Power distribution system-virtual power plant scheduling method and system based on shared energy storage

Abstract

The invention relates to the technical field of virtual power plants, and provides a power distribution system-virtual power plant scheduling method and system based on shared energy storage, and the method comprises the steps: building a multi-main-body system cooperation framework of a power distribution system operator DSO, a virtual power plant VPP and a shared energy storage system SES; based on a multi-subject system cooperation framework, an SC-Mixed hybrid game model is constructed, an upper layer is a DSO operation problem with self-income maximization as a target, a lower layer is a VPP-SES scheduling problem, and SES and VPP cooperate with each other to reduce cost; and solving the SC-Mixed mixed game model by using an alternating direction multiplier method ADMM, and realizing adaptive optimization of model solving by dynamically adjusting a penalty factor to obtain a Pareto optimal result in a Nash equilibrium state. The invention aims to balance the benefit relationship between a guider and a follower by optimizing a traditional Stackelberg game model, and construct a multi-agent coordination mechanism to relieve systematic benefit loss caused by lack of dominant coordination in a cooperative game.

Description

Technical Field

[0001] The present invention belongs to the technical field of virtual power plants, and particularly relates to a distribution system-virtual power plant scheduling method and system based on shared energy storage. Background Art

[0002] As an important organizational form for aggregating and regulating distributed energy resources (DERs) in a new type of power system, a virtual power plant (VPP) can effectively integrate multiple resources such as renewable energy generation devices, energy storage systems, and controllable loads, improving the operating efficiency and economy of the system. Current research mainly focuses on constructing the cost optimization model of a single VPP. However, a single VPP has obvious limitations in operation: on the one hand, its operation strategy often ignores the global optimization requirements of the distribution system operator (DSO), which may cause safety problems such as node voltage over-limit; on the other hand, limited by the aggregation capacity, it is difficult to meet large-scale regulation requirements such as peak shaving and valley filling at the regional level.

[0003] To solve the above problems, it is urgent to establish a collaborative optimization mechanism between multiple VPPs and the DSO. Existing research has covered aspects such as stochastic optimization, application of game theory, and electro-carbon coupling mechanism. However, in the traditional Stackelberg game, the unequal status of participants is likely to damage the interests of followers, while a pure cooperative game lacks a leading coordination mechanism, affecting the overall benefits of the system. In addition, introducing a shared energy storage system (SES) helps to enhance the flexibility of the system. The SES can provide energy and reserve services for multiple VPPs, improving resource utilization efficiency and scheduling flexibility. However, in the multi-agent collaboration considering the participation of the SES, how to reasonably model its reserve capacity constraint and ensure the privacy of all parties is still a key challenge. Summary of the Invention

[0004] To solve the problems existing in the above-mentioned prior art, the present invention provides a distribution system-virtual power plant scheduling method based on shared energy storage, including the following steps: Step S1, establish a multi-agent system collaboration framework for the distribution system operator DSO, virtual power plant VPP, and shared energy storage system SES. The DSO, as the leader, manages the participation of the VPP and SES in the transactions of the energy spot market and reserve service market. The VPP optimizes the control of distributed generators according to the electricity price set by the DSO, and the SES provides energy storage services and reserve services for the VPP; Step S2: Based on the multi-agent system cooperation framework, construct the SC-Mixed hybrid game model. The upper layer is the DSO operation problem, aiming to maximize its own benefits, and the lower layer is the VPP-SES scheduling problem, where SES and VPP cooperate with each other to reduce costs. Step S3: Use the Alternating Direction Method of Multipliers (ADMM) to solve the SC-Mixed hybrid game model, and achieve the adaptive optimization of model solution by dynamically adjusting the penalty factor, obtaining the Pareto optimal result under the Nash equilibrium state.

[0005] Based on the above solution, each VPP exchanges information data with other VPPs, DSOs or SESs. The information data includes power flow, electricity price, energy / reserve service information, buying and selling information, and operation strategies. The DSO formulates the initial electricity price according to the electricity price set in the energy market to provide a reference for VPPs and SESs. VPPs and SESs formulate the initial operation strategies and SES reserve capacity according to the initial electricity price, and send the initial operation strategies and SES reserve capacity to the DSO, and the DSO optimizes the electricity price.

[0006] Based on the above solution, step S2 includes the following steps: Step S21: The upper-layer DSO optimization model establishes the DSO objective function aiming to maximize its own profit. The DSO objective function is subject to the upper and lower limits of the electricity price, the power line transmission capacity limit constraint, the average price constraint of power purchase and sale, and the multi-period power balance constraint under the day-ahead - real-time market coupling mechanism. Step S22: The VPP acts as both a consumer and a producer in the system, and establishes the VPP objective function aiming to minimize the total cost. The VPP objective function is subject to the power flow power balance constraint, the distributed generator operation constraint, and the load limit constraint. Step S23: The SES establishes the SES objective function aiming to maximize the comprehensive income. The SES objective function is subject to the SES reserve capacity constraint and the daily charge and discharge power balance constraint.

