Microgrid dispatching optimization method and device based on improved dung beetle optimization algorithm

Abstract

The invention discloses a micro-grid dispatching optimization method and device based on an improved dung beetle optimization algorithm, and the method comprises the steps: constructing a micro-grid dispatching model, building a dual-objective function of the operation and maintenance cost of a micro-grid and the environmental protection cost, and obtaining the operation parameters and constraint conditions of the micro-grid; optimizing the micro-grid dispatching model by using an improved multi-target dung beetle optimization algorithm; the improvement on the multi-target dung beetle optimization algorithm comprises the following steps: generating an initialized population with strong diversity by adopting a mixed initialization strategy of Latin hypercube sampling and cosine similarity reverse learning; the acceptance degree of young balls and dung beetles on the optimal solution is improved; a disturbance strategy of a following mechanism of the sparrow search algorithm is fused into a dung beetle algorithm; cauchy Gaussian variation is introduced, and variation amplitude is dynamically adjusted in an algorithm iteration process. Compared with the prior art, the method has more excellent optimization precision, and has important theoretical value and practical significance for improving the operation economy and environmental friendliness of the micro-grid.

Description

A microgrid scheduling optimization method and device based on an improved dung beetle optimization algorithm Technical Field

[0001] This application relates to the field of microgrid dispatch optimization technology, and in particular to a microgrid dispatch optimization method and apparatus based on an improved dung beetle optimization algorithm. Background Technology

[0002] Driven by dual carbon targets, the power grid is rapidly transitioning towards environmentally friendly and sustainable practices. The widespread implementation of distributed generation has become an effective solution for reshaping the power grid's energy structure. Therefore, scientific and rational microgrid dispatching is crucial for improving energy utilization and promoting the large-scale application and effective absorption of renewable energy. However, microgrid optimal dispatching is mostly a complex nonlinear optimization problem, characterized by numerous decision variables and complex constraints. Against this backdrop, how to comprehensively utilize intelligent optimization technologies and distributed power source collaborative control strategies from the perspectives of overall dispatch economic cost to maximize the operational performance of microgrids has become a current focus in this field.

[0003] With the continuous advancement of computer science and technology, researchers utilize computer technology to mathematically model microgrid systems, then employ optimization algorithms for optimal scheduling, ultimately obtaining the lowest total system scheduling cost and thus determining the optimal scheduling scheme for the microgrid system. Existing algorithms for microgrid scheduling optimization mainly include genetic algorithms, particle swarm optimization, and gravity search. These swarm intelligence algorithms have been widely applied in industrial production optimization. A microgrid is a small distribution network with independent power generation and distribution capabilities, composed of photovoltaics, wind turbines, diesel generators, micro gas turbines, batteries, and daily loads. Microgrid applications cover multiple scenarios including energy, residential, industrial, and emergency response, enabling efficient power supply and energy management tailored to different environmental needs. Microgrid scheduling is a complex, high-dimensional, non-differentiable optimization problem, and its optimal solution is difficult to find using traditional methods. Currently, research on microgrid optimal scheduling mainly focuses on model optimization, scheduling strategies, and intelligent optimization algorithms. However, traditional intelligent optimization algorithms, such as particle swarm optimization, genetic algorithms, and differential evolution, often suffer from getting trapped in local optima when applied to microgrid scheduling, resulting in low optimization accuracy.

[0004] Therefore, finding an efficient optimization algorithm that combines global and local search capabilities has become a key scientific problem to be solved in the field of microgrid optimization scheduling. Summary of the Invention

[0005] This application provides a microgrid scheduling optimization method and apparatus based on an improved dung beetle optimization algorithm. Its advantage is that it uses the improved dung beetle optimization algorithm to optimize the scheduling of the microgrid. Compared with the prior art, the present invention has superior optimization accuracy and has important theoretical value and practical significance for improving the economic efficiency and environmental friendliness of microgrid operation.

[0006] The technical solution of this application is as follows:

[0007] On the one hand, this application provides a microgrid scheduling optimization method based on an improved dung beetle optimization algorithm, including the following steps:

[0008] S1: Construct a microgrid scheduling model, establish a dual objective function for microgrid operation and maintenance costs and environmental protection costs, and obtain the microgrid's operating parameters and constraints;

[0009] S2: The microgrid scheduling model is optimized using an improved multi-objective dung beetle optimization algorithm to obtain the optimal microgrid scheduling scheme;

[0010] Improvements to the multi-objective dung beetle optimization algorithm include:

[0011] (1) A hybrid initialization strategy of Latin hypercube sampling and cosine similarity back learning is adopted to generate a highly diverse initial population;

[0012] (2) Improve the acceptance of optimal solutions by the baby ball and the dung beetle;

[0013] (3) Integrate the random perturbation strategy of the sparrow search algorithm into the dung beetle algorithm;

[0014] (4) Introduce Cauchy-Gaussian mutation and dynamically adjust the mutation amplitude during the algorithm iteration process;

[0015] S3: Optimize the microgrid scheduling using the optimal microgrid scheduling scheme.

