Heterogeneous virtual power plant cluster marketization transaction optimization method based on decision fusion

Abstract

The invention provides a heterogeneous virtual power plant cluster marketization transaction optimization method based on decision fusion, and the method comprises the following steps: firstly, constructing a framework and a transaction model of a virtual power plant cluster, and building an output model; carrying out clearing simulation on the spot market, and constructing a scheduling model; performing simulation calculation on the bid-winning electric quantity through the scheduling model; and then, establishing a virtual power plant cluster collaborative operation model, and obtaining transferred electric quantity and unit transaction cost among the virtual power plants through Nash bargaining so as to formulate a transaction plan. The method is simple and low in execution cost, the transaction strategy obtained by constructing the virtual power plant cluster collaborative operation optimization model and carrying out Nash bargaining and information gap robust decision fusion can reflect heterogeneity characteristics among virtual power plants, and the deficiency of calculation efficiency can be made up by accurately modeling strong random resources, so that the method can be applied to the collaborative operation of the virtual power plants. Even if wind and light fluctuation exists in the execution period, the whole cluster can still stably operate and supply power, and the overall power generation cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of power supply, and in particular to a market-based transaction optimization method for a heterogeneous virtual power plant cluster based on decision fusion. Background Art

[0002] The construction of new power systems plays a key role in promoting the development of distributed flexible resources. In the process of building a unified national power market system, the attention paid to distributed resource aggregation entities such as virtual power plants (VPPs) has been increasing. At present, the qualifications of spot market access entities have been gradually liberalized, allowing new entities such as independent energy storage and virtual power plants to participate in market transactions.

[0003] On the basis of the above, how to coordinate the collaborative operation of various entities in the new generation of smart grid has become the main research direction, so that it can optimize energy utilization efficiency to the greatest extent, protect social interests, and achieve environmental protection effects such as emission reduction. Specifically, it has become extremely urgent to study the market-oriented trading strategy of virtual power plants.

[0004] There are some problems with the trading methods or strategies produced by the existing virtual power plant market-based trading optimization models. On the one hand, most models do not fully reflect the heterogeneous characteristics among virtual power plants. On the other hand, when modeling the uncertainty of highly random resources such as wind, solar, and electric vehicles, there are deficiencies in accuracy and computational efficiency. In addition, in the face of wind and solar output deviations, there is currently a lack of research on coordinating the allocation of contracted electricity through cluster cooperation to avoid deviation assessment risks. In an environment where strong output uncertainty and high flexibility in demand response coexist, it is crucial to optimize virtual power plant trading strategies to ensure a balance between economy and robustness.

[0005] In summary, since distributed resource aggregation entities play an important role in the construction of new power systems and a national unified power market system, in the face of a cluster power supply network composed of multiple virtual power plants including distributed wind and solar power generation entities, it is necessary to establish an optimized trading method before entering the market transaction, which can effectively reflect the heterogeneity between virtual power plants and solve the accuracy and computational efficiency problems of uncertainty modeling of strongly random resources; then, in the current complex environment, a balance between economy and robustness can be achieved to ensure the sustainable development of virtual power plants. Summary of the invention

[0006] In view of the problems existing in the prior art, the present invention provides a market-oriented transaction optimization method for heterogeneous virtual power plant clusters based on decision fusion.

[0007] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0008] The present invention comprises the following steps:

[0009] Construct the cluster architecture, operation framework and peer-to-peer distributed transaction model of heterogeneous virtual power plant clusters in the market environment;

[0010] Establishing an output model of a virtual power plant cluster based on the cluster architecture and operation framework;

[0011] According to the group architecture, operation framework and output model, the virtual power plant cluster is simulated to clear in the spot market, and a dispatch model based on security constraints is constructed with the goal of minimizing power generation costs; the winning bid power of each virtual power plant in the virtual power plant cluster is calculated by simulating the dispatch model;

[0012] Based on the output uncertainty and deviation assessment penalty of new energy power plants in the virtual power plant cluster, a virtual power plant cluster collaborative operation model with the goal of minimizing the total deviation loss of virtual power plants is established; robust decision-making with information gaps is introduced into the virtual power plant cluster collaborative operation model;

[0013] According to the point-to-point distributed transaction model, the winning bid power and the virtual power plant cluster collaborative operation model, the transferred power and unit transaction cost between virtual power plants are obtained through Nash bargaining;

[0014] A trading plan is output based on the bid-winning electricity, the transferred electricity and the unit transaction cost, and market-based transactions of the virtual power plant cluster are carried out according to the trading plan.

