual power plant operators Optimal trading power between user i and virtual power plant operator By solving equations (20) and (21) above, we can obtain the equilibrium model of the entire system based on the following equations:

[0076]

[0077] In the formula, The optimal trading power between user i and the virtual power plant operator;

[0078] Step 2: The virtual power plant-multi-user non-cooperative game economic scheduling model is as follows:

[0079]

[0080] Where: λ i,t For the introduced dual variables, the first row of constraints indicates that the electricity consumption between user i and the virtual power plant operator and between the virtual power plant operator and user i is consistent. The electricity allocation value provided by the virtual power plant operator to user i at time t for price p; Let F be the power purchase request value provided by user i to the virtual power plant at time t; F is the economic dispatch objective function of the virtual power plant operator under risk preference; corresponding to the main function part in parentheses in formula (21);

[0081] assumed To obtain the optimal solution of equation (29) above, if the equilibrium model is derived under the clearing price, we get... Therefore, we get:

[0082]

[0083] Furthermore, in step S05, the specific solution steps are as follows:

[0084] Step 1: List the Lagrange function;

[0085]

[0086] Where: λ i,t ρ is the dual variable corresponding to the constraint condition in equation (29); i is the penalty factor for the Lagrange function;

[0087] Step 2: For virtual power plant operators, update their own variables by solving the local optimization model;

[0088]

[0089] The optimization model specifically refers to the sub-optimization problem that each participant solves independently based on its own local information under the ADMM algorithm framework: for virtual power plant operators, their local optimization model is a sub-problem containing equation (21);

[0090] Step 3: For each user i, update its own variables by solving the local optimization model;

[0091]

[0092] In the formula, This is an estimate of the transaction volume between user i and the virtual power plant operator in the previous iteration;

[0093] For each user i, its local optimization model is objective of Equation (22) and constrained by user-side electrothermal behavior and comfort requirements.

[0094] Step 4: Solve the optimized solutions obtained in steps 2 and 3. The Lagrange multipliers are further updated using the value of the vth iteration.

[0095]

[0096] In the formula, This represents the estimated trading volume between users and virtual power plant operators in the v-th iteration;

[0097] Step 5: Determine the convergence of the algorithm. If the convergence requirement is not met, return to step 2.

[0098]

[0099] Where: ε is the convergence accuracy, which is taken as 10 in this paper. -3 ;

[0100] Step 6: Output the optimal solution to obtain the optimal interactive power volume between the virtual power plant operator and the HVAC user.

[0101] This invention not only provides a new approach to the scheduling and operation decision-making of virtual power plant operators under multiple uncertainties, but also offers a method for optimizing energy transactions between virtual power plant operators and energy buyers. Furthermore, the risk parameters in the model can be dynamically adjusted to flexibly manage the unexpected gains, expected profits, and extreme losses of virtual power plants. The agreement on optimal energy trading strategies between virtual power plants and users through a non-cooperative game theory method driven by variational inequalities can create a favorable trading environment for all trading parties. Attached Figure Description

[0102] Figure 1 This is a flowchart of the scheduling method of the present invention;

[0103] Figure 2 This invention provides a multi-user VPP system framework.

[0104] Figure 3 This is an electrical load data diagram provided in an embodiment of the present invention;

[0105] Figure 4 This is a renewable energy data graph provided in an embodiment of the present invention;

[0106] Figure 5 This is a graph showing the electricity purchase price data of the VPP from the grid provided in this embodiment of the invention;

[0107] Figure 6 This is a graph showing the electricity price data for VPP selling electricity to the grid, provided in an embodiment of the present invention.

[0108] Figure 7 This is a graph showing the ambient temperature data of HVAC users provided in an embodiment of the present invention;

[0109] Figure 8 This is a diagram showing the results of electricity trading between a VPP and a HVAC user, provided in an embodiment of the present invention.

[0110] Figure 9 This is a diagram showing the electricity price transaction results between VPP and HVAC users provided in this embodiment of the invention;

[0111] Figure 10 This is a diagram showing the power optimization results of VPP in scenario 1 provided in this embodiment of the invention. Detailed Implementation

[0112] The present invention is not limited to the following embodiments, and the specific implementation can be determined according to the technical solution of the present invention and the actual situation.

[0113] In this invention, it should be noted that the terms "first," "second," "third," etc., are used only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the modules or elements referred to must have a specific order and operation, and therefore should not be construed as limiting the invention.

[0114] The present invention will be further described below with reference to embodiments:

[0115] Example 1: As Figure 1 , 2 As shown, a multi-user virtual power plant scheduling method that integrates risk-based preferences or aversion preferences includes the following steps:

[0116] S01: Establish a virtual power plant operator (VPPO) model under deterministic scenarios;

[0117] S02: Considering the uncertainty of source-load-price scenario, based on the optimal value theory, the potential optimal situation is quantified and incorporated into the overall dispatch decision framework of the virtual power plant operator. Based on conditional risk value, extreme losses are quantified and incorporated into the overall dispatch decision framework of the virtual power plant operator. Finally, an economic dispatch model of the virtual power plant operator containing optimal value and conditional risk value is established.

