Multi-microgrid shared energy storage scheduling method considering the uncertainties of electric vehicle charging and discharging

Abstract

This invention discloses a multi-microgrid shared energy storage scheduling method considering the uncertainties of electric vehicle charging and discharging. First, a shared energy storage system framework incorporating microgrids is established, with a shared energy storage operator investing in and managing centralized energy storage facilities to provide energy storage services to multiple microgrid operators and electric vehicle clusters. Second, electric vehicle travel behavior is simulated and classified based on a random forest algorithm to generate high-precision charging and discharging load curves for electric vehicle clusters. Then, a Stackelberg game optimization model is constructed with the shared energy storage operator as the leader and each microgrid operator as a follower. The leader sets dynamic energy storage leasing prices, and the followers optimize their own energy storage leasing and scheduling schemes. Finally, the optimal pricing strategy and charging / discharging scheduling scheme are obtained by solving the game model. This invention effectively improves the utilization efficiency of shared energy storage and the economic efficiency of system operation, promoting the local consumption of renewable energy.

Description

Technical Field

[0001] This invention patent relates to the field of energy storage optimization scheduling technology, specifically a multi-microgrid shared energy storage scheduling method that considers the uncertainty of electric vehicle charging and discharging. Background Technology

[0002] With the overexploitation of fossil fuels and the increasing deterioration of the natural environment, energy systems are accelerating their transformation towards distributed renewable energy utilization. Distributed renewable energy generation technologies, such as wind power and photovoltaic power, have been widely applied. However, renewable energy output is characterized by significant intermittency, volatility, and randomness, and its large-scale grid connection poses a severe challenge to the safe and stable operation of the power system, most notably the difficulty in integrating renewable energy.

[0003] Meanwhile, with the rapid development of the electric vehicle industry, large-scale integration of electric vehicles into the user side has become an important trend. As a flexible mobile load, the charging and discharging behavior of electric vehicles is highly uncertain, influenced by multiple factors such as user travel patterns, battery status, and electricity price incentives. This uncertainty further exacerbates the complexity and safety and stability risks of microgrid system scheduling: on the one hand, disorderly charging of electric vehicles may exacerbate the peak-valley load difference, leading to distribution network overload; on the other hand, if electric vehicles can be rationally guided to participate in bidirectional charging and discharging, they can be used as distributed energy storage resources to assist the system in peak shaving and valley filling.

[0004] In distribution network systems containing multiple microgrids, shared energy storage, as an emerging business model, can break the traditional independent configuration model of "one microgrid, one energy storage" by providing leasing services to multiple microgrids through centralized energy storage facilities, thereby reducing energy storage investment costs and improving energy storage utilization. However, existing shared energy storage dispatch methods still have the following shortcomings: (1) The uncertainty of electric vehicle charging and discharging behavior is not accurately characterized. Simplified probability models or deterministic scenarios are often used. There is a lack of refined simulation of individual travel characteristics and participation intentions, resulting in low accuracy of electric vehicle cluster charging and discharging load prediction, which makes it difficult to provide reliable input data for shared energy storage optimization scheduling.

[0005] (2) There are different interests between shared energy storage operators and microgrid operators. Existing methods mostly adopt centralized optimization or unilateral pricing mechanisms, which fail to effectively coordinate the game relationship between multiple entities. As a result, it is difficult to achieve equilibrium between the leasing pricing of shared energy storage and the leasing decision of microgrids, and the overall economic efficiency of the system needs to be improved.

[0006] (3) Existing methods rarely consider the potential of shared energy storage operators to participate in the "low storage and high release" peak arbitrage of the distribution network, and fail to fully explore the multiple sources of revenue for shared energy storage in addition to providing microgrid energy storage services.

[0007] Therefore, there is an urgent need to propose an optimized scheduling method that can accurately predict the uncertainty of electric vehicle charging and discharging and effectively coordinate the interest game between shared energy storage operators and multiple microgrids, so as to improve the economic efficiency of system operation and the capacity for renewable energy absorption. Summary of the Invention

[0008] This invention provides a multi-microgrid shared energy storage scheduling method that considers the uncertainty of electric vehicle charging and discharging. It aims to solve the dual challenges faced by microgrid operation under the background of high proportion of renewable energy access and large-scale electric vehicle development: first, the problem of renewable energy consumption caused by the inherent intermittency and volatility of wind and solar power generation; second, the uncertainty of electric vehicle charging and discharging behavior exacerbates the complexity of system scheduling and the risk to safety and stability.

