Optical storage charging station electric vehicle ordered charging control method based on bilevel programming

Abstract

The invention discloses a light storage charging station electric vehicle ordered charging control method based on bilevel planning, and the method comprises the steps: firstly constructing a bilevel planning model of a light storage charging station, and enabling the upper layer planning model to be used for optimizing the overall energy scheduling of the charging station; the lower-layer planning model performs optimization scheduling on the charging behavior of the electric vehicle in the charging process according to the energy distribution result of the upper-layer planning model; and then, iteratively solving the bilayer planning model by adopting a dynamic alternating direction multiplier method, taking the output of the upper layer planning model as the input constraint of the lower layer planning model, and feeding back the result to the upper layer planning model by the lower layer planning model until the target function value of the upper layer planning and the lower layer planning converges or reaches a preset number of iterations, and the orderly charging power of the electric vehicle is obtained, and a control strategy is dynamically adjusted in case of operation deviation. According to the method, through collaborative optimization of upper-layer planning and lower-layer planning, the overall operation benefit of the power system is optimized, and a flexible charging strategy is provided for an electric vehicle user.

Description

Technical Field

[0001] The present invention belongs to the field of electric vehicle charging control, and in particular relates to a method for controlling electric vehicle orderly charging at a photovoltaic storage charging station based on double-layer planning. Background Art

[0002] As global attention to energy transition and environmental issues continues to grow, the use of renewable energy sources such as photovoltaics and wind power has become an integral part of the power system. At the same time, the widespread adoption of electric vehicles (EVs) has also driven enormous electricity demand, particularly in terms of charging, posing new challenges to power grids and related infrastructure. To effectively address this issue, integrated photovoltaic, energy storage, and charging power stations have emerged as a key solution. By combining photovoltaic power generation and energy storage systems with EV charging infrastructure, these power stations can meet EV charging needs while improving energy efficiency, reducing dependence on traditional power grids, and promoting the widespread use of clean energy.

[0003] With the rapid development of electric vehicles, the demand for charging is growing, especially in green energy charging facilities such as photovoltaic and energy storage charging stations. Efficiently and coordinately charging large numbers of electric vehicles has become a pressing issue. Traditional charging methods often ignore grid load, the volatility of photovoltaic power generation, the capacity limitations of energy storage devices, and the personalized charging needs of electric vehicle users (such as charging time and rate). This leads to inefficient charging station operation, poor energy utilization, and even grid congestion and battery over-discharge. While some existing charging strategies can schedule and manage electric vehicles, they often use a single planning model and fail to fully consider complex factors such as photovoltaic power generation at different time periods, the charge and discharge capacity of energy storage systems, grid load, and the charging priority of electric vehicles. This results in insufficient charging system scheduling and resource utilization, and an inability to simultaneously balance economic efficiency, efficiency, and system stability. Therefore, a new charging control strategy is urgently needed to improve energy scheduling efficiency during the charging process, achieve grid load balancing and battery health management, and dynamically adjust charging strategies to optimize overall economic benefits and user satisfaction. This new strategy requires effective coordination between photovoltaic power generation (with its intermittent and unstable nature), energy storage, and electric vehicle charging demand to ensure stable system operation and address the challenges posed by the uncertainties of charging station load, photovoltaic power generation, and energy storage capacity. Reasonable charging scheduling is crucial for reducing grid load fluctuations, ensuring stable grid operation, promoting the adoption of electric vehicles and renewable energy, and achieving carbon neutrality. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for orderly charging control of electric vehicles in a photovoltaic storage charging station based on double-layer planning, so as to solve the problems of high operating costs, low energy utilization efficiency and unreasonable charging process in the existing technology. Specifically, through the coordinated optimization of upper-level planning and lower-level planning, the overall operating efficiency of the power system is optimized, and flexible charging strategies are provided for electric vehicle users.

[0005] In order to achieve the above object, the solution of the present invention is:

[0006] A method for controlling the orderly charging of electric vehicles at a photovoltaic charging station based on a double-layer planning includes the following steps:

[0007] Step 1: Construct a two-layer planning model for the solar-storage charging station, where the upper-layer planning model is the system optimization layer and the lower-layer planning model is the charging scheduling layer;

[0008] The upper-level planning model is used to optimize the overall energy scheduling of the charging station, including the power distribution of photovoltaic power generation, energy storage equipment and the power grid; the lower-level planning model optimizes the charging behavior of the electric vehicles during the charging process based on the energy allocation results of the upper-level planning model;

[0009] Step 2: The dynamic alternating direction multiplier method is used to iteratively solve the two-level planning model. The output of the upper-level planning model is used as the input constraint of the lower-level planning model. The lower-level planning model then feeds the results back to the upper-level planning model until the objective function values ​​of the upper and lower planning converge or the preset number of iterations is reached. The optimal operation strategy of the solar-powered charging station in each time period, that is, the orderly charging power of the electric vehicle, is obtained.

[0010] Step 3: monitor photovoltaic output, energy storage status, grid parameters and electric vehicle charging requirements in real time, optimize the two-level programming model when an operation deviation is detected, and dynamically adjust the control strategy.

[0011] In step 1 above, the optimization goal of the upper-level planning model is to maximize the economic benefits of the power plant and improve the absorption capacity of renewable energy, while ensuring the load balance and sustainable operation of the power grid. Its objective function is:

[0012]

[0013] Among them, E station is the total revenue of the charging station, P pv (t) is the photovoltaic power generation power, P ess (t) is the charging and discharging power of the energy storage system, λ pv and λ ess are all power conversion factors; R sell (t) is the income from selling electricity to the grid when the surplus photovoltaic power is connected to the grid; R ev(t) the revenue from providing charging services to electric vehicles at charging stations; ess (t) is the cost of the energy storage system, including depreciation cost and maintenance cost; C pv The maintenance cost C of the photovoltaic power generation system pv =α pv P pv,rated , where α pv is the maintenance factor per unit power of the photovoltaic power generation system, P pv,rated is the rated power of the photovoltaic power generation system; C grid (t) is the cost of the power station purchasing electricity from the grid at time t, and is related to the electricity price p grid proportional to;

[0014] The upper-level decision variables include the charging power of the energy storage system at time t Discharge power of the energy storage system at time t Power P of the interaction between the power station and the grid grid (t), photovoltaic power generation power P pv (t);

[0015] The constraints include grid load balance constraints, energy storage system status constraints, photovoltaic power generation constraints, and grid power constraints.

