A multi-objective optimization-based double-layer optimization scheduling method for power distribution network with high proportion of wind and solar power

Abstract

The application discloses a kind of based on multi-objective optimization's high proportion wind light containing distribution network double-layer optimization scheduling method, belong to distribution network optimization scheduling technical field.The method includes: the double-layer optimization scheduling model is established, the minimum network loss and power quality optimum are as target in upper layer, the minimum sum of purchase power cost, power selling cost, operating cost, demand response cost and wind and light abandonment cost is as target in lower layer;Multi-objective octopus optimization algorithm is used to solve upper layer model, and after obtaining pareto frontier, only solution is selected by approximation ideal solution sorting method;Robust Benders decomposition algorithm is used to decompose original problem into main problem and subproblem for iterative solution, and double-layer optimal scheduling strategy is obtained.The application considers the security and economy of distribution network operation, effectively improves new energy consumption capacity, and is suitable for high proportion wind light access distribution network optimization scheduling.

Description

Technical Field

[0001] This invention belongs to the field of power distribution network optimization and scheduling technology, specifically involving a two-layer optimization scheduling method for power distribution networks with a high proportion of wind and solar power based on multi-objective optimization. Background Technology

[0002] With the continuous increase in the penetration rate of new energy sources in the power system, the access capacity of large-scale distributed photovoltaic and wind power generation is constantly increasing, making the random fluctuations and power imbalances of the distribution network increasingly prominent, posing a severe challenge to the structure and operation of the traditional distribution network. At the same time, affected by weather factors, the inherent intermittency and uncertainty of new energy power generation further exacerbate voltage fluctuations in the power grid, leading to a decline in power quality, which in turn adversely affects the normal power consumption of power equipment and end users within the distribution network.

[0003] In distribution networks with a high proportion of wind and solar power generation, how to formulate scientific and effective optimization dispatch strategies to improve the stability of grid operation and the reliability of power supply, reduce system operating costs, and enhance the capacity for renewable energy absorption has become a current research hotspot and challenge.

[0004] Most existing distribution network dispatching methods only consider a single economic objective or focus solely on power quality, making it difficult to coordinate multiple conflicting objectives. Furthermore, traditional optimization algorithms suffer from slow convergence, susceptibility to local optima, and inability to effectively handle the coupling of two-level variables when facing high-dimensional nonlinear, multi-constraint, and uncertain bi-level optimization problems. Therefore, there is an urgent need for an intelligent dispatching method that can balance network losses, power quality, and operational economics, and effectively handle the uncertainties of wind and solar power. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a two-layer optimization scheduling method for distribution networks with a high proportion of wind and solar power based on multi-objective optimization. This method constructs a two-layer optimization model and uses the multi-objective octopus optimization algorithm and robust Benders decomposition algorithm to solve it, which can effectively coordinate the security, power quality and economy of the distribution network and improve the renewable energy absorption capacity.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a two-layer optimization scheduling method for a distribution network with a high proportion of wind and solar power based on multi-objective optimization, comprising the following steps:

[0007] Step S1: Establish a two-layer optimal scheduling model for a distribution network with a high proportion of wind and solar power; the two-layer optimal scheduling model includes an upper-layer optimization model and a lower-layer optimization model;

[0008] The upper-level optimization model takes minimizing network loss and optimizing power quality of the distribution network as its primary optimization objectives, and sets upper-level constraints; the power quality evaluation indicators include node voltage offset level and static voltage stability.

[0009] The lower-level optimization model takes the optimal operation economy of the distribution network as the second optimization objective and sets lower-level constraints; the operation economy of the distribution network includes electricity purchase cost, electricity sales cost, operating cost, demand response cost, and wind and solar curtailment cost;

[0010] Step S2: Use the multi-objective octopus optimization algorithm to solve the upper-level optimization model and obtain the upper-level optimal solution set;

[0011] Step S3: Use the robust Benders decomposition algorithm to solve the two-level optimization problem consisting of the upper-level optimization model and the lower-level optimization model to obtain the two-level optimal scheduling strategy;

[0012] Step S4: Output the two-layer optimal scheduling strategy.

[0013] Further, in step S1, the objective function of the upper-level optimization model is:

[0014]

[0015] in, The objective function of the upper-level optimization model; This is the network loss function; Let the power quality level function be expressed as:

[0016]

[0017] in, As the voltage offset level weight, It is a voltage offset level function. As the weight for static voltage stability, This is a static voltage stability function;

[0018] The upper-level constraints include: voltage drop constraints, node active power balance constraints, node reactive power balance constraints, current definition constraints, and distribution network voltage and current limit constraints.

[0019] Further, in step S1, the objective function of the lower-level optimization model is:

[0020]

[0021] in, The objective function of the lower-level optimization model; Let t be the cost of electricity purchased by the distribution network from the main grid; Let t be the cost of electricity sold from the distribution network to the main grid. Let t be the operating cost of the distribution network at time t; Let t be the demand response cost of the flexible load in the distribution network. Let t be the cost of wind and solar power curtailment in the distribution network at time t;

[0022] The lower-level constraints include: uncertainty constraints on new energy output, power interaction constraints between the main grid and distribution grid, and demand response constraints.

