Multi-objective coordinated optimization reconfiguration method for distribution network based on improved multi-objective river-horse optimization algorithm

Abstract

The application discloses a power distribution network multi-objective collaborative optimization reconstruction method based on an improved multi-objective hippo optimization algorithm. First, a multi-objective optimization model containing power loss cost, distributed power investment cost and operation cost is constructed, and a scene-based random optimization method is adopted to model load and distributed power output uncertainty. Second, the improved multi-objective hippo optimization algorithm is adopted for solution, and the improvement strategies include: sinusoidal chaotic mapping initialization to enhance population diversity; dynamic Lévy distribution beta parameter adjustment to balance exploration and development; and sinusoidal decay factor to realize smooth transition from exploration to development. Finally, optimization is performed on decision variables including reconstructed switch state, distributed power site selection, fixed capacity and power factor, and a Pareto optimal solution set is output. The application overcomes the problems of traditional algorithms, such as easy to fall into local optimum and insufficient uncertainty processing, and improves the economy and robustness of the optimization scheme.

Description

Technical Field

[0001] This invention belongs to the field of distribution network planning, optimization operation and reconfiguration technology. Specifically, it relates to a method for multi-objective collaborative optimization of distributed generation (DG) site selection and network reconfiguration based on the improved multi-objective Hippopotamus Optimization Algorithm (Improved MOHO), considering the uncertainty of load and renewable energy output. Background Technology

[0002] As a crucial link connecting the transmission network and end users in the power system, the distribution network's safety and economic efficiency directly impact the overall power supply quality and operational effectiveness. With the increasing penetration of distributed generation (DG), especially renewable DG such as photovoltaic and wind power, in distribution networks, the operating characteristics of these networks are undergoing profound changes. Rational configuration of DG and optimization of network topology have become important means to improve system operating efficiency, reduce power loss, and improve voltage distribution.

[0003] Distribution network reconfiguration refers to the process of optimizing operation by adjusting the combination of sectionalizing switches and tie switches to alter the network structure, while meeting constraints such as radial topology and power supply reliability. Simultaneously, the location and capacity of distributed generation (DG) determine its ability to support grid operation. These two aspects are physically coupled and mutually influential in their objectives, thus necessitating coordinated optimization of DG planning and network reconfiguration.

[0004] However, existing research still faces the following technical bottlenecks in solving the multi-objective collaborative optimization and reconfiguration problem of distribution networks with DG: (1) Insufficient handling of uncertainties: The output of renewable DG is significantly affected by meteorological conditions such as sunlight and wind speed, exhibiting obvious intermittency and volatility; at the same time, load demand is also affected by factors such as user behavior and seasonal changes, resulting in strong uncertainty. Existing methods mostly use deterministic models for optimization, lacking systematic modeling of uncertainties, leading to insufficient robustness of optimization results in actual operation and difficulty in adapting to changing operating scenarios.

[0005] (2) High dimensionality of decision variables and low coding efficiency: The distribution network reconfiguration problem is essentially a high-dimensional nonlinear mixed integer programming problem. Traditional binary coding methods require independent coding of each switch state, which leads to a dramatic increase in the dimensionality of decision variables. For example, in the IEEE 69-bus system, the dimension of switch state variables is as high as 68 dimensions, the search space is huge, and a large number of codes correspond to infeasible solutions (non-radial or isolated), which seriously restricts the solution efficiency and convergence performance.

[0006] (3) Performance limitations of multi-objective optimization algorithms: Co-optimization of DG configuration and network reconfiguration involves multiple conflicting objectives, such as minimizing power loss costs, DG investment costs, and operating costs, which are typical multi-objective optimization problems. Existing metaheuristic algorithms (such as the multi-objective gray wolf optimizer MOGWO and the multi-objective dragonfly algorithm MODA) have difficulty balancing exploration and development capabilities when dealing with this type of problem. They are prone to getting trapped in local optima, resulting in insufficient convergence accuracy and distribution diversity of the Pareto front, and making it difficult to provide high-quality compromise solution sets.

