A master-slave game two-layer model solving method, system, device and storage medium

Abstract

The present invention relates to the technical field of power grid control, and in particular to a master-slave game double-layer model solving method, system, device and storage medium; the method comprises the following steps: S1, constructing a double-layer opposing game model including an upper layer and a lower layer, wherein the upper layer model of the double-layer opposing game model is a dynamic pricing model of a distribution network system operator, and the lower layer model is a VPP energy management model; the present invention realizes a dynamic balance of interests between the distribution network system operator and the VPP in the power market by establishing a master-slave game double-layer model of the distribution network system operator and each VPP in the lower layer, can coordinate the interests between the distribution network system operator and the VPP and effectively protect the privacy of the VPP, and introduces a KAN network architecture solving algorithm, which improves the solving efficiency and accuracy of the ordinary neural network model for finding a balanced solution, and plays a positive role in promoting the stable operation and efficient dispatching of the entire new power system.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid control, and in particular to a method, system, device and storage medium for solving a master-slave game double-layer model. Background Art

[0002] In recent years, with the emergence of global energy shortages and the continued deterioration of environmental pollution, the development and utilization of clean energy has not only become a key way to alleviate energy pressure, but has also become a mainstream trend in the development of the energy field. It is an important force in promoting the green transformation of the economy and society and achieving sustainable development. Although clean energy has shown great potential with its environmental friendliness and rapid development momentum, its inherent characteristics - high uncertainty in output and wide and dispersed geographical distribution - have brought unprecedented challenges to the safe and stable operation of the power grid and the efficient and coordinated dispatch of energy.

[0003] As a solution to the challenge of distributed energy grid connection, Virtual Power Plant (VPP) can integrate diversified resources such as distributed power sources, energy storage systems, controllable loads and even electric vehicles into a highly coordinated virtual entity, enabling it to seamlessly integrate into and actively participate in power market transactions and daily grid operations and scheduling. However, at the same time, the aggregation calculations of VPP are often affected by distributed energy and flexible loads, which is not conducive to auxiliary services such as voltage control, peak and frequency regulation, and congestion management. In addition, since VPP is jointly participated by multiple parties, there are serious privacy protection and information security issues in the sharing of its own physical model. Summary of the invention

[0004] In view of the problems mentioned in the prior art, the present invention proposes a master-slave game two-layer model solving method, system, device and storage medium. Taking the distribution network system operator and VPP as an example, a master-slave game model between the distribution network system operator and the VPP is established, and a deep learning network solving method based on the Kolmogorov-Arnold Network (KAN) network architecture is introduced. The method can effectively coordinate the interests between the upper-level distribution network system operator and the lower-level VPP, ensure privacy protection, avoid frequent calls to the lower-level model, and thus reduce the computing cost; through this method, not only the solving efficiency and accuracy are improved, but also a dynamic balance is achieved in the power market, which promotes the stable operation and efficient dispatch of the new power system, and provides new ideas and technical support for the optimization and development of the power market.

[0005] In order to achieve the above object, the present invention adopts the following technical solution:

[0006] The present invention provides a method for solving a master-slave game double-layer model, comprising the following steps:

[0007] S1. Construct a two-layer opposition game model including an upper layer and a lower layer. The upper layer model of the two-layer opposition game model is a dynamic pricing model of the distribution network system operator, and the lower layer model is a VPP energy management model.

[0008] S2. Build the KAN network architecture and use it to solve the two-layer opposing game model to obtain the game equilibrium solution.

[0009] As a further improvement of the present invention, the specific process of constructing the dynamic pricing model of the distribution network system operator is:

[0010] The distribution network system operator sets the purchase and sale prices of electricity with the VPP at different time periods, and then trades with the power market based on the on-grid electricity price and the power grid price. The maximum transaction net profit is obtained by summing the difference between the power sold by the distribution network system operator to the power market and the power purchased by the VPP and the power sold by the distribution network system operator. The first utility function of the game with maximizing the transaction net profit is:

[0011]

[0012] Where: The total time of the day is 24h; for The on-grid electricity price in the electricity market at the moment, for The electricity price of the power grid in the power market at the moment; for The amount of electricity sold by the distribution network system operator to the electricity market at any given moment; for The amount of electricity purchased by the distribution network system operator in the electricity market at any given moment; represents the amount of electricity sold by the jth VPP to the distribution network system operator; It represents the amount of electricity purchased by the jth VPP from the distribution network system operator; N is the total number of VPPs;

[0013] The first utility function is constrained according to a first constraint condition.