[0023] Further, step S2 specifically includes:

[0024] Step S21: Represent the upper-level optimization model as a standard multi-objective optimization problem form that includes a dual objective function and constraints;

[0025] Step S22: The multi-objective octopus optimization algorithm is used to solve the problem. The multi-objective octopus optimization algorithm simulates the global search behavior, encirclement behavior and attack behavior of octopuses in the process of foraging to update the position of individuals in the population, and saves the non-dominated solution in the iteration process through external archiving to obtain the Pareto front.

[0026] Step S23: Select a unique solution from the Pareto front using the approximation ideal solution sorting method as the optimal solution of the upper-level optimization model, and pass the upper-level decision variables in the optimal solution to the lower-level optimization model.

[0027] Furthermore, in step S22, the implementation method of the multi-objective octopus optimization algorithm is as follows:

[0028] Initialize population parameters;

[0029] Assess the fitness of each individual in the population and determine Pareto dominance;

[0030] Update the external archive and use crowding distance sorting to maintain the diversity of solutions within the archive;

[0031] The leader is selected from an external archive based on a roulette wheel selection method.

[0032] Based on the comparison between the random number and the preset probability, the global search behavior, encirclement behavior or attack behavior are executed respectively to update the individual position;

[0033] Determine if the iteration termination condition has been met; if so, output the Pareto front.

[0034] Furthermore, in step S23, selecting a unique solution using the approximation-ideal-solution sorting method specifically includes:

[0035] Construct a decision matrix and perform normalized weighting on the matrix elements;

[0036] Determine the positive and negative ideal solutions for each objective;

[0037] Calculate the Euclidean distance between each solution in the Pareto front and the positive and negative ideal solutions;

[0038] Calculate the degree of closeness between each solution and the ideal solution, and select the solution with the highest degree of closeness as the optimal solution of the upper-level optimization model.

[0039] Furthermore, step S3 specifically includes:

[0040] Step S31: Organize the lower-level optimization model into a compact form;

[0041]

[0042] in, The objective function of the lower-level optimization model is... Here, y represents the objective function of the compact lower-level optimization model, and y represents the lower-level decision variables. This represents all inequality constraints in the lower-level model. This represents all equality constraints in the lower-level model;

[0043] Step S32: Use the robust Benders decomposition algorithm to iteratively solve the problem, decomposing the original problem into a main problem and sub-problems;

[0044] The main problem is a multi-objective optimization model that incorporates the generated Benders cut plane. Solving this model yields the values ​​of the upper-level decision variables.

[0045] The subproblem is a lower-level robust linear programming model after substituting the values ​​of the upper-level decision variables, used to solve the lower-level optimal solution under the worst-case scenario of new energy output.

[0046] Generate optimality cutting planes or feasibility cutting planes based on the solutions to the subproblems, and add them to the main problem;

[0047] Repeat the iteration until the convergence condition is met, and output the optimal solution of the two layers.

[0048] Furthermore, after step S4, the method further includes:

[0049] Step S5: Perform simulation comparison analysis between the generated scheduling strategy and the comparison strategy; the comparison strategy includes the proposed strategy, demand-free response, and basic strategy; the comparison content includes: node voltage, wind power and photovoltaic absorption, network loss, load before and after demand response, and various operating costs of the distribution network.

[0050] The present invention also provides a computer device, comprising: at least one processor, at least one memory, and computer program instructions stored in the memory, which implement the above-described method when executed by the processor.

[0051] The present invention also provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the above-described method.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] 1. Balancing Safety and Economy: This invention constructs a two-layer optimization model. The upper layer aims to minimize network losses and optimize power quality to ensure the safe and stable operation of the distribution network; the lower layer aims to minimize operating costs to ensure economy. The upper and lower layers iteratively solve the problem, and the final scheduling scheme can effectively balance the contradiction between power grid safety and operational economy.

[0054] 2. Strong Multi-Objective Solving Capability: This invention employs a multi-objective octopus optimization algorithm to solve upper-level multi-objective problems. This algorithm simulates three predation behaviors of an octopus, exhibiting excellent global exploration and local exploitation capabilities. Combined with external archiving and non-dominated sorting, it can obtain a uniformly distributed Pareto front. Furthermore, the TOPSIS method is used to automatically select the optimal compromise solution from the Pareto front, avoiding biases caused by subjective weighting.

[0055] 3. Effectively handles uncertainties in new energy sources: In the lower-level optimization, this invention adopts the robust Benders decomposition algorithm to decompose the original problem into a main problem and sub-problems. Through cutting plane iteration, it can cope with the worst-case scenario of distributed photovoltaic and wind power output, and improve the robustness of the scheduling strategy in uncertain environments. Attached Figure Description

[0056] Figure 1 A flowchart illustrating the implementation of a two-layer optimization scheduling method for a high-proportion wind and solar power distribution network based on multi-objective optimization, as provided in an embodiment of the present invention.