[0009] The technical solution for implementing the present invention is as follows: A method for multi-microgrid shared energy storage scheduling that considers the uncertainties of electric vehicle charging and discharging includes the following steps: Step 1: Establish a shared energy storage system framework with microgrids. The framework includes centralized shared energy storage facilities invested and managed by shared energy storage operators, as well as multiple microgrid operators. The shared energy storage operators provide energy storage capacity leasing services to microgrid operators and provide charging and discharging services to electric vehicle clusters. Step 2: Considering the uncertainty of electric vehicle charging and discharging, by simulating the travel characteristic parameters of electric vehicles, including travel start and end time, travel distance, initial state of charge and user expectations for charging and discharging service prices, and combining the charging and discharging classification model based on the random forest algorithm, we determine the willingness and feasibility of each electric vehicle to participate in shared energy storage scheduling, and aggregate and generate the shared energy storage charging and discharging load curve of the electric vehicle cluster. Step 3: Construct a "one master, multiple followers" Stackelberg game optimization model with shared energy storage operators as the main body. In this model, the shared energy storage operator acts as the leader and formulates a dynamic energy storage leasing price strategy. Each microgrid operator acts as a follower and aims to minimize its own comprehensive electricity cost by optimizing the shared energy storage capacity leasing decision and internal resource scheduling scheme. Step 4: Solve the Stackelberg game optimization model to obtain the optimal leasing price strategy for the shared energy storage operator and the optimal charging and discharging scheduling scheme for each microgrid operator.

[0010] Furthermore, the specific process of generating the shared energy storage charging and discharging load curve of the electric vehicle cluster in step 2 is as follows: First, probabilistic modeling is performed on the arrival time, daily mileage, and initial state of charge of each electric vehicle; second, the charging and discharging behavior of electric vehicles is classified based on a random forest classification model, and the classification results include four states: vehicle idle, charging only, discharging only, and capable of participating in charging and discharging; then, the charging and discharging duration and power load of each electric vehicle are calculated; finally, the hourly time-sharing charging and discharging load curve is obtained through multiple iterative simulations and aggregations.

[0011] Furthermore, in step 3, the objective function of the shared energy storage operator aims to maximize the comprehensive operating revenue within the scheduling cycle. The comprehensive operating revenue includes energy storage service rental income, arbitrage income from participating in the "low storage, high release" peak shaving of the distribution network, and electric vehicle charging and discharging service income, minus operation and maintenance costs. Among them, energy storage service rental income includes unit power capacity rental fee and unit energy capacity rental fee.

[0012] Furthermore, the shared energy storage operator purchases and stores electricity during off-peak hours based on the time-of-use electricity price signal of the distribution network, and discharges it to the distribution network or microgrid during peak hours to realize arbitrage profits and participate in the system's peak-valley regulation.

[0013] Furthermore, in step 1, the microgrid prioritizes the use of local distributed energy to meet load demand, and injects the net surplus power into shared energy storage for storage; the shared energy storage facility, while meeting the microgrid's energy storage service and electric vehicle charging and discharging service, utilizes its surplus capacity to participate in the distribution network's peak shaving, thereby achieving "low storage and high discharge" arbitrage.

[0014] Furthermore, in step 3, the optimization objective of each microgrid operator is to minimize the comprehensive electricity cost within the scheduling cycle. The comprehensive electricity cost includes the charging and discharging fees paid to the shared energy storage operator, the electricity purchase fees from the distribution network, and the local controllable load scheduling costs. The optimization variables include the energy storage capacity, power capacity, and charging and discharging power leased by each microgrid.

[0015] Furthermore, in step 4, a genetic algorithm is used to solve the Stackelberg game optimization model: the shared energy storage operator, as the leader, uses the energy storage leasing price strategy as the decision variable and optimizes its expected total revenue through the genetic algorithm; each microgrid operator, as the follower, uses the charging and discharging power scheduling scheme as the decision variable and solves the problem of minimizing its own energy cost; the strategy equilibrium solution is obtained by iterating to Nash equilibrium.

[0016] Furthermore, the determination condition of the charging and discharging classification model based on the random forest algorithm in step 2 is as follows: the electric vehicle is determined to participate in the shared energy storage scheduling if and only if the state of charge when the vehicle arrives meets the discharge condition, the state of charge when the vehicle leaves meets the subsequent travel needs, the vehicle's parking time is sufficient to complete charging and discharging, and the user's expected compensation price is higher than the grid charging price.