[0014] As a further improvement of the present invention, the following constraints should be met between the amount of electricity sold and purchased by the distribution network system operator to the power market:

[0015]

[0016] Where: is the total amount of electricity purchased and sold by each VPP in the lower layer at time t, which is aggregated by the distribution network system operator and traded with the power market;

[0017] The electricity purchase price and electricity sales price at each time period between the distribution network system operator and the VPP should meet the following constraints:

[0018]

[0019] Where: for The on-grid electricity price in the electricity market at the moment, for The electricity price set by the distribution network system operator at that moment, for The power purchase price set by the distribution network system operator at that moment, for The electricity price of the power grid in the power market at the moment.

[0020] As a further improvement of the present invention, the specific process of constructing the VPP energy management model is:

[0021] VPP reasonably arranges internal output according to the power purchase and sales from the distribution network system operator and the output plan of distributed power sources in each period to minimize the operating cost. The second utility function of the game is to minimize the operating cost, which is expressed as:

[0022]

[0023]

[0024]

[0025]

[0026] Where: for time No. Micro gas turbine power generation cost; , , for Cost factor of gas turbine No. for time The operating cost of the energy storage power station; for The dispatch cost coefficient of the energy storage power station; for Moment demand side The interruptible load cost of unit 1; For the demand side The interruption compensation electricity price for unit No. for time The amount of electricity purchased by VPP unit No. from the distribution network system operator; for time The amount of electricity sold by VPP unit No. to the distribution network system operator; for time Output power of micro gas turbine No. for time The charging and discharging power of the energy storage power station; for time The interrupting power of the interruptible load of the unit; The total time of the day is 24h;

[0027] The second utility function is constrained according to the second constraint condition.

[0028] As a further improvement of the present invention, the second constraint condition is that when the VPP responds to the distribution network system operator's strategy, it must meet the power balance between itself and the grid, and also meet the operation restrictions of each distributed power source. The second constraint condition is expressed as follows:

[0029]

[0030]

[0031]

[0032]

[0033]

[0034]

[0035]

[0036]

[0037]

[0038]

[0039]

[0040]

[0041] Where: For VPP The maximum amount of electricity traded between the generating unit and the distribution network system operator; represents the load forecast value at time t; For MT The upward climbing rate of unit No. For MT Downward climbing rate of unit No. It means time Boolean variable of VPP unit. When the Boolean variable takes the value 0, it represents Moment VPP The unit purchases electricity from the distribution network system operator. When the value is 1, it means Moment VPP Unit No. sells electricity to the distribution network system operator; For MT Maximum output power of unit No. for time The charge state of the energy storage unit; For the previous time Moment The charge state of the energy storage unit; is the time scale, here it is 1h; The charge state of the energy storage unit at time 0; For the whole day The charge state of the energy storage unit after time Take 24; for The upper limit of the state of charge of the energy storage unit; for The lower limit of the state of charge of the energy storage unit; for The upper limit of the charging and discharging power of the energy storage power station; for The lower limit of the charging and discharging power of the energy storage power station; for The maximum energy capacity value of the energy storage power station; For the demand side The maximum interruption amount of the interruptible load of the unit; for Wind turbines in Output power at the moment; for Wind turbines in The maximum output power at the moment, here we take the predicted value.

[0042] As a further improvement of the present invention, the specific process of step S2 includes the following steps:

[0043] S21, initializing system parameters;

[0044] S22, generating a set of initial transaction electricity price sample points covering all electricity price sample intervals according to Latin hypercube sampling;

[0045] S23. For the initial transaction electricity price sample points generated in step S22, the solver is called to calculate the transaction electricity of each VPP and form a corresponding sample data set. A deep learning model is constructed according to the generated sample data set. The initial transaction electricity price sample points are input into the deep learning model to obtain the transaction electricity of each VPP.