[0016] The above grid load balance constraint is:

[0017] P grid (t)+P pv (t)+P ess (t)-P ev (t) = 0

[0018] Among them, P ev (t) is the load demand of the power station, i.e., the required charging power of the tram;

[0019] The state constraint of the energy storage system is:

[0020]

[0021] E ess (t) = E ess (t-1)+η ess P ess (t)Δt

[0022] Among them, E ess (t) is the energy storage capacity at time t, η ess is the charge and discharge efficiency of the energy storage system, Δt is the time step; and are the minimum and maximum charge and discharge power of the energy storage system, respectively, and E ess (t) is the state of charge of the energy storage battery;

[0023] The photovoltaic power generation constraint is that the power of photovoltaic power generation cannot exceed the maximum power of photovoltaic power generation.

[0024] 0≤P pv (t)≤P pv,max (t)

[0025] Among them, P pv,max (t) is the maximum photovoltaic power generation power predicted based on real-time weather forecast and sunlight intensity;

[0026] The grid power constraint is that the grid power cannot exceed the maximum carrying capacity of the grid.

[0027]

[0028] in, is the maximum carrying power of the power grid.

[0029] In step 1 above, the objective function of the lower-level planning model is,

[0030]

[0031] Among them, P target,i (t) is the target charging power of the ith trolley at time t, Price(t) is the charging price; V charge,k Score the convenience of charging; S i is the total cost of the i-th electric vehicle during the charging process; t i,start is the time when the i-th tram starts charging, t i,end is the time when the charging of the i-th electric vehicle ends; w1, w2, w3, w4 are the weight coefficients of each factor, satisfying w1+w2+w3+w4=1;

[0032] The lower-level decision variables include the charging power P of the i-th tram at time t ev,i (t), charging start time t of the i-th electric vehicle i,start , the charging end time t of the i-th electric car i,end ;

[0033] The constraints include charging time constraints, charging power constraints, electric vehicle charging demand constraints, and electric vehicle charging sequence constraints.

[0034] The above charging time constraint is,

[0035] t i,start ≤t≤t i,end

[0036] The charging power constraint is,

[0037]

[0038] in, and are the minimum and maximum charging powers of the i-th tram respectively;

[0039] The electric vehicle charging demand constraint is that the charging time of the electric vehicle must meet its charging demand, that is, each electric vehicle needs to be charged with a certain amount of electricity to meet its maximum charging capacity;

[0040] The charging order constraint of electric vehicles is that the charging order of each electric vehicle in the charging station is scheduled according to its demand and priority.

[0041] In step 2 above, the upper-level planning model is first solved to obtain the optimal power allocation scheme. The lower-level planning model optimizes the charging power and charging time for each electric vehicle based on the optimal power allocation scheme, obtains the planning results, and feeds them back to the upper-level planning model. Through iterative optimization, until the objective function values ​​of the upper and lower-level planning converge or the preset number of iterations is reached, the optimal operation strategy of the solar-storage charging station in each time period is finally obtained, that is, the orderly charging power of the electric vehicles.

[0042] The convergence conditions are set as:

[0043] and

[0044] in, and are the total benefits of the upper-level planning at the kth and k-1th iterations, respectively. and are the overall charging costs of the i-th tram at the k-th and k-1-th iterations, respectively, and ε1 and ε2 are the preset convergence thresholds.

[0045] In the above step 2, the dynamic alternating direction multiplier method is used to iteratively solve the bi-level programming model. The specific process is:

[0046] Step 21, set the initial value, including the upper decision variable x (0) and the lower-level decision variable y (0) , initialize the Lagrange multiplier λ (0) and penalty factor ρ, set convergence accuracy ε and maximum number of iterations T max ;

[0047] Step 22: Given the lower-level decision variables and Lagrange multipliers, solve the upper-level planning:

[0048]

[0049] Among them, g(x) and h(y) are constraint functions, representing the coordination constraints between the upper and lower layers;

[0050] At each iteration, the weight of the Lagrange multiplier λ is dynamically adjusted;

[0051] Step 23: Given the upper-level decision variables and Lagrange multipliers, solve the lower-level optimization problem:

[0052]

[0053] Step 24, update the Lagrange multiplier λ (k+1) ,

[0054] λ (k+1) =λ (k) +ρ(g(x (k+1) )-h(y (k+1) ))

[0055] Dynamically adjust the penalty factor ρ. If the constraint error is large, increase ρ; if the error is small, decrease ρ.

[0056] Step 25: For each iteration, calculate the error and determine whether the convergence condition is met. If the change of the objective function is less than the predetermined tolerance ε, the objective function is considered to have converged.

[0057] If the convergence condition is met, the iteration is stopped and the optimal solution is returned; otherwise, step 22 is continued.

[0058] After adopting the above scheme, the present invention has the following advantages:

[0059] (1) The present invention successfully balances the interests of the operators of photovoltaic charging stations and electric vehicle users through a two-layer planning architecture, effectively reducing operating costs and improving resource utilization efficiency, while reducing charging costs for electric vehicle users and improving their economic efficiency.

[0060] (2) The model fully considers the physical constraints in the photovoltaic storage charging station to ensure safe and reliable operation, avoids overcharging and over-discharging of the energy storage system, maximizes the utilization rate of photovoltaic power generation, and reduces dependence on the power grid.

[0061] (3) Through the dynamic alternation algorithm, an efficient and accurate optimization solution is provided, which can quickly converge and solve complex optimization problems. It has good adaptability and scalability and is suitable for photovoltaic charging stations of different sizes, promoting the intelligent management and sustainable development of charging stations.

[0062] (4) This invention not only optimizes energy scheduling and the electric vehicle charging process, reducing grid load shocks, but also improves battery health management, extending battery life, while ensuring fair distribution of charging resources and system stability. Furthermore, by prioritizing the use of renewable energy, carbon emissions are effectively reduced, providing strong support for the promotion of green energy and the sustainable development of the electric vehicle industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 It is a schematic diagram of the iterative optimization process of the two-level programming model;

[0064] Figure 2 It is the flow chart of dynamic alternating optimization algorithm;

[0065] Figure 3 It is the flow chart of tram charging scheduling;

[0066] Figure 4 It is a real-time monitoring system architecture and data flow diagram. DETAILED DESCRIPTION

[0067] The technical solutions and beneficial effects of the present invention will be described in detail below with reference to the accompanying drawings.