[0118] S03: Establish an economic dispatch model for HVAC users;

[0119] S04: Based on game theory, establish a non-cooperative game-theoretic economic dispatch model for virtual power plant operators and multiple HVAC users;

[0120] S05: Based on the ADMM algorithm, a distributed solution is provided for the non-cooperative game-theoretic economic scheduling model of virtual power plant operators and multiple HVAC users.

[0121] Specifically, in S01, the virtual power plant operator model in the deterministic scenario includes four types, specifically:

[0122] The first type: Virtual power plant operator and external power grid interaction model:

[0123]

[0124] μ buy +μ sell ≤1 (3)

[0125] in, Electricity purchased by operators from external power grids; Electricity sold by operators to external power grids; μ buy This is the status bit for the operator purchasing electricity from the external power grid; μ sellThis is the status bit for operators selling electricity to the external power grid; This represents the maximum power available for interaction with the external power grid.

[0126] The second type: Interaction model between virtual power plant operators and other users:

[0127]

[0128] in, The electricity traded between the virtual power plant operator and user i at time t; The upper limit of the transaction volume between the virtual power plant operator and user i at time t;

[0129] The third type: an energy storage system model, which constrains the charging and discharging power of the energy storage system aggregated by the virtual power plant.

[0130]

[0131] μ ech +μ edis ≤1 (11)

[0132] in, Real-time energy storage capacity for electrical energy storage; The charging efficiency of electrical energy storage; The discharge efficiency of electrical energy storage; The charging power for electrical energy storage; The discharge power of electrical energy storage; μ ech The state bit for charging the electrical energy storage; μ edis This is the state position for discharging stored electrical energy; This is the upper limit of the real-time storage capacity of electrical energy storage. This is the lower limit of the real-time energy storage capacity of electrical energy storage; The upper limit of the charging power for electrical energy storage; This represents the upper limit of the discharge power of the electrical energy storage. This represents the energy storage capacity at moment 1. The energy storage capacity at time 0; The charging power of the stored electrical energy at moment 1; This represents the energy storage capacity from time 2 to time 24. This represents the energy storage capacity at times 1 to 23. The charging power of the stored energy from time 2 to time 24; This represents the discharge power of the stored energy from time 2 to time 24. The energy storage capacity at time 24;

[0133] The fourth type: Gas turbine model, where the gas turbine burns natural gas to generate heat energy, and its heat output and constraints are as follows:

[0134]

[0135] in, For the power generation of gas turbines; The power generation efficiency of the gas turbine; The amount of gas consumed by the gas turbine; The upper limit for gas turbine power generation; H LHV It has a low calorific value for natural gas;

[0136] Fifth type: Power balance constraint

[0137]

[0138] in, Electricity purchased by operators from external power grids; Electricity sold by operators to external power grids; For the power generation of gas turbines; The discharge power of electrical energy storage; The charging power for electrical energy storage; Contribute to renewable energy forecasting; To predict load power; This represents the electricity traded between the virtual power plant operator and user i at time t.

[0139] Specifically, in step S02, the final step of establishing a virtual power plant operator economic dispatch model containing optimal value and conditional risk value is as follows:

[0140] Step 1: The objective function of the stochastic optimization model used by risk-seeking decision-makers is:

[0141]

[0142] Where x is a vector of variables; ζ is a vector of uncertain parameters; f(x,ζ) is the expected profit; E ζ {f(x,ζ)} is the expected value of f(x,ζ); B ζ {f(x,ζ)} is a statistical measure of high profits under the optimal expected profit distribution; β rs It is the risk-seeking parameter; the optimal value, abbreviated as VAB, is a statistical measure, defined as:

[0143]

[0144] Where, α VAB It is the probability parameter of VAB, α VAB ∈(0,1); VAB(α) VAB The value of x is equal to or greater than η, guaranteeing that the result is equal to or greater than η. bThe probability of profit is not less than α. VAB Maximum η b In this case, VAB(α) VAB x) can be considered as being in α VAB The lower bound of high profit under the optimal scenario of ×100% expected profit distribution allows risk-seeking decision-makers to choose α. VAB The value of is determined to achieve an ideal trade-off between total expected profit and high profit in the best-case scenario.

[0145] By simultaneously purchasing lottery tickets and insurance, decision-makers can simultaneously seek and averse to risk. This can be characterized using the novel concept of a combined risk-seeking preference or risk-averse preference. In this context, to maximize profit and minimize extreme loss under uncertainty, risk measures can be incorporated into the described risk-seeking stochastic optimization framework to construct the decision-making framework.