[0046] S24, inputting the transaction power of each VPP into the objective function of the upper model to obtain the upper objective function value of each initial transaction power price sample point;

[0047] S25, according to the upper objective function value of each initial transaction price sample point, the strategy space of the objective function in the upper model is divided into multiple key areas, and the value of the optimal solution of each key area is obtained , using the particle swarm optimization algorithm, calling the trained deep learning model to solve the local optimal transaction volume and electricity price;

[0048] S26. Call the lower-level model, and call the solver to calculate the actual transaction power according to the optimal transaction power and electricity price.

[0049] As a further improvement of the present invention, after step S26, the actual transaction power is added as an excellent sampling point to the deep learning model for iteration until the convergence condition is met and the optimal actual transaction power is output. As a further improvement of the present invention, after step S26, the actual transaction power is added as an excellent sampling point to the deep learning model for iteration until the convergence condition is met and the optimal actual transaction power is output.

[0050] A master-slave game two-layer model solving system, comprising:

[0051] A construction module is used to construct a two-layer opposing game model including an upper layer and a lower layer. The upper layer model of the two-layer opposing game model is a dynamic pricing model of a distribution network system operator, and the lower layer model is a VPP energy management model.

[0052] Solving module, building the KAN network architecture, using the KAN network architecture to solve the two-layer opposing game model and obtain the game equilibrium solution.

[0053] A master-slave game two-layer model solving device comprises a processor and a memory, wherein the processor implements the master-slave game two-layer model solving method as described above when executing a computer program stored in the memory.

[0054] A computer-readable storage medium is used to store a computer program, wherein when the computer program is executed by a processor, the method for solving the master-slave game double-layer model as described above is implemented.

[0055] Compared with the prior art, the present invention has achieved the following technical effects:

[0056] The present invention takes into account the diversity and independence of various decision-making entities, and establishes a master-slave game model between the distribution network system operator and each VPP. The distribution network system operator reasonably guides the reasonable electricity consumption within the VPP through a dynamic pricing strategy, and the VPP responds to prices by reasonably optimizing the internal energy output. The VPP aims to minimize the operating cost. The optimized trading electricity volume will affect the distribution network system operator's formulation of electricity prices. The distribution network system operator guides the VPP to reasonably purchase and sell electricity through dynamic pricing. From the perspective of interest game, the distribution network system operator increases the purchase and sale electricity volume of the VPP by formulating an optimization strategy for the purchase and sale electricity price, thereby achieving the purpose of maximizing its own profits and reducing the total operating cost of the VPP, thereby achieving a dynamic balance of interests between the operator and the VPP in the power market, coordinating the interests between the operator and the VPP in the power market and effectively protecting the privacy of the VPP.

[0057] The present invention proposes a deep learning network solving algorithm under an improved KAN network architecture. The price response behavior of VPP is fitted by adopting the KAN network architecture, and the lower-level energy management model is replaced to obtain a balanced solution. Compared with the traditional optimization algorithm, the algorithm of the present invention does not need to frequently and massively call the lower-level model, and can greatly improve the solving accuracy of the traditional neural network model. While effectively protecting the privacy of VPP, the solving efficiency and accuracy of the balanced solution are significantly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a schematic diagram of the overall process of the present invention;

[0059] Figure 2 This is a flowchart of deep learning network solution under the KAN network architecture of the present invention;

[0060] Figure 3 It is a schematic diagram of the lower-level game model under the KAN network architecture of the present invention. DETAILED DESCRIPTION