[0017] Beneficial effects: 1. This invention uses a charging and discharging classification model based on the random forest algorithm, combined with electric vehicle travel behavior simulation (arrival time, mileage, initial state of charge, etc.), to accurately generate hourly charging and discharging load curves for electric vehicle clusters participating in shared energy storage. This effectively addresses the uncertainty of electric vehicle charging and discharging and provides highly reliable input data for optimized scheduling.

[0018] 2. The accurate electric vehicle charging and discharging prediction in this invention enables shared energy storage operators to plan energy storage charging and discharging schedules more rationally, avoid the idleness or abuse of energy storage resources, and thus improve the utilization rate and operational economy of shared energy storage facilities.

[0019] 3. Under the "one master and many slaves" Stackelberg game framework, the shared energy storage operator, as the leader, can maximize its own operating profit by formulating dynamic energy storage leasing pricing strategies and actively participating in the "low storage and high release" peak arbitrage of the distribution network, while also taking into account the revenue from electric vehicle charging and discharging services.

[0020] 4. In this invention, each microgrid acts as a follower, optimizing its own energy storage capacity leasing decisions and internal wind and solar resources and load scheduling schemes based on the leasing price signals released by the shared energy storage operator. It prioritizes the consumption of local renewable energy and reduces the purchase of electricity from the distribution network at high prices, thereby effectively reducing the overall electricity cost.

[0021] 5. Through the charging and discharging scheduling of shared energy storage, the fluctuation of the net load curve of each microgrid is significantly smoothed out; the surplus renewable energy power generated by the microgrid is preferentially stored in the shared energy storage and released when the load is at its peak or when the renewable energy output is insufficient, thereby improving the local consumption capacity of renewable energy and reducing the peak-shaving pressure of the distribution network. Attached Figure Description

[0022] Figure 1 This is a flowchart of the method of the present invention.

[0023] Figure 2 These are the load curves of the EV under different charging and discharging modes in the embodiments of the present invention.

[0024] Figure 3 This describes the charging and discharging of the EV in bidirectional charging mode in an embodiment of the present invention.

[0025] Figure 4 This is a comparison diagram of the net load of microgrid 1 in an embodiment of the present invention.

[0026] Figure 5 This is a comparison chart of the net load of microgrid 2 in an embodiment of the present invention.

[0027] Figure 6 This is a comparison chart of the net load of microgrid 3 in an embodiment of the present invention.

[0028] Figure 7 The figures show the net discharge power of the microgrid and the SOC curve of the shared energy storage in this embodiment of the invention. Detailed Implementation

[0029] This invention proposes an optimized scheduling method for shared energy storage in multi-microgrid systems, considering the uncertainties of electric vehicle charging and discharging. The core of this method lies in two aspects: First, it employs combined forecasting technology, deeply integrating simulation analysis of EV travel behavior to accurately characterize its cluster charging and discharging characteristics, generating highly reliable shared energy storage charging and discharging load curves, significantly improving the accuracy of charging and discharging power prediction, and laying the foundation for the efficient utilization of shared energy storage resources. Second, it innovatively constructs a "one-master, multiple-slave" Stackelberg game framework led by a shared energy storage operator (SSO). Under this framework, the SSO refines its charging and discharging plans based on dynamic time-of-use pricing strategies and actively participates in the "low storage, high discharge" peak-shaving service of the distribution network to maximize its own operating profits. Simultaneously, this strategy effectively guides the optimization of shared energy storage capacity configuration and collaboratively optimizes the scheduling decisions of various distributed energy sources and loads within the microgrid, effectively improving the system's operational economy and promoting the multiple objectives of local renewable energy consumption.

[0030] The present invention is as follows Figure 1 As shown, the specific steps include: 1. Establish a framework for a shared energy storage system (SESS) with microgrids. The shared energy storage collaborative scheduling strategy proposed in this invention is built upon a microgrid system. The research framework primarily comprises two core operating entities: the Microgrid Operator (MGO) and the Service Provider Operator (SSO). Its main operating model can be summarized as follows: the SSO invests in and manages centralized shared energy storage facilities. These facilities, serving as a key flexibility resource, primarily provide comprehensive energy storage services to the microgrid and offer professional charging and discharging services to the large-scale EV clusters within the system. The SSO charges corresponding energy storage service fees according to predetermined rules based on the actual electricity stored or extracted by users (MGO and EV users), thus forming a sustainable commercial operation closed loop.