[0015] Optionally, the virtual power plant includes a power supply type virtual power plant, a load type virtual power plant, and a hybrid virtual power plant;

[0016] The power supply type virtual power plant aggregates distributed wind and solar power and configures energy storage equipment;

[0017] The load-based virtual power plant aggregates electric vehicle users;

[0018] The hybrid virtual power plant aggregates distributed photovoltaic and demand response users.

[0019] Optionally, the output model of the virtual power plant cluster includes a wind and solar output model, an energy storage output model, a load output model, and a distributed photovoltaic output model.

[0020] Optionally, the load output model also includes a charging load model of an electric vehicle, and the construction method steps include:

[0021] Based on the Markov chain model, the electric vehicle spatial state characteristics are constructed, and based on the electric vehicle spatial state characteristics, the charging load model is constructed with the spatial transfer quantity.

[0022] Optionally, the method for predicting the number of electric vehicles in the charging load model includes the following steps:

[0023] Screen and process the travel data of electric vehicle users and obtain statistical data;

[0024] According to the statistical data, a target time range and measurement points are set, and an initial number of electric vehicles in each area is set;

[0025] Calculating a primary transfer state of the electric vehicle at each time point based on the target time range, the measurement point, and the initial number of electric vehicles, and obtaining a primary spatial transfer frequency matrix at each time point based on the primary transfer state;

[0026] Calculating the position transfer probability matrix at each time point based on the frequency of the spatial transfer frequency matrix;

[0027] Calculating the boundary probability through the primary spatial transfer frequency matrix and the position transfer probability matrix, and introducing a chi-square test method to verify the boundary probability and the Markov properties of the matrix elements of the primary spatial transfer frequency matrix and the position transfer probability matrix;

[0028] When the chi-square test method is passed, the number of electric vehicles is calculated based on the primary space transfer frequency matrix and the position transfer probability matrix.

[0029] Optionally, the safety constraints of the scheduling model include maximum output limit of power generators, output limit of quotation sections, node power balance and line transmission capacity limit.

[0030] Optionally, a wind-solar uncertainty model based on information gap robust decision-making is established in the virtual power plant cluster collaborative operation model;

[0031] The process of solving the wind and solar uncertainty model also includes the following steps:

[0032] Build deterministic models;

[0033] Introducing non-deterministic variables into the deterministic model to obtain a non-deterministic model;

[0034] Adding constraints to the non-deterministic model to obtain a wind-solar uncertainty model; the constraints are the minimum requirements of system decision makers;

[0035] According to the wind and solar uncertainty model, the optimal solution is found in extreme scenarios.

[0036] Optionally, the steps of calculating the transferred electricity amount include:

[0037] The solution of the wind-solar uncertainty model is substituted into the virtual power plant cluster collaborative operation model to obtain the transferred electricity.

[0038] Optionally, the steps of calculating the unit transaction cost include:

[0039] A virtual power plant benefit calculation model is constructed based on the transferred electricity; the virtual power plant benefit calculation model is constructed based on the Nash bargaining theory with the objective function as follows:

[0040]

[0041] In the formula, The transaction income of the virtual power plant during cooperative operation. is the transaction income of the virtual power plant when it operates independently, A is the set of virtual power plants in the virtual power plant cluster, and i is the ordinal number of the virtual power plant;

[0042] The objective function introduces auxiliary variables and decouples the objective function to obtain the unit transaction cost.

[0043] Optionally, the transaction income during the cooperative operation of the virtual power plant includes spot income, deviation assessment fees, point-to-point transaction transmission fees, sales income from contract transfers, and purchase costs from contract transfers.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] The method of the present invention is simple and has low execution cost. By constructing a virtual power plant cluster collaborative operation optimization model, after the fusion of Nash bargaining and information gap robust decision-making, the obtained trading strategy better reflects the heterogeneous characteristics between virtual power plants, and makes up for the lack of computing efficiency by accurately modeling strongly random resources. Even if there are wind and solar fluctuations during execution, the entire cluster can still operate stably and supply power, thereby reducing the overall power generation cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0047] Figure 1 is a flow chart of the method of the present invention;

[0048] Figure 2 It is the incentive-type demand response trapezoidal price diagram of the present invention;

[0049] Figure 3 This is a flow chart of the distributed solution of Nash bargaining in the present invention. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0052] It is worth noting that the methods used in the present invention are all conventional methods unless otherwise specified; the raw materials and devices used are all conventional commercially available products, and their sources are not specifically limited unless otherwise specified.