[0061] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0062] like Figure 1 As shown, the present invention introduces a distribution system operator (DSO) and establishes a two-layer opposing game model between DSO and multiple VPPs. DSO is the setter of energy transaction prices with VPPs, responsible for trading with multiple VPPs and power trading institutions, and obtaining profits through price differences. The mutual influence of DSO dynamic pricing and VPP operation strategies is studied through the two-layer opposing game model to realize the energy management of multiple virtual power plants. DSO and VPP have different and independent status and investment operations and interest subjects. The present invention regards the owners of DSO and VPP as participants in the game. DSO mainly maximizes its own economic benefits by optimizing the dynamic electricity price strategy for purchase and sale, while VPP reasonably arranges the internal distributed energy output plan and reduces the system operation cost by receiving the electricity price strategy formulated by DSO. Therefore, when formulating the price strategy, DSO cannot only consider its own economic benefits, but also the price elasticity behavior of VPP, that is, there is an interest game relationship between the two.

[0063] At the same time, for the solution of the equilibrium solution of the two-layer opposing game model, a deep learning network solution algorithm under the improved KAN network architecture is proposed. Figure 3 As shown in the figure, a KAN network architecture is built and used to fit the price response behavior of followers, and the lower-level VPP energy management model is replaced and solved to obtain an equilibrium solution. The leader can adjust the pricing strategy and the lower-level followers can adjust the energy management strategy to achieve the optimal power market transaction and energy management effect.

[0064] The present invention will be further explained below with reference to the accompanying drawings and specific embodiments:

[0065] Step 1: Establishment of a two-level adversarial game model, including a dynamic pricing model for distribution network system operators and a VPP energy management model

[0066] The game strategy of the leader DSO is to aggregate the transaction power of each VPP in the lower layer during the optimization scheduling period, and to formulate the purchase and sale prices of electricity with VPP in each period for the purpose of maximizing economic benefits, taking into account the electricity price of the power trading agency and the price response on the VPP side, and then trade with the power trading agency based on the on-grid electricity price and the power grid electricity price, and use the price difference between the two to maximize the net profit of the transaction. The first utility function of DSO, which takes maximizing net profit as the game (including the purchase and sale costs and benefits with the power market and the lower layer), is as follows:

[0067]

[0068] Where: The total time of the day is 24h; for The on-grid electricity price in the electricity market at the moment, for The electricity price of the power grid in the power market at the moment; for The amount of electricity sold by the distribution network system operator to the electricity market at any given moment; for The amount of electricity purchased by the distribution network system operator in the electricity market at any given moment; represents the amount of electricity sold by the jth VPP to the distribution network system operator; It represents the amount of electricity purchased by the jth VPP from the distribution network system operator; N is the total number of VPPs.

[0069] In order to balance the supply and demand among the lower-level VPPs, the balance of electricity purchase and sales should meet the following constraints:

[0070]

[0071] Where: It is the total amount of electricity purchased and sold by each VPP in the lower layer at time t, which is aggregated by the distribution network system operator and traded with the power market.

[0072] In addition, in order to ensure the balance of power among the VPPs in the lower layer and to ensure that each VPP is willing to choose to trade with the DSO, the price set by the DSO should meet the following constraints:

[0073]

[0074] Where: for The on-grid electricity price in the electricity market at the moment, for The electricity price set by the distribution network system operator at that moment, for The power purchase price set by the distribution network system operator at that moment, for The electricity price of the power grid in the power market at the moment.

[0075] As shown in the above formula, in order to avoid the lower-level VPPs being unwilling to trade when considering the market electricity price and maximizing their own interests, the purchase price of electricity set in the DSO strategy cannot be greater than the grid electricity price of the power trading agency, and the sales price of electricity cannot be less than the grid electricity price of the power trading agency. In this way, VPPs will choose to trade with distribution network system operators to maximize their own interests.

[0076] The game strategy of the follower VPP is to purchase and sell electricity and the output plan of distributed energy resources (DERs) in each period, so as to reasonably arrange the internal output and minimize the operating cost. The specific output plan for each period includes the power purchased from the distribution network system operator. and electricity sold to distribution network system operators 、Output power of micro turbines (MT) , the charging and discharging power of energy storages (ES) , interruption power of interruptible load (IL) and the output power of wind turbines (WT) , denoted as , , t=1:T, Represents the set of DERs contained in the jth VPP.