[0068] The present invention provides a method for controlling the orderly charging of electric vehicles in a photovoltaic charging station based on double-layer planning. The photovoltaic charging station used in the method mainly includes a distributed photovoltaic power generation module, an energy storage system module, an electric vehicle charging management module, a grid interface module, etc. Among them, the distributed photovoltaic power generation module is used to convert solar energy into electrical energy. As the main source of clean energy, it converts solar energy into electrical energy. This part of the electrical energy can be directly supplied to local loads, including the charging needs of electric vehicles. The excess electrical energy can be fed into the energy storage system for storage or directly uploaded to the grid interface module; the energy storage system module is used to provide energy storage and scheduling functions, and the electrical energy can be controlled by charging and discharging. The station's energy balance plays a key role in energy buffering and scheduling within the system. It absorbs excess photovoltaic power and releases it when photovoltaic generation is insufficient or when the grid load is peaking, maintaining stable system operation and optimizing energy utilization. The electric vehicle charging management module is responsible for controlling the charging of the numerous electric vehicles connected to the charging station. Its charging power and schedule are regulated by the instructions of the underlying planning model. The grid interface module serves as the system's supplementary energy source and energy output channel, exchanging energy with the external grid. It draws power from the grid when local power generation and storage are insufficient, or sells excess power back to the grid to ensure load balance at the power station. Photovoltaic power generation adopts a self-generated, self-used, and grid-connected model.

[0069] The control method of the present invention mainly includes the following steps:

[0070] Step 1: Construct a two-layer planning model for the solar-storage charging station;

[0071] The two-layer planning model proposed in this invention includes two main levels: the upper planning model is the system optimization layer, and the lower planning model is the charging scheduling layer.

[0072] The upper-level planning model is used to optimize the overall energy scheduling of the charging station, including the power distribution of photovoltaic power generation, energy storage equipment, and the power grid. The main objectives of this layer are to maximize the economic benefits and energy utilization efficiency of the system while meeting the charging needs of electric vehicles and the load constraints of the power grid. The overall economic benefits take into account factors such as the depreciation cost of the energy storage system, the grid's electricity purchase costs, and the revenue from photovoltaic power generation. The decision variables of the upper-level planning include the charging and discharging power settings of the energy storage system, the grid interaction power, and other adjustable system parameters. The optimization goal of this model is to maximize the economic benefits of the power station and improve the renewable energy absorption capacity, while ensuring the load balance and sustainable operation of the power grid.

[0073] The lower-level planning model optimizes the charging behavior of each vehicle during the charging process based on the energy allocation results of the upper-level planning model. While satisfying the power constraints for interaction between the solar-powered charging station and the grid, as well as the energy storage system's charge and discharge power constraints, the lower-level planning model optimizes the allocation of charging power and time for each vehicle, taking into account constraints such as vehicle charging demand, vehicle battery characteristics, and user arrival and departure times. This includes allocating charging power to each vehicle, controlling charging duration, and prioritizing each vehicle to ensure an orderly charging process and avoid overloading and battery loss. Based on the real-time status of the charging stations, the model uses an ordered charging algorithm to prioritize and schedule the charging needs of each vehicle, preventing grid overload and ensuring a reasonable allocation of charging time and energy for each vehicle. By considering time-of-use pricing strategies and vehicle charging characteristics, the primary decision variables for the lower-level planning are the charging start time and charging power for each vehicle.

[0074] The results of the upper-level planning are used as inputs to the lower-level planning, which then feeds back into the upper-level planning. Through continuous iterative optimization, the objective function values ​​of the upper and lower levels converge or the preset number of iterations is reached. Ultimately, the optimal operating strategy for the solar-powered charging station in each time period is obtained.

[0075] The objective function of upper-level planning is:

[0076]

[0077] Among them, E station is the total revenue of the charging station, P pv (t) is the photovoltaic power generation power, P ess (t) is the charging and discharging power of the energy storage system, λ pv and λ ess are all power conversion factors; R sell (t) is the income from selling electricity to the grid when the surplus photovoltaic power is connected to the grid; R ev (t) the revenue from providing charging services to electric vehicles at charging stations; ess(t) is the cost of the energy storage system, including depreciation cost and maintenance cost; C pv The maintenance cost C of the photovoltaic power generation system pv =α pv P pv,rated , where α pv is the maintenance factor per unit power of the photovoltaic power generation system, P pv,rated is the rated power of the photovoltaic power generation system; C grid (t) is the cost of the power station purchasing electricity from the grid at time t, and is related to the electricity price p grid Directly proportional.

[0078] C ess The expression of (t) is:

[0079] C ess =α ess P ess,rated +β ess E ess,rated

[0080] Among them, α ess and β ess are the depreciation and maintenance coefficients of the energy storage system per unit power and per unit capacity, P ess,rated and E ess,rated are the rated power and rated capacity of the energy storage system respectively.

[0081] The upper-level decision variables are:

[0082] Charging power of the energy storage system at time t.

[0083] The discharge power of the energy storage system at time t.

[0084] P grid (t): The power exchanged between the power station and the grid. A positive value indicates that the power station purchases electricity from the grid, while a negative value indicates that the power station sells electricity to the grid.

[0085] P pv (t) is the photovoltaic power generation power.

[0086] The constraints are:

[0087] Grid load balancing constraints:

[0088] P grid (t)+P pv (t)+P ess (t)-P ev (t) = 0

[0089] Among them, P ev (t) is the load demand of the power station, that is, the required charging power of the tram.

[0090] State constraints of energy storage systems:

[0091]

[0092] E ess (t) = E ess (t-1)+η ess P ess (t)Δt

[0093] Among them, E ess (t) is the energy storage capacity at time t, η ess is the charge and discharge efficiency of the energy storage system, Δt is the time step; and are the minimum and maximum charge and discharge power of the energy storage system, respectively, and E ess (t) is the state of charge of the energy storage battery.

[0094] Photovoltaic power generation constraints:

[0095] The power of photovoltaic power generation cannot exceed the maximum power of photovoltaic power generation:

[0096] 0≤P pv (t)≤P pv,max (t)

[0097] Among them, P pv,max (t) is the maximum photovoltaic power generation power predicted based on real-time weather forecast and sunlight intensity.

[0098] Grid power constraints:

[0099] The power drawn from the grid cannot exceed the grid's maximum carrying capacity:

[0100]

[0101] in, is the maximum carrying power of the power grid.