[0053] It should also be noted that in the following description of the technical solution, regarding repeated parameter identifiers (such as i, j, h, etc.), if there is no separate explanation in the formula description, the relevant instructions recorded in the full text shall be used; if there is a separate explanation, its meaning shall be understood by referring to the separate explanation.

[0054] like Figure 1 As shown, this embodiment provides a market-based transaction optimization method for heterogeneous virtual power plant clusters based on decision fusion, which includes the following steps:

[0055] 1. Build a cluster architecture, operation framework and peer-to-peer (P2P) distributed transaction model for a heterogeneous virtual power plant cluster in a market environment.

[0056] Optionally, the heterogeneity of the virtual power plant cluster is reflected in the existence of power plants with different functions and types in the power plant cluster. Therefore, the virtual power plants are divided into three categories according to their functions and types, namely, power supply type virtual power plants, load type virtual power plants, and hybrid virtual power plants.

[0057] Among them, the power supply type virtual power plant aggregates distributed wind and solar power and configures energy storage equipment. Its energy storage equipment is charged when there is wind and solar power abandonment, and participates in the market to discharge during peak load hours;

[0058] Load-based virtual power plants aggregate electric vehicle users, taking into account the spatiotemporal flexibility of electric vehicles, and selectively charge them during the user's parking period to form an adjustable load;

[0059] The hybrid virtual power plant aggregates distributed photovoltaic and demand response users. The hybrid virtual power plant can adjust the energy consumption curve to a certain extent according to the photovoltaic output. Distributed photovoltaics give priority to meeting the electricity load, and the remaining power generation space participates in transactions.

[0060] The above three types of virtual power plants form a cluster and participate in spot clearing.

[0061] 2. Establish the output model of the virtual power plant cluster based on the cluster architecture and operation framework.

[0062] Among them, the output model of the virtual power plant cluster includes wind and solar output model, energy storage output model, load output model, and distributed photovoltaic output model; on this basis, a specific output model is constructed according to the entity unit, as follows:

[0063] 2.1 Power supply type virtual power plant output model, which includes wind and solar output model and energy storage output model;

[0064] Specifically, wind power output is related to wind speed and can be described by a cubic function. The model constructed is:

[0065]

[0066] Where P wp,t is wind power, P wp,rated is the rated power, s t is the wind speed, s out 、s in and rated They are cut-out wind speed, cut-in wind speed and rated wind speed respectively.

[0067] Photovoltaic output power is related to light, and shows a positive correlation. The model constructed is:

[0068]

[0069] Where P pv,t is the photovoltaic power, γ is the temperature parameter, β1 and β2 are empirical parameters, T0、T ref is the atmospheric environment temperature, the normal operating temperature of the photovoltaic panel and the reference temperature, is the light radiation intensity at time t, η pv is the reference efficiency of photovoltaic panels, S pv and N pv are the area of ​​a single photovoltaic panel and the number of photovoltaic panels respectively.

[0070] Energy storage equipment is mainly used to store electricity when wind and solar power are abandoned, and to release it when the load is peak, so as to achieve the time transfer of electric energy. During the operation of energy storage, charging and discharging cannot be carried out at the same time, and in order to ensure the continuity of operation, the capacity at the initial time and the end time is generally set to be equal. The model constructed by this is:

[0071]

[0072] In the formula, Ces,t is the energy stored at time t, C es,t-1 is the energy storage at time t-1, η es is the charge and discharge efficiency, P ch,t is the charging power at time t, P rel,t is the discharge power at time t, C es,min and C es,max are the minimum and maximum storage capacities of the energy storage device, C es,0 is the energy storage at the beginning of the cycle, C es,end is the energy storage at the end of the cycle; μ is the conversion of charge and discharge, expressed as a 0-1 variable; P ch,min and P ch,max is the minimum and maximum charging power, P rel,min and P rel,max are the minimum and maximum discharge powers.