[0007] In summary, the existing technology lacks a collaborative optimization and reconfiguration method for distribution networks that can systematically integrate uncertainty modeling, efficient dimensionality reduction coding, and high-performance multi-objective optimization algorithms. It is difficult to obtain an economical and robust reconfiguration scheme while ensuring solution efficiency. Summary of the Invention

[0008] In view of this, the present invention proposes a method and system for multi-objective collaborative optimization and reconfiguration of distribution networks based on an improved multi-objective hippo optimization algorithm, thereby solving the technical problems in the background art.

[0009] The present invention provides a multi-objective collaborative optimization and reconfiguration method for power distribution networks based on an improved multi-objective hippo optimization algorithm, comprising the following steps:

[0010] Step S1: Obtain the topology, load data, and distributed generation output data of the distribution network. Use a scenario-based stochastic optimization method to generate multiple discrete operating scenarios to model the uncertainties of load and distributed generation output.

[0011] Step S2: Construct a multi-objective fitness vector with the objective of minimizing the expected values ​​of power loss cost, distributed power source investment cost, and distributed power source operating cost;

[0012] Step S3: Generate an initial population of hippopotamus individuals using a sinusoidal chaotic mapping initialization method, which serves as the initial solution set for the improved multi-objective hippopotamus optimization algorithm;

[0013] Step S4: The improved multi-objective hippo optimization algorithm is used to optimize the decision variable vector to obtain the Pareto optimal solution set, wherein the decision variable vector includes the reconstructed switch state, the installation location of the distributed power source, the capacity of the distributed power source, and the power factor of the distributed power source; the improved multi-objective hippo optimization algorithm includes a dynamic adjustment mechanism for the beta parameter of the Lévy distribution.

[0014] Step S5: For each candidate solution, traverse all operating scenarios, perform power flow calculations, evaluate the multi-objective function, and verify the bus voltage constraint, power balance constraint, and distributed source capacity constraint.

[0015] Step S6: Output the Pareto optimal solution set to determine the optimal combination of distributed power source location, capacity setting, and network reconfiguration switching states.

[0016] Preferably, the method of generating multiple discrete operating scenarios using scenario-based stochastic optimization in step S1 includes: modeling the uncertainty of load demand using a normal distribution superimposed with random noise; modeling the uncertainty of solar distributed power output using a log-normal distribution superimposed with random noise and physical range truncation; generating an initial integrated operating scenario through Monte Carlo sampling, and simplifying the initial scenario into a typical scenario using the K-means clustering method.

[0017] Preferably, the sinusoidal chaotic mapping initialization method in step S3 uses the following mapping formula to generate the initial population: ; Where α is the control parameter, x k Let be the value of the k-th chaotic sequence.

[0018] Preferably, the dynamic adjustment mechanism for the Lévy distribution beta parameter in step S4 is as follows: the beta parameter of Lévy flight is dynamically adjusted according to the distance between the hippopotamus individual and the predator, increasing the beta value when the individual is closer to the predator and decreasing the beta value when the individual is farther from the predator.

[0019] Preferably, the improved multi-objective hippo optimization algorithm further includes: introducing a dynamic factor based on a decreasing sine function in the first exploration phase. The dynamic factor is used to adjust the balance between exploration and development in the algorithm. It takes a larger value in the early stage of algorithm iteration to encourage global exploration and a smaller value in the later stage of iteration to promote local development.

[0020] Preferably, the reconfiguration switch state in the decision variable vector in step S4 is represented by an integer ring network coding scheme, which reduces the dimension of the reconfiguration switch to be equal to the number of independent loops formed after closing the tie switch; the integer ring network coding scheme reduces the dimension of the reconfiguration switch in the IEEE 69-node system to 5 dimensions.

[0021] Preferably, when verifying the bus voltage constraint in step S5, it is necessary to ensure that the node voltage in all operating scenarios is within the preset safe range; when verifying the radial topology constraint, it is necessary to ensure that the network maintains a radial structure after reconstruction and that all loads are powered.

[0022] The present invention also provides a system for multi-objective collaborative optimization and reconfiguration of distribution networks, comprising:

[0023] The data acquisition module is used to acquire the operating parameters, topology information, and predicted data of distributed power sources and loads of the distribution network.

