A wind-solar-storage collaborative planning method based on improved bat algorithm

Abstract

The application discloses a wind-light-storage collaborative planning method based on an improved bat algorithm, and specifically comprises the following steps: constructing a double-layer optimization model; the upper layer is a planning model, and the maximum net income in the whole life cycle is taken as a target to optimize the site selection and capacity configuration of wind power, photovoltaic and energy storage systems; the lower layer is an operation model, and the minimum wind power abandonment, light abandonment and network loss are taken as targets to coordinate the operation scheduling between the renewable energy and the energy storage system to guarantee the economy and stability of system operation; and then, the improved bat algorithm is used to iteratively solve the double-layer planning model. The application can effectively improve the consumption capacity of the distribution network to the distributed renewable energy, and further improve the operation reliability and safety of the power grid under the condition of high proportion of renewable energy access.

Description

Technical Field

[0001] This invention relates to the field of power system planning and renewable energy grid connection and dispatch technology, specifically to a wind-solar-storage collaborative planning method based on an improved bat algorithm. It is applicable to the joint site selection and capacity configuration of distributed wind power, photovoltaic and energy storage systems in distribution systems, so as to improve the renewable energy absorption capacity, optimize system operating efficiency and enhance the economy and flexibility of the power grid. Background Technology

[0002] With the large-scale grid connection of renewable energy sources, the proportion of distributed energy sources such as wind power and photovoltaics in the power system is constantly increasing, bringing unprecedented challenges to the planning and operation of the distribution network. Wind power and photovoltaic power generation are characterized by strong volatility and unpredictability, which can easily cause voltage fluctuations, power flow reversals, and wind and solar curtailment, affecting the stability and operating efficiency of the power grid. To enhance the grid's regulation capabilities, energy storage systems are being widely introduced as a regulatory resource, operating in conjunction with wind power and photovoltaics.

[0003] However, the optimal allocation of wind, solar, and energy storage resources involves complex nonlinear coupling relationships, encompassing multi-dimensional decision variables such as site selection, capacity determination, and operational scheduling. Traditional planning methods often employ linear programming, single-layer optimization, or heuristic approaches, failing to balance the system's long-term economic viability and operational coordination. Furthermore, most methods only consider a single objective or static scenario, making them ill-suited to the dynamic characteristics and uncertainties of wind and solar power output across multiple scenarios. At the algorithmic level, traditional heuristic algorithms, such as standard particle swarm optimization and genetic algorithms, are prone to getting trapped in local optima when solving complex multivariate problems, exhibiting slow convergence speeds and sensitivity to model dimensionality, making it difficult to guarantee the globality and stability of the solution.

[0004] Therefore, there is an urgent need to construct a collaborative optimization method that can take into account both the planning and operation layers, possess a dynamic feedback mechanism, and adapt to the uncertainty of renewable energy output. Simultaneously, it is necessary to improve existing intelligent optimization algorithms, enhancing their global search capability and convergence efficiency in complex high-dimensional planning models. To this end, a wind-solar-storage collaborative planning method based on an improved bat algorithm is proposed. By constructing a two-layer optimization model, resource allocation and operation scheduling are coordinated, and mechanisms such as chaotic search, sinusoidal perturbation, and Lévy flight are introduced to optimize algorithm performance, thereby improving the system's renewable energy absorption capacity and operational economy, and providing efficient and intelligent technical support for the development of the distribution network. Summary of the Invention

[0005] The purpose of this invention is to solve the technical problems of planning and operation coordination difficulties caused by volatility and uncertainty in the process of renewable energy access to the distribution network; and the technical defects of slow convergence speed, low search accuracy, weak local search ability, and easy getting trapped in local optima when using the existing bat algorithm to solve the bi-level optimization model, which leads to the technical problems of low computational efficiency, unstable global optimal solution and insufficient accuracy of information transmission between upper and lower levels in the bi-level optimization model.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0007] A wind-solar-storage collaborative planning method based on an improved bat algorithm includes the following steps:

[0008] Step 1: Construct a two-layer optimization model for the distribution network that coordinates planning and operation;

[0009] The upper layer of the distribution network dual-layer optimization model is a planning model, which aims to maximize the net benefit over the entire life cycle and optimize the site selection and capacity configuration of wind power, photovoltaic and energy storage equipment; the lower layer is an operation model, which, based on the planning results of the upper layer, aims to minimize the amount of abandoned electricity and network loss and coordinate the operation scheduling among wind power, photovoltaic and energy storage equipment.

[0010] Step 2: Solve the two-layer optimization model of the distribution network that coordinates planning and operation constructed in Step 1, and finally obtain the optimal planning scheme.

[0011] In step 1, the upper-level planning model comprehensively considers electricity sales revenue, equipment investment, and operation and maintenance costs. It optimizes the site selection and capacity allocation scheme of wind power, photovoltaic, and energy storage systems under constraints such as investment budget, capacity allocation, and grid stability, so as to maximize the net revenue throughout the entire life cycle.

[0012] The objective function of the upper-level planning model is to maximize the net revenue over the entire life cycle, including electricity sales revenue and equipment life cycle costs within the planning period. The specific objective function of the planning model is as follows:

[0013] (1);

[0014] in, Indicates the time period within the planning cycle. t Revenue from electricity sales This represents the annualized net present value of the equipment investment cost. T For the planning period, The goal of optimization at the higher level is net profit;

[0015] Electricity sales revenue The calculation is as follows:

[0016] (2);

[0017] In the formula: For wind turbines i During the period t The power generation capacity; For photovoltaic units j During the period t The power generation capacity; For energy storage systems during time periods t The charging and discharging power, with positive values ​​for discharging and negative values ​​for charging; For time period t Electricity price; N and M These represent the total number of wind turbine units and photovoltaic units, respectively.

[0018] Equipment investment cost The calculation is as follows:

[0019] (3);

[0020] Equipment investment cost This is the sum of the total lifecycle costs of wind power, solar power, and energy storage, including initial investment costs and operation and maintenance costs, using the annual discount rate. Discount them all at once; i, j, e These represent wind turbines, photovoltaic systems, and energy storage equipment, respectively. , , These are 0-1 decision variables, representing the construction status of wind turbines, photovoltaics, and energy storage, respectively. A value of 1 indicates installation. , Wind turbine i With photovoltaic units j The unit investment cost; , , These refer to the service life of wind turbines, photovoltaic systems, and energy storage equipment, respectively. , , These are the annual operation and maintenance costs for wind turbines, photovoltaic systems, and energy storage equipment, respectively. , , These are the rated power of wind turbines, photovoltaic systems, and energy storage equipment, respectively. This refers to a collection of newly planned energy storage devices; Indicates the first k Investment cost per unit power capacity of energy storage equipment Indicates the first k Investment cost per unit energy capacity of energy storage equipment For the energy capacity of energy storage devices, Other initial investment costs for energy storage equipment.

[0021] The constraints of the upper-level planning model are expressed as follows:

[0022] The total investment cost of equipment during the planning period shall not exceed the budget ceiling, as specified below:

[0023] (4);

[0024] In the formula, This is the total investment cost of the equipment, which is the sum of the construction costs of wind power, photovoltaic, and energy storage equipment; The maximum allowed total investment limit;

[0025] The total installed capacity of distributed generation must be within a specific proportion of the system load to ensure stable grid operation. The constraint is expressed as follows:

[0026] (5);

[0027] In the formula, P t For the system in time period t Total load power, This represents the ratio of distributed power generation to the total system load power.