[0007] Furthermore, the solution of the SC-Mixed hybrid game model is as follows: Decompose the SC-Mixed into a multi-agent cooperation cost minimization problem and a cost allocation problem, and use the ADMM algorithm to solve the transaction problems of different stakeholders, specifically including: (1) First, for Construct the augmented Lagrangian equation:

[0008] Among them, denotes the Lagrange multiplier, denotes the penalty term, Denoted as the optimal power trading volume from i to j of the virtual power plant at time t, Denoted as the optimal power trading volume from j to i of the virtual power plant at time t, Denoted as the total cost of the i-th virtual power plant; (2) Iterative initialization: Set the initial iteration values Denoted as the maximum number of iterations; The initial value of the power trading between virtual power plants VPP is set to ; The initial setting of the Lagrange multiplier ; (3) Parameter iterative update:

[0009] Among them, Denoted as the power trading power of the , with the unit of KW, Denoted as the power trading power of the , with the unit of KW, The penalty term coefficient; (4) Convergence judgment:

[0010] Among them, Denoted as the residual convergence setting value.

[0011] On the basis of the above scheme, on the basis of the ADMM algorithm, the self-saturation characteristic of the logarithmic function is used to adjust the penalty factor in the iterative process of the algorithm, balance the convergence of the original residual and the dual residual, and introduce a modulo operation to limit the update frequency control strategy, specifically including:

[0012] Among them, Denoted as the penalty factor in the k-th iteration; and respectively denote the original residual and the dual residual in the k-th iteration, Denoted as the adjustable parameter, usually set .

[0013] On the basis of the above scheme, if the convergence is not satisfied, the improved Illinois algorithm is used to update the electricity price, and the feasible region of the electricity price is rapidly reduced in each iteration:

[0014] Among them, and respectively denote the electricity prices set for purchase and sale by the DSO in the -th iteration; Denoted as the convergence accuracy; Denoted as the Illinois method parameter; if the electricity price obtained from the iteration result meets the convergence condition, it is used as the initial parameter of the electricity price set by the DSO.

[0015] Based on the above scheme, the SES objective function includes the charging and discharging service fees charged to the VPP, the charging and discharging costs of the energy storage, the standby power capacity configuration cost, and the standby power income charged to the VPP:

[0016]

[0017] Among them, is the SES objective function; Denoted as the service income; Denoted as the charging and discharging cost of the SES; Denoted as the reserved capacity cost that the SES needs to pay to the DSO; Denoted as the reserve cost of the VPP paid by the SES, here referring to the income; and respectively denote the charging and discharging power of the i-th VPP user using the SES for charging and discharging behaviors in cycle t, with the unit of KW; and respectively denote the unit charging and discharging cost coefficients of the SES; Denoted as the number of virtual power plants; T is the operation time of a typical day; Denoted as the unit transmission cost of the energy storage during the charging and discharging process; The SES objective function is subject to the SES standby capacity constraint:

[0018]

[0019]

[0020] Among them, and respectively denote the SES capacities of the SES in cycles t and t + 1, with the unit of KWh; Denoted as the self-discharge rate of the SES; and respectively denote the transmission efficiency coefficients of the SES during the charging / discharging process; and respectively denote the upper and lower limits of the charging and discharging power of the i-th VPP when using the SES for charging and discharging behaviors, with the unit of KW; and are all Boolean variables, representing the charge and discharge states of the i-th VPP in cycle t: that is, when , , it represents the charging state; when , , it represents the discharging state; and respectively represent the maximum charge / discharge power per unit time of the SES, in KW; when the VPP calls the SES simultaneously, they are not allowed to exceed the upper limit; and respectively represent the maximum capacity and minimum capacity allowed by the SES, in KWh; The SES objective function is subject to the daily charge and discharge power balance constraint:

[0021] where represents the cycle time of the system operation; and respectively represent the power transmission efficiency coefficients of the SES during the charge / discharge process; and respectively represent the maximum charging and discharging powers of the SES per unit time, in KW.

[0022] The present invention also provides a distribution system-virtual power plant scheduling system based on shared energy storage, including: A multi-agent system cooperation framework design module, used to establish a multi-agent system cooperation framework for the distribution system operator DSO, virtual power plant VPP, and shared energy storage system SES. The DSO, as a leader, manages the transactions of the VPP and SES participating in the energy spot market and reserve service market. The VPP optimizes the distributed generator control according to the electricity price set by the DSO, and the SES provides energy storage services and reserve services for the VPP; A model construction module, based on the multi-agent system cooperation framework, constructs an SC-Mixed hybrid game model. The upper layer is the DSO operation problem, with the goal of maximizing its own profit, and the lower layer is the VPP-SES scheduling problem, where the SES and VPP cooperate with each other to reduce costs; A solution module, which uses the alternating direction method of multipliers ADMM to solve the SC-Mixed hybrid game model, and realizes the adaptive optimization of the model solution by dynamically adjusting the penalty factor, and obtains the Pareto optimal result under the Nash equilibrium state.

[0023] The solution module includes an adjustment module, which utilizes the self-saturation characteristic of the logarithmic function to adjust the penalty factor during the algorithm iteration process, balance the convergence of the original residual and the dual residual, and introduce a modulo operation to limit the update frequency control strategy; the solution module also includes a superlinear convergence module, which uses the obtained converged electricity price as the initial electricity price parameter of the DSO.