[0031] Distributed energy resources in microgrids mainly include wind turbines (WT) and photovoltaic (PV) systems. In terms of operational strategy, MGs prioritize using local wind and solar power to meet their internal load demands; the resulting net surplus power (the portion of power generation exceeding load) is injected into the SESS (System-Enhanced Energy Storage) network for storage, thereby achieving system-level power balance regulation. SSOs aim to maximize operational efficiency by designing attractive service pricing mechanisms to provide flexible and efficient energy storage services to each microgrid, effectively incentivizing their active participation in energy time-shifting. Furthermore, large-scale EV clusters, as important distributed flexibility resources, can be deeply integrated into the SESS operational framework, assisting in energy buffering and regulation by providing controlled charging and discharging services. Simultaneously, the surplus capacity of the SESS itself can actively respond to time-of-use pricing signals from the distribution network, proactively participating in system peak-valley regulation (i.e., the "low-storage, high-discharge" strategy), assisting the grid in peak shaving and valley filling while realizing arbitrage profits based on electricity market prices, further improving overall economic efficiency.

[0032] 2. Shared energy storage optimization scheduling strategy considering EV uncertainty The prediction process for EV cluster charging and discharging behavior begins with a detailed simulation of individual EV travel characteristics. First, by comprehensively simulating key travel parameters (including start and end times, travel distance, initial state of charge (SOC), and user expectations regarding charging and discharging service prices), the initial state of each EV upon arrival at the charging / discharging station is accurately calculated. Second, combining the specific charging and discharging capacity parameters of the vehicles with a pre-trained classification model, the willingness and feasibility of each EV participating in SESS (Service-Oriented Service) collaborative scheduling are determined. Based on this determination: for EVs participating in scheduling, their feasible charging or discharging duration and corresponding power load are calculated; for EVs not participating in scheduling, only the time required to complete charging and the charging load need to be calculated. Finally, by repeating the above simulation-analysis-calculation cycle a sufficient number of times, the overall charging and discharging load curve of the local EV cluster can be generated, providing high-precision input for subsequent optimized scheduling.

[0033] By performing the above steps, the EV cluster's shared energy storage charging and discharging load curve is output. This prediction process can accurately calculate the charging and discharging load on an hourly basis, providing strong support for the energy storage dispatching of the regional power grid.

[0034] Working principle 2.1 Shared Energy Storage Optimization Scheduling Strategy Considering EV Uncertainty 2.1.1 EV Arrival Time In EV travel behavior analysis, vehicle arrival time is... The probability density function is; (1) In the formula: , These are the expected value and standard deviation of the sample data, respectively. The EV departure time and the dwell time are... .

[0035] (2) 2.1.2 Daily mileage of EVs The probability density function is: (3) Where: mean , These represent the expected value and standard deviation of the sample data, respectively.

[0036] 2.1.3 Initial State of Charge of EV The initial state of charge (SOC) probability density function of the EV is: (4) In the formula: , These represent the expected value and standard deviation of the sample data, respectively.

[0037] 2.1.4 Charge / discharge duration With EVs participating in the dispatch of shared energy storage, , , , These represent four different charging and discharging behaviors of an EV: vehicle idle, charging only, discharging only, and capable of participating in charging and discharging. The charging and discharging duration is related to the power. During the EV's participation in shared energy storage charging and discharging, the charging and discharging power is assumed to be... , The maximum power is 30kW.

[0038] Analyzing the charging and discharging behavior of vehicles, by , Calculate the first During the next battery cycle, the charging and discharging times of the vehicle are calculated, and the vehicle scheduling time is calculated using equation (5). .

[0039] (5) (6) (7) In the formula: , For charge and discharge efficiency; For the first The state of charge of the vehicle when it arrives during the next battery cycle; This represents the lowest acceptable state of charge for the user. For the first The state of charge of the vehicle when it leaves during a secondary battery cycle; C represents the category of vehicle charging and discharging behavior; C represents the vehicle battery capacity. 2.2 Charging and discharging classification model based on random forest algorithm Random Forest, an advanced ensemble learning algorithm, relies on its core mechanism of constructing and combining multiple decision trees to form a strong predictive model. This algorithm generates multiple subsets of differentiated samples by performing bootstrap sampling on the original training dataset, and then trains a large number of decision trees in parallel based on these subsets. Simultaneously, during the node splitting process of each tree, only a subset of features, rather than all features, are randomly selected for optimal splitting. This dual randomness design effectively reduces the sensitivity of a single decision tree to specific data noise or outliers, significantly suppressing the model's overfitting tendency. Random Forest not only improves the overall accuracy and robustness of predictions but also exhibits excellent generalization ability, making it an effective tool for processing high-dimensional and complex data.