[0146] Step 2: Simultaneously considering the high profit under the best-case scenario and the risk under the worst-case scenario, based on the optimal value and risk metric, we propose an objective function, expressed as:

[0147]

[0148] Among them, R ζ {f(x,ζ)} is the risk measure of f(x,ζ), β ra It is the risk aversion parameter, β ra ∈(0,1); β ra A larger β indicates a higher level of risk aversion among participants. Therefore, the sum of all weight parameters in the objective function should equal 1. ra =β rs When β = 0, a risk-neutral model is obtained; when β = 0, a risk-neutral model is obtained. ra When β = 0, the risk-seeking model is obtained; when β = 0, the risk-seeking model is obtained. rs When = 0, the risk aversion model is obtained;

[0149] The objective function of the virtual power plant is the weighted sum of expected profit, optimal value, and conditional risk value, expressed as:

[0150]

[0151] Here, Ξ is the set of decision variables consisting of all variables, and the risk seeks parameter β. rs and risk aversion parameter β ra These are the weights assigned to the optimal value and the conditional risk value, respectively; pr w Let w be the probability of scenario w; For the profit of the virtual power plant in scenario w; π CVAR Conditional Value at Risk (VaR) is a measure of extreme losses.

[0152] In scenario w, the expected profit of the virtual power plant is expressed as:

[0153]

[0154] in, Electricity purchased for virtual power plants; Electricity sales volume of the virtual power plant; For virtual power plant electricity purchase price; The electricity price of the virtual power plant; p gas For gas price; p e This represents the operation and maintenance cost coefficient for energy storage. This refers to gas consumption. The charging power at time t in scenario w; Let be the discharge power at time t in scenario w;

[0155] The conditional value at risk model is represented by the constraints in the first three rows of the following equation, π VAB The value is considered as in α VAB The lower bound of high profit in the best-case scenario of ×100% expected profit, assuming the expected profit distribution is known, is used to calculate π. VAB The closed form is Equation (16). In the scenario-based stochastic optimization problem, the expected profit distribution is unknown and determined by the decision variables of the virtual power plant. In this case, by maximizing π in the objective function Equation (18) VAB And include the constraints in rows 5 through 7 of the following formula to calculate π. VAB .

[0156]

[0157] Among them, pr w Let g be the probability of scenario w; w ζ represents the loss level below ζ in scenario w; M is a sufficiently large constant used for logical constraints; Scenario selection variables, if This scenario then participates in the calculation of π. VAB w represents the scene;

[0158] Step 3: The virtual power plant economic dispatch model considering conditional risk value and optimal value is as follows:

[0159]

[0160] Where st is a constraint, and equations (1)-(14) and (19) are constraints of the virtual power plant economic dispatch model that considers conditional risk value and optimal value.

[0161] Specifically, in step S03, the steps for establishing the HVAC user economic dispatch model are as follows:

[0162] Step 1: Determine the objective function for user i

[0163]

[0164] in: π represents the operating cost of user i; TC Indicates a penalty price for discomfort due to heat; The power exchange between user i and the virtual power plant operator; The electricity trading price between user i and virtual power plant operator; PMV i,t For thermal comfort index;

[0165] Step 2: Constraints

[0166] Constraint 1: Thermal Comfort Constraint

[0167]

[0168] Among them: PMV t For user thermal comfort indicators; θ human The average temperature of human skin in a comfortable state; θ i,t Indoor temperature i represents the user's skin temperature at which comfort is achieved; M represents the human body's energy metabolism rate; R represents the indoor temperature at which the user's skin is in a comfortable state. cloth Thermal resistance of clothing; PMV min The lower limit of the user's thermal comfort index; PMV max This represents the upper limit of the user's thermal comfort index.

[0169] Constraint 2: Constraints on Electric Heating Systems

[0170]

[0171] Where: ΔT represents the time interval; C represents the heat capacity; R represents the thermal resistance; This represents the ambient temperature of the HVAC system for user i at time t. This represents the power consumption of the HVAC system for user i at time t. This represents the electrical heating power consumption of the HVAC system for user i at time t. This represents the set temperature for user i during time period t; λ i This indicates the user i's tolerance to temperature. A Boolean variable (0 or 1) representing the HVAC cooling status of user i at time t; A Boolean variable (0 or 1) representing the HVAC thermal state of user i at time t; This indicates the lower limit of the power consumption for heating, ventilation, and air conditioning (HVAC) cooling systems. This indicates the upper limit of power consumption for HVAC (Heating, Ventilation, and Air Conditioning) electric cooling. Indicates the lower limit of the heat consumption power of electric heating; Indicates the upper limit of the heat consumption of electric heating; θ i,t+1 Let θ be the indoor temperature of user i at time t+1; t The indoor temperature during the current period is equivalent to θ. i,t+1 , used in state recursion formulas; The temperature rise of user i at time t caused by cooling / heating; Heating, ventilation and air conditioning system thermal efficiency;

[0172] The first line of constraints indicates that the temperature for the next hour is a function of the indoor temperature, ambient temperature, and the temperature generated by the HVAC system within the same hour.

[0173] The second row of constraints represents the power consumption of the HVAC system;

[0174] The third constraint indicates that the initial temperature value must be set to a value determined by the user;

[0175] The upper and lower limits of indoor temperature are restricted by the constraints in the fourth row.

[0176] The fifth and sixth lines of constraints specify the power consumption constraints of the HVAC system;

[0177] The seventh line constraint prohibits simultaneous operation of HVAC cooling and heating.