[0028] In step 1, the lower-level operation model, based on the site selection and capacity determination scheme provided by the planning layer, comprehensively considers grid power balance, equipment output characteristics, node voltage constraints, and the dynamic operation characteristics of energy storage to determine the optimal coordination strategy between renewable energy and energy storage. The objective function of the lower-level operation model is to minimize the total operating cost, including the costs of wind and solar curtailment and grid losses. The specific objective function of the operation model is as follows:

[0029] (6);

[0030] In the formula, , These represent the unit costs of wind and solar power curtailment, respectively. , They are respectively t Time period i The wind turbine unit and the first j The amount of electricity wasted by each photovoltaic unit; For electricity price, This refers to the annual maximum load hours. L The total number of branch roads; branch road l electrical conductivity, , Branch roads i and branch roadsj The node voltage amplitude, For nodes ij Phase angle difference between wind turbine units i and photovoltaic units j For the device index number, and in the node voltage i , j This refers to the physical node number of the power grid.

[0031] The constraints of the lower-level operational model include node power balance constraints, distributed generation output constraints, node voltage constraints, and energy storage system operational constraints, which are specifically expressed as follows:

[0032] Power balance constraint: The real-time power flow balance of the power grid must be maintained. The constraint is expressed as follows:

[0033] (7);

[0034] In the formula, For wind turbine i Rated power; For photovoltaic j Rated power; This refers to the discharge power of the energy storage device. The charging power for energy storage devices; P L This refers to the system load power.

[0035] Distributed power generation output constraints: The real-time output of wind and solar power must meet their rated capacity limits and follow the output curve. The constraints are expressed as follows:

[0036] (8);

[0037] (9);

[0038] In the formula, , for t Actual output of wind turbines and solar power during specific time periods;

[0039] Node voltage constraints: i , j The physical nodes of the power grid are numbered, and the node voltage amplitude must meet the dynamic safe operating range. The constraints are as follows:

[0040] (10);

[0041] In the formula, for t Time period nodes i voltage amplitude, , fort Time period nodes i The upper and lower limits of the voltage amplitude.

[0042] Energy storage system operating constraints: The energy storage system must simultaneously satisfy the State of Charge (SOC) dynamic equation, capacity limits, and charge / discharge power limits, as expressed below:

[0043] (11);

[0044] (12);

[0045] (13);

[0046] The operation of energy storage systems must comply with strict dynamic constraints: among which For energy storage systems t State of charge during a period of time This represents the state of charge at the previous moment. for t Time-based charge and discharge power, and For charging and discharging efficiency; The time interval for updating the energy storage system status; and The maximum and minimum states of charge allowed by the system are the safe operating boundaries of the state of charge, preventing equipment lifespan degradation caused by overcharging or over-discharging; among the energy storage capacity parameters, For the planned energy storage system's rated capacity, Configure the actual capacity of the energy storage system to ensure that the system has the necessary redundancy to cope with load fluctuations.

[0047] In step 2, the upper-level model takes maximizing the net revenue over the entire lifecycle as its objective function and the site selection and capacity configuration of distributed wind power, photovoltaic, and energy storage systems as decision variables. This problem belongs to the integer linear programming class, so the commercial solver CPLEX is used for solving it. The upper-level planning model outputs the node numbers of the selected wind power stations, photovoltaic stations, and energy storage locations, as well as the optimal capacity corresponding to each node, and uses them as input to the lower-level operation model. The lower-level model takes minimizing the amount of abandoned electricity and network loss as its objective function and coordinates the operation scheduling between wind power, photovoltaic, and energy storage equipment. This problem involves nonlinearity and coupled variables, so an improved bat algorithm is used for solving it. The upper and lower levels interact and optimize each other through a "feedback iteration" method until convergence, and finally obtain the optimal planning scheme.

[0048] Step 2 specifically includes the following steps:

[0049] Step 2.1 Initialize basic parameters; Initialize the basic parameters of the entire system, inputting grid structure parameters, including the number of nodes, branch topology, node voltage range, and conductance; load parameters, such as load time series data and annual maximum load hours for each node; wind and solar energy resource data; parameters of wind turbines, solar generators, and energy storage devices; and electricity price parameters.

[0050] Step 2.2 Use the commercial solver CPLEX to solve the upper-level planning model. Its objective function, decision variables, and constraints are shown in formulas (1) to (5). The solver outputs the planning scheme solved in this round, including: the node numbers of the selected wind turbines, photovoltaic units, and energy storage equipment, the optimal capacity of each node, and the economic allocation strategy.

[0051] Step 2.3: The site selection and sizing plan scheme obtained from the upper-level planning model is used as input data and passed to the lower-level running model;

[0052] Step 2.4 The lower-level operation model needs to optimize the operation control strategy based on the input current planning scheme to achieve the goal of minimizing wind and solar power curtailment and grid loss. Therefore, the improved bat algorithm is used to solve the problem.

[0053] In step 2.4, the improved bat algorithm is used to solve the problem, including the following steps:

[0054] Step 2.4.1 Initialize the basic parameters of the improved bat algorithm; input the power grid structure parameters and the objective function of the lower-level operating model. F The corresponding formula (6) is used to calculate the fitness value of the objective function; the initialization settings for the bat population and algorithm parameters are specifically included: randomly generating N The initial position of the bat and speed ( i =1,2,…, N ), the location of each bat It is a D-dimensional vector representing a feasible planning strategy based on upper-level capacity configuration, and also a solution to the two-level optimization model of the distribution network that coordinates planning and operation; algorithm parameters, loudness , emit sound wave frequency Sound wave frequency range and Used to generate random factors Chaotic mapping control parameters Random walk parameters of Levy flight Pulse emission rate Chaotic inertial weights Range of inertia weight and and maximum number of iterations ;

[0055] Step 2.4.2 Based on the objective function of the lower-level running model F Calculate the fitness value of each individual bat and mark the optimal location of the bats in the current population. The speed and position of the bat are iteratively updated according to the rules defined by the following formula;

[0056] For each bat i In the t iterations (t=1,2,…) Perform the following operations when:

[0057] (1) Update the sound wave frequency The original formula is as follows:

[0058] (14);

[0059] For the first i The frequency of sound waves emitted by an individual bat during this iteration; and The lower and upper limits of the set sound wave frequency; For interval Random numbers.

[0060] This invention considers sinusoidal chaotic mapping technology to improve the acoustic frequency update equation. Sinusoidal perturbations, with their finite amplitude and smooth changes, can enhance the search jump capability without introducing drastic fluctuations, thereby improving the global convergence of the bat algorithm. By superimposing sinusoidal perturbation terms in each iteration, the acoustic frequency acquires a certain degree of volatility, which can control the search jump amplitude and increase population diversity. Therefore, utilizing the non-repetitive, ergodic, and hybrid properties of chaotic mapping, sinusoidal chaotic mapping is used to optimize the bat's acoustic frequency. The update rule for each bat's acoustic frequency is as follows:

[0061] (15);

[0062] In the formula, Indicates the first t In the next iteration, the bat i The frequency of the sound wave; These are the control coefficients for the chaotic mapping.

[0063] (2) Update speed The original formula is as follows:

[0064] (16);

[0065] For the first iThe bat in the t The velocity vector during round iteration; This is the velocity vector from the previous iteration; For the bat in the previous iteration i Location; This represents the current local optimum (position) within the population.