[0040] To explore in depth the willingness of EV users to participate in SESS and their behavioral decision-making mechanism, this study adopted a two-stage data-driven approach. First, through a structured random questionnaire survey, the subjective participation intentions of users and their related background information were systematically collected. Second, multi-source actual operation data (including specific charging and discharging history records and real-time vehicle state of charge information) were integrated to conduct a refined analysis of the charging and discharging decision-making behavior of individual users. Finally, by constructing a feature dataset for the model, the charging and discharging behavior decision-making patterns of users were systematically classified based on this feature dataset. (1) If This indicates that the vehicle can participate in the discharge scheduling upon arrival. (2) If This indicates that the vehicle's state of charge when it leaves can meet the user's subsequent travel needs. (3) Assuming the vehicle participates in shared energy storage scheduling, calculate the following respectively. and ,like This indicates that the vehicle can complete charging and discharging within the parking time. (4) Charging and discharging compensation: Statistics show that users who intend to participate in shared energy storage dispatch expect a higher compensation price than the charging price. Therefore, the compensation price will be... Charging price on the grid side To make a comparison, if This indicates that the user is participating in the discharge process. The charging price is based on the peak-hour electricity price in a certain city. Take 1.2 yuan / (kW·h).

[0041] Therefore, if and only if , , , Vehicles must participate in shared energy storage scheduling when the vehicle is in use; otherwise, they will not participate.

[0042] This represents the state of charge of the vehicle upon arrival. This represents the state of charge of the vehicle when it leaves. This refers to the duration of vehicle downtime.

[0043] 2.2.1 EV Charge / Discharge Capacity Prediction By executing the EV charge / discharge prediction process, the system outputs the EV cluster's shared energy storage charge / discharge load curve. This prediction process can accurately calculate the charge / discharge load on an hourly basis, providing strong support for the energy storage dispatch of the regional power grid.

[0044] Based on the EV charge / discharge load curve, we obtain t The charging and discharging power of EVs participating in shared energy storage during specific time periods , .

[0045] 3. Establish an optimization model with shared energy storage operators as the main body. This invention constructs a Stackelberg game optimization framework dominated by Shared Energy Storage Operators (SSOs). In this framework, the shared energy storage operator acts as the leader, dynamically formulating its energy storage service leasing price strategy (which is subdivided into unit power capacity leasing fee and unit energy capacity leasing price), and transmitting price signals to its subordinate Management Optimizers (MGOs). In response, each microgrid acts as a follower, using the leasing price structure issued by the SSO as a key input parameter in its operational optimization model. The core objective is to minimize the expected overall electricity cost of its microgrid system, solving for the optimal shared energy storage capacity leasing decision and the coordinated scheduling scheme of its internal resources (such as wind and solar turbines and local loads).

[0046] 3.1 Shared Energy Storage Game Theory Optimization Scheduling Model The main body is SSO, and the slave bodies are various microgrids.

[0047] 3.1.1 Shared Energy Storage Objective Function SSO (Storage Storage Service) utilizes a day-ahead market mechanism to provide energy storage capacity leasing services, and its operational strategy is deeply integrated with electricity market dynamics. Specifically, SSO publishes energy storage leasing bids to microgrids during the day-ahead phase, while actively participating in system peak-valley regulation using the distribution network's time-of-use pricing mechanism: arbitrage is achieved by purchasing and storing electricity during off-peak hours and discharging it to the grid or microgrid during peak hours (i.e., the "low-storage, high-discharge" strategy). Within this framework, SSO's optimization objective is to maximize its overall operational revenue over a complete dispatch cycle, which comprises both energy storage service leasing revenue and revenue from participating in electricity market arbitrage. The specific objective function is as follows: : (8) (9) (10) (11) (12) In the formula: For energy storage leasing fees; To profit from peak shaving; To benefit from participating in EVs; For operation and maintenance costs; , The prices per unit of energy capacity and per unit of power capacity are respectively. , Shared energy storage t Constant charge / discharge levels; , These refer to the electricity purchased and sold between shared energy storage and the distribution network; , Each day t Time-of-use electricity purchase and sale prices on the distribution network; For the first i The charging and discharging fees paid by the individual micro-network; , For the respective i The energy storage capacity and power capacity of individual microgrid leases; Operation and maintenance cost per unit power of energy storage; , They are respectively t Electricity price for EV charging and discharging during specific time periods.