[0057] Figure 2 This is a comparison curve of photovoltaic power consumption in an embodiment of the present invention;

[0058] Figure 3 This is a comparison curve of wind power consumption in an embodiment of the present invention;

[0059] Figure 4 This is a comparison chart of load curves before and after demand response in an embodiment of the present invention;

[0060] Figure 5 This is a comparison curve of network loss over 24 hours for different strategies in an embodiment of the present invention;

[0061] Figure 6 This is a full node voltage distribution diagram at 21:00 in an embodiment of the present invention. Detailed Implementation

[0062] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0063] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0064] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0065] like Figure 1 As shown in the figure, this embodiment provides a two-layer optimization scheduling method for a distribution network with a high proportion of wind and solar power based on multi-objective optimization, and its implementation steps are as follows.

[0066] Step S1: Establish a two-layer optimal scheduling model for a distribution network with a high proportion of wind and solar power; the two-layer optimal scheduling model includes an upper-layer optimization model and a lower-layer optimization model.

[0067] The upper-level optimization model takes minimizing network loss and optimizing power quality of the distribution network as its primary optimization objectives and sets upper-level constraints. The power quality evaluation indicators include node voltage offset level and static voltage stability.

[0068] The lower-level optimization model takes the optimal operation economy of the distribution network as the second optimization objective and sets lower-level constraints; the operation economy of the distribution network includes electricity purchase cost, electricity sales cost, operating cost, demand response cost, and wind and solar curtailment cost.

[0069] Step S2: Use the multi-objective octopus optimization algorithm to solve the upper-level optimization model and obtain the upper-level optimal solution set.

[0070] Step S21: Represent the upper-level optimization model as a standard multi-objective optimization problem form that includes a dual objective function and constraints.

[0071] Step S22: The multi-objective octopus optimization algorithm is used to solve the problem. The multi-objective octopus optimization algorithm simulates the global search behavior, encirclement behavior and attack behavior of octopuses in the process of foraging to update the position of individuals in the population, and saves the non-dominated solutions in the iteration process through external archiving to obtain the Pareto front.

[0072] The implementation method of the multi-objective octopus optimization algorithm is as follows:

[0073] 1) Initialize population parameters;

[0074] 2) Assess the fitness of each individual in the population and determine Pareto dominance;

[0075] 3) Update external archives and use congestion distance sorting to maintain the diversity of solutions within the archive;

[0076] 4) Select individual leaders from external archives based on the roulette wheel method;

[0077] 5) Based on the comparison between the random number and the preset probability, perform global search, encirclement, or attack actions respectively to update the individual's position;

[0078] 6) Determine if the iteration termination condition has been met. If so, output the Pareto front.

[0079] Step S23: Select a unique solution from the Pareto front using the approximation ideal solution sorting method as the optimal solution of the upper-level optimization model, and pass the upper-level decision variables in the optimal solution to the lower-level optimization model.

[0080] The method of selecting a unique solution using the approximation of the ideal solution sorting method specifically includes:

[0081] 1) Construct a decision matrix and perform normalization and weighting on the matrix elements;

[0082] 2) Determine the positive and negative ideal solutions for each objective;

[0083] 3) Calculate the Euclidean distance between each solution in the Pareto front and the positive and negative ideal solutions;

[0084] 4) Calculate the closeness of each solution to the ideal solution, and select the solution with the highest closeness as the optimal solution of the upper-level optimization model.

[0085] Step S3: Use the robust Benders decomposition algorithm to solve the two-layer optimization problem consisting of the upper-layer optimization model and the lower-layer optimization model to obtain the two-layer optimal scheduling strategy.

[0086] Step S31: Organize the lower-level optimization model into a compact form.

[0087] Step S32: The robust Benders decomposition algorithm is used for iterative solution, decomposing the original problem into a main problem and sub-problems. The main problem is an upper-level multi-objective optimization model that embeds the generated Benders cutting plane, and the upper-level decision variable values ​​are obtained after solving it. The sub-problems are lower-level robust linear programming models that substitute the upper-level decision variable values, used to solve the lower-level optimal solution under the worst-case scenario of renewable energy output.

[0088] Generate optimal or feasible cutting planes based on the solutions to the subproblems and add them to the main problem.

[0089] Repeat the iteration until the convergence condition is met, and output the optimal solution of the two layers.

[0090] Step S4: Output the two-layer optimal scheduling strategy.

[0091] Step S5: Perform simulation comparison analysis between the generated scheduling strategy and the comparison strategy; the comparison strategy includes the proposed strategy, demand-free response, and basic strategy; the comparison content includes: node voltage, wind power and photovoltaic absorption, network loss, load before and after demand response, and various operating costs of the distribution network.

[0092] The following section will further elaborate on the relevant aspects of this method.

[0093] I. Using a dual model, design objective functions and constraints for different optimization objectives.