[0016] Furthermore, microgrid operation and maintenance costs include the operation and maintenance costs of power generation equipment, battery charging and discharging costs, and electricity purchase and sales costs.

[0017] Furthermore, the environmental protection costs of microgrids include the costs of treating pollutants from grid interactions and the costs of treating pollutants generated by power generation equipment.

[0018] Furthermore, between steps S1 and S2, there is also a step: initializing the operating parameters of the microgrid.

[0019] Furthermore, in the improvement of the multi-objective dung beetle optimization algorithm (1), Latin hypercube sampling is used to generate the dung beetle initial population, and the steps are as follows:

[0020] Determine the sampling size H, i.e., the dung beetle population size;

[0021] The position variable d of each dung beetle i Domain interval [d l i ,d u i Divide the data into H equal intervals, that is:

[0022] d l i =d0 i <d1 i <···<d H i =d u i

[0023] Generate an H×n matrix A, where each column is a randomized sort of the sequence 1, 2, ..., H;

[0024] Each row of matrix A corresponds to a hypercube, and each hypercube generates one sample;

[0025] While using Latin hypercube sampling to generate the initial population, a cosine similarity back-learning strategy is introduced to optimize the population structure.

[0026] Furthermore, in the improvement of the multi-objective dung beetle optimization algorithm (2), the acceptance of the juvenile ball and the thieving dung beetle to the local optimal position or the global optimal position is dynamically changed according to the fitness of the dung beetle.

[0027] Furthermore, in the improvement of the multi-objective dung beetle optimization algorithm (3), the optimal position of the discoverer in the sparrow search algorithm is defined as the global optimal position in the dung beetle optimization algorithm. The perturbation probability is adaptively changed according to the iteration depth, and a greedy strategy is used to decide whether to retain the perturbation. The perturbation probabilities are as follows:

[0028]

[0029] Where t represents the current iteration number and T represents the maximum iteration number.

[0030] Furthermore, in the improvement of the multi-objective dung beetle optimization algorithm (4), after each random perturbation of the sparrow search algorithm, it is determined whether the population is premature. The determination method is as follows: if the optimal solution of the population remains unchanged after 5 iterations, it is determined that the algorithm is premature and the Cauchy-Gaussian mutation strategy is adopted for the entire population to make the population jump out of the local optimal solution.

[0031] On another front, this application provides a microgrid scheduling optimization device based on an improved multi-objective dung beetle optimization algorithm, comprising a processor and a memory, wherein the memory stores a computer program, and when the computer program is called and executed by the processor, it implements the steps in the method described above.

[0032] In summary, the beneficial effects of this application are as follows: Compared with single-objective optimization, multi-objective optimization not only preserves the independence and conflict of objectives, but also provides more comprehensive decision-making information, finding more suitable solutions for complex scenarios in microgrids where economic and environmental protection need to be dynamically balanced. Compared with traditional multi-objective dung beetle optimization algorithms, the joint strategy-improved multi-objective dung beetle optimization algorithm proposed in this invention not only has a more reasonable transition between the global exploration and local development stages, but also further enhances the utilization of information within the population. This makes it easier for the proposed method to escape local optima and find potential global optimal solutions in microgrid scheduling.

[0033] This application proposes a joint strategy-improved multi-objective dung beetle optimization algorithm for microgrid scheduling. Based on existing multi-objective dung beetle optimization algorithms, this method first introduces a hybrid initialization improvement strategy combining Latin hypercube and cosine similarity back-learning to generate a more uniform and higher-quality initial solution set. Then, it improves the acceptance of optimal solutions by the baby ball and the thieving dung beetle, enhancing the algorithm's search capability. Next, it integrates a perturbation strategy based on randomness from the sparrow search algorithm into the multi-objective dung beetle algorithm, improving its convergence speed and accuracy. Finally, it introduces Cauchy-Gaussian mutation to increase the probability of escaping local optima, enhancing the algorithm's optimization accuracy. The organic combination of these strategies enhances the algorithm's spatial exploration capability and optimization accuracy. Applying this method to solve the mathematical model of the constructed microgrid yields the optimal scheduling scheme, ensuring the minimization of the total cost of microgrid optimization scheduling. Attached Figure Description

[0034] Figure 1 is a schematic diagram of a typical microgrid system;

[0035] Figure 2 is a flowchart of the DBO algorithm;

[0036] Figure 3 is a flowchart of the JSI-MODBO algorithm in an embodiment of this application;

[0037] Figure 4 is a graph showing the predicted output power and load demand data of WT and PV in the embodiments of this application;

[0038] Figure 5 is a real-time electricity price data chart under the demand response policy in the embodiments of this application;

[0039] Figure 6 is a comparison of the Pareto fronts of different multi-objective algorithms in the embodiments of this application;

[0040] Figure 7 is a diagram of the microgrid scheduling results obtained by JSI-MODBO in the embodiments of this application. Detailed Implementation

[0041] The specific embodiments of this application are described in detail below with reference to the accompanying drawings.