[0048] 3.1.2 Constraints The microgrid satisfies the following constraints: (13) (14) (15) (16) In the formula: , These represent the charging and discharging power of the microgrid. This represents the total power capacity.

[0049] The constraints for EVs participating in shared energy storage are: (17) (18) (19) Shared energy storage state of charge constraints are: (20) (twenty one) (twenty two) In the formula: For shared energy storage t State of charge at the end of time; This represents the initial state of charge of the energy storage. To share energy storage capacity.

[0050] Shared energy storage power balance constraints: (twenty three) Maximum tie-line power is The constraints for purchasing electricity for shared energy storage are: (twenty four) (25) 4. Solution and Steps of the Model 4.1 Solving the Master-Slave Game Optimization Model The master-slave game model can be established as follows: (26) The leader's strategy is to adjust the rental price specifically as follows: Followers adjust energy storage leasing capacity specifically as follows: .

[0051] , This indicates the leased energy storage capacity and power capacity of each microgrid. , This indicates the charging and discharging power of each microgrid.

[0052] The objective functions of the participants are respectively expressed as: and When the game reaches Nash equilibrium, neither side can gain more profit simply by changing their strategies. This is the strategy equilibrium solution.

[0053] This invention applies a genetic algorithm to solve the master-slave game optimization model between the constructed SSO and MGO. In this model, the SSO, as the leader, aims to maximize the expected total revenue within the scheduling cycle; the optimal day-ahead energy storage leasing pricing strategy can be obtained through the genetic algorithm. For each microgrid, as a follower, the optimization objective is to minimize its own comprehensive energy cost within the same cycle; the solution process will output the optimal charging and discharging power scheduling scheme for it. The simulation of the above game model uses the prediction scenario set generated by the random forest algorithm proposed in the patent as the input scenario data, ensuring the authenticity of the model input. The time-of-use electricity purchase and sale price parameters of the distribution network involved in the model are detailed in Table 1.

[0054] Table 1 Time-of-use electricity pricing

[0055] Detailed analysis process: The random forest algorithm is used to predict the charging and discharging power of EVs. A classification model is established using historical charging data and survey data on users' willingness to participate in shared energy storage interactions as training sets to predict the charging or discharging behavior of EVs at future moments. The predicted cases involve three charging and discharging modes, including ordered charging, disordered charging, and participation in shared energy storage charging and discharging. 341 EVs responded to the shared energy storage scheduling, with 22 EVs idle, 196 EVs charging only, 123 EVs charging and discharging, and 0 EVs discharging only.

[0056] Load curves under different charge and discharge modes are as follows Figure 2 As shown.

[0057] Disorderly charging can lead to peak load overlap between 08:00 and 11:00. In contrast, the one-way orderly charging mode can shift the charging load to 12:00-17:00 through reasonable scheduling. This is because the grid load is relatively large between 09:00 and 11:00, so we try to avoid peak electricity consumption periods and choose to charge during periods with lower electricity prices.

[0058] In bidirectional charging, EV-assisted shared energy storage scheduling guides the load to charge and discharge reasonably during the 07:00-11:00 period, reducing load congestion during peak hours and smoothing the electricity load curve. During the 12:00-17:00 period, the load participating in shared energy storage scheduling is higher than the load participating in unidirectional charging. This is because the grid load is lower and the electricity price is lower during this period. The electricity price for participating in shared energy storage is lower than the grid electricity price, and there is more energy interaction between EVs and shared energy storage.

[0059] The charging and discharging status of EVs participating in the shared energy storage dispatch mode is as follows: Figure 3 As shown.

[0060] Between 8:00 and 11:00, the area experiences high electricity consumption, and EVs discharge to the shared energy storage. At 9:00, the discharge power reaches its maximum, approximately 1600kW, at which time the number of EVs is the highest. Subsequently, between 12:00 and 15:00, the number of vehicles charging increases rapidly, and around 14:00, the EV charging power reaches its maximum of 1900kW, at which time 142 vehicles are charging.

[0061] The scheduling strategy of the microgrid was analyzed, and the results are as follows: Figures 4 to 6 As shown.