[0178] Constraint 3: Constraints on electricity trading with virtual power plant operators

[0179]

[0180] In the formula, Let be the electrical power exchanged between user i and user j at time t; The upper limit of the power exchanged between user i and user j; and This is the lower limit of the power consumption exchanged between user i and user j;

[0181] Constraint 4: Power Balance Constraint

[0182]

[0183] in, Let be the electrical power exchanged between user i and user j at time t; This represents the power consumption of the HVAC system for user i at time t. These represent the electrical power consumption of the HVAC system for user i at time t.

[0184] Specifically, in step S04, the specific steps are as follows:

[0185] Step 1: Considering the clearing price between user i and virtual power plant operators Optimal trading power between user i and virtual power plant operator By solving equations (20) and (21) above, we can obtain the equilibrium model of the entire system based on the following equations:

[0186]

[0187] In the formula, The optimal trading power between user i and the virtual power plant operator;

[0188] Step 2: The virtual power plant-multi-user non-cooperative game economic scheduling model is as follows:

[0189]

[0190] Where: λ i,t For the introduced dual variables, the first row of constraints indicates that the electricity consumption between user i and the virtual power plant operator and between the virtual power plant operator and user i is consistent. The electricity allocation value provided by the virtual power plant operator to user i at time t for price p; Let F be the power purchase request value provided by user i to the virtual power plant at time t; F is the economic dispatch objective function of the virtual power plant operator under risk preference; corresponding to the main function part in parentheses in formula (21);

[0191] assumed To obtain the optimal solution of equation (29) above, if the equilibrium model is derived under the clearing price, we get... Therefore, we get:

[0192]

[0193] Specifically, in step S05, the specific solution steps are as follows:

[0194] Step 1: List the Lagrange function;

[0195]

[0196] Where: λ i,t ρ is the dual variable corresponding to the constraint condition in equation (29); i is the penalty factor for the Lagrange function;

[0197] Step 2: For virtual power plant operators, update their own variables by solving the local optimization model;

[0198]

[0199] Step 3: For each user i, update its own variables by solving the local optimization model;

[0200]

[0201] In the formula, This is an estimate of the transaction volume between user i and the virtual power plant operator in the previous iteration;

[0202] Step 4: Solve the optimized solutions obtained in steps 2 and 3. The Lagrange multipliers are further updated using the value of the vth iteration.

[0203]

[0204] In the formula, This represents the estimated trading volume between users and virtual power plant operators in the v-th iteration;

[0205] Step 5: Determine the convergence of the algorithm. If the convergence requirement is not met, return to step 2.

[0206]

[0207] Where: ε is the convergence accuracy, which is taken as 10 in this paper. -3 ;

[0208] Step 6: Output the optimal solution to obtain the optimal interactive power volume between the virtual power plant operator and the HVAC user.

[0209] Implementation Cases

[0210] The method described in this invention, within the Matlab compilation environment, establishes a variational inequality-driven non-cooperative game model between a virtual power plant (VPP) and three users (HVAC) using the Yalmip toolbox. It then calls the Gurobi and Cplex solvers to solve the interaction of electricity consumption and pricing between the VPP and the three users (HVAC). This invention considers energy sharing between the VPP and the three users, where both the VPP and the three users are pure electric models; the VPP is equipped with renewable energy units; and the gas price purchased by the VPP from the gas grid is 3.2 yuan / m³. 3 VPP energy storage device parameters are shown in Table 1; VPP gas turbine operating parameters are shown in Table 2; VPP source-load-price data are shown in Table 3. Figures 3-6 As shown, the ambient temperatures of the three users are as follows: Figure 7 As shown.

[0211] By performing a distributed solution to the non-cooperative game-theoretic economic scheduling model of a virtual power plant operator (VPP) and multiple HVAC users, the optimized interactive electricity volume and price between the VPP and the three HVAC users, the VPP's scheduling plan, and the operating costs of the VPP and the three HVAC users can be obtained, such as... Figure 8 , Figure 9 , Figure 10 As shown in Table 3.

[0212] Depend on Figure 8 Analysis shows that all three users purchase electricity from the virtual power plant, and the trends are basically consistent. During the periods of 0:00-9:00 and 17:00-24:00, the VPP's own electrical load level is low, while its renewable energy output is high. The VPP has more surplus electricity while meeting its own power balance, so the electricity sold by the VPP to the users increases during these periods. However, during the period of 10:00-16:00, the VPP's own electrical load is high, while its renewable energy output is low. The VPP has less surplus electricity while meeting its own power balance, so the electricity sold by the VPP to the users decreases during this period.

[0213] Depend on Figure 9 It is evident that the VPP and the three Users negotiate the optimal internal energy trading price through a non-cooperative game theory method driven by variational inequalities. This method creates a favorable trading environment for all participants, ensuring that the electricity trading price is strategically positioned between the grid time-of-use price and the feed-in price at different times. This, in turn, promotes beneficial energy trading among the participants, helping to improve the economic efficiency of each entity.

[0214] Since VPP optimizes 10 scenarios, this article takes scenario 1 as an example for analysis. Figure 10 It can be seen that there is varying degrees of electrical energy interaction between VPP and Users throughout the day; the renewable energy output of VPP also covers the entire dispatch period; around 12:00 noon and 20:00 in the evening, the demand for electricity is relatively high, so the gas turbine units of VPP output more to supplement the electrical energy; at the same time, VPP tends to purchase electricity from the upper-level grid during periods of relatively low electricity demand to reduce its own operating costs; the energy storage in VPP is charged during periods of low electricity demand and discharged during periods of high electricity demand to improve the economic benefits of the system.