[0077] The VPP utility function is the second utility function based on minimizing operating costs, and takes into account multiple costs, including electricity purchase costs, gas turbine costs, energy storage power station operating costs, and demand-side interruptible load costs, as shown in the following formula:

[0078]

[0079]

[0080]

[0081]

[0082] Where: for time No. Micro gas turbine power generation cost; , , for Cost factor of gas turbine No. for time The operating cost of the energy storage power station; for The dispatch cost coefficient of the energy storage power station; for Moment demand side The interruptible load cost of unit 1; For the demand side The interruption compensation electricity price for unit No. for time The amount of electricity purchased by VPP unit No. from the distribution network system operator; for time The amount of electricity sold by VPP unit No. to the distribution network system operator; for time Output power of micro gas turbine No. for time The charging and discharging power of the energy storage power station; for time The interrupting power of the interruptible load of the unit; The total time of the day is 24h.

[0083] When VPP responds to the strategy formulated by DSO, it meets the operation constraints of each distributed power source. When VPP responds to the strategy formulated by the distribution network system operator, it also needs to balance its own power with the grid. The constraints that need to be met are shown in the following formula:

[0084]

[0085]

[0086]

[0087]

[0088]

[0089]

[0090]

[0091]

[0092]

[0093]

[0094]

[0095]

[0096] Where: For VPP The maximum amount of electricity traded between the generating unit and the distribution network system operator; represents the load forecast value at time t; For MT The upward climbing rate of unit No. For MT Downward climbing rate of unit No. It means time Boolean variable of VPP unit. When the Boolean variable takes the value 0, it represents Moment VPP The unit purchases electricity from the distribution network system operator. When the value is 1, it means Moment VPP Unit No. sells electricity to the distribution network system operator; For MT Maximum output power of unit No. for time The charge state of the energy storage unit; For the previous time Moment The charge state of the energy storage unit; is the time scale, here it is 1h; The charge state of the energy storage unit at time 0; For the whole day The charge state of the energy storage unit after time Take 24; for The upper limit of the state of charge of the energy storage unit; for The lower limit of the state of charge of the energy storage unit; for The upper limit of the charging and discharging power of the energy storage power station; for The lower limit of the charging and discharging power of the energy storage power station; for The maximum energy capacity value of the energy storage power station; For the demand side The maximum interruption amount of the interruptible load of the unit; for Wind turbines in Output power at the moment; for Wind turbines in The maximum output power at the moment, here we take the predicted value, and the above related constraints do not take into account the system network loss.

[0097] Based on the game strategy of the leader DSO and the game strategy of the follower VPP, a two-layer opposition game model can be established as shown in the following formula:

[0098]

[0099]

[0100] Where: , Take 24h; The quotation strategy space for DSO; It is the average output of all power generation entities in each VPP unit; The first The output of unit No. The first The strategic space of Unit No.

[0101] In the above formula, the DSO side is the leader and the VPP side is the follower. DSO and VPP respectively formulate the optimization strategy space with the goal of maximizing net profit and minimizing operating costs. The income of DSO is jointly affected by the two factors of the purchase and sale price and the VPP transaction volume: if the difference between the purchase and sale prices increases, the income of DSO will also increase with the increase of VPP shared electricity. At the same time, the response of VPP to electricity prices will also have an effect on the income of DSO: when the purchase price rises, the purchase volume of VPP will decrease accordingly; when the sales price decreases, the sales volume of VPP will also decrease, which will in turn lead to a decrease in the amount of electricity shared between VPPs. It can be seen that there is a game relationship of interests between operators and VPPs. Therefore, in order to meet their respective needs, operators must fully consider the response behavior of VPP to price changes and formulate the most appropriate electricity price strategy by seeking Nash equilibrium solutions.

[0102] Step 2: An improved KAN network architecture is introduced to simulate the lower-level VPP energy management model, aiming to clearly show the mapping relationship between transaction price and transaction volume, which is difficult to intuitively reflect. This method effectively avoids frequent calls to the lower-level model due to relying solely on intelligent algorithms to solve Nash equilibrium solutions, thereby significantly improving the efficiency of the solution process and reducing the computational cost.