[0062] Analyze microgrid 1. Figure 4 It includes the charging and discharging power, net load curves of each microgrid, and the net load curve of the microgrid after participating in the shared energy storage charging and discharging.

[0063] Microgrid 1 effectively mitigated its net load fluctuations by integrating SESS (Search and Execution System). The specific analysis process is as follows: During the peak daytime PV output period of 10:00-16:00, the net load is positive because renewable energy generation significantly exceeds local load demand. At this time, the microgrid prioritizes using this surplus power to charge the shared energy storage, resulting in a positive net load curve after discharge at its grid connection point. It is worth noting that around 15:00, with the natural decay of PV output and continuous charging, the net load value after discharge begins to show a downward trend. Entering the evening peak load period of 18:00-21:00, both the distribution network and shared energy storage are in the high-price range of time-of-use electricity pricing. To cope with this cost pressure, Microgrid 1 moderately reduces controllable load through optimized scheduling to lower overall electricity costs. During the off-peak nighttime period of 0:00-7:00, with local renewable energy output essentially disappearing, Microgrid 1 mainly relies on the low-priced electricity from the distribution network to meet its basic load demand.

[0064] The net load curve of microgrid 2 is as follows Figure 5 As shown.

[0065] analyze Figure 5 It is evident that the net load curve of Microgrid 2 exhibits a wide range of variation, primarily concentrated in three time periods: 08:00-10:00, 11:00-13:00, and 19:00-21:00. During these periods, the microgrid utilizes shared energy storage to purchase or sell electricity to meet its own power balance, thus reducing the curve's fluctuation range. In the periods of 14:00-18:00 and 22:00-24:00, the load fluctuation range is smaller, and the microgrid chooses to reduce its charging and discharging power, which to some extent reduces the energy storage capacity. Microgrid 2 prioritizes the absorption of new energy sources and utilizes shared energy storage to smooth out load fluctuations.

[0066] The net load curve of microgrid 3 is as follows Figure 6 As shown.

[0067] analyze Figure 6As can be seen from the charge and discharge curves of Microgrid 3, there are differences in the absorption of loads and renewable energy sources such as wind and solar power within the microgrid. This will increase the differences in the smoothing effect of shared energy storage to some extent. The net load curve smoothing in Microgrid 3 is more obvious. During periods of higher distribution network electricity prices (09:00-11:00 and 19:00-21:00), load fluctuations can be reduced through the scheduling of shared energy storage.

[0068] Figure 7 The net discharge power and shared energy storage SOC curves for the microgrid. Figure 7 Analysis of the energy storage state-of-charge curves reveals the typical intraday charging and discharging modes and driving factors of SESS under microgrid collaborative operation. Figure 7 It can be seen that during the peak photovoltaic output period from 11:00 to 14:00, the SOC value of shared energy storage shows a rapid downward trend. This is mainly due to a dual effect: on the one hand, SESS responds to high-price signals and sells electricity to the distribution network for arbitrage; on the other hand, large-scale EV clusters are centrally connected and use SESS for charging. Subsequently, during the period from 14:00 to 18:00, a large amount of surplus renewable energy power (especially photovoltaic) in the microgrid is injected into the shared energy storage for storage, driving the SOC to rebound. Entering the evening peak period from 19:00 to 22:00, facing the decline in renewable energy output in the microgrid, high local load, and peak distribution network electricity price, the shared energy storage prioritizes discharge scheduling, which not only meets the critical load demand of the microgrid but also sells electricity to the high-price distribution network to maximize operating revenue. During the off-peak period from 22:00 to 5:00 at night, the shared energy storage makes full use of the low-price electricity and fewer renewable energy sources in the distribution network for charging, significantly reducing its energy replenishment cost.

[0069] It is worth noting that the State of Charge (SOC) reached its daily peak around 11:00 AM, thanks to the combined effect of microgrid photovoltaic power input and EV cluster discharge. Around midnight, the SOC returned to near its initial level, marking the completion of a full intraday charge-discharge cycle. Overall, the shared energy storage charging and discharging behavior is highly coordinated with microgrid energy balance: its core function is to absorb microgrid power surplus, compensate for load deficits, and provide EV charging and discharging services; simultaneously, its operating strategy is deeply coupled with electricity price signals, achieving arbitrage through "low storage, high discharge" to optimize overall economic efficiency.