[0024] Compared with the prior art, the present invention has the following beneficial effects: 1. A three-level collaborative system of "DSO-VPP-SES" is constructed. Based on the coordinated operation of multiple VPPs dominated by the DSO and the cooperative participation of the SES, by integrating the advantages of the DSO, VPPs, and SES, the overall system benefit is maximized, not only improving the economic benefit but also enhancing the stability and security of the power grid. 2. A Stackelberg cooperative hybrid game framework is proposed, with the DSO leading the electricity price mechanism. Through the Stackelberg game, the transaction electricity price between the DSO and the virtual power plant VPP is optimized, and the Nash bargaining mechanism is used to ensure the fair distribution of interests among all participating parties in the cooperation, solving the problem of damaged interests of followers in the traditional Stackelberg game, and at the same time avoiding the loss of the overall system benefit caused by the lack of a leading coordination mechanism in a pure cooperative game. 3. An Illinois method with superlinear convergence and an ADMM solution strategy with adaptive penalty factor adjustment are proposed. The outer layer processes the game equilibrium, and the inner layer optimizes the operation strategy. Based on the adaptive parameter adjustment mechanism, parallel computing and distributed solution of the optimization sub-problems of each subject are efficiently completed, achieving the Pareto optimal result under the Nash equilibrium state. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is the flowchart of the method of the present invention; Figure 2 It is the schematic diagram of the collaborative information interaction framework of the DSO-VPP-SES multi-agent system; Figure 3 It is the schematic diagram of the SC-Mixed game interaction model; Figure 4 It is the schematic diagram of the distributed solution algorithm process; Figure 5 It is the architecture diagram of the system of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0026] The present invention will be further described below in conjunction with specific embodiments, but the implementation manners of the present invention are not limited thereto.

[0027] Embodiment 1 As Figure 1As shown in the figure, this embodiment discloses a dispatching method for a distribution system-virtual power plant based on shared energy storage, which specifically includes the following steps: S1: Construct a multi-agent collaborative system including a Distribution System Operator (DSO), a Virtual Power Plant (VPP, vpp, VPPi, vppi, VPP i , vpp i ) and a Shared Energy Storage System (SES). As shown in Figure 2 the figure, as the leader, the DSO manages and controls the VPP and SES to participate in the energy spot market and the reserve service market for energy and reserve capacity trading. This trading is restricted by relevant documents and operators to ensure the fairness of multi-agent market trading.

[0028] The DSO is mainly responsible for resetting the day-ahead electricity price according to the hierarchical electricity price, coordinating the VPP and SES to complete electricity trading at the best price signal when participating in market trading. During the process of pricing and trading with the VPP, the DSO aims to maximize its own profit; the SES can provide energy storage services and reserve services for the VPP. Due to the volatility and uncertainty of renewable energy, the SES assists the VPP in storing part of the energy when the generator output power is high to reduce power waste. The SES sets reserve capacity to prevent certain extreme situations from occurring and can provide operation support for the VPP and DSO to ensure the security of system operation. The goal of the SES is to maximize its revenue (i.e., energy trading profit and reserve service profit). The VPP will re-optimize the control operation strategy of each Distributed Generator (DG) according to the price set by the DSO and the SES to minimize its own operation cost.

[0029] Assume that the above-mentioned agents are geographically adjacent and connected by power lines. Each VPP can provide electricity for load demand and interact with other VPPs, DSOs or SESs through information data exchange, including power flow, electricity price, energy / reserve service information, buying and selling information, and operation strategies. During operation, the DSO needs to formulate an initial electricity price according to the electricity price set in the energy market to provide a reference for Virtual Power Producers (VPPs) and the Shared Energy Storage System (SES). The VPPs and SESs formulate initial operation strategies and SES reserve capacity according to the electricity price, send them to the DSO and optimize the electricity price.

[0030] Since the SES can effectively integrate distributed energy and balance load fluctuations, it plays a crucial role in system operation. Therefore, first establish a multi-modal SES reserve capacity model: When the supercapacitor is in the charging state, energy reserve can be provided by reducing the charging power or switching from charging to the discharging state. The upward expansion of the reserve can be achieved by increasing the charging power. On the contrary, when the supercapacitor is in the discharging state, the method to achieve the upward expansion of the reserve is to increase the discharging power. While for the downward reduction of the reserve, the discharging power is reduced or switched from discharging to the charging state. Therefore, fully considering the characteristics of the reserve capacity of the supercapacitor in the charging or discharging state, the upper and lower limits of the reserve capacity are shown in formulas (1) - (4): (1) (2) (3) (4) where i represents the index of the virtual power plant unit (VPP); i , represents the number of virtual power plants. represents the index of the time slot in the operating time domain; and respectively represent the SES reserve power increased and decreased at time t (by the i-th VPP); and respectively represent the upper and lower limits of the charging and discharging power of the i-th VPP when using SES to describe the charging and discharging behaviors, with the unit of KW; and are both Boolean variables, respectively representing the charging and discharging states of the i-th VPP within time t; when , it represents the charging state; when , it represents the discharging state; and respectively represent the maximum charging power and maximum discharging power allowed per unit time of SES. When multiple VPPs use the shared energy storage system simultaneously, it is not allowed to exceed the maximum upper limit, with the unit of KW; and respectively represent the upper and lower limit reserve powers that can be transferred to the i-th VPP at time t, with the unit of KW.