[0103] like Figure 2 As shown, the specific steps are as follows:

[0104] a: Initialize system parameters, including setting the on-grid electricity price and grid electricity price in the power market, determining the relevant parameters of each VPP, including the type, capacity and constraints of DERs, and configuring the relevant parameters of the deep learning network architecture.

[0105] b: Generate initial sample points. According to Latin Hypercube Sampling (LHS), generate a set of initial transaction electricity price sample points covering all electricity price sample intervals for simulating the lower-level energy management process.

[0106] c: Call the solver for the initial transaction electricity price sample points generated in the previous step to calculate the transaction electricity of each VPP and form the corresponding sample data set.

[0107] d: Train the deep learning model based on the generated sample data set, input the initial transaction electricity price sample points into the deep learning model, and obtain the transaction electricity volume of each VPP. The model input layer is the purchase electricity price of 24 periods under a one-hour time scale. and electricity price The output layer is the sales volume of each VPP in 24 hours. Purchase electricity .

[0108] e: According to the upper objective function value of each sample point calculated in step d, the strategy space of the upper model objective function is divided into multiple key areas, and the value of the optimal solution for each key area is obtained. (The subscript l represents the number of key areas). The particle swarm optimization algorithm is used to call the trained deep learning model to solve the local optimal transaction volume and electricity price.

[0109] f: Call the lower-level model, and based on the optimal transaction volume and electricity price, call the solver to calculate the actual transaction volume.

[0110] g: Add the above actual transaction power as excellent sampling points to the data set of the deep learning model, and then retrain the model with the new data set. At the same time, calculate the true objective function value of the upper model , and with For comparison, if , then let .

[0111] h: meets the convergence condition After that, stop the iteration and output the game equilibrium solution. If the convergence condition is not met, return to continue the iteration. For tolerance, set it to 5%.

[0112] Based on the same inventive concept, an embodiment of the present invention also provides a master-slave game two-layer model solving system. Since the principle of solving the problem by the master-slave game two-layer model solving system is similar to the aforementioned master-slave game two-layer model solving method, the implementation of the master-slave game two-layer model solving system can refer to the implementation of the master-slave game two-layer model solving method, and the repeated parts will not be repeated.

[0113] In specific implementation, the master-slave game two-layer model solving system provided by the embodiment of the present invention specifically includes:

[0114] A construction module is constructed to construct a two-layer opposing game model including an upper layer and a lower layer, wherein the upper layer game model of the two-layer opposing game model is a dynamic pricing model of a distribution network system operator, and the lower layer game model is a VPP energy management model;

[0115] Solving module, building the KAN network architecture, using the KAN network architecture to solve the two-layer opposing game model and obtain the game equilibrium solution.

[0116] Correspondingly, an embodiment of the present invention also provides a master-slave game two-layer model solving device, including a processor and a memory, wherein when the processor executes the computer program stored in the memory, it implements the master-slave game two-layer model solving method provided by the embodiment of the present invention.

[0117] For more specific processes of the above method, please refer to the corresponding contents disclosed in the aforementioned embodiments, which will not be repeated here.

[0118] Correspondingly, an embodiment of the present invention further provides a computer-readable storage medium for storing a computer program, wherein when the computer program is executed by a processor, the method for solving the master-slave game two-layer model as provided in the embodiment of the present invention is implemented.

[0119] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the systems, devices, and storage media disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method part.