[0066] This invention considers chaotic inertia weights to improve the velocity update equation. During iteration, the Bat Algorithm is prone to getting trapped in local optima, often hindering its accurate identification of the global optimum. To address this issue, a chaotic inertia weight strategy is proposed to enhance the original velocity update mechanism. By introducing chaotic inertia weights, the algorithm's ability to avoid local optima is improved, thereby enhancing global search performance. The chaotic inertia weights generated in each iteration are represented as follows:

[0067] (17);

[0068] In the formula: , These represent the upper and lower limits of the inertia weight parameter, respectively. This represents the maximum total number of iterations. This represents the current iteration number. In the interval The internally randomly generated chaotic perturbation factor.

[0069] A chaotic inertia weighting strategy is introduced into the velocity update equation. In each iteration, the chaotic inertia weight is calculated to enhance the dynamic adjustment of the bat velocity. The improved velocity update equation incorporating the chaotic inertia weight is expressed as follows:

[0070] (18);

[0071] in, The chaotic inertial weights change dynamically with iteration, thus dynamically adjusting the bat's flight speed.

[0072] (3) Update location For each bat i In the t iterations (t=1,2,…) When the location is updated, a solution to the new two-level optimization model of the distribution network that coordinates planning and operation is generated. The original location update formula is as follows:

[0073] (19);

[0074] For the first i Only one bat t The position vector during round iteration.

[0075] This invention considers the Lévy flight strategy and improves the position update equation. Lévy flight introduces large-scale, sporadic jumps and frequent, abrupt changes in direction during the search process, helping individual bats avoid getting trapped in local optima and expanding the search space. Therefore, optimization performance in high-dimensional space is significantly improved, thereby enhancing the overall algorithm performance. This invention improves the bat algorithm by introducing the random walk characteristic of Lévy flight. After applying the Lévy flight strategy, the position update formula for each individual bat is as follows:

[0076] (20);

[0077] in For Lévy's random walk path; It is a gamma function; The scaling parameter determines the probability of large jumps in the step size, and is usually set to a value that corresponds to the scaling parameter. , The closer it is to 1, the greater the jump in the search; The closer it gets to 2, the more stable the jump amplitude becomes.

[0078] Step 2.4.3 Perform local search and update. To enhance the algorithm's local search capability in the solution space, a local perturbation is set to conduct a small-scale search on some individual bats, appropriately guiding the solution to converge towards the vicinity of the optimal solution.

[0079] Specifically, generating random numbers If satisfied If the current optimal bat is selected, an optimal individual is chosen, and a local solution is generated near the selected optimal individual using the following local perturbation formula (21). Otherwise, a new solution satisfying the objective function is updated according to the position update formula (20). The local perturbation formula is as follows:

[0080] (twenty one);

[0081] in, for Random numbers in an interval The average sound wave loudness of the entire bat colony during this iteration.

[0082] Step 2.4.4 Perform a global search and update. Generate random numbers. If satisfied If the fitness value of the objective function is greater than the new solution in step 2.4.3, then the solution is accepted, and the sound wave loudness is reduced according to the rules of equation (23-24). and increase pulse emission rate .

[0083] (1) Original sound wave loudness The updated formula is shown below:

[0084] (twenty two);

[0085] in, Let be the loudness of the sound wave of bat i at the t-th iteration; The loudness of bat i's sound wave in the next iteration; This is the sound wave loudness attenuation coefficient.

[0086] With improved sound wave frequency Similar to the updated equation, this invention considers sinusoidal chaotic mapping technology to improve sound wave loudness. Update equation:

[0087] (twenty three);

[0088] in, These are the control coefficients for the chaotic mapping.

[0089] (2) The pulse emissivity update equation is as follows:

[0090] (twenty four);

[0091] in, It is the pulse emission rate enhancement factor.

[0092] Step 2.4.5 Select the best solution with the highest fitness as the global optimum and update its position. Sort the fitness values ​​of all bat individuals in the population, find the position of the bat with the highest fitness value, which is the current best position, and thus the optimal solution. During the iteration process, this solution is the optimization result of the lower-level running model.

[0093] Step 2.4.6 determines whether the maximum number of iterations has been reached and outputs the optimal solution. If the maximum number of iterations has been reached, or the global optimal solution has not significantly improved in consecutive iterations, the iteration is terminated, and the current wind-solar-storage configuration and its operating results are output, i.e., the solution of the desired two-layer optimization model for the distribution network that coordinates planning and operation. Otherwise, return to steps 2.4.2-2.4.5 for iterative calculation. The solution of this iteration, i.e., the lower-level optimization result, is input into the upper-level planning model and fed back to the planning layer to update the planning configuration nodes, power, and capacity, for a new round of iteration. Then, jump to step 2.2 to use the commercial solver CPLEX to complete the optimization calculation of the upper-level planning model.

[0094] Compared with the prior art, the present invention has the following technical effects:

[0095] 1) By constructing a two-layer optimization model, the rationality of wind, solar and energy storage resource allocation and the economic efficiency of system operation are improved; by constructing a two-layer optimization model with upper and lower layers working together, the upper layer realizes the economical allocation of site and capacity, and the lower layer realizes the operation scheduling and energy collaborative control, which effectively improves the overall return on investment and renewable energy consumption efficiency of the system.

[0096] 2) Improvements to the traditional bat algorithm to enhance convergence speed and global optimization capability. This invention introduces chaotic mapping, chaotic inertial weights, and Lévy flight strategy to improve the traditional bat algorithm, effectively avoiding local optima, improving globality and convergence speed, and making it suitable for high-dimensional complex optimization problems.

[0097] 3) The optimization strategy proposed in this invention can significantly reduce the curtailment rate of wind and solar power, reduce network losses, and improve the stability of node voltage, thereby enhancing the operational reliability and flexibility of the distribution network under a high proportion of distributed energy access. Attached Figure Description

[0098] The present invention will be further described below with reference to the accompanying drawings and embodiments:

[0099] Figure 1 This is a flowchart of the present invention for solving the two-layer optimization model of distribution network based on the improved bat algorithm for planning and operation coordination;

[0100] Figure 2 A schematic diagram of the topology of the IEEE 33-node test system for simulation verification;

[0101] Figure 3 This is a curve showing the change in electricity prices;

[0102] Figure 4 and Figure 5 Typical daily power output curves for wind and solar power systems in different seasons;

[0103] Figure 6 A schematic diagram comparing the convergence characteristic curves of different algorithms;

[0104] Figure 7 The node voltage distribution diagrams for different planning scenarios. Detailed Implementation

[0105] A wind-solar-storage collaborative planning method based on an improved bat algorithm includes the following steps:

[0106] Step 1: Construct a two-layer optimization model for the distribution network that coordinates planning and operation;

[0107] The upper layer of the distribution network dual-layer optimization model is a planning model, which aims to maximize the net benefit over the entire life cycle and optimize the site selection and capacity configuration of wind power, photovoltaic and energy storage equipment; the lower layer is an operation model, which, based on the planning results of the upper layer, aims to minimize the amount of abandoned electricity and network loss and coordinate the operation scheduling among wind power, photovoltaic and energy storage equipment.

[0108] In step 1, the upper-level planning model comprehensively considers electricity sales revenue, equipment investment, and operation and maintenance costs. Under the constraints of investment budget, capacity allocation, and stable grid operation, it optimizes the site selection and capacity allocation scheme of wind power, photovoltaic, and energy storage systems to maximize net revenue throughout the entire life cycle.

[0109] The objective function of the upper-level planning model is to maximize the net revenue over the entire life cycle, including electricity sales revenue and equipment life cycle costs within the planning period. The specific objective function of the planning model is as follows:

[0110] (1);

[0111] in, Indicates the time period within the planning cycle. t Revenue from electricity sales This represents the annualized net present value of the equipment investment cost. T For the planning period, The goal of the upper-level optimization is net profit.