[0073] 2.2 Load-type virtual power plant output model, which includes a charging load model of electric vehicles, and the construction method steps include:

[0074] First, charging data is closely related to the driving status of electric vehicles, such as the number of vehicles and mileage. Therefore, based on the Markov chain model, the spatial state characteristics of electric vehicles are constructed, and the charging load model is constructed based on the spatial transfer quantity.

[0075] Among them, the Markov chain model is as follows:

[0076] 2.2.1 Taking the number of electric vehicle users transferring in different spaces as the prediction object, the current spatial state of the electric vehicle is recorded as The space state at the next moment is The conditional probability formula of Markov state transition is:

[0077]

[0078] Where P is the conditional probability function, p ij is the spatial transition probability from time t to time t+1, i represents the spatial state at time t, and j represents the spatial state at time t+1.

[0079] 2.2.2 User travel destinations can be divided into home (Home, H), work (Work, W), mall (Mall, M), and others (O), so the spatial transfer frequency matrix can be expressed as:

[0080]

[0081] In the formula, F t,t+1 is the frequency matrix of the electric vehicle’s primary spatial transfer, N t and N t+1is the state of the electric car in a certain space at time t and time t+1, s h The electric car is in the residential area at a certain moment; similarly, s o 、s w 、s m , n a,b is the number of electric vehicles transferred from area a to area b in a certain period, expressed as n h,w For example, it is the number of electric vehicles transferred from residential areas to work areas; n is the spatial transfer frequency, and its subscripts represent the two end points of the spatial transfer area, such as the spatial transfer frequency from residential to residential is n h,h .

[0082] 2.2.3 Based on the frequency of one-time spatial transfer, the probability of one-time spatial transfer of electric vehicles in different spatial areas can be calculated. h,w For example, the specific calculation formula is as follows:

[0083]

[0084] In the formula, p h,w is the probability of electric vehicles in residential areas transferring to work areas. Similarly, the probability of electric vehicles in other spatial areas transferring can be calculated to obtain a spatial transfer probability matrix, as follows:

[0085]

[0086] Where P t,t+1 is a spatial transfer probability matrix. After k transfers, the state probability distribution of the electric vehicle can be calculated by the following formula:

[0087]

[0088] Where P0 is the initial state.

[0089] 2.2.4 Based on the above content, the vehicle space characteristics can be predicted. Therefore, the number of electric vehicles needs to be predicted in order to predict the corresponding load. Therefore, the steps of the electric vehicle number prediction method include:

[0090] Screen and process the travel data of electric vehicle users and obtain statistical data, which mainly includes the user's travel chain, the start and end time of the trip;

[0091] According to the statistical data, the target time range is set to one day, one day is divided into 24 measurement points (t = 1, 2, 3, ..., 24), and the initial number of electric vehicles in each area is set;

[0092] Based on the target time range, measurement points, and the initial number of electric vehicles, the primary transfer state of the electric vehicles at each time point is calculated on the basis of the above conclusions, and the primary spatial transfer frequency matrix at each time point is obtained based on the primary transfer state;

[0093] Calculate the position transfer probability matrix at each time point based on the frequency of the spatial transfer frequency matrix;

[0094] The boundary probability is calculated through a spatial transfer frequency matrix and a position transfer probability matrix, and the chi-square test method is introduced to verify the Markov properties of the boundary probability and the matrix elements of the spatial transfer frequency matrix and the position transfer probability matrix; the chi-square test formula is:

[0095]

[0096] Where P j is the boundary probability; 2 is the chi-square value.

[0097] When the chi-square test method is passed, the number of electric vehicles is calculated based on the primary spatial transfer frequency matrix and the position transfer probability matrix.

[0098] After determining the number of electric vehicles, the daily mileage, parking time and other parameters of electric vehicles are fitted and generated to determine the 24-hour charging load and the dispatchable period of each electric vehicle. The daily mileage approximately satisfies the log-normal distribution, and the parking time approximately satisfies the normal distribution. The above calculation results are integrated to obtain the electric vehicle charging load, that is, the electric vehicle output model is solved. Combined with the data of other unit load usage scenarios, the load-type virtual power plant output model can be obtained.