[0042] A specific embodiment of this application provides a microgrid scheduling optimization method based on an improved dung beetle optimization algorithm, comprising the following steps:

[0043] Construct a microgrid scheduling model.

[0044] This application illustrates a typical microgrid, as shown in Figure 1, which includes a photovoltaic (PV) system, a wind turbine (WT), a diesel generator (DG), a micro gas turbine (MT), a battery (BT), and daily loads. The PV, WT, DG, and MT are the power generation equipment.

[0045] A dual objective function for microgrid operation and maintenance costs and environmental protection costs is established, and the operating parameters and constraints of the microgrid are obtained.

[0046] Microgrid operation and maintenance costs include the operation and maintenance costs of diesel generators and micro gas turbines, battery charging and discharging costs, and electricity purchase and sales costs. The economic objective function can be expressed as follows:

[0047]

[0048] In the formula, C DG (t) represents the operating and management cost of DG during time period t. C MT (t) represents the operating and management cost of MT during time period t. C BT (t) represents the operating and management cost of BT during time period t. C GIRD (t) represents the interaction cost between the microgrid and the main grid during time period t. The expressions for each part are as follows:

[0049]

[0050] In the formula, P DG (t) represents the output of DG at a certain moment. K DG.OM α represents the maintenance cost coefficient of DG. α, β, and γ represent the fuel cost coefficients of DG.

[0051]

[0052] In the formula, P MT (t) represents the output of MT at a certain moment. K MT.OMV represents the MT maintenance cost coefficient. C represents the natural gas price. LH This represents the lower heating value of natural gas.

[0053]

[0054] In the formula, P BT (t) represents the output of BT at a certain moment. K BT This represents the BT maintenance cost coefficient.

[0055]

[0056] In the formula, c buy (t) represents the electricity price of the main power grid during time period t. P buy (t) represents the power purchased by the microgrid and the main grid during time period t. sell (t) represents the electricity price of the microgrid during time period t. P sell (t) represents the electricity sales power of the microgrid and the main grid during time period t.

[0057] The environmental protection costs of microgrids include the pollution treatment costs of main grid interconnection, the pollution treatment costs of micro gas turbines, and the pollution treatment costs of diesel engines. The objective function for environmental protection costs can be expressed by the following formula:

[0058]

[0059] In the formula, C GIRD.EN (t) represents the pollutant treatment cost of the main network interaction during time period t. C MT.EN (t) represents the pollutant treatment cost at time t (MT). C DG.EN (t) represents the cost of treating the pollutants generated by DE during time period t. The expressions for each part are as follows:

[0060]

[0061] In the formula, C k q represents the cost coefficient for treating pollutant class k. GIRD.k This represents the emissions of pollutants of type k generated by the operation of a large power grid. q MT.k This represents the emissions of pollutants of type k generated during the operation of MT. q DG.k P represents the emissions of type k pollutants generated during DE operation. DG (t) represents the output power of the diesel generator during time period t.

[0062] In this microgrid dispatch optimization scenario, there is a significant mutual constraint between operation and maintenance costs (Objective 1) and environmental protection costs (Objective 2): if the focus is on reducing operation and maintenance costs (such as increasing the output ratio of traditional diesel generators and micro gas turbines, or increasing the amount of electricity purchased from the main grid during periods of low electricity prices), it often leads to an increase in pollutant emissions, thereby increasing environmental protection costs; conversely, if the focus is on strictly controlling pollution (such as prioritizing the consumption of clean energy such as photovoltaic and wind power, and reducing the output of high-emission power sources), it may lead to an increase in operation and maintenance costs due to factors such as unstable output of clean energy and energy storage charging and discharging losses.

[0063] The core objective of this invention is to find a Pareto optimal solution that balances economic efficiency and environmental friendliness. Specifically, it seeks a set of scheduling schemes that minimizes the total operating cost of the microgrid (the sum of operation and maintenance costs and environmental protection costs) without causing excessive environmental pollution (environmental costs are controlled within a reasonable range, meeting pollutant emission constraints). This compromise solution must balance economic efficiency and environmental friendliness, avoiding both neglecting pollution control in pursuit of low costs and preventing uncontrolled operation and maintenance costs due to overemphasis on environmental protection. Ultimately, it adapts to the complex dynamic operation of microgrids, achieving synergistic optimization of economic efficiency and environmental protection.

[0064] Since multi-objective optimization does not have a unique "optimal value," but only a set of Pareto optimal values ​​(compromise solutions), each combination of values ​​in the set cannot optimize one objective without sacrificing another. Ultimately, an optimal value must be selected based on decision preferences. Common methods include weighted summation, ideal point method, and analytic hierarchy process (AHP). This invention uses the weighted summation method to select an optimal value. The solution with the lowest total operating cost is considered optimal; therefore, the "optimal solution" is:

[0065]

[0066] The microgrid dispatch optimization steps are as follows: The entire microgrid dispatch optimization process includes: ① Initializing the parameters of distributed power sources, energy storage and loads in the microgrid; ② Using the multi-objective dung beetle optimization algorithm to solve the microgrid dispatch optimization model and obtain the optimal dispatch scheme; ③ Based on the optimal dispatch scheme obtained by optimization, obtaining its operation and maintenance cost and environmental protection cost, that is, completing the microgrid dispatch optimization objective.