[0102] By solving the above objective function and constraints, the optimal power allocation solution P can be obtained. pv (t), and P ess (t), thus providing input to the lower charging scheduling layer.

[0103] The goal of the lower-level planning is to provide an optimal charging strategy for each tram, optimizing the cost or charging time of the tram. By considering the time-of-use electricity price policy and the charging needs of the tram, the lower-level planning aims to determine the optimal charging period and charging power for each tram.

[0104] The objective function of the lower-level planning is:

[0105]

[0106] Among them, P target,i (t) is the target charging power of the ith trolley at time t, Price(t) is the charging price; V charge,k Score the convenience of charging, calculated based on geographic location and the availability of charging equipment.

[0107] S i is the total cost of the i-th electric vehicle during the charging process;

[0108] t i,start is the time when the i-th tram starts charging, t i,end End charging time for the i-th tram;

[0109] w1, w2, w3, w4 are the weight coefficients of each factor, satisfying w1+w2+w3+w4=1.

[0110] The lower-level decision variables include:

[0111] P ev,i (t): Charging power of the i-th tram at time t.

[0112] t i,start : Charging start time of the i-th tram.

[0113] t i,end : Charging completion time of the i-th tram.

[0114] Constraints:

[0115] Charging time constraint: t i,start ≤t≤t i,end

[0116] Charging power constraints:

[0117] in, and are the minimum and maximum charging powers of the i-th tram respectively.

[0118] Tram charging demand constraint: The charging time of trams must meet their charging demand, that is, each tram needs to be charged with a certain amount of electricity to meet its maximum charging capacity:

[0119]

[0120] Electric vehicle charging sequence constraints:

[0121] The charging order of each electric vehicle in the charging station should be scheduled according to its demand and priority. The charging order of each vehicle can be determined by:

[0122]

[0123] The electric vehicle charging strategy will be dispatched according to electricity prices, avoiding peak electricity price periods to reduce charging costs.

[0124] The calculation formula of charging convenience is as follows: Charging convenience V charge,i It can be expressed in the form of a weighted sum, taking into account the following influencing factors:

[0125] V charge,i =w v1 ·D distance,i +w v2 ·T wait,i +w v3 ·T charge,i +w v4 ·A availability,i

[0126] D distance,i is the distance indicator from tram i to the nearest charging station.

[0127] T wait,i is the waiting time indicator of tram i, that is, the time the tram spends queuing at the charging station.

[0128] T charge,i It is the charging time indicator of tram i, which is related to the charging power, the charging demand of the tram battery and the charging rate of the charging facility.

[0129] A availability,i It is an availability scoring indicator for charging facilities, taking into account the idle rate of charging piles, equipment maintenance, etc.

[0130] w v1 ,w v2 ,w v3 ,w v4 is the weight coefficient of each factor, satisfying w v1 +w v2 +w v3 +w v4 =1.

[0131] The distance between the tram and the charging station directly affects the convenience of charging. The shorter the distance, the more convenient charging. Convenience can be expressed as the inverse of the distance:

[0132]

[0133] in:

[0134] d i is the distance from tram i to the nearest charging station.

[0135] α is the coefficient of influence of distance on convenience, which is a positive constant.

[0136] The longer the waiting time, the lower the convenience. This can be represented by the inverse of the waiting time:

[0137]

[0138] in:

[0139] t wait,i is the waiting time of tram i (in minutes).

[0140] β is the influence coefficient of waiting time on convenience, which is a positive constant.

[0141] The shorter the charging time, the higher the convenience. In order to simplify the calculation, the charging time T is usually charge,i Convert to an inverse value and multiply by a weight coefficient:

[0142]

[0143] in:

[0144] t charge,i is the time required for electric vehicle i to complete charging.

[0145] γ is the coefficient of the impact of charging time on convenience, which is usually a positive constant.

[0146] The availability of charging piles indicates the degree of idleness of charging facilities. The lower the usage rate of charging piles, the higher the convenience. The availability can be calculated by the idle rate of charging piles:

[0147]

[0148] in:

[0149] ava i is the idle rate of the charging pile when tram i arrives (the idle rate is the ratio of the idle time of the charging pile to the total time).

[0150] δ is the coefficient of the effect of usability on convenience, which is usually a positive constant.

[0151] The results obtained from the upper-level planning ( P grid (t), P pv (t), P ev (t)) is used as the input of the lower-level planning, and the result of the lower-level planning (charging power P of each electric vehicle) ev,i (t)) is fed back to the upper-level planning. Through continuous iterative optimization, the objective function values ​​of the upper and lower-level planning converge or the preset number of iterations is reached. The convergence condition is set as:

[0152] and

[0153] in, and are the total benefits of the upper-level planning at the kth and k-1th iterations, respectively. and are the overall charging costs of the i-th tram at the k-th and k-1-th iterations, respectively, and ε1 and ε2 are the preset convergence thresholds.

[0154] Finally, the optimal operation strategy of the photovoltaic charging station in each time period is obtained, that is, the orderly charging power of the tram.

[0155] Step 2: The dynamic alternating direction multiplier method (ADMM) is used to iteratively solve the two-layer model. The upper and lower layers are optimized collaboratively by adaptively adjusting the Lagrange multiplier and penalty factor. The output of the upper layer model serves as the input constraint for the lower layer model, and the charging demand and results of the lower layer model are fed back to the upper layer model. The lower layer planning model optimizes the charging behavior of the electric vehicles during the charging process based on the energy allocation results of the upper layer model. The results of the upper layer planning serve as the input of the lower layer planning, and the results of the lower layer planning are fed back to the upper layer planning. Through continuous iterative optimization, the objective function values ​​of the upper and lower layers converge or the preset number of iterations is reached, ultimately achieving the globally optimal operation strategy.

[0156] In step 2, an improved dynamic alternating algorithm is used to solve the above two-level optimization problem, that is, a collaborative solution is achieved by alternately optimizing the upper and lower level planning.

[0157] The dynamic alternating algorithm gradually optimizes the operating efficiency of the integrated photovoltaic storage and charging power station and the charging strategy of the electric vehicle by alternately updating the decision variables of the upper and lower layers and combining the ideas of the Lagrange multiplier method. Each iteration includes:

[0158] (1) Upper-level optimization: Optimize the charging and discharging strategies of the energy storage system based on the decision-making results of the lower level.