[0094] Step S11: Define the objective function of the upper-level optimization model:

[0095] The upper-level model aims to ensure the safe and reliable operation of the distribution network by optimizing its power quality. Power quality can typically be comprehensively assessed from two key dimensions: voltage stability and voltage deviation level. To further measure the distribution network's ability to maintain system stability and safe operation under power fluctuation conditions, this method uses static voltage stability as an important component of the evaluation system.

[0096] The objective function of the upper-level optimization model is The optimization objectives are to minimize distribution network losses and achieve optimal power quality. Power quality assessment indicators include node voltage offset level and static voltage stability, expressed as:

[0097]

[0098] In the formula: —Network loss function, kWh; —Power quality level function; —Voltage offset level weight; —Voltage offset level; —Static voltage stability; — Static voltage stability weight.

[0099]

[0100] Where: T—dispatch cycle duration, h; Ω—the set of all transmission lines in the distribution network; —The current value flowing through line ij in the distribution network at time t, in kA; — The resistance value of line ij, in Ω.

[0101]

[0102] Where: K—the number of nodes in the distribution network; Vk,t—the per-unit value of the node voltage at the k-th node at time t, pu; ΔV—the benchmark value for judging node voltage deviation.

[0103]

[0104] In the formula: —Static voltage stability of line ij at time t; —The net active power load of the j-th node at time t, in MW; —The net reactive load of the j-th node at time t, in MW; —The reactance value of line ij; —Voltage value of the i-th node at time t in the power distribution network; —All feeders The maximum value in kV.

[0105] Step S12: Set the constraints for the upper-level optimization model:

[0106] Voltage drop constraint:

[0107]

[0108] Nodal active power balance constraints:

[0109]

[0110] Nodal reactive power balance constraints:

[0111]

[0112] Current definition constraint:

[0113]

[0114] Distribution network voltage and current limiting constraints:

[0115]

[0116] In the formula: —Active and reactive power flowing through branch ij at time t, MW and Mvar; —Active power and reactive power injected at node j at time t, MW and Mvar. —The upper limit of the square value of the line current. ; — The upper and lower limits of the squared value of the node voltage, in V.

[0117] Step S13: Define the objective function of the lower-level optimization model:

[0118] The lower-level optimization model needs to determine the actual output of high-proportion distributed photovoltaic and various dispatch resources in the distribution network based on information such as wind and solar data, load demand data, and electricity price parameters, so as to ensure the economic minimization of the distribution network operation.

[0119] The objective function of the lower-level optimization model is :

[0120]

[0121] In the formula: — Cost of electricity purchased from the main grid by the distribution network at time t, in ten thousand yuan; —Cost of electricity sold from the distribution network to the main grid at time t, in ten thousand yuan; — Distribution network operating cost at time t, in ten thousand yuan; — Cost of flexible load demand response of distribution network at time t, in ten thousand yuan; — Cost of wind and solar power curtailment in the distribution network at time t, in ten thousand yuan.

[0122] The specific definitions of various cost functions are as follows:

[0123] Distribution network electricity purchase cost

[0124]

[0125] In the formula: —The unit electricity price of the distribution network purchasing electricity from the main grid at time t, in yuan / kWh; —Power purchased by the distribution network from the main grid at time t, in MW.

[0126] Electricity sales cost of distribution network

[0127]

[0128] In the formula: —The unit electricity price at time t for the distribution network to sell electricity to the main grid, in yuan / kWh; —Power output from the distribution network to the main grid at time t, in MW.

[0129] Operating costs of power distribution networks:

[0130]

[0131] In the formula: —DG unit electricity operating cost coefficient and generating capacity at time t, in MW;

[0132] Demand response cost of distribution network:

[0133]

[0134] In the formula: —The unit load demand response cost coefficient at time t; —Load response power at time t, in MW.

[0135] Costs of wind and solar power curtailment in power distribution networks:

[0136]

[0137] In the formula: — Cost coefficient of wind and solar power curtailment per unit of electricity in the distribution network at time t; — Curtailment power of wind turbines and distributed photovoltaic power at time t, in MW.

[0138] Step S14: Set the constraints for the lower-level optimization model:

[0139] Uncertainty Constraints of New Energy Sources: The uncertainty of output from distributed photovoltaic and wind turbine units is one of the important factors affecting power grid dispatching. Its uncertainty constraints are as follows:

[0140]

[0141] In the formula: , —The active power injected by the i-th distributed photovoltaic and wind turbine unit at time t, in MW; , — The predicted output deviation of the i-th distributed photovoltaic and wind turbine at time t; , —Confidence level of the probability that the constraint condition is true; , — The reactive power output of the i-th distributed photovoltaic and wind turbine unit at time t, Mvar; , — The power factor angle of the i-th distributed photovoltaic and wind turbine at time t.

[0142] Power interaction constraints between main and distribution networks:

[0143]

[0144] In the formula: —Maximum power purchased by the distribution network from the main grid per unit time, in MW; —Maximum power output of the distribution network to the main grid per unit time, in MW.