[0099] 2.3 Hybrid virtual power plant output model, whose main body is set as a smart house scenario, including distributed photovoltaic and smart home. The photovoltaic output model has been constructed in the aforementioned wind and solar output model. It is set that the user's smart home can participate in incentive demand response, and can interrupt, transfer, reduce, and increase the load according to the photovoltaic output. The specific model formula is:

[0100] R min ≤ΔR h ≤R h ,h=1;

[0101] 0≤ΔR h ≤R h -R h-1 ,h=2,3,...,H;

[0102]

[0103] In the formula, R min is the minimum capacity of the incentive-based demand response, ΔRh is the incentive demand response capacity of users participating in the hth stage, R h is the user capacity in the hth stage, R is the cumulative response capacity in the H period, H is the total period of user participation in incentive demand response, and h is the stage divided according to the response price in the total period; Figure 2 As shown, The response prices at different stages.

[0104] The wind power output model and photovoltaic output model of the hybrid virtual power plant are combined to construct the corresponding hybrid virtual power plant output model.

[0105] 3. Based on the cluster architecture, operation framework and output model, a clearing simulation of the virtual power plant cluster in the spot market is carried out.

[0106] A dispatch model based on security constraints is constructed with the goal of minimizing power generation costs. Its objective function is constructed as follows:

[0107]

[0108] In the formula, is the electricity energy cost of the generator, The capacity of the bidding segment for the generator, is the declared price, t is the time, i is the number of generators, b is the number of quotation segments, T is the time set, I is the generator set, and B is the quotation segment set.

[0109] Adding constraints to the cost objective function forms a complete dispatch model. The constraints include the maximum output limit of the generator, the output limit of the quotation section, the node power balance and the line transmission capacity limit.

[0110] Among them, the constraints on the maximum output limit of the power generator are:

[0111]

[0112] In the formula, p i,t is the output of the ith generator at time t, and They are the lower and upper limits of the power generation capacity.

[0113] The constraints of the output limit in the quotation section are:

[0114]

[0115] In the formula, p i,b is the declared output of the i-th power producer in the b-th quotation period, It is the maximum output in this quotation segment.

[0116] The node power balance and line transmission capacity constraints are:

[0117]

[0118] P xy,min ≤P xy,t ≤P xy,max ;

[0119] Where P xy,t is the transmission power of branch (x, y), (x, y) represents a directed branch from node x to node y, x is the parent node of y; w y is the set of child nodes of node y, P yz,t is the transmission power of branch (y,z), is the other load of node b, and are the power consumption and power generation of VPP respectively; P xy,min and P xy,max are the line transmission capacity limits of branches (x, y) respectively.

[0120] The bidding power of each virtual power plant in the virtual power plant cluster is calculated through scheduling model simulation.

[0121] 4. Based on the output uncertainty and deviation assessment penalty of new energy power plants in the virtual power plant cluster, a virtual power plant cluster collaborative operation model is established with the goal of minimizing the total deviation loss of the virtual power plant.

[0122] Among them, the transaction basis is the point-to-point transaction model established in the first preparation stage. Its decentralized characteristics can enable fast and complete transactions between virtual power plants in the cluster. Although the expected winning bid electricity of each virtual power plant in the future is known, considering the uncertainty of the internal entities of the virtual power plant, in order to avoid deviation assessment penalties, the transferred electricity and unit transaction cost (i.e., unit electricity transaction price) between the corresponding entities must be obtained. Therefore, a virtual power plant cluster collaborative operation model is needed to pursue the maximization of benefits under the cooperation model and balance the interests of all parties.

[0123] On the basis of the above, the virtual power plant cluster collaborative operation model of this embodiment is constructed based on Nash bargaining, and its basic model is:

[0124]

[0125] In the formula, H i is the benefit after cooperation of person i in the negotiation, is the gain of person i in the negotiation before cooperation.

[0126] The bargaining strategy (unresolved variables) in the negotiation process is the transferred electricity and transaction price between virtual power plants. Its profit function contains the product of unresolved variables, which makes the optimization problem multiple variable coupling, non-convex and nonlinear, and difficult to solve. In addition, the players are operated by independent operators, have the intention to protect privacy, and cannot fully share information. Therefore, a distributed solution is used to solve the Nash bargaining model, and the virtual power plant cluster P2P transaction optimization is decomposed into two sub-problems: Sub-problem 1 is a virtual power plant cluster collaborative operation optimization model with the goal of minimizing the total deviation loss of the virtual power plant; Sub-problem 2 is a benefit measurement model with the goal of individual benefits of the virtual power plant.

[0127] Among them, sub-problem 1 also involves wind and solar uncertainty modeling; therefore, the specific steps for solving it are as follows:

[0128] Firstly, the robust decision with information gap is explicitly introduced into the collaborative operation model of virtual power plant clusters, thus establishing a wind-solar uncertainty model based on robust decision with information gap.