[0031] The above-mentioned multi-modal SES reserve capacity model proposed by the present invention fully considers three dimensions: power dynamic regulation, mode intelligent switching, and SOC safety protection, and constructs various different operating modes of SES, namely: increasing or decreasing the discharge power; increasing and decreasing the charging power; mutual conversion between the charging state and the discharging state. Since the energy storage system is limited by its own capacity and does not have infinite reserve transmission capacity, the upper and lower limits of the energy storage capacity must be considered, and the response market access rules need to be introduced to increase the responsibility for the market mechanism (i.e., increasing the power reserve capacity). According to the existing data, most existing independent system operators (ISOs) at home and abroad require that the energy storage system participate in the market operation for at least four hours. When the energy storage system uses the reserve capacity within a continuous period, it is necessary to ensure that the reserve capacity does not exceed the upper and lower limits of the entire energy storage capacity. When the reserve capacity of the energy storage system is scheduled for a long time, it may not be able to provide energy power and reserve power for subsequent times. The present invention sets that the reserve capacity of the energy storage system should have the delivery ability at least continuously for four hours above and below the limit values, as shown in formulas (5) to (8): (5) (6) (7) (8) Wherein: and respectively represent the maximum values of the upward and downward reserved capacities called by VPPi at time t, with the unit of KWh; represents the start time of the upward and downward called reserved capacities; and respectively represent the SES capacities at times t and t + 1, with the unit of KWh; and respectively represent the maximum allowable capacity and the minimum capacity of SES, with the unit of KWh.

[0032] S2: Based on the constructed DSO-VPPs-SES multi-agent collaborative system, establish a multi-layer model of Stackelberg-Cournot Mixed Game (SC-Mixed), such as Figure 3As shown in the figure, the upper layer represents the DSO operation problem, and the lower layer describes the VPPs-SES scheduling problem. This hybrid game model divides the system into two levels by finding the balance point between SCs: the DSO is in the role of the leader, aiming to maximize its own benefits and optimize the electricity price for the lower layer; at the same time, the DSO interacts with VPPs in terms of electric energy and formulates its scheduling strategy to ensure the economy and stability of power supply. In the lower layer model, SES and VPPs cooperate with each other to reduce costs. According to the transmitted electricity price, the virtual power plant plans to change the generator set strategy and energy trading, which is fed back to the upper layer DSO as the initial strategy to provide a basis for its further decision-making. It includes the following steps: S21: Establish the upper layer optimization model of DSO, specifically including: S211: Aiming to maximize the total profit of DSO maximize, establish the objective function as shown in formulas (9)~(12): (9) (10) (11) (12) Among them, represents the income of DSO from selling electric energy to VPP during the intraday operation stage; represents the cost of DSO purchasing the remaining electric energy from VPPs during the intraday operation stage; represents the reserve capacity cost of SES, recorded as the income of DSO; represents the time in a typical day; represents the unit period of the operation time; and respectively represent the purchased and sold power quantities of DSO to VPPi, with the unit of KW; and respectively represent the selling electricity price and purchasing electricity price formulated by DSO; represents the grid electricity price at time t; represents the on-grid electricity price at time t; and respectively represent the power purchased and sold by the upper layer grid to DSO, with the unit of KW.

[0033] S212: The constraint conditions that the objective function of the upper layer optimization model of DSO needs to satisfy are: (1) The upper and lower limit constraint conditions of the electricity price, as shown in formulas (13) and (14): (13) (14) Among them, and respectively represent the selling electricity price and the purchasing electricity price set for the DSO; represents the grid tariff price at time t; represents the on-grid electricity price at time t.

[0034] (2) Power line transmission capacity limit constraint conditions, as shown in formulas (15) - (19): (15) (16) (17) (18) (19) Among them, represents the upper limit of the transmission power of the distribution network line, with the unit of KW; and respectively represent the electricity purchase and sales parameters of the i-th VPP at time t; and are both Boolean variables; and respectively represent the upper limits of the power of electricity purchase and sales, with the unit of KW.

[0035] (3) Average price constraint for electricity purchase and sales To prevent the DSO from abusing market power and inducing price spikes, this mechanism constructs a dynamic electricity price constraint model based on bilateral transactions, adding an average price constraint on the electricity buying and selling prices of the DSO, as shown in formulas (20) and (21): (20) (21) Among them, and respectively represent the average values of the electricity prices at which the DSO purchases and sells electricity from / to the VPP at time t.

[0036] (5) Multi-period power balance constraint of the DSO under the day-ahead - real-time market coupling mechanism, as shown in formula (22): (22) Among them, and respectively represent the power purchased and sold by the upper-level grid from / to the DSO, with the unit of KW; and respectively represent the power of the energy purchased and sold by the i-th VPP at time t, with the unit of KW.