[0070] In summary, the above are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A multi-microgrid shared energy storage scheduling method considering the uncertainty of electric vehicle charging and discharging, characterized in that, Includes the following steps: Step 1: Establish a shared energy storage system framework with microgrids. The framework includes centralized shared energy storage facilities invested and managed by shared energy storage operators, as well as multiple microgrid operators. The shared energy storage operators provide energy storage capacity leasing services to microgrid operators and provide charging and discharging services to electric vehicle clusters. Step 2: Considering the uncertainty of electric vehicle charging and discharging, by simulating the travel characteristic parameters of electric vehicles, including travel start and end time, travel distance, initial state of charge and user expectations for charging and discharging service prices, and combining the charging and discharging classification model based on the random forest algorithm, we determine the willingness and feasibility of each electric vehicle to participate in shared energy storage scheduling, and aggregate and generate the shared energy storage charging and discharging load curve of the electric vehicle cluster. Step 3: Construct a Stackelberg game optimization model with shared energy storage operators as the main body. In this model, the shared energy storage operators act as leaders and formulate dynamic energy storage leasing pricing strategies. Each microgrid operator acts as a follower and aims to minimize its own comprehensive electricity costs by optimizing the shared energy storage capacity leasing decision and internal resource scheduling scheme. Step 4: Solve the Stackelberg game optimization model to obtain the optimal leasing price strategy for the shared energy storage operator and the optimal charging and discharging scheduling scheme for each microgrid operator.

2. The method according to claim 1, characterized in that, The specific process for generating the shared energy storage charging and discharging load curve of the electric vehicle cluster in step 2 is as follows: First, probabilistic modeling is performed on the arrival time, daily mileage, and initial state of charge of each electric vehicle; second, the charging and discharging behavior of the electric vehicles is classified based on the random forest classification model, and the classification results include four states: vehicle idle, charging only, discharging only, and capable of participating in charging and discharging; then, the charging and discharging duration and power load of each electric vehicle are calculated; finally, the hourly time-sharing charging and discharging load curve is obtained through multiple iterative simulations and aggregations.

3. The method according to claim 1, characterized in that, In step 3, the objective function of the shared energy storage operator is to maximize the comprehensive operating revenue within the scheduling cycle. The comprehensive operating revenue includes energy storage service rental income, arbitrage income from participating in the "low storage, high release" peak shaving of the distribution network, and electric vehicle charging and discharging service income, minus operation and maintenance costs. Among them, energy storage service rental income includes unit power capacity rental fee and unit energy capacity rental fee.

4. The method according to claim 3, characterized in that, The shared energy storage operator purchases and stores electricity during off-peak hours based on the time-of-use electricity price signal of the distribution network, and discharges it to the distribution network or microgrid during peak hours to realize arbitrage profits and participate in the system's peak-valley regulation.

5. The method according to claim 1, characterized in that, In step 1, the microgrid prioritizes the use of local distributed energy to meet load demand, and injects the net surplus power into shared energy storage for storage. In addition to meeting the microgrid's energy storage service and electric vehicle charging and discharging service, the shared energy storage facility uses its surplus capacity to participate in the distribution network's peak shaving, thereby achieving "low storage and high discharge" arbitrage.

6. The method according to claim 1, characterized in that, In step 3, the optimization objective of each microgrid operator is to minimize the comprehensive electricity cost within the scheduling cycle. The comprehensive electricity cost includes the charging and discharging fees paid to the shared energy storage operator, the electricity purchase fees from the distribution network, and the local controllable load scheduling costs. The optimization variables include the energy storage capacity, power capacity, and charging and discharging power leased by each microgrid.

7. The method according to claim 1, characterized in that, In step 4, a genetic algorithm is used to solve the Stackelberg game optimization model: the shared energy storage operator, as the leader, uses the energy storage leasing price strategy as the decision variable and optimizes its expected total revenue through the genetic algorithm; each microgrid operator, as the follower, uses the charging and discharging power scheduling scheme as the decision variable and solves the problem of minimizing its own energy cost; the strategy equilibrium solution is obtained by iterating to Nash equilibrium.

8. The method according to claim 1 or 2, characterized in that, The determination criteria for the charge-discharge classification model based on the random forest algorithm in step 2 are as follows: the electric vehicle is determined to participate in the shared energy storage scheduling if and only if the state of charge when the vehicle arrives meets the discharge conditions, the state of charge when the vehicle leaves meets the subsequent travel needs, the vehicle's parking time is sufficient to complete the charge-discharge process, and the user's expected compensation price is higher than the grid charging price.

Images