[0024] The scene generation module is used to generate multiple discrete running scenes according to the method described in claim 7;

[0025] The algorithm control module is used to execute the improved multi-objective hippo optimization algorithm according to any one of claims 1 to 7;

[0026] The fitness evaluation module is used to evaluate the multi-objective expected cost vector of candidate solutions based on scenario probabilities.

[0027] The constraint checking module is used to verify whether the candidate reconfiguration schemes meet the radial topology constraints, bus voltage constraints, and power balance constraints.

[0028] The results output module is used to output the optimal distributed power supply configuration parameters and reconfigurable switch states.

[0029] Preferably, the algorithm control module further includes:

[0030] A sinusoidal chaotic initialization unit is used to generate an initial population of individual hippos using a sinusoidal chaotic mapping.

[0031] The dynamic parameter adjustment unit is used to dynamically adjust the beta parameter of the Lévy distribution based on the distance between the hippopotamus individual and the predator.

[0032] Explore and develop balancing units to introduce and develop dynamic factor adjustment algorithms based on sinusoidal function decrement.

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] Strong uncertainty handling capability: This invention adopts a scenario-based stochastic optimization method to accurately capture the uncertainty of load and DG output, making the optimization scheme more robust and adaptable in actual complex operating environments.

[0035] High coding efficiency: This invention adopts an integer ring network coding scheme, which greatly reduces the dimensionality of the reconstruction problem (such as reducing the dimensionality of the IEEE 69-node system from 68 dimensions to 5 dimensions), effectively reducing the search space and significantly improving computational efficiency.

[0036] Superior algorithm performance: This invention significantly enhances the global exploration capability and convergence accuracy of the MOHO algorithm through a hybrid improvement strategy including sinusoidal chaotic initialization, dynamic Lévy beta adjustment, and sinusoidal decay balance factor, ensuring that the algorithm can find a Pareto front solution set with better convergence and uniformity.

[0037] Optimal economic benefits: This invention comprehensively evaluates the benefits of DG access from an economic perspective by simultaneously minimizing network loss costs, DG investment costs, and DG operating costs, and provides an optimal configuration scheme with more practical guidance. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 This is an overall flowchart of the power distribution network reconfiguration method provided in an embodiment of the present invention.

[0040] Figure 2 This is a functional block diagram of a power distribution network reconfiguration system provided in an embodiment of the present invention.

[0041] Figure 3 This is a flowchart of the improved multi-objective hippo optimization algorithm described in step S2 of the present invention.

[0042] Figure 4 A schematic diagram of the probability distribution model for load uncertainty and DG output uncertainty provided in the embodiments of the present invention. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are merely some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0044] This invention provides a multi-objective collaborative optimization and reconfiguration method for power distribution networks based on the Improved Multi-Objective Hippopotamus Optimization Algorithm (Improved MOHO). Figure 1 This is an overall flowchart of the power distribution network reconfiguration method provided in an embodiment of the present invention. Figure 1 As shown, the method includes the following four main steps:

[0045] Step S1, Initialization and Scene Generation;

[0046] This step aims to acquire basic data on the distribution network and model the uncertainties of load and distributed generation (DG) output to generate typical operating scenarios for subsequent optimization and evaluation.

[0047] S1.1 Data Acquisition and Encoding

[0048] First, the topology, line parameters, load data, and predicted output data of the distribution network are obtained. In this embodiment, the IEEE 69-node distribution network system is used as an example. This system includes 69 nodes, 68 sectionalizing switches, and 5 tie switches, and the initial network has a radial structure.

[0049] To improve optimization efficiency and reduce the proportion of infeasible solutions, this invention employs an integer-type ring network coding scheme for reconstructing switch states. Specifically, each independent loop formed after closing a tie switch is considered as a coding dimension. For the IEEE 69-node system, a total of 5 independent loops are formed, thus reducing the dimension of the decision variables for reconstructing switch states from the traditional 68 dimensions to 5 dimensions. Each integer variable represents the segment switch number that needs to be opened in that loop, thereby ensuring that the reconstructed network always satisfies the radial topology constraint and that there are no isolated islands.