[0145] Demand response constraints:

[0146] Demand response mechanisms, through consultation with users and with reasonable compensation, guide users to proactively reduce electricity consumption during peak load periods or critical periods of supply and demand tension in the power system. This effectively avoids system overload or operational risks caused by insufficient peak-shaving capacity. Simultaneously, demand response plays a crucial role in load regulation by "peak shaving and valley filling," helping to reduce the peak level of the load curve, alleviate grid operational pressure, and thus improve the overall flexibility and rapid response capability of the power system. This not only enhances the operational stability of the power grid but also effectively reduces overall energy supply costs during the optimization of resource allocation.

[0147]

[0148] In the formula: —The demand response at time t can reduce the maximum power value, in MW; —The number of times the distribution network demand response occurs at time t; —Maximum number of reductions in demand response.

[0149] II. Solving the upper-level optimization model using the multi-objective octopus optimization algorithm.

[0150] Step S21: Reorganize the upper-level model into a standard multi-objective optimization problem:

[0151] The upper-level model function can be summarized as follows:

[0152]

[0153] in, The objective function is to minimize network loss. Let the objective function be the power quality optimization function. For the safety constraints of the power distribution network, Constrained by various trends.

[0154] Step S22: Solve the multi-objective optimization problem using the Octopus Optimization Algorithm:

[0155] The multi-objective octopus algorithm is a swarm intelligence multi-objective optimization algorithm that simulates the predatory behavior of octopuses. Its core biomimetic logic originates from three core behaviors in the octopus's foraging process:

[0156] 1. Global search behavior: Octopuses patrol open waters and explore the distribution of potential prey over a wide area, which corresponds to the global exploration capability of the algorithm and avoids getting trapped in local optima;

[0157] 2. Encirclement behavior: After spotting prey, the octopus adjusts its position to move closer to the prey, narrowing the search range, which corresponds to the local exploitation capability of the algorithm;

[0158] 3. Attack behavior: The octopus launches a precise tentacle attack near its prey to complete the capture. This corresponds to the algorithm's fine search near the optimal solution, which accelerates convergence.

[0159] The solution process for the multi-objective octopus algorithm is as follows:

[0160] 1. Algorithm parameter initialization: Set the core operating parameters of the algorithm, specifically including: population size. Maximum number of iterations Global search step size factor α, encirclement probability β, attack probability γ, and maximum external archive capacity. 2. Initial population generation: Within the feasible region of the decision variables, an initial octopus population is constructed using a random generation method. The position vectors of the individual octopuses are:

[0161]

[0162] In the formula: D represents the dimension of the decision variable; Let be the D-th dimension decision variable for the i-th individual.

[0163] 2. Fitness assessment and Pareto dominance determination: For each octopus individual in the population, calculate the dual objective function value to complete the fitness assessment; based on the Pareto dominance criterion of minimizing the optimization problem, determine the dominance relationship between individuals: for two feasible solutions... and If both conditions are met:

[0164]

[0165] but Pareto Domination An individual is called a non-dominated solution if it is not dominated by any other individual in the population.

[0166] 3. External archive update: External archive A is used to store all non-dominated solutions generated during the iteration process and is the core carrier for constructing the Pareto front. Its update rule is as follows:

[0167] 1) For each individual in the current population, if it is dominated by any solution in the archive, it is discarded directly;

[0168] 2) If the individual controls some of the solutions in the archive, delete the controlled solutions in the archive and save the individual to the archive;

[0169] 3) If the individual does not interact with any of the solutions in the archive, then save it directly into the archive;

[0170] 4) If the archive size exceeds the maximum capacity Then, crowding distance sorting is used, the individual with the smallest crowding distance is deleted, and the uniformly distributed non-dominated solutions are retained to ensure the diversity of the Pareto front.

[0171] The crowding distance is calculated as follows: For the i-th solution in the archive, its crowding distance is the sum of the normalized distances of the objective values ​​of two adjacent solutions on the two objective function axes, as shown in the formula:

[0172]

[0173] In the formula: These are the maximum and minimum values ​​of the k-th objective function, respectively; These are the k-th objective function values ​​of the two adjacent solutions to the i-th solution.

[0174] 4. Leader Selection: To guide the octopus population's search direction, a roulette wheel method is used to select a leader for each individual from an external archive. The probability of selection is positively correlated with the individual's distance from crowding, as shown in the formula:

[0175]

[0176] In the formula: The current size of the external archive; Let be the crowding distance of the i-th solution in the archive. This strategy can preferentially select solutions in sparsely distributed regions as leaders, ensuring the uniformity of the Pareto front.

[0177] 5. Individual location update

[0178] Based on three types of octopus predatory behaviors, the iterative update of individual positions is completed. The step factor α dynamically decays with the number of iterations, so as to balance the global exploration and local exploitation capabilities of the algorithm:

[0179]

[0180] In the formula: t is the current number of iterations.