[0159] (2) Lower-level optimization: Optimize the charging strategy of the electric vehicle based on the decision-making results of the upper level.

[0160] (3) Lagrange multiplier update: used to adjust the constraint differences between the upper and lower layers.

[0161] Through these steps, ADMM can effectively decouple the complex two-layer optimization problem and ultimately achieve the coordinated optimization of integrated photovoltaic storage and charging power stations and electric vehicle charging.

[0162] The specific steps include: S201, initialization, S202, upper layer problem optimization, S203, lower layer problem optimization, S204, Lagrange multiplier update, S205, convergence check.

[0163] The upper level goal is to maximize the overall benefits of the integrated photovoltaic storage and charging power station, including reducing operating costs and improving energy efficiency, taking into account the costs of energy storage systems, photovoltaic power generation, purchasing electricity from the grid, and the revenue from selling electricity. Its mathematical model is:

[0164]

[0165] Among them, x is the upper-level decision variable, such as the charging and discharging power of the energy storage system, and y is the decision variable of the lower-level planning (the charging power of the tram).

[0166] The goal of the lower layer is to minimize the charging cost or charging time while meeting the user's charging needs. Taking into account the charging time of the tram and the time-of-use electricity price strategy, the objective function is:

[0167]

[0168] The following problem is solved by combining constraints using a Lagrange multiplier method:

[0169]

[0170] Where g(x) and h(y) are constraint functions, and λ is the Lagrange multiplier.

[0171] Step 1: Initialization

[0172] Set the initial value:

[0173] Upper-level decision variable x (0) and the lower-level decision variable y (0) .

[0174] Initialize the Lagrange multiplier λ (0) and penalty factor ρ.

[0175] Set the convergence accuracy ε and the maximum number of iterations T max .

[0176] Step 2: Upper layer optimization update

[0177] Given the lower-level decision variables and Lagrange multipliers, solve the upper-level program:

[0178]

[0179] Here g(x) and h(y) are constraint functions, representing the coordination constraints between the upper and lower layers.

[0180] At each iteration, the weight of the Lagrange multiplier λ is dynamically adjusted to suit the optimization requirements of the system at different stages. This can be achieved by automatically adjusting λ by calculating the gradient or error of the current objective function, accelerating the convergence process. A feasibility check algorithm is used to ensure that each updated solution is feasible, avoiding unnecessary infeasible solutions.

[0181] Step 3: Lower layer optimization update

[0182] Given the upper-level decision variables and Lagrange multipliers, solve the lower-level optimization problem:

[0183]

[0184] The goal of the lower-level problem is to optimize the charging time and power of the electric vehicle based on the decision variables of the upper level (such as energy storage power, photovoltaic power generation, etc.).

[0185] The more efficient augmented Lagrangian optimization algorithm handles the nonlinear constraints of electric vehicle charging, avoiding the slow convergence issues often encountered with traditional ADMMs. Incorporating dynamic programming and taking into account the changing charging needs of users, this approach allows for more rational scheduling of charging power and time periods, reducing charging costs.

[0186] Step 4: Lagrange multiplier update

[0187] Update the Lagrange multiplier λ (k+1) :

[0188] λ (k+1) =λ (k) +ρ(g(x (k+1) )-h(y (k+1) ))

[0189] Dynamically adjust the penalty factor ρ to adapt to changes in the problem at each update. If the constraint error is large, increase ρ to strengthen the constraint; if the error is small, decrease ρ to reduce the computational burden.

[0190] Step 5: Feasibility check and convergence determination

[0191] For each iteration, calculate the error and determine whether the convergence condition is met:

[0192] If the change of the objective function is less than a predetermined tolerance ε, the objective function is considered to have converged.

[0193] ||g(x (k+1) )-h(y (k+1) )||≤ε

[0194] If the convergence condition is met, stop the iteration and return to the optimal solution; otherwise, continue to step 2.

[0195] The improved dynamic alternation algorithm improves performance in the following ways:

[0196] Adaptive Lagrange multiplier adjustment: Dynamically adjust λ and penalty factor ρ to accelerate convergence and ensure algorithm stability.

[0197] Enhanced optimization algorithms: More efficient optimization methods (such as augmented Lagrangian method and interior point method) are used to handle nonlinear constraints and improve the computational efficiency of lower-level optimization.

[0198] Dynamic programming strategy: In the optimization of electric vehicle charging, the idea of ​​dynamic programming is used to optimize the time period and power scheduling, reduce charging costs, and improve user satisfaction.

[0199] Based on the results of upper- and lower-level planning, a charging control instruction set for the solar-powered energy storage charging station is constructed. Power instructions for interaction between the solar-powered energy storage charging station and the grid are sent to the grid access device, controlling the energy storage system to operate according to the predetermined charging and discharging power strategy. Charging power curve instructions are also sent to the charging equipment of each tram, achieving precise control of the tram charging process.

[0200] A real-time monitoring system is established, and a sensor network is used to collect and monitor photovoltaic power generation, energy storage system status (including remaining capacity, charge and discharge current and voltage, etc.), electric vehicle charging status (such as charging power, battery capacity, etc.), and grid operating parameters (such as voltage, frequency, etc.) in real time.

[0201] When the monitoring system detects a deviation or anomaly between the actual operating status and the predetermined strategy, it immediately initiates the optimization and adjustment mechanism. For example, if photovoltaic power generation suddenly increases or decreases beyond the predicted range, the power balance constraints in the upper-level planning model are reassessed, and the energy storage system's charging and discharging power strategy is adjusted to fully utilize or store excess photovoltaic power. If a tram leaves early or charging needs change, the relevant constraints in the lower-level planning model are promptly updated, and the charging strategy for the remaining trams is re-optimized to ensure that the photovoltaic and energy storage charging station can always operate safely, efficiently, and economically.