[0112] Electricity sales revenue The calculation is as follows:

[0113] (2);

[0114] In the formula: For wind turbines i During the period t The power generation capacity; For photovoltaic units j During the period t The power generation capacity; For energy storage systems during time periods t The charging and discharging power, with positive values ​​for discharging and negative values ​​for charging; For time period t Electricity price; N and M These represent the total number of wind turbine units and photovoltaic units, respectively.

[0115] Equipment investment cost The calculation is as follows:

[0116] (3);

[0117] Equipment investment cost This is the sum of the total lifecycle costs of wind power, solar power, and energy storage, including initial investment costs and operation and maintenance costs, using the annual discount rate. Discount them all at once; i, j, e These represent wind turbines, photovoltaic systems, and energy storage equipment, respectively. , , These are 0-1 decision variables, representing the construction status of wind turbines, photovoltaics, and energy storage, respectively. A value of 1 indicates installation. , Wind turbine i With photovoltaic units j The unit investment cost; , , These refer to the service life of wind turbines, photovoltaic systems, and energy storage equipment, respectively. , , These are the annual operation and maintenance costs for wind turbines, photovoltaic systems, and energy storage equipment, respectively. , , These are the rated power of wind turbines, photovoltaic systems, and energy storage equipment, respectively. This refers to a collection of newly planned energy storage devices; Indicates the first k Investment cost per unit power capacity of energy storage equipment Indicates the first k Investment cost per unit energy capacity of energy storage equipment For the energy capacity of energy storage devices, Other initial investment costs for energy storage equipment.

[0118] The constraints of the upper-level planning model are expressed as follows:

[0119] The total investment cost of equipment during the planning period shall not exceed the budget ceiling, as specified below:

[0120] (4);

[0121] In the formula, This is the total investment cost of the equipment, which is the sum of the construction costs of wind power, photovoltaic, and energy storage equipment; The maximum allowed total investment limit;

[0122] The total installed capacity of distributed generation must be within a specific proportion of the system load to ensure stable grid operation. The constraint is expressed as follows:

[0123] (5);

[0124] In the formula, Pt For the system in time period t Total load power, This represents the ratio of distributed power generation to the total system load power.

[0125] The lower-level operation model is based on the site selection and capacity determination scheme provided by the planning layer. It comprehensively considers grid power balance, equipment output characteristics, node voltage constraints, and the dynamic operation characteristics of energy storage to determine the optimal coordination strategy between renewable energy and energy storage. The objective function of the lower-level operation model is to minimize the total operating cost, including wind and solar curtailment costs and grid loss costs. The specific objective function of the operation model is as follows:

[0126] (6);

[0127] In the formula, , These represent the unit costs of wind and solar power curtailment, respectively. , They are respectively t Time period i The wind turbine unit and the first j The amount of electricity wasted by each photovoltaic unit; For electricity price, This refers to the annual maximum load hours. L The total number of branch roads; branch road l electrical conductivity, , Branch roads i and branch roads j The node voltage amplitude, For nodes ij Phase angle difference between wind turbine units i and photovoltaic units j For the device index number, and in the node voltage i , j This refers to the physical node number of the power grid.

[0128] The constraints of the lower-level operational model include node power balance constraints, distributed generation output constraints, node voltage constraints, and energy storage system operational constraints, which are specifically expressed as follows:

[0129] Power balance constraint: The real-time power flow balance of the power grid must be maintained. The constraint is expressed as follows:

[0130] (7);

[0131] In the formula, For wind turbine i Rated power; For photovoltaic jRated power; This refers to the discharge power of the energy storage device. The charging power for energy storage devices; P L This refers to the system load power.

[0132] Distributed power generation output constraints: The real-time output of wind and solar power must meet their rated capacity limits and follow the output curve. The constraints are expressed as follows:

[0133] (8);

[0134] (9);

[0135] In the formula, , for t Actual output of wind turbines and solar power during specific time periods;

[0136] Node voltage constraints: i , j The physical nodes of the power grid are numbered, and the node voltage amplitude must meet the dynamic safe operating range. The constraints are as follows:

[0137] (10);

[0138] In the formula, for t Time period nodes i voltage amplitude, , for t Time period nodes i The upper and lower limits of the voltage amplitude.

[0139] Energy storage system operating constraints: The energy storage system must simultaneously satisfy the State of Charge (SOC) dynamic equation, capacity limits, and charge / discharge power limits, as expressed below:

[0140] (11);

[0141] (12);

[0142] (13);

[0143] The operation of energy storage systems must comply with strict dynamic constraints: among which For energy storage systems t State of charge during a period of time This represents the state of charge at the previous moment. for t Time-based charge and discharge power, and For charging and discharging efficiency; The time interval for updating the energy storage system status; and The maximum and minimum states of charge allowed by the system are the safe operating boundaries of the state of charge, preventing equipment lifespan degradation caused by overcharging or over-discharging; among the energy storage capacity parameters, For the planned energy storage system's rated capacity, Configure the actual capacity of the energy storage system to ensure that the system has the necessary redundancy to cope with load fluctuations.

[0144] Step 2: Solve the two-layer optimization model of the distribution network integrating planning and operation constructed in Step 1. The upper-layer planning model's problem falls under the integer linear programming category, so the commercial solver CPLEX is used. The upper-layer planning model outputs the node numbers of the selected wind power stations, photovoltaic stations, and energy storage locations, along with the optimal capacity for each node, which serves as input to the lower-layer operation model. The lower-layer operation model's problem involves nonlinearity and coupled variables, so an improved bat algorithm is used. The upper and lower layers interact and optimize through a "feedback iteration" method until convergence, ultimately obtaining the optimal planning scheme.

[0145] The upper-level model aims to maximize the net revenue over the entire lifecycle, using the site selection and capacity configuration of distributed wind power, photovoltaic, and energy storage systems as decision variables. This problem falls under the integer linear programming category, and therefore the commercial solver CPLEX is used. The upper-level planning model outputs the node numbers of the selected wind power stations, photovoltaic stations, and energy storage locations, along with the optimal capacity for each node, which serves as input to the lower-level operation model. The lower-level model aims to minimize abandoned power and grid losses, coordinating the operation and scheduling of wind power, photovoltaic, and energy storage devices. This problem involves nonlinearity and coupled variables, so an improved bat algorithm is used for solving it. The upper and lower levels interact and optimize through a "feedback iteration" method until convergence, ultimately obtaining the optimal planning scheme. Specific steps include:

[0146] Step 2.1 Initialize basic parameters; Initialize the basic parameters of the entire system, inputting grid structure parameters such as the number of nodes, branch topology, node voltage range, conductance, etc.; load parameters such as load time series data and annual maximum load hours for each node; wind and solar energy resource data; wind turbine, solar generator and energy storage equipment parameters; and electricity price parameters.

[0147] Step 2.2 uses the commercial solver CPLEX to solve the upper-level planning model. Its objective function, decision variables, and constraints are shown in formulas (1-5) in Step 1. The solver outputs the planning scheme obtained in this round, including: the node numbers of the selected wind turbines, photovoltaic units, and energy storage equipment, the optimal capacity of each node, and the economic allocation strategy.

[0148] Step 2.3 The site selection and sizing plan scheme obtained from the upper-level planning model is used as input data and passed to the lower-level running model.

[0149] Step 2.4 The lower-level operation model needs to optimize the operation control strategy based on the input current planning scheme to achieve the goal of minimizing wind and solar power curtailment and grid loss. Therefore, the improved bat algorithm is used to solve the problem.