[0037] S22: Virtual power plant agent modeling, specifically including: S221: A single VPP acts as both a consumer and a producer in the system, aiming to minimize the total cost , including unit operation cost, unit maintenance cost, reserve power dispatch cost, carbon footprint compensation cost, electricity buying and selling cost, and SES service cost, and establish the objective function as shown in formulas (23) and (24): (23) (24) Among them, represents the electricity cost coefficient of distributed energy resources (DERs), including gas boiler (GB), gas turbine (GT), electric vehicle (EV), photovoltaic (PV), and wind turbine (WT); represents the output power of DG in the i-th VPP at time t, in kW; represents the recovery cost coefficient of DG; and respectively represent the reserve cost coefficients when the i-th VPP calls the up or down reserve capacity of SES at time t; represents the type number of pollutants emitted during unit operation; represents the unit in the i-th VPP emission coefficient of pollutant k.

[0038] S222: Establish the constraints of the virtual power plant, including: (1) Power flow power balance constraint, as shown in formula (25): (25) Among them, and respectively represent the output power of the PV unit and the WT in VPPi at time t, in kW; represents the output power of the EV in VPPi at time t, in kW; represents the exchange power between VPPi and other VPPs at time t, in kW; represents the power consumption of the air conditioner (AC) at time t, in kW; represents the load power of VPPi at time t, in kW.

[0039] (2) The constraint of the operating power of the Distributed Generator (DG) is as shown in formula (26): (26) Among them, is expressed as the upper limit power of the distributed generator, with the unit of KW.

[0040] (3) The load limit constraint is described as shown in Equation (27): (27) Among them, and respectively represent the flexible load and the rigid load of VPPi at time t.

[0041] To more accurately describe the power adjustment ability of the virtual power plant power plan, the automatic response ability of all VPP flexible power loads is through the adjustable ratio per cycle and the total ratio adjustment to control. Then the flexible power load adjustment constraint of the user aggregator is described as shown in Equations (28) to (30): (28) (29) (30) Among them, represents the power load deviation adjustment of VPPi at time t; represents the flexible power load adjusted by VPPi at time t; represents the maximum allowable ratio of power load adjustment at time t; represents the ratio of the total power load adjustment in VPPi during a day; and The larger they are, the more flexible the VPP adjusts the power load, and the greater the DR ability of the VPP.

[0042] S23: Build a shared energy storage system model, specifically including: S231: The shared energy storage system takes the maximum comprehensive income as the objective function, including: the charge and discharge service fee charged to the VPP, the charge and discharge cost of the energy storage, the reserve power capacity configuration cost, and the reserve power income charged to the VPP, as shown in Formulas (31) to (32): (31) (32) Among them, is the SES objective function; represents the service income; represents the charge and discharge cost of the SES; represents the reserve capacity cost that the SES needs to pay to the DSO; Reserve costs of the virtual power plant (VPP) represented as SES payments, here referring to revenue; and respectively represent the charging and discharging powers of the i-th VPP user using SES for charging and discharging behaviors in cycle t, with the unit of KW; and respectively represent the unit charging and discharging cost coefficients of SES; represents the number of virtual power plants; T is the operation time of a typical day; represents the unit transmission cost during the charging and discharging processes of the energy storage.

[0043] S222: Establish SES constraint conditions, including: (1) SES reserve capacity constraint, as shown in formulas (33) - (35): (33) (34) (35) Among them, and respectively represent the SES capacities of SES in cycles t and t + 1, with the unit of KWh; represents the self-discharge rate of SES; and respectively represent the power transmission efficiency coefficients during the charging / discharging processes of SES; and respectively represent the upper and lower limits of the charging and discharging powers of the i-th VPP when using SES for charging and discharging behaviors, with the unit of KW; and are both Boolean variables, representing the charging and discharging states of the i-th VPP in cycle t: that is, when , , it represents the charging state; when , , it represents the discharging state; and respectively represent the maximum charging / discharging powers of SES per unit time, with the unit of KW; when VPPs call SES simultaneously, they are not allowed to exceed the upper limit; and respectively represent the maximum and minimum capacities allowed for SES, with the unit of KWh.

[0044] (2) To ensure that SES provides continuous and stable energy services, a typical-day charging and discharging power balance constraint is introduced, that is, the charging power and discharging power of the energy storage system need to meet the energy neutral condition, as shown in formula (36): (36) Among them, represents a cycle time of the system operation; and respectively represent the power transmission efficiency coefficients of the SES during the charging / discharging process; and respectively represent the maximum charging and discharging powers of the SES per unit time, with the unit of KW.

[0045] Regarding the interest-driven decision-making mechanism among the DSO, VPP, and SES, a collaborative optimization framework based on the Stackelberg-Cournot hybrid game is constructed. In this model, the DSO is the leader and the VPP-SES alliance is the follower. The game elements include: participants, strategies, and payoffs, as shown in formula (37): (37) Among them, represents the set of participants, including the DSO, SES, and VPP; represents the pricing strategy of the DSO; represents the SES strategy; represents strategy.

[0046] S3: Solving the DSO-VPP-SES collaborative optimization model based on the Stackelberg-Cournot hybrid game, specifically including: S31: Transforming the SC-Mixed game model into two sub-problems, including the multi-agent collaborative cost minimization problem and the cost allocation problem, and solving them sequentially.