[0202] like Figure 1As shown in the figure, the proposed bi-level planning model iterative optimization process achieves global optimal scheduling of PV-storage charging stations through dynamic interaction between upper and lower-level models. The process begins with initialization of the upper-level planning model. This stage sets the predicted PV output curve, the initial state of charge of the energy storage system, and the grid interaction power boundary, while also generating an initial set of charging power constraints. The upper-level planning model, with the overall economic efficiency of the PV-storage charging station as its objective, solves an optimization problem encompassing PV absorption, energy storage scheduling, and grid power purchase and sales strategies, and outputs power allocation plans for each time period to the lower-level model. After receiving the charging power limits provided by the upper-level model, the lower-level planning model uses a multi-objective optimization algorithm to assign optimal charging time periods and power to each vehicle based on real-time charging queue information and user demand characteristics. It also calculates the actual charging load curve and feeds it back to the upper-level planning model. The upper and lower-level planning models are collaboratively optimized using the dynamic alternating direction multiplier method (ADMM). In each iteration, the upper-level model adjusts the power allocation strategy based on the actual load feedback from the lower-level model, while the lower-level model simultaneously adjusts the charging plan to adapt to the updated constraints. During the iteration process, two sets of key convergence indicators are continuously monitored: one is the difference between the upper and lower objective functions (reflecting the progress of economic optimization), and the other is the matching error between power distribution and actual load (indicating system feasibility). The iteration is terminated when both are below the preset threshold. The final output includes the global optimal solution for energy storage charging and discharging plan, photovoltaic absorption plan and electric vehicle charging strategy, and is sent to the charging station execution terminal through the communication module. The entire process ensures that the system maintains the optimal operating state in a dynamic environment through a closed-loop feedback mechanism, ultimately achieving efficient energy scheduling of the photovoltaic storage charging station and optimized control of electric vehicle charging.

[0203] like Figure 2As shown in the figure, the dynamic alternating optimization algorithm proposed in this paper is used to solve the bi-level planning problem of photovoltaic charging stations. Its core process includes five key stages: initialization, upper-level optimization, lower-level optimization, Lagrange multiplier update, and convergence determination. First, the algorithm initializes the upper-level decision variables (such as energy storage charging and discharging power and grid interaction power), lower-level decision variables (such as vehicle charging power), and Lagrange multipliers, and sets a convergence threshold and a maximum number of iterations. It then enters an iterative loop. In the upper-level optimization stage, based on the current lower-level charging demand and Lagrange multipliers, the upper-level model is solved with the goal of minimizing operating costs, outputting the optimal power allocation solution. In the lower-level optimization stage, based on the power constraints provided by the upper layer and the goals of user charging cost and convenience, the charging time and power of each vehicle are optimized, and the actual charging demand is fed back to the upper-level grid load balance constraint. Next, the algorithm dynamically updates the Lagrange multiplier and adjusts the penalty factor by comparing the difference between the upper and lower power allocations. If the constraint error is large, the penalty factor is increased to strengthen the constraint; otherwise, it is reduced to reduce the computational burden. Finally, the algorithm checks whether the change in the objective function and the constraint error meet the preset convergence criteria. If so, it outputs the optimal strategy; otherwise, it proceeds to the next iteration. This process significantly improves convergence speed and stability through adaptive adjustment mechanisms and nonlinear constraint processing techniques, making it suitable for real-time optimization and scheduling of large-scale solar-powered charging stations. This flowchart intuitively illustrates the algorithm's closed-loop optimization mechanism, highlighting the dynamic process of interaction and dynamic adjustment between upper and lower layers.

[0204] The dynamic alternating direction multiplier method (ADMM) achieves efficient solution of the bilevel programming model through a decomposition and coordination mechanism. The algorithm initialization phase sets initial values ​​for upper-level decision variables (such as energy storage charging and discharging power and grid interaction power), lower-level decision variables (such as vehicle charging power and charging time window), as well as Lagrange multipliers and penalty factors. Convergence accuracy and maximum iteration thresholds are also configured. The upper-level optimization phase, based on charging load data fed back from the lower level, updates the energy storage charging and discharging strategy and grid interaction power while satisfying grid balance, energy storage status, and PV output constraints. The lower-level optimization phase dynamically adjusts the charging power and timing of each vehicle based on the power allocation constraints provided by the upper level, combined with vehicle charging demand, time-of-use electricity prices, and convenience scores, to minimize user costs. The Lagrange multiplier update phase dynamically adjusts the multiplier weights and penalty factors by calculating the constraint error between the upper and lower level objective functions. If the error is large, the constraint strength is strengthened to accelerate convergence. The convergence determination phase checks whether the change in the upper and lower layer objective functions is less than a preset threshold or the maximum number of iterations has been reached. If so, the iteration is terminated and the optimal control instructions are output. Otherwise, the algorithm returns to the upper optimization phase and continues the cycle. Through adaptive parameter adjustment and real-time error feedback, the algorithm ensures rapid convergence to the global optimal solution under complex multi-constraint conditions.

[0205] Cooperate Figure 3 As shown in the figure, in the operation scenario of the photovoltaic charging station, the lower-level model's electric vehicle charging scheduling is sophisticated and complex. In the initial stage, the lower-level model receives information from the upper-level model on the interaction power constraints between the photovoltaic charging station and the power grid, as well as the charging and discharging power constraints of the energy storage system. For each electric vehicle entering the charging station, the charging demand information it carries is taken into consideration. This demand is closely related to factors such as the current state of charge of the vehicle battery and the driver's preset mileage requirements to ensure that the electric vehicle charging demand constraints are met. This constraint accurately controls the matching relationship between the vehicle's charge amount, charging time, and power, ensuring that the vehicle has sufficient power for subsequent operation.

[0206] The battery characteristics of trams constitute one of the key constraints in scheduling. The minimum and maximum charging power allowed by the battery, namely the charging power constraint, strictly limits the power allocation range during the charging process. Any operation outside this range may cause battery performance degradation and shorten its lifespan. Therefore, the scheduling process always follows this power boundary for reasonable allocation.

[0207] The arrival and departure times of users are also crucial. They not only define the time span during which vehicles can charge but also profoundly influence the order in which they are charged. Based on factors such as the urgency of the vehicle's charging needs and the priority levels set by the vehicle owner, the system uses a pre-defined algorithm to determine the charging order for each vehicle. This ensures that the charging needs of multiple vehicles are met efficiently within a limited timeframe, avoiding idle resources and wasted resources.

[0208] The time-of-use electricity pricing strategy plays a cost control role in charging scheduling. Given the differences in electricity prices at different times, the system cleverly arranges electric vehicles to obtain higher charging power or extend charging time during periods of low electricity prices based on the electricity price fluctuation curve, while meeting the basic charging needs of vehicles and other constraints. It also appropriately reduces charging time during peak electricity price periods, thereby reducing charging costs for vehicle owners and maximizing economic benefits.