[0050] S1.2 Uncertainty Modeling and Scene Generation

[0051] At the start of the optimization process, the algorithm randomly generates an initial set of hippo individuals (i.e., candidate solutions). These individuals are uniformly distributed within the search space, representing the initial population of solutions to the problem. The position of each hippo individual...

[0052]

[0053] Assuming the load forecast error follows a normal distribution, simulations are performed by superimposing random noise with a mean of zero and a specific standard deviation onto the baseline load factor during scenario generation. It is also assumed that the DG output fluctuation follows a log-normal (or normal) distribution. Simulations are performed by superimposing random noise onto the DG predicted output, and a pruning function is used to limit its output to the physical range of [0.0, 1.0]. Eighty hundred initial integrated operating scenarios are generated through Monte Carlo sampling, and then simplified into ten representative baseline scenarios using K-means clustering. These ten scenarios are assigned corresponding probabilities Ps, significantly reducing the computational burden of subsequent optimization processes.

[0054]

[0055] To accurately describe the uncertainties in load and DG output, this invention employs a scenario-based stochastic optimization method. The specific modeling approach is as follows:

[0056] Load uncertainty modeling: It is assumed that the load forecasting error follows a normal distribution. Based on the baseline load factor, random noise with a mean of zero and a specific standard deviation (e.g., 5% of the baseline load) is added for simulation, as shown in the following equation:

[0057]

[0058] As the baseline load; This is the random error term that follows a normal distribution.

[0059] Modeling the uncertainty of DG output: For solar photovoltaic (DG) generators, it is assumed that their output fluctuations follow a log-normal distribution. Random noise is superimposed on the predicted DG output, and a clipping function is used to limit the output value to a physical range of [0.0, 1.0] (expressed in per-unit values), as shown in the following equation:

[0060]

[0061] Contribute to DG's predictions; It is a random multiplicative factor that follows a log-normal distribution.

[0062] Eighty hundred initial integrated operation scenarios were generated using Monte Carlo sampling. Subsequently, the K-means clustering algorithm was used to simplify these scenarios into 10 representative typical scenarios, each assigned a corresponding probability P. s (s=1,2,l,10), and satisfy sums=1^10 P s= 1. This scenario reduction strategy greatly reduces the computational burden of subsequent optimization calculations.

[0063] Step S2, parameter adaptive optimization;

[0064] This step aims to optimize the decision variables using an improved MOHO algorithm. Figure 3 This is a flowchart of the improved multi-objective hippo optimization algorithm described in step S2 of the present invention. Figure 3 As shown, this algorithm introduces three key improvement strategies based on the standard MOHO algorithm.

[0065] 1) The core mechanism of standard MOHO

[0066] The standard MOHO algorithm simulates three behavioral patterns of hippos: river / pond location updates (Phase 1), predator defense (Phase 2), and predator escape (Phase 3). Within a multi-objective framework, the algorithm evaluates fitness through non-dominated sorting and crowding distance, and manages the Pareto non-dominated solution set through an external archive.

[0067] 2) Improvement strategies to enhance algorithm adaptability

[0068] This invention introduces the following key improvements to the standard MOHO algorithm to enhance its global optimization and exploration capabilities:

[0069] Strategy 1: Sinusoidal Chaotic Map Initialization

[0070] This invention employs a sinusoidal chaotic mapping x k+1 = a \sin(p ixk This method replaces traditional random number generation. It leverages the nonlinear characteristics of chaotic systems to ensure that the initial population more comprehensively covers the entire search space, thereby effectively improving the algorithm's global exploration capability.

[0071]

[0072] Strategy 2: Dynamic adjustment of Lévy beta parameters

[0073] To give the algorithm more flexible exploration behavior in the predator defense phase (Phase 2), the beta parameter of Lévy flight in this invention is no longer fixed, but is dynamically adjusted according to the distance between the hippo and the "predator" through a dynamic Lévy beta function: the closer the individual is to the predator, the larger the beta value, prompting the hippo to make longer and more aggressive jumps to escape the local area; conversely, when the distance is far, the beta decreases, which is conducive to fine exploration of the local area.