[0150] Step 2.4.1 Initialize the basic parameters of the improved bat algorithm. Input the power grid structure parameters and the objective function of the lower-level operating model. F The corresponding formula (6) is used to calculate the fitness value of the objective function; the initialization settings for the bat population and algorithm parameters are specifically included: randomly generating N The initial position of the bat and speed ( i =1,2,…, N ), the location of each bat It is a D-dimensional vector representing a feasible planning strategy based on upper-level capacity configuration, and also a solution to the two-level optimization model of the distribution network that coordinates planning and operation; algorithm parameters, loudness , emit sound wave frequency Sound wave frequency range and Used to generate random factors Chaotic mapping control parameters Random walk parameters of Levy flight Pulse emission rate Chaotic inertial weights Range of inertia weight and and maximum number of iterations .

[0151] Step 2.4.2 Based on the objective function of the lower-level running model F Calculate the fitness value of each individual bat and mark the optimal location of the bats in the current population. The speed and position of the bat are iteratively updated according to the rules defined by the following formula.

[0152] For each bat i In the t iterations (t=1,2,…) Perform the following operations when:

[0153] (1) Update the sound wave frequency The original formula is as follows:

[0154] (14);

[0155] For the first i The frequency of sound waves emitted by an individual bat during this iteration; and The lower and upper limits of the set sound wave frequency; For interval Random numbers.

[0156] This invention considers sinusoidal chaotic mapping technology to improve the acoustic frequency update equation. Sinusoidal perturbations, with their finite amplitude and smooth changes, can enhance the search jump capability without introducing drastic fluctuations, thereby improving the global convergence of the bat algorithm. By superimposing sinusoidal perturbation terms in each iteration, the acoustic frequency acquires a certain degree of volatility, which can control the search jump amplitude and increase population diversity. Therefore, utilizing the non-repetitive, ergodic, and hybrid properties of chaotic mapping, sinusoidal chaotic mapping is used to optimize the bat's acoustic frequency. The update rule for each bat's acoustic frequency is as follows:

[0157] (15);

[0158] In the formula, Indicates the first t In the next iteration, the bat i The frequency of the sound wave; These are the control coefficients for the chaotic mapping.

[0159] (2) Update speed The original formula is as follows:

[0160] (16);

[0161] For the first i The bat in the t The velocity vector during round iteration; This is the velocity vector from the previous iteration; For the bat in the previous iteration i Location; This represents the current local optimum (position) within the population.

[0162] This invention considers chaotic inertia weights to improve the velocity update equation. During iteration, the Bat Algorithm is prone to getting trapped in local optima, often hindering its accurate identification of the global optimum. To address this issue, a chaotic inertia weight strategy is proposed to enhance the original velocity update mechanism. By introducing chaotic inertia weights, the algorithm's ability to avoid local optima is improved, thereby enhancing global search performance. The chaotic inertia weights generated in each iteration are represented as follows:

[0163] (17);

[0164] In the formula: , These represent the upper and lower limits of the inertia weight parameter, respectively. This represents the maximum total number of iterations. This represents the current iteration number. For in the interval The internally randomly generated chaotic perturbation factor.

[0165] A chaotic inertia weighting strategy is introduced into the velocity update equation. In each iteration, the chaotic inertia weight is calculated to enhance the dynamic adjustment of the bat velocity. The improved velocity update equation incorporating the chaotic inertia weight is expressed as follows:

[0166] (18);

[0167] in, The chaotic inertial weights change dynamically with iteration, thus dynamically adjusting the bat's flight speed.

[0168] (3) Update location For each bat i In the t iterations (t=1,2,…) When the location is updated, a solution to the new two-level optimization model of the distribution network that coordinates planning and operation is generated. The location update formula is as follows:

[0169] (19);

[0170] For the first i Only one bat t The position vector during round iteration.

[0171] This invention considers the Lévy flight strategy and improves the position update equation. Lévy flight introduces large-scale, sporadic jumps and frequent, abrupt changes in direction during the search process, helping individual bats avoid getting trapped in local optima and expanding the search space. Therefore, optimization performance in high-dimensional space is significantly improved, thereby enhancing the overall algorithm performance. This invention improves the bat algorithm by introducing the random walk characteristic of Lévy flight. After applying the Lévy flight strategy, the position update formula for each individual bat is as follows:

[0172] (20);

[0173] in For Lévy's random walk path; It is a gamma function; The scaling parameter determines the probability of large jumps in the step size, and is usually set to a value that corresponds to the scaling parameter. , The closer it is to 1, the greater the jump in the search; The closer it gets to 2, the more stable the jump amplitude becomes.

[0174] Step 2.4.3 Perform local search and update. To enhance the algorithm's local search capability in the solution space, a local perturbation is set to conduct a small-scale search on some individual bats, appropriately guiding the solution to converge towards the vicinity of the optimal solution.

[0175] Specifically, generating random numbers If satisfied If the current optimal bat is selected, an optimal individual is chosen, and a local solution is generated near the selected optimal individual using the following local perturbation formula (21). Otherwise, a new solution satisfying the objective function is updated according to the position update formula (20). The local perturbation formula is as follows:

[0176] (twenty one);

[0177] in, for Random numbers in an interval The average sound wave loudness of the entire bat colony during this iteration.

[0178] Step 2.4.4 Perform a global search and update. Generate random numbers. If satisfied If the fitness value of the objective function is greater than the new solution in step 2.4.3, then the solution is accepted, and the sound wave loudness is reduced according to the rules of equation (23-24). and increase pulse emission rate .

[0179] (1) Original sound wave loudness The updated formula is shown below:

[0180] (twenty two);

[0181] in, Let be the loudness of the sound wave of bat i at the t-th iteration; The loudness of bat i's sound wave in the next iteration; This is the sound wave loudness attenuation coefficient.

[0182] With improved sound wave frequency Similar to the updated equation, this invention considers sinusoidal chaotic mapping technology to improve sound wave loudness. Update equation:

[0183] (twenty three);

[0184] in, These are the control coefficients for the chaotic mapping.

[0185] (2) The pulse emissivity update equation is as follows:

[0186] (twenty four);

[0187] in, It is the pulse emission rate enhancement factor.

[0188] Step 2.4.5 compares the fitness values ​​of all individual bats, selects the best solution with the highest fitness value as the global optimum, and updates its position. The fitness values ​​of all bats in the population are sorted, and the position of the bat with the highest fitness value is found; this is the current best position, and thus the optimal solution. During the iteration process, this solution is the optimization result of the lower-level running model.

[0189] Step 2.4.6 determines whether the maximum number of iterations has been reached and outputs the optimal solution. If the maximum number of iterations has been reached, or if simulation tests various energy storage configuration schemes show that the relative error of the upper-level objective function between two consecutive iterations is less than 0.5%, the iteration process terminates, and the current wind-solar-storage configuration and its operating results are output, i.e., the solution of the desired two-layer optimization model for the distribution network that coordinates planning and operation. Otherwise, return to steps 2.4.2-2.4.5 for iterative calculation. The solution of this iteration, i.e., the lower-level optimization result, is input into the upper-level planning model and fed back to the planning layer to update the planning configuration nodes, power, and capacity, for a new round of iteration. Then, jump to step 2.2 to use the commercial solver CPLEX to complete the optimization calculation of the upper-level planning model.