[0209] Furthermore, the charging convenience score is integrated into the scheduling process as an auxiliary decision-making tool. It comprehensively considers factors such as the distance from the owner to the nearest charging station, the waiting time at the charging station, the actual charging time, and the availability score of the charging facilities. Each factor is calculated based on a preset weight coefficient. The resulting score intuitively reflects the convenience of charging the vehicle at the charging station, helping the system select the optimal charging time and power for each vehicle, thereby comprehensively improving the user experience.

[0210] Finally, after coordinating and calculating the above-mentioned factors in a complex manner, the system determines the charging start time, charging power, and charging end time for each electric vehicle, enabling the electric vehicles to charge in an orderly and efficient manner at the solar storage charging station, ensuring the smooth operation of the entire charging station and achieving dual optimization of energy utilization and user needs.

[0211] Cooperate Figure 4 As shown in the figure, in the operating environment of the photovoltaic storage charging station, the real-time monitoring system plays a vital role, and its architecture presents a multi-node, all-round coverage layout. First, high-precision light sensors and power monitoring equipment are deployed in the photovoltaic power generation area. These sensors can capture key information such as light intensity and angle in real time, and accurately calculate the photovoltaic power generation based on the light data. The collected data will be continuously transmitted to the central control unit via wired or wireless transmission along the pre-set data link. At the same time, the energy storage system is equipped with a series of dedicated sensors for monitoring the remaining capacity of the battery. The voltage and current sensors provide real-time feedback on the charging and discharging current and voltage. These parameters not only reflect the current energy reserve status of the energy storage system, but also provide a basis for judging the health of the battery. The data is also aggregated to the central control unit in a high-speed and stable transmission manner.

[0212] In the electric vehicle charging area, each charging station is integrated with a charging power sensor and a communication module that interacts with the vehicle's battery management system. This allows for accurate measurement of each electric vehicle's real-time charging power and access to key charging status information, such as battery capacity. This decentralized data, collected from each charging station, is ultimately aggregated to a central control unit. Furthermore, grid parameter monitors are installed at the interface with the external power grid to closely monitor operating parameters such as voltage and frequency, providing rapid feedback in the event of any abnormalities, such as grid fluctuations.

[0213] As the core hub of the entire real-time monitoring system, the central control unit receives massive amounts of data from each of the aforementioned areas and then integrates, analyzes, and stores the data. On the one hand, it evaluates the operating status of the photovoltaic charging station in real time based on preset rules and algorithms to determine whether the current operation complies with the predetermined strategy. On the other hand, when the monitoring system detects a deviation or abnormality between the actual operating status and the predetermined strategy, such as when photovoltaic power generation suddenly exceeds or falls far below the predicted value, a tram leaves early, or charging demand suddenly changes, the central control unit will immediately initiate an optimization and adjustment mechanism. Depending on the type of abnormality, it will send targeted instructions to the energy storage system, tram charging management module, or grid interface module to adjust the energy storage system's charging and discharging power strategy to balance energy, re-optimize the charging strategy for the remaining trams, or coordinate and communicate with the grid to ensure that the photovoltaic charging station always maintains the optimal operating state of safety, efficiency, and economy, ensuring the smooth operation of the entire system.

[0214] In summary, the present invention provides a method for controlling the orderly charging of electric vehicles in a photovoltaic charging station based on double-layer planning, which mainly has the following improvements:

[0215] First, two-tier planning architecture design

[0216] The present invention constructs a mathematical model for collaborative optimization of the upper layer (system optimization layer) and the lower layer (charging scheduling layer), and achieves a dynamic balance between global energy scheduling and local charging control through hierarchical decoupling. The upper-layer planning model aims to minimize the overall operating costs of the photovoltaic storage charging station and maximize the consumption of renewable energy, and optimizes photovoltaic power generation power, energy storage charging and discharging strategies, and grid interaction power; the lower-layer planning model is based on the power constraints allocated by the upper layer, with the goal of minimizing user charging costs and maximizing convenience, and dynamically optimizes the charging power, time window, and priority sorting of trams.

[0217] Second, dynamic alternating optimization algorithm

[0218] This paper uses a modified augmented Lagrangian alternating direction multiplier method (ADMM) to achieve efficient iterative convergence of the upper and lower models by adaptively adjusting the Lagrangian multipliers and penalty factors. This algorithm incorporates dynamic programming concepts to address nonlinear constraints on electric vehicle charging (such as battery charge and discharge rate limits and user demand fluctuations), improving solution efficiency and strategy robustness.

[0219] Third, multi-objective collaborative constraint mechanism

[0220] The present invention sets the following constraints:

[0221] Grid load balancing constraints: Real-time coordination of the power balance of photovoltaic power generation, energy storage charging and discharging, grid power purchase and sales, and tram charging needs.

[0222] Battery health management constraints: Avoid overcharging / overdischarging of the energy storage system and electric vehicle batteries through charging power time optimization and charge and discharge depth limitation.

[0223] User demand elasticity constraint: Combined with time-of-use electricity prices and charging convenience scores (based on distance, waiting time, charging time, and facility availability), this system optimizes both user satisfaction and system economics.

[0224] Fourth, real-time monitoring and dynamic adjustment system

[0225] The present invention constructs a sensor network to collect photovoltaic output, energy storage status, grid parameters and tram charging data in real time. When the actual operation deviates from the predetermined strategy, the re-optimization mechanism of the double-layer model is triggered to ensure the real-time optimal operation of the system.

[0226] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0227] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A method for controlling the orderly charging of electric vehicles in a photovoltaic charging station based on a double-layer planning, characterized by The steps include: Step 1: Construct a two-layer planning model for the solar-storage charging station, where the upper-layer planning model is the system optimization layer and the lower-layer planning model is the charging scheduling layer; The upper-level planning model is used to optimize the overall energy scheduling of the charging station, including the power distribution of photovoltaic power generation, energy storage equipment and the power grid; the lower-level planning model optimizes the charging behavior of the electric vehicles during the charging process based on the energy allocation results of the upper-level planning model; Step 2: The dynamic alternating direction multiplier method is used to iteratively solve the two-level planning model. The output of the upper-level planning model is used as the input constraint of the lower-level planning model. The lower-level planning model then feeds the results back to the upper-level planning model until the objective function values ​​of the upper and lower planning converge or the preset number of iterations is reached. The optimal operation strategy of the solar-powered charging station in each time period, that is, the orderly charging power of the electric vehicle, is obtained. Step 3: monitor photovoltaic output, energy storage status, grid parameters, and electric vehicle charging requirements in real time, and optimize the two-level programming model when an operating deviation is detected.