[0120] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0121] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0122] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0123] The master-slave game two-layer model solving method, system, device and storage medium provided by the present invention are introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for general technical personnel in this field, according to the idea of ​​the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A method for solving a master-slave game two-layer model, characterized in that: The following steps are involved: S1. Construct a two-layer opposition game model including an upper layer and a lower layer. The upper layer model of the two-layer opposition game model is a dynamic pricing model of the distribution network system operator, and the lower layer model is a VPP energy management model. The specific process of constructing the dynamic pricing model of the distribution network system operator is as follows: the distribution network system operator formulates the purchase and sale prices of electricity with the VPP in each period, and then trades with the power market based on the on-grid electricity price and the power grid electricity price. The maximum transaction net profit is obtained by summing the price difference between the power sold by the distribution network system operator to the power market and the power purchased by the VPP and the power sold by the distribution network system operator. The first utility function of the game with maximizing the transaction net profit is: Where: The total time of the day is 24h; for The on-grid electricity price in the electricity market at the moment, for The electricity price of the power grid in the power market at the moment; for The amount of electricity sold by the distribution network system operator to the electricity market at any given moment; for The amount of electricity purchased by the distribution network system operator in the electricity market at any given moment; represents the amount of electricity sold by the jth VPP to the distribution network system operator; It represents the amount of electricity purchased by the jth VPP from the distribution network system operator; N is the total number of VPPs; constraining the first utility function according to a first constraint condition; The specific process of constructing the VPP energy management model is as follows: VPP reasonably arranges internal output according to the power purchase and sales from the distribution network system operator and the output plan of distributed power sources in each period to minimize the operating cost. The second utility function of the game is to minimize the operating cost, which is expressed as: Where: for time No. Micro gas turbine power generation cost; , , for Cost factor of gas turbine No. for time The operating cost of the energy storage power station; for The dispatch cost coefficient of the energy storage power station; for Time demand side The interruptible load cost of unit 1; For the demand side The interruption compensation electricity price for unit No. for time The amount of electricity purchased by VPP unit No. from the distribution network system operator; for time The amount of electricity sold by VPP unit No. to the distribution network system operator; for time Output power of micro gas turbine No. for time The charging and discharging power of the energy storage power station; for time The interrupting power of the interruptible load of the unit; The total time of the day is 24h; constraining the second utility function according to a second constraint; S2. Build the KAN network architecture and use it to solve the two-layer opposing game model to obtain the game equilibrium solution.

2. A method for solving a master-slave game double-layer model according to claim 1, characterized in that: The distribution network system operator should meet the following constraints between the electricity sold to the electricity market and the electricity purchased: Where: is the total amount of electricity purchased and sold by each VPP in the lower layer at time t, which is aggregated by the distribution network system operator and traded with the power market; The purchase and sale prices of electricity between the distribution network system operator and the VPP at each time period should meet the following constraints: Where: for The on-grid electricity price in the electricity market at the moment, for The electricity price set by the distribution network system operator at that moment, for The power purchase price set by the distribution network system operator at that moment, for The electricity price of the power grid in the power market at the moment.

3. According to claim 1, a method for solving a master-slave game double-layer model is characterized in that: The second constraint is that when the VPP responds to the distribution network system operator's strategy, it must meet the power balance between itself and the grid, and also meet the operating restrictions of each distributed power source. The second constraint is expressed as follows: Where: For VPP The maximum amount of electricity traded between the generating unit and the distribution network system operator; represents the load forecast value at time t; For MT The upward climbing rate of unit No. For MT Downward climbing rate of unit No. It means time Boolean variable of VPP unit. When the Boolean variable takes the value 0, it represents Moment VPP The unit purchases electricity from the distribution network system operator. When the value is 1, it means Moment VPP Unit No. sells electricity to the distribution network system operator; For MT Maximum output power of unit No. for time The charge state of the energy storage unit; For the previous time Moment The charge state of the energy storage unit; is the time scale, here it is taken as 1h; The charge state of the energy storage unit at time 0; For the whole day The charge state of the energy storage unit after time Take 24; for The upper limit of the state of charge of the energy storage unit; for The lower limit of the state of charge of the energy storage unit; for The upper limit of the charging and discharging power of the energy storage power station; for The lower limit of the charging and discharging power of the energy storage power station; for The maximum energy capacity value of the energy storage power station; For the demand side The maximum interruption amount of the interruptible load of the unit; for Wind turbines in Output power at the moment; for Wind turbines in The maximum output power at the moment, here we take the predicted value.