[0190] The economic performance and system operation under different planning schemes were further analyzed. To evaluate the effectiveness of the proposed method, the following three sets of comparisons were designed:

[0191] Scenario 1: Only distributed generation is considered, without configuring an energy storage system;

[0192] Scenario 2: Considering the configuration of distributed generation and energy storage systems simultaneously, the standard Bat algorithm is used for solution;

[0193] Scenario 3: Considering the configuration of distributed generation and energy storage systems simultaneously, an improved bat algorithm is used to solve the problem;

[0194] Example:

[0195] like Figure 1 As shown, the method proposed in this invention was simulated and verified using the IEEE 33-node test system, and its topology is as follows. Figure 2 As shown. The relevant parameters for energy storage systems and distributed power sources are listed in Tables 1 and 2, respectively. Figure 3 This is the electricity price change curve. Figure 4 and Figure 5 Typical daily power output curves for wind and solar power systems in different seasons. Figure 6 A comparison of the convergence characteristic curves of different algorithms. Figure 7 The node voltage distribution for different planning scenarios.

[0196] Table 1 Energy Storage System Parameters

[0197]

[0198] Table 2 Parameters of Distributed Generation System

[0199]

[0200] In typical operating scenarios during spring and autumn, the convergence of the objective function indicates that the model parameters have reached their optimal state. Figure 6 For the iterative process of different algorithms, the Particle Swarm Optimization (PSO) algorithm converges to the optimal solution on the 39th iteration, the standard Bat Algorithm requires 28 iterations, while the improved Bat Algorithm converges in only 17 iterations with higher convergence accuracy. Analysis shows that the Bat Algorithm outperforms the PSO algorithm, mainly due to dynamic frequency adjustment and local perturbation. By enhancing the jump capability of the solution space, it effectively avoids premature convergence in high-dimensional non-convex scheduling problems, verifying the excellent adaptability of the Bat Algorithm in the optimization of complex energy systems. The improved Bat Algorithm employs chaotic mapping technology to optimize the pulse frequency, amplitude, and inertia weight of individual bats, enhancing the algorithm's global search capability and preventing iterative processes from getting trapped in local optima and affecting the global result. Furthermore, the Levy Flight strategy is introduced to expand the search space, thereby improving the algorithm's convergence speed and solution accuracy. Table 3 compares the configuration results of wind power, photovoltaics, and energy storage under different scenarios.

[0201] Table 3 Wind-Solar-Storage Resource Allocation Schemes in Different Scenarios

[0202]

[0203] As shown in Table 3, in Scenario 1 without energy storage, wind turbines with capacities of 635kW and 350kW were installed at nodes 15 and 17, respectively, and 180kW photovoltaic units were configured at the same nodes. The comparison shows that Scenario 2 and Scenario 3 have higher wind and photovoltaic installed capacities, indicating that the configuration of the energy storage system effectively improves the system's ability to absorb distributed energy. Therefore, reasonable configuration of energy storage helps to improve the utilization rate and absorption level of distributed renewable energy.

[0204] In scenarios 2 and 3, wind power and solar power are installed at nodes 9, 15, and 24, respectively, but the planned capacity configurations differ because the methods used to solve the lower-level models in the two scenarios are different. Different algorithms produce different optimal operating strategies, which in turn feed back to the upper-level models, affecting the planned capacity configuration results. The economic comparison analysis results for each planning scenario are shown in Table 4.

[0205] Table 4. Economic Comparison Analysis under Different Planning Scenarios

[0206]

[0207] As shown in Table 4, Scenario 1 has the lowest total cost, mainly because it lacks an energy storage system, resulting in a relatively low installed capacity for wind and solar power, thus minimizing initial investment costs. Correspondingly, its operating costs and annual electricity sales revenue are also lower. In contrast, Scenario 2 and Scenario 3, which include energy storage systems, show significantly reduced network loss costs, decreasing by 25.7% and 32.9% respectively compared to Scenario 1. Analysis shows that the coordinated operation of wind, solar, and energy storage effectively reduces network losses. Although configuring energy storage increases initial investment costs, the operational benefits in reducing network losses and curtailment can gradually offset these additional costs over the long term, leading to a decrease in total system cost. Furthermore, although the investment cost of Scenario 3 is 2.2% higher than that of Scenario 2, its lower operating costs ultimately reduce the total system cost by 4.1%. Analysis shows that Scenario 3, with its higher capacity wind, solar, and energy storage, while increasing investment costs, effectively reduces operating costs through optimized operating strategies. Simultaneously, the increased installed capacity boosts power generation, resulting in a 4.9% increase in annual electricity sales revenue for Scenario 3 compared to Scenario 2. Therefore, from a long-term planning perspective, Scenario 3 is more economical. The improved bat algorithm effectively enhances the system's capacity to absorb distributed energy, demonstrating the effectiveness of the proposed bi-level planning model and its solution method. The wind and solar energy utilization rates under the three planning scenarios are shown in Table 5.

[0208] Table 5 Renewable Energy Utilization Efficiency under Different Planning Scenarios

[0209]

[0210] As shown in Table 5, in Scenario 1, the utilization rates of photovoltaic (PV) and wind power are 83.67% and 87.23%, respectively. In Scenario 2, by configuring an energy storage system, the utilization rates of PV and wind power are increased to 85.32% and 89.45%, respectively. In Scenario 3, after adopting the improved Bat Algorithm, the utilization rate of PV further increases to 90.27%, and the utilization rate of wind power reaches 92.18%. The data shows that the reasonable configuration of energy storage systems can significantly reduce wind and solar curtailment, and the improved Bat Algorithm can coordinate source-storage operation strategies, thereby improving the grid connection efficiency of renewable energy.

[0211] The integration of distributed generation typically increases the voltage at various nodes in a distribution network, while the configuration of energy storage systems further improves voltage distribution characteristics. The node voltage distribution under three planning scenarios is as follows: Figure 7 As shown in the figure. The results show that the system has the best voltage regulation performance after adopting the improved bat algorithm. In Scenario 1 without energy storage system, the lowest node voltage is 0.908 pu; in Scenario 2 using the standard bat algorithm, the lowest node voltage is increased to 0.936 pu; in Scenario 3 using the improved bat algorithm, this value is further increased to 0.937 pu.

[0212] The above results demonstrate that the proposed improved bat algorithm can effectively improve system voltage stability and power quality, and is particularly suitable for distribution networks with large-scale distributed power sources, thereby enhancing the reliability of power grid operation.

[0213] Based on the planning scheme of Scenario 3, a numerical example quantitatively analyzes the impact of prosumer behavior on system performance. Three trading modes are set up for comparative study: Mode 1 is a scenario without prosumer participation, where all electricity demand is supplied by the main grid; Mode 2 is a standard scenario, where prosumers exist but their participation is limited, and demand is met only through grid purchases; Mode 3 is an optimized prosumer-consumption model, where prosumers not only consume photovoltaic power locally but also regulate load through energy storage systems and participate in point-to-point electricity trading based on the planning scheme. The comparison results of the revenue under different trading modes are shown in Table 6.

[0214] Table 6. Comparison of revenues under different power generation trading models;

[0215]

[0216] Table 6 shows that under trading mode 1 without producer-consumer participation, the distribution network has the highest electricity sales revenue, but the operating revenue of distributed generation and energy storage systems is the lowest, and the user's electricity purchase cost is the highest. Compared with mode 1, the daily net revenue of the distribution network in modes 2 and 3 decreases by 61.3% and 33.4% respectively, while the operating revenue of distributed generation and energy storage systems increases by 76.7% and 58.7% respectively. The user's electricity purchase cost in modes 2 and 3 is reduced by 6.7% and 6.9% respectively compared with mode 1. Analysis shows that mode 2 has the lowest electricity sales revenue for the distribution network, but significantly increases the operating revenue of distributed generation and energy storage systems, and reduces the user's electricity purchase cost. In contrast, although the distribution network's electricity sales revenue increases in mode 3, the operating revenue of distributed generation and energy storage systems decreases. Analysis shows that distributed energy trading can effectively promote the consumption of renewable energy, improve the operating efficiency of distributed generation and energy storage systems, and ensure user revenue. Mode 3, through collaborative optimization of the electricity price structure, increases the revenue of the distribution network, reduces the user's electricity purchase cost, and achieves a win-win situation for operators and electricity consumers.