[0129] Construct a deterministic model; first, assume that there are no non-deterministic variables in the system, and construct an initial optimization model based on the wind and solar forecast values. The simplified formula is as follows:

[0130]

[0131] Where f(u,o) is the objective function; H(u,o) and G(u,o) are the equality constraint and inequality constraint respectively; u is the uncertain variable; o is the decision variable.

[0132] This optimization model is the basis of the virtual power plant cluster collaborative operation optimization model with the goal of minimizing the total deviation loss of the virtual power plant.

[0133] On the basis of the above, we introduce uncertain variables into the deterministic model, and obtain the uncertain model (introducing uncertain variables, i.e., wind and solar output fluctuations):

[0134]

[0135] In the formula, is the predicted value of u; α is the uncertainty coefficient (α≥0); For u and The deviation between

[0136] Add constraints to the non-deterministic model to obtain the wind and solar uncertainty model; the constraints are the minimum requirements of the system decision makers. Among them, the minimum requirements of the system decision makers are understood as the robust response strategy sought to avoid the maximum economic losses caused by uncertainty. Therefore, the system equations are sorted as:

[0137]

[0138] In the formula, f0 is the optimization target value of the determined model, β a The degree of deviation from the calculated results of the determined model.

[0139] The robust decision value can ensure that the expected value does not exceed (1+β a )f0, it can be clearly seen that the above model is a two-layer optimization model: the lower layer (i.e. constraint ①) is the cost generated when the uncertain variables fluctuate within the expected range; and the upper layer (i.e. α) is the maximum output deviation when the lower layer is established.

[0140] According to the uncertainty model of wind and solar, the optimal solution is found in extreme scenarios; according to the change direction of the uncertain variables, it is judged whether it is an extreme scenario when constraint ② is equal. When the uncertain variables take the endpoint value, the extreme scenario can be satisfied, and constraint ① is established at any time. After conversion, the optimization model is rewritten as:

[0141]

[0142] Where μ is a 0-1 variable, which is used to indicate that one of the positive and negative extreme scenarios meets the decision maker's requirements.

[0143] The total deviation loss of the virtual power plant in sub-question 1 includes deviation assessment penalty fees, deviation electricity opportunity costs (i.e., the income that the virtual power plant should have obtained if there was no deviation in wind and solar power output), P2P grid access fees, etc.

[0144] Sub-question 1 emphasizes the overall benefits. The profit transfers within the cluster offset each other. Therefore, Sub-question 1 only results in the amount of electricity transferred between virtual power plants, but cannot obtain the transaction price results.

[0145] The specific objective function is as follows:

[0146]

[0147] In the formula, C pe,i is the penalty fee for deviation assessment, C tra,i For P2P network fee, C opp,i is the opportunity cost, d t,i is the deviation assessment quantity (the portion less than a% of the winning bid quantity is the deviation assessment quantity), c pe,t,i is the penalty coefficient for deviation assessment at time t, c tra,i is the P2P network fee coefficient, p x-y,t,i is the transaction power between VPPs, c lmp,t,i It is the settlement price at the node where the VPP is located at time t.

[0148] P2P transactions should ensure that the transmission power between VPPs is within the transmission channel capacity limit and ensure power balance. The constraints are similar to those in the third step and will not be repeated here.

[0149] Furthermore, an auxiliary variable can be optionally introduced to decouple sub-problem 1. After negotiation between the transferor and the transferee, the transaction volume should be balanced. The reference balance formula is:

[0150] p t,i-j =-p t,j-i ;

[0151] In the formula, p t,i-j and p t,j-i is the amount of electricity that virtual power plant i and virtual power plant j expect to trade with each other. When the two are equal, it means that both parties have reached a consensus on the amount of electricity traded. The above objective function is rewritten by introducing the Lagrangian function as follows:

[0152]

[0153] Where, L i,pro1 is the Lagrangian function of subproblem 1, l i,t,pro1 is the Lagrange multiplier of subproblem 1, ρ i,pro1 is the penalty factor for subproblem 1, is the square of the two-norm.

[0154] Therefore, the transfer amount of electricity between virtual power plants can be obtained by using the above model through auxiliary variables such as the expected amount of electricity to be traded with the other party, Lagrange multiplier, penalty factor, and second norm.