2. The method for controlling orderly charging of electric vehicles at a photovoltaic charging station based on double-layer planning as claimed in claim 1, characterized in that: In step 1, the optimization goal of the upper-level planning model is to maximize the economic benefits of the power plant and improve the absorption capacity of renewable energy, while ensuring the load balance and sustainable operation of the power grid. Its objective function is: Among them, E station is the total revenue of the charging station, P pv (t) is the photovoltaic power generation power, P ess (t) is the charging and discharging power of the energy storage system, λ pv and λ ess are all power conversion factors; R sell (t) is the income from selling electricity to the grid when the surplus photovoltaic power is connected to the grid; R ev (t) the revenue from providing charging services to electric vehicles at charging stations; ess (t) is the cost of the energy storage system, including depreciation cost and maintenance cost; C pv The maintenance cost C of the photovoltaic power generation system pv =α pv P pv,rated , where α pv is the maintenance factor per unit power of the photovoltaic power generation system, P pv,rated is the rated power of the photovoltaic power generation system; C grid (t) is the cost of the power station purchasing electricity from the grid at time t, and is related to the electricity price p grid proportional to; The upper-level decision variables include the charging power of the energy storage system at time t Discharge power of the energy storage system at time t Power P of the interaction between the power station and the grid grid (t), photovoltaic power generation power P pv (t); The constraints include grid load balance constraints, energy storage system status constraints, photovoltaic power generation constraints, and grid power constraints.

3. The method for controlling the orderly charging of electric vehicles at a photovoltaic charging station based on double-layer planning as claimed in claim 2, characterized in that: The grid load balancing constraint is, P grid (t)+P pv (t)+P ess (t)-P ev (t)=0 Among them, P ev (t) is the load demand of the power station, i.e., the required charging power of the tram; The state constraint of the energy storage system is: From ess (t)=E ess (t-1)+η ess P ess (t)Δt Among them, E ess (t) is the energy storage capacity at time t, η ess is the charge and discharge efficiency of the energy storage system, Δt is the time step; and are the minimum and maximum charge and discharge power of the energy storage system, respectively, and E ess (t) is the state of charge of the energy storage battery; The photovoltaic power generation constraint is that the power of photovoltaic power generation cannot exceed the maximum power of photovoltaic power generation. 0≤P pv (t)≤P pv,max (t) Among them, P pv,max (t) is the maximum photovoltaic power generation power predicted based on real-time weather forecast and sunlight intensity; The grid power constraint is that the grid power cannot exceed the maximum carrying capacity of the grid. in, is the maximum carrying power of the power grid.

4. The method for controlling the orderly charging of electric vehicles at a photovoltaic and energy storage charging station based on double-layer planning as claimed in claim 1, characterized in that: In step 1, the objective function of the lower-level planning model is: Among them, P target,i (t) is the target charging power of the ith trolley at time t, Price(t) is the charging price; V charge,k Score the convenience of charging; S i is the total cost of the i-th electric vehicle during the charging process; t i,start is the time when the i-th tram starts charging, t i,end is the time when the charging of the i-th electric vehicle ends; w1, w2, w3, w4 are the weight coefficients of each factor, satisfying w1+w2+w3+w4=1; The lower-level decision variables include the charging power P of the i-th tram at time t ev,i (t), charging start time t of the i-th electric vehicle i,start , the charging end time t of the i-th electric car i,end ; The constraints include charging time constraints, charging power constraints, electric vehicle charging demand constraints, and electric vehicle charging sequence constraints.

5. The method for controlling orderly charging of electric vehicles at a photovoltaic charging station based on double-layer planning as claimed in claim 4 is characterized in that: The charging time constraint is, t i,start ≤t≤t i,end The charging power constraint is, in, and are the minimum and maximum charging powers of the i-th tram respectively; The electric vehicle charging demand constraint is that the charging time of the electric vehicle must meet its charging demand, that is, each electric vehicle needs to be charged with a certain amount of electricity to meet its maximum charging capacity; The charging order constraint of electric vehicles is that the charging order of each electric vehicle in the charging station is scheduled according to its demand and priority.

6. The method for controlling orderly charging of electric vehicles at a photovoltaic and energy storage charging station based on double-layer planning as claimed in claim 1, characterized in that: In step 2, the upper-level planning model is first solved to obtain the optimal power allocation solution; The lower-level planning model optimizes the charging power and charging time for each electric vehicle based on the optimal power allocation scheme, obtains a planning result, and feeds it back to the upper-level planning model. Through iterative optimization, until the objective function values ​​of the upper and lower-level planning converge or a preset number of iterations is reached, the optimal operation strategy of the solar-powered charging station in each time period is finally obtained, that is, the orderly charging power of the electric vehicles. The convergence conditions are set as: in, and are the total benefits of the upper-level planning at the kth and k-1th iterations, respectively. and are the overall charging costs of the i-th tram at the k-th and k-1-th iterations, respectively, and ε1 and ε2 are the preset convergence thresholds.

7. The method for controlling orderly charging of electric vehicles at a photovoltaic and energy storage charging station based on double-layer planning as claimed in claim 1, characterized in that: In step 2, the dynamic alternating direction multiplier method is used to iteratively solve the bi-level programming model. The specific process is: Step 21, set the initial value, including the upper decision variable x (0) and the lower-level decision variable y (0) , initialize the Lagrange multiplier λ (0) and penalty factor ρ, set convergence accuracy ε and maximum number of iterations T max ; Step 22: Given the lower-level decision variables and Lagrange multipliers, solve the upper-level planning: Among them, g(x) and h(y) are constraint functions, representing the coordination constraints between the upper and lower layers; At each iteration, the weight of the Lagrange multiplier λ is dynamically adjusted; Step 23: Given the upper-level decision variables and Lagrange multipliers, solve the lower-level optimization problem: Step 24, update the Lagrange multiplier λ (k+1) , l (k+1) =λ (k) +ρ(g(x (k+1) )-h(y (k+1) )) Dynamically adjust the penalty factor ρ. If the constraint error is large, increase ρ; if the error is small, decrease ρ. Step 25: For each iteration, calculate the error and determine whether the convergence condition is met. If the change of the objective function is less than the predetermined tolerance ε, the objective function is considered to have converged. If the convergence condition is met, the iteration is stopped and the optimal solution is returned; otherwise, step 22 is continued.

Images