[0217] This invention proposes a wind-solar-storage collaborative planning method based on an improved bat algorithm. Through collaborative optimization of upper and lower layer models, it achieves efficient operation of distributed power sources and energy storage. Simulation results show that configuring an energy storage system can significantly reduce wind and solar power curtailment, thereby effectively reducing system operating costs; the collaborative configuration of wind, solar, and energy storage resources effectively reduces distribution system network losses. Simultaneously, the improved bat algorithm demonstrates significant advantages in improving the solution performance of the two-layer model. Compared with traditional heuristic algorithms, the improved bat algorithm can solve complex source-storage planning problems more efficiently, effectively improving global search capabilities, accelerating convergence speed, and avoiding local optima.

Claims

1. A wind-solar-storage collaborative planning method based on an improved bat algorithm, characterized in that, Includes the following steps: Step 1: Construct a two-layer optimization model for the distribution network that coordinates planning and operation; The upper layer of the two-layer optimization model for the power distribution network is a planning model, which aims to maximize the net revenue throughout the entire life cycle and optimize the site selection and capacity configuration of wind power, photovoltaic and energy storage equipment. The lower layer is the operation model, which, based on the planning results received from the upper layer, coordinates the operation and scheduling between wind power, photovoltaic and energy storage equipment with the goal of minimizing abandoned electricity and grid loss. Step 2: Solve the two-layer optimization model of the distribution network that coordinates planning and operation constructed in Step 1, and finally obtain the optimal planning scheme; The objective function of the upper-level planning model is to maximize the net revenue over the entire life cycle, including electricity sales revenue and equipment life cycle costs within the planning period. The specific objective function of the planning model is as follows: （1）； in, Indicates the time period within the planning cycle. t Revenue from electricity sales This represents the annualized net present value of the equipment investment cost. T For the planning period, The goal of optimization at the higher level is net profit; Electricity sales revenue The calculation is as follows: （2）； In the formula: For wind turbines i During the period t The power generation capacity; For photovoltaic units j During the period t The power generation capacity; For energy storage systems during time periods t The charging and discharging power, with positive values ​​for discharging and negative values ​​for charging; For time period t Electricity price; N and M These represent the total number of wind turbine units and photovoltaic units, respectively. Equipment investment cost The calculation is as follows: （3）； Equipment investment cost This is the sum of the total lifecycle costs of wind power, solar power, and energy storage, including initial investment costs and operation and maintenance costs, using the annual discount rate. Discount them all at once; i, j, e These represent wind turbines, photovoltaic systems, and energy storage equipment, respectively. , , These are 0-1 decision variables, representing the construction status of wind turbines, photovoltaics, and energy storage, respectively. A value of 1 indicates installation. , Wind turbine i With photovoltaic units j The unit investment cost; , , These refer to the service life of wind turbines, photovoltaic systems, and energy storage equipment, respectively. , , These are the annual operation and maintenance costs for wind turbines, photovoltaic systems, and energy storage equipment, respectively. , , These are the rated power of wind turbines, photovoltaic systems, and energy storage equipment, respectively. This refers to a collection of newly planned energy storage devices; Indicates the first k Investment cost per unit power capacity of energy storage equipment Indicates the first k Investment cost per unit energy capacity of energy storage equipment For the energy capacity of energy storage devices, Other initial investment costs for energy storage equipment.

2. The method according to claim 1, characterized in that, In step 1, the upper-level planning model comprehensively considers electricity sales revenue, equipment investment, and operation and maintenance costs. It optimizes the site selection and capacity allocation scheme of wind power, photovoltaic, and energy storage systems under constraints such as investment budget, capacity allocation, and grid stability, so as to maximize the net revenue throughout the entire life cycle.

3. The method according to claim 2, characterized in that, The constraints of the upper-level planning model are expressed as follows: The total investment cost of equipment during the planning period shall not exceed the budget ceiling, as specified below: （4）； In the formula, This is the total investment cost of the equipment, which is the sum of the construction costs of wind power, photovoltaic, and energy storage equipment; The maximum allowed total investment limit; The total installed capacity of distributed generation must be within a specific proportion of the system load to ensure stable grid operation. The constraint is expressed as follows: （5）； In the formula, P t For the system in time period t Total load power, This represents the ratio of distributed power generation to the total system load power.

4. The method according to claim 1, characterized in that, In step 1, the lower-level operation layer model, based on the site selection and capacity determination scheme provided by the planning layer, comprehensively considers grid power balance, equipment output characteristics, node voltage constraints, and dynamic operation characteristics of energy storage to determine the optimal coordination strategy between renewable energy and energy storage. The objective function of the lower-level operation layer model is to minimize the total operating cost, including the costs of wind and solar curtailment and grid loss. The specific objective function of the operation model is as follows: （6）； In the formula, , These represent the unit costs of wind and solar power curtailment, respectively. , They are respectively t Time period i The wind turbine unit and the first j The amount of electricity wasted by each photovoltaic unit; For electricity price, This refers to the annual maximum load hours. L The total number of branch roads; branch road l electrical conductivity, , Branch roads i and branch roads j The node voltage amplitude, For nodes ij Phase angle difference between wind turbine units i and photovoltaic units j For the device index number, and in the node voltage i , j This refers to the physical node numbering of the power grid.

5. The method according to claim 4, characterized in that, The constraints of the lower-level operational model include node power balance constraints, distributed generation output constraints, node voltage constraints, and energy storage system operational constraints, which are specifically expressed as follows: Power balance constraint: The real-time power flow balance of the power grid must be maintained. The constraint is expressed as follows: （7）； In the formula, For wind turbine i Rated power; For photovoltaic j Rated power; This refers to the discharge power of the energy storage device. The charging power for energy storage devices; P L This refers to the system load power. Distributed power generation output constraints: The real-time output of wind and solar power must meet their rated capacity limits and follow the output curve. The constraints are expressed as follows: （8）； （9）； In the formula, , for t Actual output of wind turbines and solar power during specific time periods; Node voltage constraints: i , j The physical nodes of the power grid are numbered, and the node voltage amplitude must meet the dynamic safe operating range. The constraints are as follows: （10）； In the formula, for t Time period nodes i voltage amplitude, , for t Time period nodes i The upper and lower limits of the voltage amplitude at the point; Energy storage system operating constraints: The energy storage system must simultaneously satisfy the State of Charge (SOC) dynamic equation, capacity limits, and charge / discharge power limits, as expressed below: （11）； （12）； （13）； The operation of energy storage systems must comply with strict dynamic constraints: among which For energy storage systems t State of charge during a period of time The state of charge at the previous moment. for t Time-based charge and discharge power, and For charging and discharging efficiency; The time interval for updating the energy storage system status; and The maximum and minimum states of charge allowed by the system are the safe operating boundaries of the state of charge, preventing equipment lifespan degradation caused by overcharging or over-discharging; among the energy storage capacity parameters, For the planned energy storage system's rated capacity, Configure the actual capacity of the energy storage system to ensure that the system has the necessary redundancy to cope with load fluctuations.