[0155] Sub-problem 2 uses the contractually transferred electricity between virtual power plants as input to construct a virtual power plant benefit calculation model to achieve cooperative benefit distribution among players.

[0156] The steps for calculating unit transaction costs include:

[0157] A virtual power plant benefit calculation model is constructed based on the transferred electricity. The virtual power plant benefit calculation model is constructed based on the Nash bargaining theory and the objective function is:

[0158]

[0159] In the formula, The transaction income of the virtual power plant during cooperative operation. is the transaction income of the virtual power plant when it operates independently, and i is the ordinal number of the virtual power plant;

[0160] The transaction income during independent operation includes spot income and deviation assessment fees, and the transaction income during cooperative operation includes spot income, deviation assessment fees, P2P network fees, contract transfer (sell) income, and contract transfer (buy) costs. The transaction income during cooperative operation is as follows:

[0161]

[0162] In the formula, I g,i is the spot income, C pe,i is the deviation assessment fee, C tra,i For network fee, I con,i is the contract transfer (sale) income, C con,i is the contract transfer (buy) cost. Logarithmize the objective function to convert continuous multiplication into continuous addition, and the objective function can be rewritten as:

[0163]

[0164] The objective function introduces auxiliary variables and decouples the objective function to obtain the unit transaction cost. The unit transaction cost balance can be referred to as follows:

[0165] c t,i-j =c t,j-i ;

[0166] In the formula, c t,i-j and c t,j-i is the electricity price that virtual power plant i and virtual power plant j expect to trade with each other. When the two are equal, it means that both parties have reached a consensus on the transaction volume.

[0167] The Lagrangian function is introduced to rewrite the above objective function as follows:

[0168]

[0169] Where, L i,pro2 is the Lagrangian function of subproblem 2, l i,t,pro2 is the Lagrange multiplier of subproblem 2, ρ i,pro2 is the penalty factor for sub-problem 2. Based on Figure 3 The solution process shown in the figure finally obtains the equilibrium solution, that is, the transferred electricity and unit transaction cost between each VPP.

[0170] Finally, based on the winning bid electricity, transferred electricity and unit transaction cost obtained in the above steps, a trading plan for the virtual power plant cluster in a certain period of time in the future is formulated, and the market-oriented transaction of the virtual power plant cluster is carried out according to the trading plan. Since this embodiment adopts Nash bargaining and information gap robust decision-making, the trading plan obtained under the method of this embodiment can enable the entire cluster to operate stably and supply electricity even if there are fluctuations in wind and solar power during execution, thereby reducing the overall power generation cost.

[0171] After the integration of Nash bargaining and information gap robust decision-making, an optimized transaction method that is more in line with market transaction conditions is obtained. It not only meets individual interest demands and achieves a balanced distribution of cooperation benefits, but also ensures a balance between economy and robustness.

[0172] Finally, it should be noted that the above content is only used to illustrate the technical solution of the present invention, rather than to limit the scope of protection of the present invention. Simple modifications or equivalent substitutions of the technical solution of the present invention by ordinary technicians in this field do not deviate from the essence and scope of the technical solution of the present invention.

Claims

1. A market-based transaction optimization method for heterogeneous virtual power plant clusters based on decision fusion, characterized by: The steps include: Construct the cluster architecture, operation framework and peer-to-peer distributed transaction model of heterogeneous virtual power plant clusters in the market environment; Establishing an output model of a virtual power plant cluster based on the cluster architecture and operation framework; According to the group architecture, operation framework and output model, a clearing simulation is performed on the virtual power plant cluster in the spot market, and a scheduling model based on security constraints is constructed with the goal of minimizing power generation costs; Calculate the winning bid power of each virtual power plant in the virtual power plant cluster through simulation of the scheduling model; Based on the output uncertainty and deviation assessment penalty of new energy power plants in the virtual power plant cluster, a virtual power plant cluster collaborative operation model with the goal of minimizing the total deviation loss of virtual power plants is established; robust decision-making with information gaps is introduced into the virtual power plant cluster collaborative operation model; According to the point-to-point distributed transaction model, the winning bid power and the virtual power plant cluster collaborative operation model, the transferred power and unit transaction cost between virtual power plants are obtained through Nash bargaining; A trading plan is output based on the bid-winning electricity, the transferred electricity and the unit transaction cost, and market-based transactions of the virtual power plant cluster are carried out according to the trading plan.