[0067] An improved multi-objective dung beetle optimization algorithm is used to optimize the microgrid scheduling model and obtain the optimal scheduling scheme for the microgrid.

[0068] The Dung Beetle Optimizer (DBO), proposed by Jianka Xue and Bo Shen in 2022, is a novel swarm intelligence optimization algorithm inspired by the dung beetle's ball-rolling, dancing, foraging, stealing, and reproductive behaviors. This algorithm considers both global exploration and local exploitation, resulting in fast convergence and high accuracy, effectively solving complex optimization problems. This paper will explain the algorithm's principles and its program implementation.

[0069] In DBO (Dungeon Bottom), each dung beetle's location corresponds to a solution. Dung beetles exhibit five foraging behaviors: rolling (rolling dung into a ball and using celestial cues for navigation to align the ball in a straight line); dancing (allowing them to reorient themselves); foraging (some adult dung beetles emerge from the ground to search for food); stealing (some dung beetles, known as thieves, steal dung balls from other dung beetles); and reproduction (in nature, dung beetles roll their dung balls to safe locations to reproduce). Therefore, the dung beetle population in the algorithm is divided into four groups: rolling dung beetles, brooding dung beetles, baby dung beetles, and thieves, in a ratio of 6:6:7:11. Dung beetles are constantly changing their course due to various natural environmental influences. They initially search for safe foraging locations, and the brooding balls are laid in known safe areas. The dung beetles that grow into adults are called baby dung beetles. Baby dung beetles will forage in the best foraging areas. Dung beetles will also search for food based on the location of other dung beetles and the best foraging areas.

[0070] The four methods for updating dung beetle locations are as follows:

[0071] (1) Dung beetle

[0072] Dung beetles use the sun for navigation to ensure their dung balls roll along a straight path. Natural factors such as light intensity and wind affect their movement. The dung beetle updates its position as follows:

[0073]

[0074] Where t represents the current iteration number; x i (t) represents the position information of the i-th dung beetle in the t-th iteration; α is a natural coefficient indicating whether it deviates from the original direction, assigned as −1 or 1 according to the probability method; k represents the deflection coefficient, which is taken as 0.1 in this patent; b represents a constant, which is taken as 0.3 in this patent; X w This represents the worst-case position globally, and Δx is used to simulate changes in light intensity.

[0075] When a dung beetle encounters an obstacle and cannot move forward, it needs to dance to adjust its direction. The formula for the dung beetle's dance to update its position is defined as follows:

[0076]

[0077] In the formula, θ∈[0, π] represents the deflection angle. When θ is equal to 0, π / 2 or π, the position of the dung beetle will not be updated.

[0078] (2) Brooding ball

[0079] The brooding ball uses a boundary selection strategy to simulate the oviposition area of ​​a female dung beetle. The oviposition area is defined as follows:

[0080]

[0081] Among them, X ∗ Lb represents the current local optimum. ∗ and Ub ∗ represents the lower and upper bounds of the spawning zone, respectively; t and T represent the current iteration number and the maximum iteration number, respectively; Lb and Ub represent the lower and upper bounds of the optimization problem, respectively.

[0082] During the iteration process, the position of the brooding ball changes dynamically, and is defined as follows:

[0083]

[0084] Among them, B i (t) represents the position information of the i-th brooding ball in the t-th iteration, and b1 and b2 represent two independent random vectors of size 1×D, where D represents the dimension of the optimization problem.

[0085] (3) Dung beetle

[0086] An optimal foraging area needs to be established to guide dung beetle larvae in finding food and to simulate their foraging behavior. The optimal foraging area is defined as follows:

[0087]

[0088] Among them, X b Lb represents the current local optimum. b and Ub b These represent the lower and upper limits of the optimal foraging zone, respectively.

[0089] The location of the dung beetle has been updated as follows:

[0090]

[0091] Where, x i(t) represents the position information of the i-th dung beetle in the t-th iteration, C1 represents a random number following a normal distribution, and C2∈(0,1) represents a random vector.

[0092] (4) Dung beetle

[0093] X b This is the optimal location for food competition, therefore the dung beetle's location update method is as follows:

[0094]

[0095] Where, x i (t) represents the position information of the i-th dung beetle in the t-th iteration, g represents a random vector of size 1×D that follows a normal distribution, and S represents a constant value.

[0096] The flowchart of the dung beetle algorithm is shown in Figure 2.

[0097] An improved multi-objective dung beetle optimization algorithm is proposed using a joint strategy, which is defined as the Joint Strategy Improved Multi-Objective Dung Beetle Optimization Algorithm (JSI-MODBO).