[0074]

[0075] Strategy 3: Exploratory Development of Balance Factor y_1 Based on Sinusoidal Decay

[0076] To achieve a smoother, more adaptive balance between exploration and development, this invention introduces the function `calculate_y1_sine_decay` in the first exploration phase, generating a y_{1\_dynamic} factor based on a decreasing sine function. This factor has a large value in the early stages of the algorithm (when the number of iterations t is small), encouraging global exploration; as the number of iterations increases, its value gradually decreases, prompting the algorithm to shift its focus to local development, thereby accelerating convergence.

[0077]

[0078] Specifically, the steps include the following:

[0079] S2.1 Sinusoidal Chaotic Map Initialization

[0080] To enhance the diversity of the initial population and prevent the algorithm from getting trapped in local optima too early, this invention uses sinusoidal chaotic mapping instead of the traditional random initialization method. The iterative formula for sinusoidal chaotic mapping is shown in equation (3):

[0081]

[0082] Where α is a control parameter, typically ranging from 0.9 to 1.0; πx k Let be the value of the k-th chaotic sequence, with a value range of (0,1). The initial population of hippopotamus individuals generated through this mapping can more evenly cover the entire search space.

[0083] S2.2 Dynamic Lévy beta parameter adjustment

[0084] In the predator defense phase (Phase 2) of the standard MOHO algorithm, hippopotamus individuals update their position through Lévy flight to escape predators. To enhance the algorithm's adaptability in this phase, this invention introduces a dynamic Lévy beta parameter adjustment mechanism. This mechanism dynamically adjusts the beta parameter of the Lévy distribution based on the distance d between the hippopotamus individual and the "predator," as shown in the following formula:

[0085]

[0086] Where, β min and β max These represent the minimum and maximum values ​​of the β parameter, respectively, and λd is the decay coefficient. The closer the individual is to the predator (the smaller d is), the greater the β... dynThe larger the β value, the greater the probability of Lévy flight producing a longer stride, prompting individuals to make large jumps to escape localized areas; when the distance is greater, β... dyn The algorithm is relatively small and focuses on fine-grained local search.

[0087] S2.3 Exploratory Development of Balance Factor Based on Sinusoidal Decay

[0088] To achieve a smooth and adaptive transition between the exploration and development phases of the algorithm, this invention introduces a dynamic factor y1 based on a decreasing sine function in the first exploration phase of the algorithm. dynamic (t). The calculation method for this factor is shown in the formula:

[0089]

[0090] Where t is the current iteration number, T max This represents the maximum number of iterations. In the initial stage of the algorithm, t, y_ 1, dynamic A larger value encourages individual hippos to conduct large-scale global exploration; as the number of iterations increases, this factor gradually decreases, prompting the algorithm to shift its search focus to local exploration, thereby accelerating convergence.

[0091] S2.4 Decision Variable Vectors and Optimization Process

[0092] In the improved MOHO algorithm, the position of each hippopotamus corresponds to a decision variable, as shown in the following formula: Where Switchl (l=1,2,...L) represents the reconfiguration switch state of the integer ring network code, and L is the number of independent loops (L=5 in the IEEE 69-node system); LOC DG DG installation location (busbar number); Cap DG DG installation capacity; PF DG This represents the power factor for DG operation. The algorithm iteratively updates the position of individual hippos, continuously approaching the Pareto optimal front.

[0093] S3: Multi-objective assessment and record updating

[0094] The candidate solutions (hippo individuals) generated by the algorithm are transformed from fitness vectors into Pareto front solution sets, while ensuring the convergence and diversity of the optimal solution set.

[0095] Each hippopotamus individual X no longer has a single fitness value, but is represented by a cost vector f(X) = [F1(X), F2(X), F3(X)] containing all objective function values. The evaluation of objective functions F1 (energy loss cost) and F3 (DG operating cost) must be based on the expected value of the scenario, that is, a probability-weighted summation of the objective values ​​under all Ns scenarios.