[0047] Based on the price signal coordination architecture, the upper-layer distribution system operator (DSO) transmits dynamic energy price parameters to the lower-layer virtual power plant-shared energy storage alliance (VPPs-SES) through two-way information interaction. This collaborative mechanism realizes the minimization of the system comprehensive cost on the premise of ensuring the interest balance of multiple agents. Its cooperative optimization process can adopt the Nash bargaining game model to solve the Pareto optimal solution of the multi-agent alliance, as shown in formula (38): (38) Among them, represents the costs of the VPP and SES participating in the negotiation respectively; represents the costs obtained by the VPP and SES not participating in the negotiation, that is, the negotiation breakdown point.

[0048] Since this model is non-convex and non-linear, it is difficult to solve directly and needs to be transformed into two sub-problems (i.e., the multi-agent collaborative cost minimization problem and the cost allocation problem) to be solved sequentially. The specific decomposition steps are as follows: Sub - problem 1: The multi - agent cooperation cost minimization problem, as shown in Equation (39): (39) where, represents the total cost of multi - agent cooperation benefits.

[0049] Sub - problem 2: The cost allocation problem, as shown in Equation (40): (40) where, represents the exchange electricity price between VPPi and VPPj; represents the optimal solution of the electricity trading volume between i and j of the virtual power plant obtained from sub - problem 1; represents the minimum value of the cooperation cost obtained through sub - problem 1.

[0050] S32: A distributed algorithm that combines super - linear convergence characteristics and dynamic parameter optimization, specifically including: Using the Alternating Direction Method of Multipliers (ADMM) algorithm to solve the trading problems of different stakeholders can protect the privacy of each entity.

[0051] First, initialize the parameters. Import the initial load of the VPP and the initial electricity price set by the DSO into the system. After one round of optimization by VPPs - SES, the results are fed back to the DSO. The DSO will optimize and adjust the electricity price according to the strategy and transmit it to VPPs - SES; after one round of iteration, a criterion is formulated through Equation (43) to determine whether the convergence condition is reached. If the convergence condition is reached, skip the iteration and output the final electricity price and scheduling strategy. The specific steps are as follows: (1)Construct the augmented Lagrangian equation for as shown in Equation (41): (41) where, represents the Lagrange multiplier, represents the penalty term, represents the optimal electricity trading volume from i to j of the virtual power plant at time t, represents the optimal electricity trading volume from j to i of the virtual power plant at time t, represents the total cost of the i - th virtual power plant.

[0052] (2)Iterative initialization: Set the initial iteration values represents the maximum number of iterations; the initial value of the electricity trading between virtual power plants (VPPs) is set to ; the initial setting of the Lagrange multiplier .

[0053] (3) The parameter iterative update is shown in Equation (42): (42) Wherein, represents the power trading power of the th, with the unit of KW, represents the power trading power of the th, with the unit of KW, The th penalty term coefficient.

[0054] (4) Determine the convergence, as shown in Equation (43): (43) Wherein, represents the residual convergence set value.

[0055] For the alternating direction multiplier method (ADMM) in the distributed optimization scenario, there are the following technical problems, including: ① The convergence speed is restricted by the sensitivity of the penalty factor ρ, and the iterative efficiency is low; ② It is difficult to guarantee the global nature of the solution in non-convex problems; ③ The frequent adjustment of the penalty factor ρ in the traditional manual parameter adjustment mode lacks theoretical support and does not meet the requirements of engineering practice. To solve the above problems, the present invention innovatively introduces a dual adaptive mechanism, realizes the adaptive optimization of the ρ value through a dynamic penalty factor adjustment algorithm, uses the self-saturation characteristic of the logarithmic function to adjust the penalty factor in the algorithm iteration process, balances the convergence of the original residual and the dual residual, and introduces a modulo operation to limit the update frequency control strategy to ensure that the iteration process effectively avoids the risk of the algorithm diverging due to falling into an infeasible domain. This composite adjustment mechanism not only improves the robustness of the solution but also greatly improves the calculation efficiency, as shown in Equation (44): (44) In the formula, represents the penalty factor in the kth iteration; and respectively represent the original residual and the dual residual in the kth iteration, represents an adjustable parameter, usually set to .

[0056] If the convergence condition of Equation (43) is not satisfied, then determine whether the vpp purchase and sales power meet the conditions and then execute the improved Illinois algorithm to update the electricity price. This algorithm has a superlinear convergence speed, superior to the interval linear convergence characteristic of the bisection method, and the specific implementation process is as Figure 4 shown. represents the number of iterations of the Illinois method. Each iteration quickly shrinks the feasible domain of the electricity price, and its convergence criterion is shown in Equation (45): (45) Among them, and respectively represent the electricity prices for purchase and sale set by DSO in the i-th iteration; represents the convergence accuracy; represents the Illinois method parameter; if the electricity price obtained from the iteration result meets the convergence condition, it is used as the initial parameter of the electricity price set by DSO.