[0098] Initializing the population: The mathematical model of the dung beetle optimization algorithm is described as follows: Let the population size of the dung beetles be P, and the initial dung beetle population be X(0) = {X1, X2, ...} X P If the dimension of the feasible solution space is d, then the i-th dung beetle in the solution space can be represented as X. i = [x i1 , x i2 , , x id The initial dung beetle swarm in the dung beetle optimization algorithm is randomly generated within a certain search range and evenly distributed in the solution space.

[0099] Randomly forming the initial population refers to the optimization of the output of each distributed generation unit at each time step during the microgrid dispatch optimization process. That is, each dispatch scheme uses one dung beetle X i This indicates that the output of each power generation unit at each moment is represented by x. i This means that the output of each power generation device and the number of dung beetles P are determined, and they are combined to form a dung beetle swarm. That is, one dung beetle is used to represent a solution for a set of microgrid scheduling situations. Within a certain range of search parameters, each dung beetle in the swarm is determined using a random rand method (i.e., the working state of the power generation unit is used as the value corresponding to the particle in the swarm), forming the initial population.

[0100] Improvements to the multi-objective dung beetle optimization algorithm include:

[0101] Improvement (1) For microgrid scheduling optimization, the dung beetle optimization algorithm balances the global and local search performance of the algorithm well. However, the initial population distribution of the dung beetle optimization algorithm is not uniform enough, which may affect the convergence speed of the algorithm and cause the algorithm to get stuck in local optima, ultimately resulting in poor optimization accuracy. In view of the above-mentioned defects of the dung beetle optimization algorithm, this embodiment proposes a hybrid initialization strategy of Latin hypercube sampling and cosine similarity back learning to generate a highly diverse initial population.

[0102] Latin hypercube sampling is a statistical method for sampling in multidimensional parameter spaces, widely used in uncertainty quantification, sensitivity analysis, and engineering simulation. The key to Latin hypercube sampling is stratifying the input probability distribution and drawing one sample from each stratum. This ensures a more uniform distribution of the dung beetles' initial positions, avoiding premature convergence. The steps of Latin hypercube sampling are as follows:

[0103] Determine the sampling size H, i.e., the dung beetle population size;

[0104] The position variable d of each dung beetle i Domain interval [d l i ,d u i Divide the data into H equal intervals, that is:

[0105] d l i =d0 i <d1 i <···<d H i =d u i

[0106] Generate an H×n matrix A, where each column is a randomized sort of the sequence 1, 2, ..., H;

[0107] Each row of matrix A corresponds to a hypercube, and each hypercube generates one sample;

[0108] While using Latin hypercube sampling to generate the initial population, a cosine similarity back-learning strategy is introduced to optimize the population structure, which helps to improve the diversity of the initial population and thus further improve the quality of the initial solution. The mathematical expression of the cosine similarity back-learning strategy is shown in the following equation:

[0109]

[0110] Where, x i *This represents the generated reverse solution; Lb and Ub represent the lower and upper bounds of the variable, respectively; N represents the population size.

[0111] Improvement (2) Improve the acceptance of the optimal solution by the dung beetle and the dung beetle; In the basic dung beetle optimization algorithm, the dung beetle updates its position based on the local optimal position, and the dung beetle also updates its position based on the global optimal position. It can be seen that the dung beetle and the dung beetle will keep moving closer to the local optimal position and the global optimal position. However, in the early stage of the algorithm, the search range of the group should be expanded as much as possible. Although moving quickly to the local optimal position or the global optimal position can make the algorithm converge quickly, there is also a certain probability that the algorithm will stagnate and fail to converge to the global optimal solution. Therefore, a strategy is designed to dynamically change the acceptance of the dung beetle and the dung beetle to the local optimal position or the global optimal position based on the strength of the dung beetle's search ability (fitness). The specific method is as follows: The dung beetle group is sorted according to its search ability (fitness value). The dung beetle and the dung beetle update their positions according to their own search ability as follows:

[0112]

[0113] Among them, rank i (t) represents the rank of the i-th dung beetle in the t-th iteration; pop represents the size of the dung beetle population.

[0114] Improvement (3) Integrating the perturbation strategy of the Sparrow Search Algorithm's randomization mechanism into the Dung Beetle Algorithm; By integrating the perturbation strategy of the Sparrow Search Algorithm's randomization mechanism, the Sparrow Algorithm has the characteristics of high convergence accuracy, fast convergence speed, and strong robustness. In terms of function optimization problems, it is superior to swarm intelligence algorithms such as particle swarm optimization and gray wolf optimization. Therefore, the follower position update mechanism in the Sparrow Search Algorithm is introduced into the Dung Beetle Optimization Algorithm as a perturbation strategy, and the optimal position of the discoverer in the Sparrow Search Algorithm is defined as the global optimal position in the Dung Beetle Optimization Algorithm. If the perturbation is too strong, the group is prone to fall into a chaotic and disordered state; if the perturbation is too weak, the perturbation strategy will not play a role; therefore, the perturbation probability is adaptively changed according to the iteration depth, and a greedy strategy is used to decide whether to retain this perturbation. The perturbation probability is as follows:

[0115]

[0116] Where t represents the current iteration number and T represents the maximum iteration number.