[0096]

[0097] Non-dominated sorting and archive management: Domination judgment: The algorithm uses the Pareto dominance principle to compare the quality of solutions. If solution A is no worse than solution B on all objectives and is better than solution B on at least one objective, then A dominates B. Non-dominated sorting: In each iteration, the algorithm performs non-dominated sorting on the population, dividing it into different Pareto lines. External archive maintenance: The algorithm maintains a separate external archive specifically for storing all non-dominated solutions discovered throughout the optimization process. Update mechanism: Newly generated non-dominated solutions are added to the archive; if any existing solution in the archive is dominated by a new solution, it is removed. Diversity and leader selection: Diversity maintenance: To ensure the diversity of the solution set in the archive, the algorithm initiates a pruning mechanism when the number of solutions in the archive exceeds the preset maximum capacity. This mechanism identifies and removes solutions located in crowded regions by calculating the crowding distance of each solution in the objective space. Leader selection: When guiding the population movement, the algorithm selects a "leader" from the archive. To encourage exploration of different frontier areas, MOHO tends to select leaders from solutions located in sparser areas of the archive, using crowding information and roulette wheel selection to increase the chances of expanding the search into unexplored areas.

[0098] Specifically, this step aims to perform multi-objective evaluation of candidate solutions and maintain the Pareto optimal solution set through non-dominated sorting and archive management mechanisms.

[0099] Energy loss cost F1: As shown in the formula, it represents the expected cost of network active power loss under all operating scenarios.

[0100] ;

[0101] P s Let C be the probability of the s-th scene. loss P is the cost per unit of electrical energy loss. loss,s Let T be the active power loss in the s-th scenario, and T be the running time (usually taken as 8760 hours per year).

[0102] DG investment cost F2: As shown in the formula, this is the annual value investment cost of the DG equipment.

[0103]

[0104] Among them, C inv,i Cap is the unit capacity investment cost. DG , i Let r be the installed capacity of the i-th DG, r be the discount rate, and n be the service life of the equipment.

[0105] DG operating cost F3: The expected value of DG operation and maintenance costs under all operating scenarios.

[0106] S3.2 Constraint Verification

[0107] For each candidate solution, it is necessary to traverse all Ns running scenarios, perform power flow calculations, and verify the following running constraints:

[0108] Node voltage constraint: All node voltages U j It must be kept within a safe range, as shown in the formula:

[0109] Branch power constraint: The transmission power of each branch shall not exceed its capacity limit.

[0110] DG capacity constraint: The actual output of the DG must not exceed its installed capacity.

[0111] Radial topology constraint: The reconstructed network must maintain a radial structure, and all load nodes must be powered, with no islands.

[0112] S3.3 Non-dominated sorting and record management

[0113] In each iteration, the algorithm performs a non-dominated sort of the population, dividing the solutions into different Pareto Fronts. Simultaneously, a separate external archive is maintained to store all non-dominated solutions discovered throughout the optimization process. When the number of solutions in the archive exceeds a preset maximum capacity, a pruning mechanism is initiated: by calculating the crowding distance of each solution in the target space, solutions located in crowded regions are removed to ensure the diversity of the Pareto front distribution.

[0114] In the leader selection strategy, the algorithm tends to select leaders from solutions located in sparser regions of the archive, using a roulette wheel selection mechanism to increase the chance of the search expanding into unexplored areas.

[0115] S4: Refactoring Scheme Output

[0116] After the algorithm finishes running, it outputs the final non-dominated solution set stored in an external archive. This solution set constitutes the Pareto front of the multi-objective optimization problem of the distribution network. Decision-makers can select a final reconfiguration scheme from this front based on actual operational needs and preferences.

[0117] The final solution includes the following specific details:

[0118] Reconfigure switch states: Determine the segment switch numbers that need to be turned on in each independent loop to form the optimal radial topology.

[0119] DG configuration scheme: Determine the installation location (bus number), optimized capacity (in MV), and operating power factor of the DG.

[0120] Objective function value: Output the expected values ​​of the corresponding power loss cost, DG investment cost, and DG operating cost.

[0121] Performance constraints: Safety constraints such as minimum output node voltage ensure that the solution meets the safety requirements under all operating scenarios.

[0122] This invention employs scenario-based stochastic optimization to accurately capture the uncertainties of load and DG, making the optimization scheme more robust and adaptable in real-world complex operating environments.