[0057] In summary, by constructing a three-level collaborative system of "DSO-VPP-SES", a Stackelberg cooperative mixed game (SC-Mixed game) framework is proposed. The DSO dominates the electricity price mechanism, optimizes the transaction electricity price between the DSO and the virtual power plant through the Stackelberg game, calculates the optimal operation strategies of multiple virtual power plants, and realizes the electricity-energy and information collaborative interaction among multiple entities; an Illinois method with superlinear convergence and an ADMM solution strategy with adaptive penalty factor adjustment are proposed. The algorithm adopts a two-layer iterative structure: the outer layer processes the game equilibrium, and the inner layer optimizes the operation strategy. Based on the adaptive parameter adjustment mechanism, parallel computing and distributed solution of the optimization sub-problems of each entity are efficiently completed, and the Pareto optimal result under the Nash equilibrium state is achieved.

[0058] Embodiment 2 This embodiment discloses a distribution system-virtual power plant dispatching system based on shared energy storage, which is used to implement the distribution system-virtual power plant dispatching method described in Embodiment 1, as Figure 5 shown, including: A multi-agent system cooperation framework design module, which is used to establish a multi-agent system cooperation framework for the distribution system operator DSO, the virtual power plant VPP, and the shared energy storage system SES. The DSO, as the leader, manages the participation of the VPP and SES in the transactions of the energy spot market and the reserve service market. The VPP optimizes the distributed generator control according to the electricity price set by the DSO, and the SES provides energy storage services and reserve services for the VPP; A model construction module, which constructs an SC-Mixed hybrid game model based on the multi-agent system cooperation framework. The upper layer is the DSO operation problem, with the goal of maximizing its own profit, and the lower layer is the VPP-SES dispatching problem, where the SES and the VPP cooperate with each other to reduce costs; A solution module, which uses the alternating direction method of multipliers ADMM to solve the SC-Mixed hybrid game model, and realizes the adaptive optimization of the model solution by dynamically adjusting the penalty factor, and obtains the Pareto optimal result under the Nash equilibrium state.

[0059] The solution module includes a regulation module, which uses the self-saturation characteristics of the logarithmic function to adjust the penalty factor in the algorithm iteration process, balances the convergence of the original residual and the dual residual, and uses a modular operation to limit the update frequency control strategy; the solution module also includes a superlinear convergence module, which uses the obtained converged electricity price as the initial electricity price parameter of the DSO.

[0060] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0061] Although the above describes the specific implementation methods of the present invention, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A dispatching method for a distribution system-virtual power plant based on shared energy storage, characterized in that It includes the following steps: Step S1, establish a multi-agent system cooperation framework for the Distribution System Operator (DSO), Virtual Power Plant (VPP), and Shared Energy Storage System (SES). The DSO, as the leader, manages the participation of the VPP and SES in the transactions of the energy spot market and reserve service market. The VPP optimizes the control of distributed generators according to the electricity price set by the DSO, and the SES provides energy storage services and reserve services for the VPP. Step S2, based on the multi-agent system cooperation framework, construct an SC-Mixed hybrid game model. The upper layer is the DSO operation problem, aiming to maximize its own benefits, and the lower layer is the VPP-SES scheduling problem, where the SES and VPP cooperate with each other to reduce costs. Step S3, use the Alternating Direction Method of Multipliers (ADMM) to solve the SC-Mixed hybrid game model, and achieve the adaptive optimization of the model solution by dynamically adjusting the penalty factor, obtaining the Pareto optimal result in the Nash equilibrium state.

2. A dispatching method for a distribution system-virtual power plant based on shared energy storage according to claim 1, characterized in that Each VPP interacts with other VPPs, DSOs, or SESs through information data exchange. The information data includes power flow, electricity price, energy / reserve service information, buying and selling information, and operation strategies. The DSO formulates the initial electricity price according to the electricity price set in the energy market to provide a reference for the VPP and SES. The VPP and SES formulate the initial operation strategies and SES reserve capacity according to the initial electricity price, and send the initial operation strategies and SES reserve capacity to the DSO, and the DSO optimizes the electricity price.

3. A dispatching method for a distribution system-virtual power plant based on shared energy storage according to claim 1, characterized in that, Step S2 includes the following steps: Step S21, the upper-layer DSO optimization model establishes a DSO objective function aiming to maximize its own profit. The DSO objective function is subject to the upper and lower limits of the electricity price, the power line transmission capacity limit constraint, the average price constraint of power purchase and sale, and the multi-period power balance constraint under the day-ahead - real-time market coupling mechanism. Step S22, the VPP acts as both a consumer and a producer in the system, and establishes a VPP objective function aiming to minimize the total cost. The VPP objective function is subject to the power flow power balance constraint, the distributed generator operation constraint, and the load limit constraint. Step S23, the SES establishes an SES objective function aiming to maximize the comprehensive income. The SES objective function is subject to the SES reserve capacity constraint and the daily charge and discharge power balance constraint.