[0117] The follower position update strategy in the sparrow search algorithm is as follows:

[0118]

[0119] Where: X i,j (t) represents the position information of the j-th dimension of the i-th sparrow in the t-th iteration; α represents a random number within the range [0, 1]; ST represents the safety value, which is taken as 0.7 in this patent; Q represents a random number obeying the normal distribution within the range [0, 1]; L represents a 1×d matrix of all 1s; T represents the maximum number of iterations.

[0120] In the early stage of iteration, the perturbation probability is relatively large, which can give full play to the role of the perturbation of the sparrow following mechanism, so as to improve the population quality and the optimization performance of the algorithm; in the later stage of iteration, the perturbation probability gradually decreases, which has the effect of accelerating the convergence speed of the algorithm. The specific perturbation steps are as follows: a) Calculate the perturbation probability pv of the i-th individual according to the above formula; b) Generate a random number r ∈ (0, 1). If r < pv, go to step c), otherwise the i-th individual does not undergo perturbation; c) Generate a random number r1 ∈ (0, 1). If r1 < ST, generate a new solution according to the second formula in the sparrow algorithm, otherwise generate a new solution according to the first formula. After a large number of experiments, it is proved that the effect is the best when ST = 0.7. This means that 70% of the individuals in the population mutate towards the optimal position of the population, and 30% of the individuals mutate towards a position worse than themselves; d) Calculate the fitness value of the new solution. If it is better than the fitness value before perturbation, replace the solution before perturbation with the new solution, otherwise this perturbation is invalid.

[0121] Improvement (4) is to introduce Cauchy-Gaussian mutation to deal with the multi-peak situation, dynamically adjust the mutation amplitude during the algorithm iteration process, increase the probability of the algorithm jumping out of the local optimal solution, and enhance the optimization accuracy of the algorithm. After each perturbation of the sparrow search algorithm following mechanism, it is judged whether the population is premature. The judgment method is: if the optimal solution remains unchanged after 5 iterations of the population, it is determined that the algorithm has fallen into stagnation (premature), and the Cauchy-Gaussian mutation strategy is adopted for the entire population to make the population jump out of the local optimal solution. The Cauchy-Gaussian mutation method is as follows:

[0122]

[0123] Where, x i new is the position after the initial position x is perturbed and updated, t is the current iteration number, and T is the maximum number of iterations

[0124] This invention proposes a joint strategy improved multi-objective dung beetle optimization algorithm (JSI-MODBO). The main improvements of this algorithm include the following aspects: (1) The initial solution of the population is updated by a hybrid initialization improvement strategy of Latin hypercube and cosine similarity back learning to form a uniform and diverse initial population and enhance the global search capability; (2) The acceptance of the optimal solution by the juvenile ball and the thieving dung beetle is improved so that the dung beetle can expand the search range of the population in the early iteration and enhance the global optimization performance of the algorithm; (3) The perturbation strategy of the random mechanism of the sparrow search algorithm is integrated to further enhance the algorithm's ability to jump out of local optima and enhance the optimization accuracy and convergence speed of the algorithm; (4) Cauchy-Gaussian mutation is introduced. This mechanism combines the long-tail characteristics of Cauchy distribution with the local search advantage of Gaussian distribution. By dynamically adjusting the mutation amplitude during the algorithm iteration process, it can maintain the stability of local fine search in the later stage of optimization and trigger a large-amplitude random perturbation when trapped in a local optimum. This dual characteristic effectively breaks the algorithm's dependence on local optima, significantly increasing the probability of escaping local extrema. Simultaneously, by reasonably balancing exploration and exploitation capabilities, it further reduces optimization errors, ultimately achieving a simultaneous improvement in both optimization accuracy and global convergence efficiency. The flowchart of the JSI-MODBO algorithm is shown in Figure 3.

[0125] The optimal scheduling scheme for microgrids is used to optimize the control of microgrid scheduling.

[0126] The specific steps for microgrid dispatch optimization are as follows:

[0127] Microgrid parameter initialization: Initialize the microgrid's operating parameters, including the operating parameters of DG, MT, BT, WT and PV power forecast data, and daily load.

[0128] JSI-MODBO initialization: First, the population is initialized using Latin hypercube sampling and cosine similarity back-learning. Then, the fitness of the initial population is calculated to generate the initial Pareto front, preparing for JSI-MODBO iterations.

[0129] Iterative optimization: First, update the position of the dung beetle and the foraging dung beetle according to their position update strategies. Then, calculate the acceptance of the optimal solution by the improved juvenile dung beetle and the stealing dung beetle, update the position of the juvenile dung beetle and the stealing dung beetle, and then calculate the perturbation probability. If the perturbation condition is met, perturb the dung beetle. Finally, determine whether the optimal solution has changed after 5 iterations. If it has not changed, use Cauchy-Gaussian mutation to perturb the entire population.

