Microgrid optimal dispatching method based on adaptive scheduling and mutation particle swarm algorithm

Abstract

The microgrid optimization scheduling method based on the particle swarm algorithm with adaptive modulation and mutation aims to improve the operation efficiency and environmental protection performance of the microgrid. This method innovatively proposes an adaptive parameter adjustment method and mutation mechanism, overcomes the problems of the traditional particle swarm algorithm's easy premature maturity and lack of population diversity, and significantly enhances the global search capability and optimization effect. Taking the operation cost and environmental protection cost as the optimization goals, a mathematical model of the microgrid in the grid-connected operation mode is constructed; the particle swarm algorithm with adaptive modulation and mutation is used for scheduling optimization; and the superiority of this method under different load demands and resource distribution conditions is verified through comparative simulation experiments. The present invention is suitable for microgrid optimization scheduling scenarios including distributed power sources, energy storage systems, etc., and can maximize economic and environmental benefits while ensuring power supply reliability.

Description

Technical Field

[0001] The present invention relates to the technical field of optimal dispatching of power systems, and in particular to a microgrid optimal dispatching method based on an adaptive modulation and mutation particle swarm algorithm. Background Art

[0002] With the adjustment of the global energy structure and the promotion of low-carbon economy, the traditional centralized power supply system has gradually shifted to a distributed power supply mode. As a typical representative of the distributed power supply system, microgrids have played an important role in improving energy efficiency, reducing power transmission losses and enhancing system resilience. By integrating a variety of distributed energy sources (such as photovoltaic power generation, wind power generation, diesel generators, micro gas turbines and energy storage devices, etc.), microgrids can not only operate independently, but also be connected to the main grid, and have high flexibility and controllability in the power system.

[0003] In the actual operation of microgrids, scheduling issues are the core to ensure their safe, reliable and economical operation. Scheduling objectives usually include reducing operating costs, reducing environmental pollution, and improving power supply reliability. However, distributed power sources have different power generation characteristics. For example, photovoltaic and wind power generation are greatly affected by weather and have uncertainty in power generation; while diesel generators and micro gas turbines have relatively high operating costs and are accompanied by pollutant emissions. Therefore, how to reasonably schedule these distributed power sources while ensuring power supply demand and achieve a balance between operating costs and environmental protection costs has become a key issue in microgrid scheduling.

[0004] For the scheduling problem of microgrids, traditional optimization methods such as linear programming and dynamic programming have certain applicability in small-scale problems, but when faced with large-scale nonlinear problems with multiple objectives and multiple constraints, they have the limitations of low computational efficiency and difficulty in obtaining the global optimal solution. With the rise of intelligent optimization algorithms, particle swarm optimization (PSO), as an optimization algorithm based on swarm intelligence, has been widely used in optimization problems due to its simple calculation, fast convergence speed, and strong global search ability. However, when solving multi-objective optimization problems, the traditional PSO algorithm has defects such as premature convergence and insufficient population diversity, which leads to its limited effect in complex microgrid scheduling.

[0005] Compared with the existing technology, the technical differences are as follows:

[0006] Technical comparison with patent CN118589588A_A microgrid optimization dispatching system based on improved particle swarm algorithm;

[0007] 1. Algorithm used in CN118589588A

[0008] Patent CN118589588A proposes a microgrid optimization dispatching system based on an improved particle swarm algorithm. The method mainly uses a genetic algorithm to improve the particle swarm algorithm and introduces a crossover mutation operation. The speed and position of the particles are updated through the improved particle swarm algorithm. The specific steps include:

[0009] Selection: A part of particles is selected to participate in the crossover operation based on the fitness value, using the roulette wheel selection method.

[0010] Crossover: Perform a two-point crossover operation on the selected particles to generate new particles.

[0011] Mutation: Perform mutation operation on the particles after crossover, randomly change some positions of particles with a preset probability, and enhance population diversity.

[0012] This method uses the classic genetic algorithm to improve the particle swarm algorithm in microgrid scheduling optimization, which can increase the population diversity of the particle swarm algorithm to a certain extent and enhance the global search ability of the particle swarm algorithm. However, its mutation probability is a preset constant, which fails to fully consider the local search of the particle swarm algorithm (PSO) in the later stage, and does not fundamentally solve the disadvantage of PSO's premature convergence. In addition, in the case of large and complex microgrids, the optimization efficiency is limited.

[0013] 2. Method of the present invention

[0014] This paper proposes a microgrid optimization scheduling method based on adaptive adjustment and mutation particle swarm algorithm, which realizes the scheduling optimization of the microgrid system by adaptively adjusting the inertia weight, learning factor and mutation probability of PSO. The core innovations of the method include:

[0015] Adaptive parameter adjustment mechanism: This algorithm introduces an adaptive parameter adjustment mechanism to dynamically adjust the inertia weight ω, the individual learning factor c1, and the group learning factor c2, and flexibly change the search behavior of particles according to the number of iterations. In the early stage of the algorithm, a larger inertia weight and a smaller learning factor help enhance the global search capability and avoid premature convergence; in the later stage of the iteration, a smaller inertia weight and a larger learning factor can strengthen the local search and enable particles to approach the optimal solution faster. This dynamic adjustment mechanism greatly improves the adaptability of the algorithm at different stages.