4. A dispatching method for a distribution system-virtual power plant based on shared energy storage according to claim 3, characterized in that The solution of the SC-Mixed hybrid game model is as follows: decompose the SC-Mixed into a multi-agent cooperation cost minimization problem and a cost allocation problem, and use the ADMM algorithm to solve the transaction problems of different stakeholders, specifically including: (1)First, for construct the augmented Lagrangian equation: Among them, is expressed as a Lagrange multiplier, is expressed as a penalty term, is expressed as the optimal amount of power trading between i and j of the virtual power plant at time t, is expressed as the optimal amount of power trading between j and i of the virtual power plant at time t, represents the total cost of the i-th virtual power plant; (2) Iterative initialization: Set the initial iterative value is represented as the maximum number of iterations; the initial value of the power transaction between virtual power plants (VPPs) is set to ; the initial setting of the Lagrange multiplier ; (3) Parameter iterative update: Among them, is expressed as the electricity trading power, with the unit of KW, is expressed as the electricity trading power, with the unit of KW, is expressed as the penalty term coefficient; (4) Convergence judgment: Among them, is expressed as the residual convergence set value.

5. A dispatching method for a distribution system-virtual power plant based on shared energy storage according to claim 4, characterized in that, Based on the ADMM algorithm, use the self-saturation characteristic of the logarithmic function to adjust the penalty factor in the algorithm iteration process, balance the convergence of the primal residual and the dual residual, and introduce a modulo operation to limit the update frequency control strategy, specifically including: Among them, is expressed as the penalty factor in the k-th iteration; and are respectively expressed as the primal residual and the dual residual in the k-th iteration, is expressed as an adjustable parameter, usually set .

6. A dispatching method for a distribution system-virtual power plant based on shared energy storage according to claim 5, characterized in that, If the convergence is not satisfied, use the improved Illinois algorithm to update the electricity price, and quickly narrow the electricity price feasible region in each iteration: Among them, and respectively represent the electricity prices set for purchase and sale by DSO in the i-th iteration; represents the convergence accuracy; represents the Illinois method parameter; if the electricity price obtained from the iteration result meets the convergence condition, it is used as the initial parameter of the electricity price set by DSO.

7. A dispatching method for a distribution system-virtual power plant based on shared energy storage according to claim 3, characterized in that, The SES objective function includes the charging and discharging service fee charged to the VPP, the energy storage charging and discharging cost, the backup power capacity configuration cost and the backup power income charged to the VPP: Among them, is the SES objective function; is expressed as service revenue; is expressed as the charge and discharge cost of SES; is expressed as the reserved capacity cost that SES needs to pay to the DSO; is expressed as the reserve cost of the VPP paid by SES, referring to revenue here; and respectively represent the charge and discharge power of the i-th VPP user using SES for charge and discharge behavior in period t, with the unit of KW; and respectively represent the unit charge and discharge cost coefficient of SES; represents the number of virtual power plants; T is the operation time of a typical day; is expressed as the unit transmission cost of energy storage during charge and discharge; The SES objective function is subject to the SES spare capacity constraint: Wherein, and represent the SES capacities of the SES in cycles t and t + 1 respectively, with the unit of KWh; represents the self-discharge rate of the SES; and represent the power transmission efficiency coefficients of the SES during the charge / discharge process respectively; and represent the upper and lower limits of the charge / discharge power of the i-th VPP when using the SES for charge / discharge behavior respectively, with the unit of KW; and are both Boolean variables, representing the charge / discharge state of the i-th VPP in cycle t: that is, when , , it represents the charging state; when , , it represents the discharging state; and represent the maximum charge / discharge power of the SES per unit time respectively, with the unit of KW; when the VPPs call the SES simultaneously, they are not allowed to exceed the upper limit; and represent the maximum capacity and the minimum capacity allowed for the SES respectively, with the unit of KWh; The SES objective function is subject to the daily charge and discharge power balance constraint: Among them, represents a cycle time of the system operation; and respectively represent the power transmission efficiency coefficients of the SES during the charging / discharging process; and respectively represent the maximum charging and discharging powers of the SES per unit time, with the unit of KW.

8. A distribution system-virtual power plant dispatching system based on shared energy storage, characterized in that include: The multi-agent system collaboration framework design module is used to establish a multi-agent system collaboration framework for distribution system operators DSO, virtual power plants VPP and shared energy storage systems SES. DSO acts as a guide to manage VPP and SES's participation in transactions in the energy spot market and reserve service market. VPP optimizes distributed generator control according to the electricity price set by DSO, and SES provides energy storage services and reserve services for VPP. The model building module builds the SC-Mixed mixed game model based on the multi-agent system collaboration framework. The upper layer is the DSO operation problem, with the goal of maximizing its own revenue, and the lower layer is the VPP-SES scheduling problem, where SES and VPP cooperate with each other to reduce costs; The solution module uses the alternating direction multiplier method ADMM to solve the SC-Mixed mixed game model, and realizes adaptive optimization of the model solution by dynamically adjusting the penalty factor to obtain the Pareto optimal result under the Nash equilibrium state.

9. The dispatching system of a distribution system-virtual power plant based on shared energy storage according to claim 8, characterized in that, The solution module includes a regulation module, which uses the self-saturation characteristics of the logarithmic function to adjust the penalty factor in the algorithm iteration process, balances the convergence of the original residual and the dual residual, and quotes a modular operation to limit the update frequency control strategy.

10. A distribution system-virtual power plant dispatching system based on shared energy storage according to claim 9, characterized in that, The solution module further includes a superlinear convergence module, which uses the obtained converged electricity price as an initial electricity price parameter of the DSO.

Images