[0130] Algorithm iteration termination: Determine if the algorithm iteration is complete. If not, restart the iteration optimization. Additionally, obtain the Pareto optimal front and optimal scheduling scheme, and the calculation is complete.

[0131] The following simulation experiment demonstrates a model of microgrid optimization scheduling using a multi-objective dung beetle optimization algorithm with a joint strategy improvement. To ensure experimental fairness, uniform experimental parameters were set: the population size and maximum number of iterations for all algorithms were 100 and 50, respectively. Table 1 shows the microgrid operating parameters, including PV, WT, DE, MT, and BT. Table 2 shows the greenhouse gas emission-related processing parameters. Figure 4 shows the predicted output power and load demand data for WT and PV. Figure 5 shows the real-time electricity price data under the demand response policy.

[0132] Table 1 Microgrid Operating Parameters

[0133]

[0134] Table 2 Pollutant Emission Coefficients

[0135]

[0136] This simulation applies the JSI-MODBO, NSPSO, NSDBO, and NSGA-II algorithms to microgrid optimal scheduling and compares their optimization effects. The Pareto fronts of these four optimization algorithms for microgrid optimal scheduling are shown in Figure 6. Figure 6 shows that the data points for the JSI-MODBO algorithm are concentrated in the range of operating costs (430-550 yuan) and environmental costs (60-80 yuan). In contrast, the other three algorithms avoid wasting optimization resources in irrelevant ranges, improving the spatial search breadth and optimization accuracy of multi-objective optimization algorithms. The scheduling results in Figure 7 show that during the nighttime to early morning period from 01:00 to 08:00, the photovoltaic power generation system is in a shutdown state, wind power output is at a low level, and the grid purchase price is at its lowest. Based on economic considerations, the system purchases a moderate amount of electricity from the grid, performs a small amount of battery charging, gradually increases DG output, and maintains stable MT output to meet base load demand. From 08:00 to 13:00, as the output of photovoltaic and wind power gradually increases, the main grid electricity purchase price rises accordingly. The system dynamically adjusts its electricity purchase strategy, gradually reducing the amount of electricity purchased from the main grid and ultimately selling electricity back to the main grid to maximize economic benefits. During this process, the batteries switch from charging to discharging mode, and distributed generation (DG) and medium-duty transmission (MT) maintain system power balance and stable operation by dynamically adjusting their output in real time. From 13:00 to 17:00, the output of photovoltaic and wind power continues to decline to its lowest point. With sufficient system power supply, the batteries perform a small amount of charging. To optimize operating costs, the system appropriately reduces DG output while maintaining stable MT operation, increasing the amount of electricity sold to the main grid while ensuring power balance. From 17:00 to 24:00, as the output of photovoltaic and wind power further decreases, the batteries release energy during periods of higher electricity prices, working in conjunction with the stable-operating DG and MT to construct a multi-energy complementary operation mode, effectively meeting system load demands.

[0137] Table 3 presents the final costs of the scheduling results generated by the four algorithms mentioned above, comparing the total scheduling costs of JSI-MODBO with those of NSPSO, NSDBO, and NSGA-II. The data shows that compared to the NSPSO, NSDBO, and NSGA-II algorithms, the total scheduling cost of the JSI-MODBO algorithm is reduced by approximately 6.09%, 10.81%, and 19.76%, respectively. This result demonstrates that the JSI-MODBO algorithm exhibits significant cost control advantages when dealing with resource optimization scheduling problems in complex scenarios. This algorithm improves the economic efficiency of microgrid system operation through the rational scheduling of various distributed energy sources.

[0138] Table 3 Cost Comparison

[0139]

[0140] In this example, to verify the stability of the JSI-MODBO algorithm's effect on microgrid scheduling optimization, Table 4 presents quantitative analysis data of the scheduling results from 30 independent runs of the four algorithms. The analysis focuses on three key indicators: optimal cost, maximum cost, and average cost. The data shows that the average cost of the JSI-MODBO algorithm is reduced by 6.82%, 14.51%, and 19.29% compared to the NSPSO, NSDBO, and NSGA-Ⅱ algorithms, respectively. Its lower average cost deviation indicates that JSI-MODBO can maintain more stable performance across multiple runs, further confirming the algorithm's significant advantage in stability from a quantitative perspective.

[0141] Table 4 Stability Comparison

[0142]

[0143] In this invention, a hybrid initialization improvement strategy combining Latin hypercubes and cosine similarity back-learning generates an initial population that allows the algorithm to obtain a larger search range with uniform distribution and a better initial solution set in the early stages. This helps maintain the diversity of the early population and achieves better spatial search capabilities. Furthermore, the algorithm improves the acceptance of optimal solutions by the baby ball and the dung beetle, avoiding premature entrapment in local optima and enhancing the algorithm's search ability. Then, the perturbation strategy of the sparrow search algorithm's randomization mechanism is integrated into the multi-objective dung beetle algorithm. Utilizing the unique perturbation mechanism of the sparrow algorithm's randomization mechanism, the algorithm can further escape local optima, enhancing its convergence accuracy. Finally, Cauchy-Gaussian mutation is introduced. By dynamically adjusting the mutation amplitude during algorithm iteration, the probability of escaping local extrema is significantly improved. Simultaneously, by reasonably balancing exploration and exploitation capabilities, the optimization error is further reduced, ultimately achieving a simultaneous improvement in both optimization accuracy and global convergence efficiency. In summary, the JSI-MODBO algorithm demonstrates excellent performance in solving the proposed microgrid multi-objective optimization scheduling model, thus providing a new method and technology for practical microgrid scheduling optimization.