[0016] Adaptive adjustment of mutation operation and probability: The present invention adopts mutation operation to enhance the diversity of the population and prevent particles from falling into the local optimal solution. The mutation operation randomly selects some dimensions of the particles for mutation, and the mutation probability pm is dynamically adjusted with the number of iterations. The quality of the particles is judged by the Pareto dominance concept to achieve the goal of multi-objective optimization.

[0017] Multi-objective optimization processing and Pareto dominance mechanism: The present invention processes multi-objective optimization problems by introducing the Pareto dominance mechanism, converts multiple conflicting optimization objectives into a Pareto optimal solution set, and uses an external archive library to store non-dominated solutions.

[0018] 3. Comparison of technical advantages

[0019] Convergence speed: The adaptive adjustment and mutation particle swarm algorithm in the present invention avoids these problems by dynamically adjusting parameters and improves the overall convergence efficiency. Although the combination of crossover and mutation can enhance population diversity, the crossover operation may lead to some invalid or redundant solutions, especially in the later stage of optimization, the effectiveness of the crossover operation is reduced, which may slow down the convergence speed or increase the computational burden.

[0020] Balance global search and local search: The adaptive adjustment and mutation particle swarm algorithm in the present invention can automatically adjust the search strategy according to the number of iterations, achieving a more intelligent balance between global search and local development. Although the crossover operation in CN118589588A can generate new solutions by exchanging part of the information of particles and increase the diversity of solutions, its control over global search and local search is relatively fixed.

[0021] Stability: The method of the present invention reduces excessive random operations through adaptive parameter adjustment and controlled mutation operations, making the solution set more stable. The control of mutation can avoid the instability and excessive randomness of the solution, especially in the late optimization stage, and effectively enhances the convergence stability of the algorithm. The crossover operation in CN118589588A easily introduces more randomness, which may cause large fluctuations in the solution set, especially in the stage close to convergence, the crossover operation will destroy the excellent solution that has been found, affecting the convergence stability.

[0022] In summary, the microgrid optimization scheduling method based on the adaptive adjustment and mutation particle swarm algorithm of the present invention is better than CN118589588A in optimization effect in algorithm convergence speed, balanced global search and local search, and stability, and significantly improves population diversity and optimization efficiency, and can provide a more economical and reliable scheduling scheme for microgrid optimization scheduling.

[0023] Technical comparison with patent CN116316855A_A wind-solar-storage microgrid dispatching method based on PSO algorithm

[0024] 1. Algorithm used in CN116316855A

[0025] Patent CN116316855A proposes a wind, solar and energy storage microgrid dispatch optimization solution based on the PSO algorithm. The method is mainly based on the day-ahead forecast data of renewable energy power and load power, and uses the particle swarm optimization algorithm (PSO) to formulate day-ahead dispatch plans for the microgrid on-grid and off-grid conditions. The specific steps include:

[0026] Establish a prediction model for storage microgrids.

[0027] The daily dispatch plan of microgrid is formulated based on the prediction model of storage microgrid.

[0028] The PSO algorithm is used to optimize the daily dispatching data of the microgrid in off-grid and grid-connected states.

[0029] This method uses the classic particle swarm algorithm in microgrid optimization scheduling, which can solve the optimization problem in microgrid scheduling to a certain extent. However, its optimization process is mainly aimed at single-objective microgrid optimization scheduling, and fails to fully consider other indicators in the actual operation of the microgrid. It also has defects such as premature convergence, insufficient population diversity and unstable convergence speed.

[0030] 2. Method of the present invention

[0031] The present invention proposes a multi-objective optimization scheduling method for microgrids based on adaptive parameter adjustment and mutation particle swarm algorithm. The core innovation is to achieve multi-objective optimization in microgrid scheduling problems through adaptive parameter adjustment mechanism, mutation operation and Pareto dominance mechanism, especially for the comprehensive consideration of operating cost and environmental protection cost. The following are the main method steps and core innovations of the present invention:

[0032] Adaptive parameter adjustment mechanism: Based on the traditional PSO algorithm, the present invention adaptively adjusts the inertia weight ω and the individual learning factor c1 and the group learning factor c2, so that the algorithm can dynamically adjust the search behavior at different iteration stages. In this way, a larger inertia weight in the early stage of optimization enhances the global search capability and prevents particles from falling into the local optimum; while in the later stage of optimization, a smaller inertia weight helps to strengthen the local search, speed up the convergence speed, and find a better solution.