[0215] Table 3 shows the operating costs of the VPP and the three users. As can be seen from Table 3, all three users purchase electricity from the VPP. The three users pay the VPP operator RMB 992.78, RMB 1075.00, and RMB 1113.45 respectively, while the VPP operator charges a total of RMB 3181.23 in electricity transaction fees from the three users. Ultimately, the total costs for the VPP and the three users are RMB 7783.12, RMB 990.21, RMB 1073.02, and RMB 1112.49 respectively.

[0216] To verify the effectiveness of the multi-user virtual power plant scheduling method proposed in this invention, which comprehensively considers different risk preferences, the following four sets of cases were set up to compare and analyze the scheduling results with different risk preferences:

[0217] Case 1: Risk Neutral (β) rs =β ra =0);

[0218] Case 2: Risk-seeking (β) rs =0.2, β ra =0, α VaB =0.1);

[0219] Case 3: Risk Aversion (β) rs =0,β ra =0.2, α CVaR =0.9);

[0220] Case 4: Taking into account both risk pursuit and risk aversion (β) rs =0.2, β ra =0.2, α VaB =0.1, α CVaR =0.9), which is the strategy in this paper (i.e., the method described in this invention), and the numerical simulation results mentioned above are all based on this.

[0221] Table 4 shows the scheduling results for Cases 1-4 considering different risk preferences. As can be seen from Table 4, Case 2 employs a risk-seeking scheduling strategy, while Case 3 employs a risk-averse strategy. Therefore, Case 2 has the lowest cost, while Case 3 has the highest cost. Case 1 is a risk-neutral scheduling strategy, and Case 4 is a strategy that comprehensively considers both risk-seeking and risk-averse factors. Therefore, their costs fall between those of Case 2 and Case 3. This indicates that the proposed strategy is reasonable and feasible, meaning that the multi-user virtual power plant scheduling method that integrates risk-seeking and risk-averse preferences described in this invention is reasonable and feasible.

[0222] In summary, the multi-user virtual power plant scheduling method for integrating risk-seeking preferences and risk-aversion preferences provided by this invention demonstrates that the risk-seeking preferences and risk-aversion preferences of VPP operators can be fully satisfied simultaneously, and the economic benefits of all stakeholders in the multi-user VPP are improved. Furthermore, risk parameters can be dynamically adjusted, flexibly managing unexpected gains, expected profits, and extreme losses of the VPP.

[0223] The above technical features constitute various embodiments of the present invention, which have strong adaptability and implementation effect. Unnecessary technical features can be added or removed according to actual needs to meet the needs of different situations.

[0224] Table 1 Basic Operating Parameters of Gas Turbine

[0225]

[0226] Table 2 System Energy Storage Parameters

[0227]

[0228] Table 3 shows the operating costs of VPP and 3 users.

[0229] cost VPP User1 User2 User3 Non-cooperative game non-transaction cost / yuan 10964.35 -2.57 -1.98 -0.96 Electricity transaction cost / yuan -3181.23 992.78 1075.00 1113.45 Final cost of non-cooperative game / yuan 7783.12 990.21 1073.02 1112.49

[0230] Table 4 shows the scheduling results for Cases 1-4 considering different risk preferences.

[0231] Case VAB (yuan) CVAR (RMB) Cost (RMB) 1 9343.60 -10984.97 10984.52 2 -9993.47 -10978.83 10781.50 3 9343.60 -11873.20 11163.18 4 -10000.00 -11871.46 10964.35

Claims

1. A multi-user virtual power plant scheduling method that integrates risk-based preference or aversion preference, characterized in that, Includes the following steps: Establish a virtual power plant operator model under deterministic scenarios; Establish an economic dispatch model for virtual power plant operators that includes optimal value and conditional risk value; Establish an economic dispatch model for HVAC users; Based on game theory, a non-cooperative game-theoretic economic scheduling model is established for virtual power plant operators and multiple HVAC users. Based on the ADMM algorithm, a distributed solution is performed for the non-cooperative game-theoretic economic scheduling model of virtual power plant operators and multiple HVAC users.