[0144] Another specific embodiment of this application provides a microgrid scheduling optimization device based on an improved multi-objective dung beetle optimization algorithm, including a processor and a memory. The memory stores a computer program, and when the computer program is called and executed by the processor, it implements the steps in the method described in the above embodiment.

[0145] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several modifications and improvements can be made without departing from the inventive concept of this application, and these all fall within the protection scope of this application.

Claims

1. A microgrid scheduling optimization method based on an improved dung beetle optimization algorithm, characterized in that, Includes the following steps: S1: Construct a microgrid scheduling model, establish a dual objective function for microgrid operation and maintenance costs and environmental protection costs, and obtain the microgrid's operating parameters and constraints; S2: The microgrid scheduling model is optimized using an improved multi-objective dung beetle optimization algorithm to obtain the optimal microgrid scheduling scheme; Improvements to the multi-objective dung beetle optimization algorithm include: (1) using a hybrid initialization strategy of Latin hypercube sampling and cosine similarity back learning to generate a highly diverse initial population; (2) Improve the acceptance of the optimal solution by the baby ball and the dung beetle; (3) Integrate the perturbation strategy of the sparrow search algorithm with the random mechanism into the dung beetle algorithm; (4) Introduce Cauchy-Gaussian mutation and dynamically adjust the mutation amplitude during the algorithm iteration process; S3: Optimize the microgrid scheduling by using the microgrid optimal scheduling scheme.

2. The microgrid scheduling optimization method based on the improved dung beetle optimization algorithm according to claim 1, characterized in that, Microgrid operation and maintenance costs include the operation and maintenance costs of power generation equipment, battery charging and discharging costs, and electricity purchase and sales costs.

3. The microgrid scheduling optimization method based on the improved dung beetle optimization algorithm according to claim 1, characterized in that, The environmental protection costs of microgrids include the cost of treating pollutants from the main grid and the cost of treating pollutants generated by power generation equipment.

4. The microgrid scheduling optimization method based on the improved dung beetle optimization algorithm according to claim 1, characterized in that, Between steps S1 and S2, there is also a step: initializing the operating parameters of the microgrid.

5. The microgrid scheduling optimization method based on the improved dung beetle optimization algorithm according to claim 1, characterized in that, In the improvement of the multi-objective dung beetle optimization algorithm (1), Latin hypercube sampling is used to generate the dung beetle initial population. The steps are as follows: determine the sampling size H, that is, the dung beetle population size; and set the position variable d of each dung beetle. i Domain interval [d l i ,d u i Divide the data into H equal subintervals, i.e.: d l i =d0 i <d1 i <···<d H i =d u i Generate an H×n matrix A, where each column is a randomized sequence of numbers 1, 2, ..., H; each row of matrix A corresponds to a hypercube, and each hypercube generates one sample; while using Latin hypercube sampling to generate the initial population, a cosine similarity back-learning strategy is introduced to optimize the population structure.

6. The microgrid scheduling optimization method based on the improved dung beetle optimization algorithm according to claim 1, characterized in that, In the improvement of the multi-objective dung beetle optimization algorithm (2), the acceptance of the juvenile ball and the thieving dung beetle to the local optimal position or the global optimal position is dynamically changed according to the fitness of the dung beetle.

7. The microgrid scheduling optimization method based on the improved dung beetle optimization algorithm according to claim 1, characterized in that, In the improvement of the multi-objective dung beetle optimization algorithm (3), the optimal position of the discoverer in the sparrow search algorithm is defined as the global optimal position in the dung beetle optimization algorithm. The perturbation probability is adaptively changed according to the iteration depth, and a greedy strategy is used to decide whether to retain the perturbation. The perturbation probability is as follows: ;in; t represents the current iteration number, and T represents the maximum iteration number.

8. The microgrid scheduling optimization method based on the improved dung beetle optimization algorithm according to claim 3, characterized in that, In the improvement of the multi-objective dung beetle optimization algorithm (4), after each random perturbation of the sparrow search algorithm, it is determined whether the population is premature. The determination method is: if the optimal solution of the population remains unchanged after 5 iterations, it is determined that the algorithm is premature. The Cauchy-Gaussian mutation strategy is adopted for the entire population to make the population jump out of the local optimal solution.

9. A microgrid scheduling optimization device based on an improved multi-objective dung beetle optimization algorithm, characterized in that, It includes a processor and a memory, the memory storing a computer program, which, when executed by the processor, implements the steps of the method as described in any one of claims 1-8.

Images