[0181] The specific update rules for the three types of predatory behaviors in the algorithm are as follows:

[0182] (1) Global search behavior

[0183] When the random number rand>β+γ, the global search behavior is executed, and the update formula is:

[0184]

[0185] In the formula: rand is a uniform random number in the interval [0,1]; are the upper bound vector and lower bound vector of the decision variable, respectively.

[0186] (2) Encircling behavior

[0187] When the random number γ<rand≤β+γ, the encircling behavior is executed to move closer to the leader individual, and the update formula is:

[0188]

[0189] (3) Attack behavior

[0190] When the random number rand≤γ, the attack behavior is executed to perform fine search near the leader, and the update formula is:

[0191]

[0192] In the formula: L is the local search step vector, which is 5%~10% of the feasible interval of the decision variable.

[0193] After the position update, truncation and repair are performed on the decision variables that exceed the boundary, and infeasible solutions are verified and repaired through power flow calculation, so as to ensure that all updated individuals meet the constraint conditions.

[0194] 6. Iteration termination judgment When any of the following conditions is met, the algorithm terminates the iteration and outputs the Pareto front in the external archive:

[0195] 1) The current number of iterations reaches the maximum number of iterations Tmax;

[0196] 2) After 50 consecutive iterations, the non-dominated solutions in the external archive have not been updated, and the Pareto front has converged.

[0197] Step S23: Obtain the unique solution of the upper-level optimization model using the approximation ideal solution sorting method:

[0198] The Top-to-Ideal Solution Ranking (TOPSIS) method calculates and evaluates the combined distance of each solution in the solution set to both the positive and negative ideal solutions. A solution is considered superior if it is closer to the positive ideal solution and farther from the negative ideal solution. By combining TOPSIS decision-making with the multi-objective octopus algorithm, the solution can be made closer to the optimal solution, thus balancing the relationship between the dual objectives of power quality and network loss in the upper-level model.

[0199] The specific process is as follows:

[0200] 1. Calculate the weight values ​​to construct the decision matrix. The calculation formula is as follows:

[0201]

[0202] In the formula: —The row and column elements corresponding to the decision matrix; —The weights of the target n; — Pareto solution set elements; M — Number of solutions in the Pareto solution set.

[0203] 2. Obtain the positive optimal solution for objective n from the decision matrix. With negative optimal solution In this method, the positive optimal solution to the bi-objective problem is to minimize the network loss and power quality index in the upper-level optimization model, that is:

[0204]

[0205] In the formula: — the set of the nth column of the decision matrix.

[0206] 3. Calculate the Euclidean distance between the biobjective solution value and the positive and negative optimal solutions in the decision matrix, i.e.:

[0207]

[0208] In the formula: N — the number of objectives to be solved.

[0209] 4. The closeness fm between the objective value and the optimal solution under each decision scheme is calculated as follows:

[0210]

[0211] Sort by proximity The larger the value, the better the decision-making option. The decision scheme with the maximum value is taken as the optimal solution of the upper-level optimization model, and the reference variables of the upper-level model in this scheme are used. It is passed to the lower-level optimization model.

[0212] The dual-layer optimization scheduling strategy for a high-proportion wind and solar power distribution network based on multi-objective optimization as described in claim 1 is characterized by employing the robust Benders decomposition algorithm to solve the dual-layer optimal strategy, comprising the following steps:

[0213] Step S31: Organize the next-level model into a compact form:

[0214]

[0215] In the formula: y—reference variable of the lower-level model; G(y), H(y)—constraints of the lower-level model.

[0216] Step S32: Solve the two-layer optimal strategy using the robust Benders decomposition algorithm:

[0217] The robust Benders decomposition algorithm breaks down a bi-level optimization problem into a main problem and subproblems, and achieves the interaction between solutions at different levels through cutting plane iteration. The specific process is as follows:

[0218] 1. Initialize parameters.

[0219] 2. Solve the robust Benders master problem, which is a mixed-integer linear programming problem where the upper-level multi-objective optimization model is embedded into the generated Benders cutting plane.

[0220]

[0221] Where x is the upper-level decision variable, θ is the dual variable representing the estimated optimal objective value of the lower-level subproblem, and F(x,y) is the lower-level objective function. Solving the main problem yields the optimal solution. .

[0222] 3. Solve the robust Benders subproblem

[0223] Substituting the solution xk of the main problem into the lower-level model, a robust linear programming subproblem is constructed to find the optimal solution for the lower-level model under the worst-case scenario of renewable energy output:

[0224]

[0225] in, The objective function of the lower-level optimization model is... Here, y represents the objective function of the compact lower-level optimization model, and y represents the lower-level decision variables. This represents all inequality constraints in the lower-level model. This represents all equality constraints in the lower-level model. Strong duality theory is used to eliminate uncertainties, improve solution efficiency, and obtain the optimal solution to the subproblem. .

[0226] 4. Generation and updating of cutting planes

[0227] If the subproblem is feasible, generate the optimality cutting plane:

[0228] If the subproblem is infeasible, generate a feasibility cutting plane:

[0229] Add the generated cutting plane to set Ω, and update the upper and lower bounds: .