6. The method according to claim 1, 2, 4, or 5, characterized in that, In step 2, the upper-level planning model takes maximizing the net revenue over the entire life cycle as the objective function, and the site selection and capacity configuration of distributed wind power, photovoltaic and energy storage systems as decision variables. An integer linear programming solver is used to solve the upper-level planning model to obtain the node numbers of the selected wind power stations, photovoltaic stations and energy storage locations and the optimal capacity corresponding to each node; and the obtained optimal capacity is used as the input of the lower-level operation layer model.

7. The method according to claim 6, characterized in that, The lower-level operation layer model takes minimizing abandoned power and grid loss as its objective function, coordinates the operation scheduling among wind power, photovoltaic and energy storage equipment, and uses an improved bat algorithm to solve the lower-level operation layer model; the upper-level planning model and the lower-level operation layer model interact and optimize each other through feedback iteration until convergence, and finally obtain the optimal planning scheme.

8. The method according to claim 7, characterized in that, Step 2 specifically includes the following steps: Step 2.1: Initialize the system's basic parameters. The input basic parameters include: power grid structure parameters, load parameters, wind and solar energy resource data, wind turbine parameters, solar turbine parameters, energy storage device parameters, and electricity price parameters. Among them, the power grid structure parameters include the number of nodes, branch topology, node voltage range, and conductance; the load parameters include the load time series data of each node and the annual maximum load hours. Step 2.2: Use the solver CPLEX to solve the upper-level planning model. Its objective function, decision variables, and constraints are shown in formulas (1) to (5). The solver outputs the planning scheme solved in this round, including: the node numbers of the selected wind turbines, photovoltaic units, and energy storage equipment, the optimal capacity of each node, and the economic allocation strategy. Step 2.3: Use the site selection and capacity planning scheme solved by the upper-level planning model as input data and pass it to the lower-level running layer model; Step 2.4: Based on the input current planning scheme, use the improved bat algorithm to solve the lower-level operation layer model and optimize the operation control strategy to minimize wind and solar power curtailment and grid loss.

9. The method according to claim 8, characterized in that, In step 2.4, the improved bat algorithm is used to solve the problem, including the following steps: Step 2.4.1 Initialize the basic parameters of the improved bat algorithm; input the power grid structure parameters and the objective function of the lower-level operation layer model. F The corresponding formula (6) is used to calculate the fitness value of the objective function; the initialization settings for the bat population and algorithm parameters are specifically included: randomly generating N The initial position of the bat and speed ,in i =1,2,…, N The location of each bat It is a D-dimensional vector representing a feasible planning strategy based on upper-level capacity configuration, and also a solution to the two-level optimization model of the distribution network that coordinates planning and operation; algorithm parameters, loudness , emit sound wave frequency Sound wave frequency range and Used to generate random factors Chaotic mapping control parameters Random walk parameters of Levy flight Pulse emission rate Chaotic inertial weights Range of inertia weight and and maximum number of iterations ; Step 2.4.2 Based on the objective function of the lower-level running layer model F Calculate the fitness value of each individual bat and mark the optimal location of the bats in the current population. The speed and position of the bat are iteratively updated according to the rules defined by the following formula; For each bat i In the t Perform the following operations during the next iteration: (1) Update the sound wave frequency The original formula is as follows: （14）； For the first i The frequency of sound waves emitted by an individual bat during this iteration; and The lower and upper limits of the set sound wave frequency; For interval Random numbers; Optimize the sound wave frequencies of bats; the update rules for the sound wave frequencies of each bat are as follows: （15）； In the formula, Indicates the first t In the next iteration, the bat i The frequency of the sound wave; These are the control coefficients for the chaotic mapping; (2) Update speed The original formula is as follows: （16）； For the first i The bat in the t The velocity vector during round iteration; This is the velocity vector from the previous iteration; For the bat in the previous iteration i Location; The position is the current local optimum in the population. By introducing chaotic inertia weights, the algorithm's ability to avoid local optima is improved, thereby enhancing global search performance. The chaotic inertia weights generated in each iteration are represented as follows: （17）； In the formula: , These represent the upper and lower limits of the inertia weight parameter, respectively. This represents the maximum total number of iterations. This represents the current iteration number. For in the interval Internally randomly generated chaotic perturbation factor; A chaotic inertial weight strategy is introduced into the velocity update equation; in each iteration, the chaotic inertial weight is calculated to enhance the dynamic adjustment of the bat velocity; the improved velocity update equation including the chaotic inertial weight is expressed as follows: （18）； in, The chaotic inertial weights change dynamically with iteration, thus dynamically adjusting the bat's flight speed. (3) Update location For each bat i In the t In the next iteration, a location update is performed, generating a solution for a new two-layer optimization model of the distribution network that coordinates planning and operation. The original location update formula is as follows: （19）； For the first i Only one bat t The position vector during round iteration; The bat algorithm is improved by incorporating the random walk characteristic of Lévy flight; after applying the Lévy flight strategy, the update formula for the position of each individual bat is as follows: （20）； in For Lévy's random walk path; It is a gamma function; The scaling parameter determines the probability of large jumps in the step size; Step 2.4.3 Perform local search and update; To enhance the algorithm's local search capability in the solution space, local perturbation is set to conduct a small-scale search on some individual bats, appropriately guiding the solution to converge toward the vicinity of the optimal solution; Specifically, generating random numbers If satisfied If the current optimal bat is selected, an optimal individual is chosen, and a local solution is generated near the selected optimal individual using the following local perturbation formula (21). Otherwise, a new solution satisfying the objective function is updated according to the position update formula (20). The local perturbation formula is as follows: （21）； in, for Random numbers in an interval The average sound wave loudness of the entire bat colony during this iteration; Step 2.4.4 Perform a global search and update; generate random numbers. If satisfied If the fitness value of the objective function is greater than the new solution in step 2.4.3, then the solution is accepted, and the sound wave loudness is reduced according to the rules of equation (23-24). and increase pulse emission rate ; (1) Original sound wave loudness The updated formula is shown below: （22）； in, Let be the loudness of the sound wave of bat i at the t-th iteration; The loudness of bat i's sound wave in the next iteration; This is the sound wave loudness attenuation coefficient; With improved sound wave frequency Similarly, the updated equation improves the loudness of the sound wave. Update equation: （23）； in, These are the control coefficients for the chaotic mapping; (2) The pulse emissivity update equation is as follows: （24）； in, It is the pulse emission rate enhancement factor; Step 2.4.5 Select the best solution with the highest fitness as the global optimum and update its position; sort the fitness values ​​of all bat individuals in the population, find the position of the bat with the highest fitness value, which is the current best position, and thus the optimal solution. During the iteration process, this solution is the optimization result of the lower-level running layer model; Step 2.4.6 determines whether the maximum number of iterations has been reached and outputs the optimal solution. If the maximum number of iterations has been reached, or the global optimal solution has not significantly improved in multiple consecutive iterations, the iteration is terminated, and the current wind-solar-storage configuration and its operating results are output, i.e., the solution of the two-layer optimization model of the distribution network with coordinated planning and operation. Otherwise, it returns to steps 2.4.2-2.4.5 for iterative calculation. The solution of this iteration, i.e. the lower-level optimization result, is input into the upper-level planning model and fed back to the planning layer to update the planning configuration nodes, power and capacity, and a new round of iteration is carried out. Then, it jumps to step 2.2 to use the CPLEX solver to complete the optimization calculation of the upper-level planning model.

Images