2. The multi-user virtual power plant scheduling method based on comprehensive risk-based preference or aversion preference as described in claim 1, characterized in that, In deterministic scenarios, there are four types of virtual power plant operator models, specifically: The first type: Virtual power plant operator and external power grid interaction model: m buy +m sell ≤1 (3) in, Electricity purchased by operators from external power grids; Electricity sold by operators to external power grids; μ buy This is the status bit for the operator purchasing electricity from the external power grid; μ sell This is the status bit for operators selling electricity to the external power grid; This represents the maximum power available for interaction with the external power grid. The second type: Interaction model between virtual power plant operators and other users: in, The electricity traded between the virtual power plant operator and user i at time t; The upper limit of the transaction volume between the virtual power plant operator and user i at time t; The third type: an energy storage system model, which constrains the charging and discharging power of the energy storage system aggregated by the virtual power plant. m ech +m edis ≤1 (11) in, Real-time energy storage capacity for electrical energy storage; The charging efficiency of electrical energy storage; The discharge efficiency of electrical energy storage; The charging power for electrical energy storage; The discharge power of electrical energy storage; μ ech The state bit for charging the electrical energy storage; μ edis This is the state position for discharging stored electrical energy; This is the upper limit of the real-time storage capacity of electrical energy storage. This is the lower limit of the real-time energy storage capacity of electrical energy storage; The upper limit of the charging power for electrical energy storage; This represents the upper limit of the discharge power of the electrical energy storage. This represents the energy storage capacity at moment 1. The energy storage capacity at time 0; The charging power of the stored electrical energy at moment 1; This represents the energy storage capacity from time 2 to time 24. This represents the energy storage capacity at times 1 to 23. The charging power of the stored energy from time 2 to time 24; This represents the discharge power of the stored energy from time 2 to time 24. The energy storage capacity at time 24; The fourth type: Gas turbine model, where the gas turbine burns natural gas to generate heat energy, and its heat output and constraints are as follows: in, For the power generation of gas turbines; The power generation efficiency of the gas turbine; The amount of gas consumed by the gas turbine; The upper limit for gas turbine power generation; H LHV It has a low calorific value for natural gas; Fifth type: Power balance constraint in, Electricity purchased by operators from external power grids; Electricity sold by operators to external power grids; For the power generation of gas turbines; The discharge power of electrical energy storage; The charging power for electrical energy storage; Contribute to renewable energy forecasting; To predict load power; This represents the electricity traded between the virtual power plant operator and user i at time t.

3. The multi-user virtual power plant scheduling method based on comprehensive risk-based preference or aversion preference as described in claim 2, characterized in that, Finally, an economic dispatch model for virtual power plant operators, incorporating optimal values ​​and conditional risk values, is established. The specific steps are as follows: Step 1: The objective function of the stochastic optimization model used by risk-seeking decision-makers is: Where x is a vector of variables; ζ is a vector of uncertain parameters; f(x,ζ) is the expected profit; E ζ {f(x,ζ)} is the expected value of f(x,ζ); B ζ {f(x,ζ)} is a statistical measure of high profits under the optimal expected profit distribution; β rs It is the risk-seeking parameter; the optimal value, abbreviated as VAB, is a statistical measure, defined as: Where, α VAB It is the probability parameter of VAB, α VAB ∈(0,1); VAB(α) VAB The value of x is equal to or greater than η, guaranteeing that the result is equal to or greater than η. b The probability of profit is not less than α. VAB Maximum η b In this case, VAB(α) VAB x) is considered as being in α VAB The lower bound of high profit under the optimal scenario of ×100% expected profit distribution; risk-seeking decision-makers choose α. VAB The value of is determined to achieve an ideal trade-off between total expected profit and high profit in the best scenario; Step 2: Simultaneously considering the high profit under the best-case scenario and the risk under the worst-case scenario, based on the optimal value and risk metric, we propose an objective function, expressed as: Among them, R ζ {f(x,ζ)} is the risk measure of f(x,ζ), β ra It is the risk aversion parameter, β ra ∈(0,1); β ra A larger β indicates a higher level of risk aversion among participants. Therefore, the sum of all weight parameters in the objective function should equal 1. ra =β rs When β = 0, a risk-neutral model is obtained; when β = 0, a risk-neutral model is obtained. ra When β = 0, the risk-seeking model is obtained; when β = 0, the risk-seeking model is obtained. rs When = 0, the risk aversion model is obtained; The objective function of the virtual power plant is the weighted sum of expected profit, optimal value, and conditional risk value, expressed as: Here, Ξ is the set of decision variables consisting of all variables; risk seeker parameter β rs and risk aversion parameter β ra These are the weights assigned to the optimal value and the conditional risk value, respectively; pr w Let w be the probability of scenario w; For the profit of the virtual power plant in scenario w; π CVAR Conditional risk value; In scenario w, the expected profit of the virtual power plant is expressed as: in, Electricity purchased for virtual power plants; Electricity sales volume of the virtual power plant; For virtual power plant electricity purchase price; The electricity price of the virtual power plant; p gas For gas price; p e This represents the operation and maintenance cost coefficient for energy storage. This refers to gas consumption. The charging power at time t in scenario w; Let be the discharge power at time t in scenario w; The Conditional Value at Risk (VaR) model represents the expected value of extreme losses suffered by a virtual power plant under uncertain scenarios. The VaR model is represented by the constraints in the first three rows of the following equation: π VAB The value is considered as in α VAB The lower bound of high profit in the best-case scenario of ×100% expected profit, assuming the expected profit distribution is known, is used to calculate π. VAB The closed form is Equation (16). In the scenario-based stochastic optimization problem, the expected profit distribution is unknown and determined by the decision variables of the virtual power plant. In this case, by maximizing π in the objective function Equation (18) VAB And include the constraints in rows 5 through 7 of the following formula to calculate π. VAB ; Among them, pr w Let g be the probability of scenario w; w ζ represents the loss level below ζ in scenario w; M is a sufficiently large constant used for logical constraints; Choose variables for the scene, if This scenario then participates in the calculation of π. VAB w represents the scene; ζ represents a vector of uncertain parameters; Step 3: The virtual power plant economic dispatch model considering conditional risk value and optimal value is as follows: Where st is a constraint, and equations (1)-(14) and (19) are constraints of the virtual power plant economic dispatch model that considers conditional risk value and optimal value.