[0230] 5. Convergence Judgment

[0231] If |UB−LB|≤η, the iteration terminates and the optimal solution of the two layers is output; otherwise, let k=k+1 and return to step 2 to continue the iteration.

[0232] The two-layer optimization scheduling strategy for a high-proportion wind and solar power distribution network based on multi-objective optimization as described in claim 1 is characterized by generating the scheduling strategy and comparing it with strategies obtained by other methods, including the following steps:

[0233] Step S41: Set parameters in the IEEE 33-node distribution network:

[0234] Simulation analysis was conducted using an IEEE 33-node distribution network. Line impedance was adopted as the standard 33-node line impedance value. Peak and valley electricity prices for each time period were set, along with the unit operating cost and unit curtailment cost of distributed photovoltaic power (PV). The unit operating cost and unit curtailment cost of wind turbines were also included. Unit charging and discharging cost coefficients for energy storage devices were also specified. Installed capacity and voltage per-unit limits were also considered. The simulation included wind turbine installed capacity and distributed PV installed capacity.

[0235] Step S42: Compare the scheduling strategy obtained by other methods:

[0236] To verify the effectiveness of the proposed method, three scenarios—the proposed strategy, demand-free response, and the basic strategy—are compared. The comparison includes: node voltage, wind and solar power absorption, network losses, load before and after demand response, and various costs. To demonstrate the superiority of this method, results are also compared with those of the Weighted Sum Genetic Algorithm (WSGA) and the Multi-Objective Particle Swarm Optimization (MOPSO) algorithm. The comparison includes: distribution network operation network loss costs, power quality improvement, renewable energy absorption rate, and total distribution network operation costs.

[0237] 1. Set parameters:

[0238] This method uses an IEEE 33-node distribution network for simulation analysis. Peak and valley electricity prices for each time period are shown in Table 1. The simulation period for this example is 24 hours with a step size of 1 hour. The line impedance uses the standard 33-node line impedance value. The unit operating cost of distributed photovoltaic (PV) is 0.3 yuan / kWh, and the unit cost of curtailment is 0.74 yuan / kWh; the unit operating cost of wind turbines is 0.25 yuan / kWh, and the unit cost of wind curtailment is 0.7 yuan / kWh; the unit charging cost coefficient for energy storage devices is 0.1 yuan / kWh, and the unit discharging cost coefficient is 0.09 yuan / kWh; the installed capacity is 1.6 MW, and the per-unit voltage limits are 1.05 and 0.95 pu, respectively. The installed capacity of wind turbines is 1.5 MW, and the installed capacity of distributed PV is 1.2 MW.

[0239] Table 1 Peak-valley electricity prices for different time periods

[0240]

[0241] 2. Comparative Analysis:

[0242] To verify the effectiveness of the method proposed in this invention, this embodiment compares three scenarios: the proposed strategy, no demand response, and the basic strategy. The comparison includes: node voltage, wind and solar power absorption, network losses, load before and after demand response, and various costs. The results are as follows: Figure 2-6 As shown in Table 2, the method proposed in this invention achieves optimal voltage quality and power quality for the distribution network compared to other scenarios. It also improves the absorption rate of new energy sources, reduces the cost of wind and solar power curtailment, effectively balances the supply and demand of the power grid by reducing load, and enhances the stability and security of the grid. Furthermore, it contributes to energy conservation, emission reduction, and emergency response. The data in the table also demonstrates that this method effectively reduces various overall costs.

[0243] Table 2 Comparison of various costs

[0244]

[0245] This embodiment also provides a computer device, including: at least one processor, at least one memory, and computer program instructions stored in the memory, which implement the above-described method when executed by the processor.

[0246] This embodiment also provides a computer-readable storage medium storing computer program instructions that, when executed by a processor, implement the above-described method.

[0247] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0248] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0249] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0250] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0251] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A two-layer optimization scheduling method for a distribution network with a high proportion of wind and solar power based on multi-objective optimization, characterized in that, Includes the following steps: Step S1: Establish a two-layer optimal scheduling model for a distribution network with a high proportion of wind and solar power; the two-layer optimal scheduling model includes an upper-layer optimization model and a lower-layer optimization model; The upper-level optimization model takes minimizing network loss and optimizing power quality of the distribution network as its primary optimization objectives, and sets upper-level constraints; the power quality evaluation indicators include node voltage offset level and static voltage stability. The lower-level optimization model takes the optimal operation economy of the distribution network as the second optimization objective and sets lower-level constraints; the operation economy of the distribution network includes electricity purchase cost, electricity sales cost, operating cost, demand response cost, and wind and solar curtailment cost; Step S2: Use the multi-objective octopus optimization algorithm to solve the upper-level optimization model and obtain the upper-level optimal solution set; Step S3: Use the robust Benders decomposition algorithm to solve the two-level optimization problem consisting of the upper-level optimization model and the lower-level optimization model to obtain the two-level optimal scheduling strategy; Step S4: Output the two-layer optimal scheduling strategy.