2. The market-oriented transaction optimization method for heterogeneous virtual power plant clusters based on decision fusion according to claim 1 is characterized by: Virtual power plants include power supply type virtual power plants, load type virtual power plants, and hybrid virtual power plants; The power supply type virtual power plant aggregates distributed wind and solar power and configures energy storage equipment; The load-based virtual power plant aggregates electric vehicle users; The hybrid virtual power plant aggregates distributed photovoltaic and demand response users.

3. The market-oriented transaction optimization method for heterogeneous virtual power plant clusters based on decision fusion according to claim 1 is characterized by: The output model of the virtual power plant cluster includes a wind and solar output model, an energy storage output model, a load output model, and a distributed photovoltaic output model.

4. The method for optimizing market-based transactions of heterogeneous virtual power plant clusters based on decision fusion according to claim 3 is characterized by: The load output model also includes a charging load model of an electric vehicle, and the construction method steps include: Based on the Markov chain model, the electric vehicle spatial state characteristics are constructed, and based on the electric vehicle spatial state characteristics, the charging load model is constructed with the spatial transfer quantity.

5. The market-oriented transaction optimization method for heterogeneous virtual power plant clusters based on decision fusion according to claim 4 is characterized by: The method for predicting the number of electric vehicles in the charging load model comprises the following steps: Screen and process the travel data of electric vehicle users and obtain statistical data; According to the statistical data, a target time range and measurement points are set, and an initial number of electric vehicles in each area is set; Calculating a primary transfer state of the electric vehicle at each time point based on the target time range, the measurement point, and the initial number of electric vehicles, and obtaining a primary spatial transfer frequency matrix at each time point based on the primary transfer state; Calculating the position transfer probability matrix at each time point based on the frequency of the spatial transfer frequency matrix; Calculating the boundary probability through the primary spatial transfer frequency matrix and the position transfer probability matrix, and introducing a chi-square test method to verify the boundary probability and the Markov properties of the matrix elements of the primary spatial transfer frequency matrix and the position transfer probability matrix; When the chi-square test method is passed, the number of electric vehicles is calculated based on the primary space transfer frequency matrix and the position transfer probability matrix.

6. The market-oriented transaction optimization method for heterogeneous virtual power plant clusters based on decision fusion according to claim 1 is characterized by: The safety constraints of the dispatching model include the maximum output limit of the power generator, the output limit of the quotation section, the node power balance and the line transmission capacity limit.

7. The method for optimizing market-based transactions of heterogeneous virtual power plant clusters based on decision fusion according to claim 1 is characterized by: A wind-solar uncertainty model based on information gap robust decision-making is established in the virtual power plant cluster collaborative operation model; The process of solving the wind and solar uncertainty model also includes the following steps: Build deterministic models; Introducing non-deterministic variables into the deterministic model to obtain a non-deterministic model; Adding constraints to the non-deterministic model to obtain a wind-solar uncertainty model; the constraints are the minimum requirements of system decision makers; According to the wind and solar uncertainty model, the optimal solution is found in extreme scenarios.

8. The method for optimizing market-based transactions of heterogeneous virtual power plant clusters based on decision fusion according to claim 7 is characterized by: The calculation method of the transferred electricity volume comprises the following steps: The solution of the wind-solar uncertainty model is substituted into the virtual power plant cluster collaborative operation model to obtain the transferred electricity.

9. The method for optimizing market-based transactions of heterogeneous virtual power plant clusters based on decision fusion according to claim 8 is characterized by: The steps of calculating the unit transaction cost include: A virtual power plant benefit calculation model is constructed based on the transferred electricity; the virtual power plant benefit calculation model is constructed based on the Nash bargaining theory with the objective function as follows: In the formula, The transaction income of the virtual power plant during cooperative operation. is the transaction income of the virtual power plant when it operates independently, A is the set of virtual power plants in the virtual power plant cluster, and i is the ordinal number of the virtual power plant; The objective function introduces auxiliary variables and decouples the objective function to obtain the unit transaction cost.

10. The method for optimizing market-based transactions of heterogeneous virtual power plant clusters based on decision fusion according to claim 9 is characterized by: The trading income during the cooperative operation of virtual power plants includes spot income, deviation assessment fees, point-to-point transaction transmission fees, sales income from contract transfers, and purchase costs from contract transfers.

Images