4. A method for solving a master-slave game double-layer model according to claim 1, characterized in that: The specific process of step S2 includes the following steps: S21, initializing system parameters; S22, generating a set of initial transaction electricity price sample points covering all electricity price sample intervals according to Latin hypercube sampling; S23. For the initial transaction electricity price sample points generated in step S22, the solver is called to calculate the transaction electricity of each VPP and form a corresponding sample data set. A deep learning model is constructed according to the generated sample data set. The initial transaction electricity price sample points are input into the deep learning model to obtain the transaction electricity of each VPP. S24, inputting the transaction power of each VPP into the objective function of the upper model to obtain the upper objective function value of each initial transaction power price sample point; S25, according to the upper objective function value of each initial transaction price sample point, the strategy space of the objective function in the upper model is divided into multiple key areas, and the value of the optimal solution of each key area is obtained , using the particle swarm optimization algorithm, calling the trained deep learning model to solve the local optimal transaction volume and electricity price; S26. Call the lower-level model, and call the solver to calculate the actual transaction power according to the optimal transaction power and electricity price.

5. A method for solving a master-slave game double-layer model according to claim 4, characterized in that: After step S26, the actual transaction volume is added as an excellent sampling point into the deep learning model for iteration until the convergence condition is met, and the optimal actual transaction volume is output.

6. A master-slave game two-layer model solving system, characterized in that: include: A construction module is used to construct a two-layer opposing game model including an upper layer and a lower layer. The upper layer model of the two-layer opposing game model is a dynamic pricing model of a distribution network system operator, and the lower layer model is a VPP energy management model. The specific process of constructing the dynamic pricing model of the distribution network system operator is as follows: the distribution network system operator formulates the purchase and sale prices of electricity with the VPP in each period, and then trades with the power market based on the on-grid electricity price and the power grid electricity price. The maximum transaction net profit is obtained by summing the price difference between the power sold by the distribution network system operator to the power market and the power purchased by the VPP and the power sold by the distribution network system operator. The first utility function of the game with maximizing the transaction net profit is: Where: The total time of the day is 24h; for The on-grid electricity price in the electricity market at the moment, for The electricity price of the power grid in the power market at the moment; for The amount of electricity sold by the distribution network system operator to the electricity market at any given moment; for The amount of electricity purchased by the distribution network system operator in the electricity market at any given moment; represents the amount of electricity sold by the jth VPP to the distribution network system operator; It represents the amount of electricity purchased by the jth VPP from the distribution network system operator; N is the total number of VPPs; constraining the first utility function according to a first constraint condition; The specific process of constructing the VPP energy management model is as follows: VPP reasonably arranges internal output according to the power purchase and sales from the distribution network system operator and the output plan of distributed power sources in each period to minimize the operating cost. The second utility function of the game is to minimize the operating cost, which is expressed as: Where: for time No. Micro gas turbine power generation cost; , , for Cost factor of gas turbine No. for time The operating cost of the energy storage power station; for The dispatch cost coefficient of the energy storage power station; for Time demand side The interruptible load cost of unit 1; For the demand side The interruption compensation electricity price for unit No. for time The amount of electricity purchased by VPP unit No. from the distribution network system operator; for time The amount of electricity sold by VPP unit No. to the distribution network system operator; for time Output power of micro gas turbine No. for time The charging and discharging power of the energy storage power station; for time The interrupting power of the interruptible load of the unit; The total time of the day is 24h; constraining the second utility function according to a second constraint; Solving module, building the KAN network architecture, using the KAN network architecture to solve the two-layer opposing game model and obtain the game equilibrium solution.

7. A master-slave game two-layer model solving device, characterized in that: It comprises a processor and a memory, wherein when the processor executes the computer program stored in the memory, the method for solving the master-slave game two-layer model as described in any one of claims 1 to 5 is implemented.

8. A computer-readable storage medium, characterized in that: Used to store a computer program, wherein when the computer program is executed by a processor, the method for solving the master-slave game two-layer model as described in any one of claims 1 to 5 is implemented.

Images