[0123] This invention employs integer-type ring network coding, which significantly reduces the dimensionality of the reconstruction problem (e.g., from 68 dimensions to 5 dimensions), effectively shrinks the search space, and significantly improves computational efficiency.

[0124] This invention significantly enhances the global exploration capability and convergence accuracy of MOHO through a hybrid improvement strategy, including sinusoidal chaotic initialization and dynamic Lévy beta adjustment, ensuring that the algorithm can find a Pareto front solution set with better convergence and uniformity.

[0125] By simultaneously minimizing network loss costs, DG investment costs, and DG operating costs, this study comprehensively evaluates the economic benefits of DG access and provides a more practically guiding optimal configuration solution.

[0126] Figure 2 This is a functional block diagram of a power distribution network reconfiguration system provided in an embodiment of the present invention. Figure 2 As shown, the present invention also provides a multi-objective cooperative optimization and reconfiguration system for distribution networks to implement the above method, the system comprising:

[0127] Data acquisition module: used to acquire the operating parameters, topology information, and predicted data of distributed power sources and loads of the distribution network.

[0128] Scene generation module: Used to generate multiple discrete operating scenarios based on the uncertainty model of load and DG, and to reduce the number of scenarios through clustering algorithm.

[0129] Algorithm control module: used to execute the above-mentioned improved multi-objective hippo optimization algorithm. This module further includes a sinusoidal chaotic initialization unit, a dynamic parameter adjustment unit, and an exploration-expansion balancing unit.

[0130] Fitness evaluation module: Used to evaluate the multi-objective expected cost vector of candidate solutions based on scenario probabilities and perform power flow calculation.

[0131] Constraint checking module: used to verify whether candidate reconfiguration schemes meet radial topology constraints, bus voltage constraints, and power balance constraints.

[0132] The results output module is used to output the optimal DG configuration parameters and the reconfiguration switch status.

[0133] In summary, the multi-objective collaborative optimization and reconfiguration method and system for distribution networks based on the improved multi-objective Hippo Optimization (MOHO) algorithm proposed in this invention effectively handles the uncertainties of load and distributed generation (DG) through scenario generation technology, significantly reduces the dimensionality of decision variables by employing integer ring network coding, and significantly improves the global optimization capability and convergence accuracy of the MOHO algorithm through improved strategies such as sinusoidal chaotic initialization, dynamic Lévy beta adjustment, and sinusoidal decay balance factor. This method can obtain Pareto front solution sets with better convergence and more uniform distribution, providing a collaborative optimization and reconfiguration scheme for distribution networks with DG that is both economical and robust.

[0134] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0135] It should be noted that the purpose of disclosing the embodiments is to help further understand the present invention; however, those skilled in the art will understand that various substitutions and modifications are possible without departing from the spirit and scope of the present invention and the appended claims. Therefore, the present invention should not be limited to the content disclosed in the embodiments, and the scope of protection of the present invention is defined by the claims.

Claims

1. A multi-objective collaborative optimization and reconfiguration method for distribution networks based on an improved multi-objective hippo optimization algorithm, characterized in that: Includes the following steps: Step S1: Obtain the topology, load data, and distributed generation output data of the distribution network. Use a scenario-based stochastic optimization method to generate multiple discrete operating scenarios to model the uncertainties of load and distributed generation output. Step S2: Construct a multi-objective fitness vector with the objective of minimizing the expected values ​​of power loss cost, distributed power generation investment cost, and distributed power generation operating cost; Step S3: Generate an initial population of hippopotamus individuals using a sinusoidal chaotic mapping initialization method, which serves as the initial solution set for the improved multi-objective hippopotamus optimization algorithm; Step S4: The improved multi-objective hippo optimization algorithm is used to optimize the decision variable vector to obtain the Pareto optimal solution set, wherein the decision variable vector includes the reconstructed switch state, the installation location of the distributed power source, the capacity of the distributed power source, and the power factor of the distributed power source; the improved multi-objective hippo optimization algorithm includes a dynamic adjustment mechanism for the beta parameter of the Lévy distribution. Step S5: For each candidate solution, traverse all operating scenarios, perform power flow calculations, evaluate the multi-objective function, and verify the bus voltage constraint, power balance constraint, and distributed source capacity constraint. Step S6: Output the Pareto optimal solution set to determine the optimal combination of distributed power source location, capacity setting, and network reconfiguration switching states.