4. The multi-user virtual power plant scheduling method based on comprehensive risk-based preference or aversion preference as described in claim 2 or 3, characterized in that, The steps to establish an economic dispatch model for HVAC users are as follows: Step 1: Determine the objective function for user i in: π represents the operating cost of user i; TC Indicates a penalty price for discomfort due to heat; The power exchange between user i and the virtual power plant operator; The electricity trading price between user i and virtual power plant operator; PMV i,t For thermal comfort index; Step 2: Constraints Constraint 1: Thermal Comfort Constraint Among them: PMV t For user thermal comfort indicators; θ human The average temperature of human skin in a comfortable state; θ i,t Indoor temperature i represents the user's skin temperature at which comfort is achieved; M represents the human body's energy metabolism rate; R represents the indoor temperature at which the user's skin is in a comfortable state. cloth Thermal resistance of clothing; PMV min The lower limit of the user's thermal comfort index; PMV max This represents the upper limit of the user's thermal comfort index. Constraint 2: Constraints on Electric Heating Systems Where: ΔT represents the time interval; C represents the heat capacity; R represents the thermal resistance; This represents the ambient temperature of the HVAC system for user i at time t. This represents the power consumption of the HVAC system for user i at time t. This represents the electrical heating power consumption of the HVAC system for user i at time t. This represents the set temperature for user i during time period t; λ i This indicates the user i's tolerance to temperature. A Boolean variable representing the HVAC cooling status of user i at time t; A Boolean variable representing the HVAC thermal state of user i at time t; This indicates the lower limit of the power consumption for heating, ventilation, and air conditioning (HVAC) cooling systems. This indicates the upper limit of power consumption for HVAC (Heating, Ventilation, and Air Conditioning) electric cooling. Indicates the lower limit of the heat consumption power of electric heating; Indicates the upper limit of the heat consumption of electric heating; θ i,t+1 Let θ be the indoor temperature of user i at time t+1; t The indoor temperature during the current period is equivalent to θ. i,t+1 , used in state recursion formulas; The temperature rise of user i at time t caused by cooling / heating; Heating, ventilation and air conditioning system thermal efficiency; Constraint 3: Constraints on electricity trading with virtual power plant operators In the formula, Let be the electrical power exchanged between user i and user j at time t; The upper limit of the power exchanged between user i and user j; and This is the lower limit of the power consumption exchanged between user i and user j; Constraint 4: Power Balance Constraint in, Let be the electrical power exchanged between user i and user j at time t; This represents the power consumption of the HVAC system for user i at time t. These represent the electrical power consumption of the HVAC system for user i at time t.

5. The multi-user virtual power plant scheduling method based on comprehensive risk-based preference or aversion preference as described in claim 2 or 3, characterized in that, The specific steps for establishing a non-cooperative game-theoretic economic scheduling model based on game theory, involving a virtual power plant operator and multiple HVAC users, are as follows: Step 1: Considering the clearing price between user i and virtual power plant operators The optimal power transaction between user i and the virtual power plant operator can be obtained by solving equations (20) and (21) above; the equilibrium model of the entire system is obtained according to the following equations: In the formula, The optimal trading power between user i and the virtual power plant operator; Step 2: The virtual power plant-multi-user non-cooperative game economic scheduling model is as follows: Where: λ i,t For the introduced dual variables, the first row of constraints indicates that the electricity consumption between user i and the virtual power plant operator and between the virtual power plant operator and user i is consistent. The electricity allocation value provided by the virtual power plant operator to user i at time t for price p; Let F be the power purchase request value that user i provides to the virtual power plant at time t; F is the economic dispatch objective function of the virtual power plant operator under risk preference. assumed To obtain the optimal solution of equation (29) above, if the equilibrium model is derived under the clearing price, we get... Therefore, we get:

6. The multi-user virtual power plant scheduling method based on comprehensive risk-based preference or aversion preference as described in claim 4, characterized in that, The specific steps for establishing a non-cooperative game-theoretic economic scheduling model based on game theory, involving a virtual power plant operator and multiple HVAC users, are as follows: Step 1: Considering the clearing price between user i and virtual power plant operators The optimal power transaction between user i and the virtual power plant operator can be obtained by solving equations (20) and (21) above; the equilibrium model of the entire system is obtained according to the following equations: In the formula, The optimal trading power between user i and the virtual power plant operator; Step 2: The virtual power plant-multi-user non-cooperative game economic scheduling model is as follows: Where: λ i,t For the introduced dual variables, the first row of constraints indicates that the electricity consumption between user i and the virtual power plant operator and between the virtual power plant operator and user i is consistent. The electricity allocation value provided by the virtual power plant operator to user i at time t for price p; Let F be the power purchase request value that user i provides to the virtual power plant at time t; F is the economic dispatch objective function of the virtual power plant operator under risk preference. assumed To obtain the optimal solution of equation (29) above, if the equilibrium model is derived under the clearing price, we get... Therefore, we get:

7. The multi-user virtual power plant scheduling method based on comprehensive risk-based preference or aversion preference as described in claim 2 or 3, characterized in that, The ADMM algorithm is used to perform a distributed solution for the non-cooperative game-theoretic economic scheduling model of virtual power plant operators and multiple HVAC users. The specific solution steps are as follows: Step 1: List the Lagrange function; Where, π TC Indicates a penalty price for thermal discomfort; λ i,t ρ is the dual variable corresponding to the constraint condition in equation (29); i is the penalty factor for the Lagrange function; Step 2: For virtual power plant operators, update their own variables by solving the local optimization model; Step 3: For each user i, update its own variables by solving the local optimization model; In the formula, This is an estimate of the transaction volume between user i and the virtual power plant operator in the previous iteration; Step 4: Solve the optimized solutions obtained in steps 2 and 3. The Lagrange multipliers are further updated using the value of the vth iteration. In the formula, This represents the estimated trading volume between users and virtual power plant operators in the v-th iteration; Step 5: Determine the convergence of the algorithm. If the convergence requirement is not met, return to step 2. Where: ε is the convergence accuracy; Step 6: Output the optimal solution to obtain the optimal interactive power volume between the virtual power plant operator and the HVAC user.

8. The multi-user virtual power plant scheduling method based on comprehensive risk-based preference or aversion preference as described in claim 4, characterized in that, The ADMM algorithm is used to perform a distributed solution for the non-cooperative game-theoretic economic scheduling model of virtual power plant operators and multiple HVAC users. The specific solution steps are as follows: Step 1: List the Lagrange function; Where, π TC Indicates a penalty price for thermal discomfort; λ i,t ρ is the dual variable corresponding to the constraint condition in equation (29); i is the penalty factor for the Lagrange function; Step 2: For virtual power plant operators, update their own variables by solving the local optimization model; Step 3: For each user i, update its own variables by solving the local optimization model; In the formula, This is an estimate of the transaction volume between user i and the virtual power plant operator in the previous iteration; Step 4: Solve the optimized solutions obtained in steps 2 and 3. The Lagrange multipliers are further updated using the value of the vth iteration. In the formula, This represents the estimated trading volume between users and virtual power plant operators in the v-th iteration; Step 5: Determine the convergence of the algorithm. If the convergence requirement is not met, return to step 2. Where: ε is the convergence accuracy; Step 6: Output the optimal solution to obtain the optimal interactive power volume between the virtual power plant operator and the HVAC user.

9. The multi-user virtual power plant scheduling method based on comprehensive risk-based preference or aversion preference as described in claim 5, characterized in that, The ADMM algorithm is used to perform a distributed solution for the non-cooperative game-theoretic economic scheduling model of virtual power plant operators and multiple HVAC users. The specific solution steps are as follows: Step 1: List the Lagrange function; Where, π TC Indicates a penalty price for thermal discomfort; λ i,t ρ is the dual variable corresponding to the constraint condition in equation (29); i is the penalty factor for the Lagrange function; Step 2: For virtual power plant operators, update their own variables by solving the local optimization model; Step 3: For each user i, update its own variables by solving the local optimization model; In the formula, This is an estimate of the transaction volume between user i and the virtual power plant operator in the previous iteration; Step 4: Solve the optimized solutions obtained in steps 2 and 3. The Lagrange multipliers are further updated using the value of the vth iteration. In the formula, This represents the estimated trading volume between users and virtual power plant operators in the v-th iteration; Step 5: Determine the convergence of the algorithm. If the convergence requirement is not met, return to step 2. Where: ε is the convergence accuracy; Step 6: Output the optimal solution to obtain the optimal interactive power volume between the virtual power plant operator and the HVAC user.

10. The multi-user virtual power plant scheduling method based on comprehensive risk-based preference or aversion preference as described in claim 6, characterized in that, The ADMM algorithm is used to perform a distributed solution for the non-cooperative game-theoretic economic scheduling model of virtual power plant operators and multiple HVAC users. The specific solution steps are as follows: Step 1: List the Lagrange function; Where, π TC Indicates a penalty price for thermal discomfort; λ i,t ρ is the dual variable corresponding to the constraint condition in equation (29); i is the penalty factor for the Lagrange function; Step 2: For virtual power plant operators, update their own variables by solving the local optimization model; Step 3: For each user i, update its own variables by solving the local optimization model; In the formula, This is an estimate of the transaction volume between user i and the virtual power plant operator in the previous iteration; Step 4: Solve the optimized solutions obtained in steps 2 and 3. The Lagrange multipliers are further updated using the value of the vth iteration. In the formula, This represents the estimated trading volume between users and virtual power plant operators in the v-th iteration; Step 5: Determine the convergence of the algorithm. If the convergence requirement is not met, return to step 2. Where: ε is the convergence accuracy; Step 6: Output the optimal solution to obtain the optimal interactive power volume between the virtual power plant operator and the HVAC user.

Images