2. The two-layer optimization scheduling method for a distribution network with a high proportion of wind and solar power based on multi-objective optimization according to claim 1, characterized in that, In step S1, the objective function of the upper-level optimization model is: in, The objective function of the upper-level optimization model; This is the network loss function; Let the power quality level function be expressed as: in, As the voltage offset level weight, It is a voltage offset level function. As the weight for static voltage stability, This is the static voltage stability function; The upper-level constraints include: voltage drop constraints, node active power balance constraints, node reactive power balance constraints, current definition constraints, and distribution network voltage and current limit constraints.

3. The two-layer optimization scheduling method for a distribution network with a high proportion of wind and solar power based on multi-objective optimization according to claim 1, characterized in that, In step S1, the objective function of the lower-level optimization model is: in, The objective function of the lower-level optimization model; Let t be the cost of electricity purchased by the distribution network from the main grid; Let t be the cost of electricity sold from the distribution network to the main grid. Let t be the operating cost of the distribution network at time t; Let t be the demand response cost of the flexible load in the distribution network. Let t be the cost of wind and solar power curtailment in the distribution network. The lower-level constraints include: uncertainty constraints on new energy output, power interaction constraints between the main grid and distribution grid, and demand response constraints.

4. The two-layer optimization scheduling method for a distribution network with a high proportion of wind and solar power based on multi-objective optimization according to claim 1, characterized in that, Step S2 specifically includes: Step S21: Represent the upper-level optimization model as a standard multi-objective optimization problem form that includes a dual objective function and constraints; Step S22: The multi-objective octopus optimization algorithm is used to solve the problem. The multi-objective octopus optimization algorithm simulates the global search behavior, encirclement behavior and attack behavior of octopuses in the process of foraging to update the position of individuals in the population, and saves the non-dominated solution in the iteration process through external archiving to obtain the Pareto front. Step S23: Select a unique solution from the Pareto front using the approximation ideal solution sorting method as the optimal solution of the upper-level optimization model, and pass the upper-level decision variables in the optimal solution to the lower-level optimization model.

5. The two-layer optimization scheduling method for a distribution network with a high proportion of wind and solar power based on multi-objective optimization according to claim 4, characterized in that, In step S22, the implementation method of the multi-objective octopus optimization algorithm is as follows: Initialize population parameters; Assess the fitness of each individual in the population and determine Pareto dominance; Update the external archive and use crowding distance sorting to maintain the diversity of solutions within the archive; The leader is selected from an external archive based on a roulette wheel selection method. Based on the comparison between the random number and the preset probability, the global search behavior, encirclement behavior or attack behavior are executed respectively to update the individual position; Determine if the iteration termination condition has been met; if so, output the Pareto front.

6. The two-layer optimization scheduling method for a distribution network with a high proportion of wind and solar power based on multi-objective optimization according to claim 4, characterized in that, In step S23, the method of selecting a unique solution by approximating the ideal solution specifically includes: Construct a decision matrix and perform normalized weighting on the matrix elements; Determine the positive and negative ideal solutions for each objective; Calculate the Euclidean distance between each solution in the Pareto front and the positive and negative ideal solutions; Calculate the degree of closeness between each solution and the ideal solution, and select the solution with the highest degree of closeness as the optimal solution of the upper-level optimization model.

7. The two-layer optimization scheduling method for a distribution network with a high proportion of wind and solar power based on multi-objective optimization according to claim 1, characterized in that, Step S3 specifically includes: Step S31: Organize the lower-level optimization model into a compact form; in, The objective function of the lower-level optimization model is... Here, y represents the objective function of the compact lower-level optimization model, and y represents the lower-level decision variables. This represents all inequality constraints in the lower-level model. This represents all equality constraints in the lower-level model; Step S32: Use the robust Benders decomposition algorithm to iteratively solve the problem, decomposing the original problem into a main problem and sub-problems; The main problem is a multi-objective optimization model that incorporates the generated Benders cut plane. Solving this model yields the values ​​of the upper-level decision variables. The subproblem is a lower-level robust linear programming model after substituting the values ​​of the upper-level decision variables, used to solve the lower-level optimal solution under the worst-case scenario of new energy output. Generate optimal or feasible cutting planes based on the solutions to the subproblems and add them to the main problem; Repeat the iteration until the convergence condition is met, and output the optimal solution of the two layers.

8. The two-layer optimization scheduling method for a distribution network with a high proportion of wind and solar power based on multi-objective optimization according to claim 1, characterized in that, Following step S4, the following is also included: Step S5: Perform simulation comparison analysis between the generated scheduling strategy and the comparison strategy; the comparison strategy includes the proposed strategy, demand-free response, and basic strategy; the comparison content includes: node voltage, wind power and photovoltaic absorption, network loss, load before and after demand response, and various operating costs of the distribution network.

9. A computer device, characterized in that, include: At least one processor, at least one memory, and computer program instructions stored in the memory, which, when executed by the processor, implement the method as described in any one of claims 1-8.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by a processor, the method described in any one of claims 1-8 is implemented.

Images