2. The method for multi-objective collaborative optimization and reconfiguration of distribution networks based on the improved multi-objective hippo optimization algorithm according to claim 1, characterized in that, The step S1 describes using a scenario-based stochastic optimization method to generate multiple discrete operating scenarios, including: modeling the uncertainty of load demand using a normal distribution superimposed with random noise; modeling the uncertainty of solar distributed power output using a log-normal distribution superimposed with random noise and physical range truncation; generating an initial integrated operating scenario through Monte Carlo sampling, and simplifying the initial scenario into a typical scenario using the K-means clustering method.

3. The method for multi-objective collaborative optimization and reconfiguration of distribution networks based on the improved multi-objective hippo optimization algorithm according to claim 1, characterized in that, The sinusoidal chaotic mapping initialization method described in step S3 uses the following mapping formula to generate the initial population: ； Where α is the control parameter, x k Let be the value of the k-th chaotic sequence.

4. The method for multi-objective collaborative optimization and reconfiguration of distribution networks based on the improved multi-objective hippo optimization algorithm according to claim 1, characterized in that, The dynamic adjustment mechanism for the Lévy distribution beta parameter described in step S4 is as follows: the beta parameter of Lévy flight is dynamically adjusted according to the distance between the hippopotamus individual and the predator. The beta value is increased when the individual is closer to the predator and decreased when the individual is farther away from the predator.

5. The method for multi-objective collaborative optimization and reconfiguration of distribution networks based on the improved multi-objective hippo optimization algorithm according to claim 1, characterized in that, The improved multi-objective hippo optimization algorithm also includes: introducing a dynamic factor based on a decreasing sine function in the first exploration phase. The dynamic factor is used to adjust the balance between exploration and development in the algorithm. It takes a larger value in the early stage of algorithm iteration to encourage global exploration and a smaller value in the later stage of iteration to promote local development.

6. The method for multi-objective collaborative optimization and reconfiguration of distribution networks based on the improved multi-objective hippo optimization algorithm according to claim 1, characterized in that, In step S4, the reconfiguration switch state in the decision variable vector is represented by an integer ring network coding scheme, which reduces the dimension of the reconfiguration switch to be equal to the number of independent loops formed after closing the tie switch; the integer ring network coding scheme reduces the dimension of the reconfiguration switch in the IEEE 69-node system to 5 dimensions.

7. The method for multi-objective collaborative optimization and reconfiguration of distribution networks based on the improved multi-objective hippo optimization algorithm according to claim 1, characterized in that, When verifying the bus voltage constraint in step S5, it is necessary to ensure that the node voltage in all operating scenarios is within the preset safe range; when verifying the radial topology constraint, it is necessary to ensure that the network maintains a radial structure after reconstruction and that all loads are powered.

8. A system for multi-objective collaborative optimization and reconfiguration of distribution networks, characterized in that, include: The data acquisition module is used to acquire the operating parameters, topology information, and predicted data of distributed power sources and loads of the distribution network. The scene generation module is used to generate multiple discrete running scenes according to the method described in claim 7; The algorithm control module is used to execute the improved multi-objective hippo optimization algorithm according to any one of claims 1 to 7; The fitness evaluation module is used to evaluate the multi-objective expected cost vector of candidate solutions based on scenario probabilities. The constraint checking module is used to verify whether the candidate reconfiguration schemes meet the radial topology constraints, bus voltage constraints, and power balance constraints. The results output module is used to output the optimal distributed power supply configuration parameters and reconfigurable switch states.

9. The system for multi-objective collaborative optimization and reconfiguration of distribution networks according to claim 8, characterized in that, The algorithm control module also includes: A sinusoidal chaotic initialization unit is used to generate an initial population of individual hippos using a sinusoidal chaotic mapping. The dynamic parameter adjustment unit is used to dynamically adjust the beta parameter of the Lévy distribution based on the distance between the hippopotamus individual and the predator. Explore and develop balancing units to introduce and develop dynamic factor adjustment algorithms based on sinusoidal function decrement.

Images