[0033] Dynamic mutation operation: The present invention introduces a mutation operation, which randomly selects some dimensions of particles for mutation, thereby enhancing the diversity of the population and preventing the particle group from tending to be consistent during the iteration process and losing the global search ability. The key innovation of the mutation operation is that the mutation probability pm is dynamically adjusted with the number of iterations. In the early stage, a higher mutation probability increases the exploratory nature of the solution space and prevents premature convergence; in the later stage, reducing the mutation probability prompts the solution to converge to the global optimum. This mechanism enables the mutation operation to more flexibly adapt to the needs of different optimization stages and improve the global and local search balance of the algorithm.

[0034] Pareto dominance mechanism and external archive management: For the multi-objective optimization problem of microgrids, the present invention adopts the Pareto dominance mechanism to deal with multiple conflicting optimization objectives (such as operation and maintenance costs and environmental protection costs). The fitness value of each particle is evaluated by the Pareto dominance rule to ensure that multiple objectives can be optimized simultaneously.

[0035] 3. Comparison of technical advantages

[0036] Enhance the balance between global and local search: The present invention introduces an adaptive parameter adjustment mechanism to dynamically adjust the inertia weight and learning factor, so that particles have different search capabilities at different stages. In the early stage of the algorithm, a larger inertia weight ensures the ability of global search and avoids premature convergence; in the later stage, the inertia weight gradually decreases, improving the local search capability and accelerating the algorithm to converge to the optimal solution. Compared with CN116316855A, the present invention can intelligently switch between global and local searches to improve the quality of the solution.

[0037] Population diversity: The present invention uses dynamic mutation operations to randomly select some dimensions of particles for mutation, and dynamically adjusts the mutation probability with the number of iterations. In the early stages of iteration, a higher mutation probability enhances the ability to explore the solution space, increases population diversity, and avoids premature convergence; in the later stages, the mutation probability gradually decreases, preventing invalid disturbances of particle solutions in the later stages of iteration, and prompting the solution to gradually converge to the global optimum. In CN116316855A, as the iterations proceed, the positions and velocities of the particles gradually converge, and the population diversity is insufficient, which can easily lead to the algorithm falling into a local optimal solution. Especially in complex microgrid scheduling problems, relying solely on individual and group optimal solutions can easily ignore better solution areas.

[0038] Adaptability: The present invention improves the adaptability of the algorithm to handle complex constraints through the combination of adaptive parameter adjustment and mutation mechanism. The adaptive mechanism can dynamically adjust the search behavior of the particle swarm, so that it can flexibly search the solution space under the premise of satisfying the constraints, reducing the generation of infeasible solutions. At the same time, the introduction of the Pareto dominance mechanism and the external archive library further enhances the adaptability to the multi-objective constraints of complex systems, and improves the proportion of feasible solutions and optimization efficiency of the algorithm. Compared with CN116316855A, traditional PSO usually adopts penalty functions and other methods when dealing with complex constraints, but the processing effect is relatively limited, especially when there are many and complex constraints, it is easy to generate infeasible solutions, reducing the optimization efficiency.

[0039] In summary, the microgrid multi-objective optimization scheduling method based on the adaptive variable-participation particle swarm algorithm of the present invention is superior to CN112909981A in terms of algorithm enhancement of the balance between global and local search, population diversity and adaptability, and can provide a more economical and reliable optimization scheduling scheme for microgrid optimization scheduling. Summary of the invention

[0040] The invention aims to solve the problem of multi-objective optimization in microgrid scheduling, especially to ensure the reliability of system power supply while optimizing operation and maintenance costs and environmental protection costs. Microgrids contain a variety of distributed power sources, such as photovoltaic power generation, wind power generation, diesel generators, etc. Their power generation characteristics are different, and scheduling objectives (such as economy and environmental protection) are often contradictory. Existing scheduling methods mostly use classical algorithms such as linear programming and dynamic programming. Although they have certain effects on small-scale problems, they are prone to fall into local optimality when facing complex nonlinear and multi-objective optimization problems, with low computational efficiency and difficulty in weighing multiple objectives. PSO is widely used in optimization problems because of its strong global search ability and simple calculation. However, in the multi-objective scheduling problem of microgrids, the PSO algorithm is prone to premature convergence and insufficient population diversity, and it is difficult to effectively handle conflicting multi-objective optimization requirements. In order to overcome these limitations, the present invention proposes a microgrid optimization scheduling method based on an adaptive parameter adjustment and mutation particle swarm algorithm. By introducing an adaptive parameter adjustment mechanism and mutation operation, the global search ability and population diversity of the algorithm are enhanced, and the premature convergence problem is avoided. At the same time, the Pareto dominance rule is used to deal with multi-objective optimization problems and generate a Pareto frontier solution set, which provides a variety of optimization schemes for microgrid scheduling and helps decision makers make the best choice between economic benefits and environmental benefits.

[0041] To achieve the above object, the technical solution adopted by the present invention is:

[0042] The microgrid optimization scheduling method based on adaptive scheduling and mutation particle swarm algorithm includes the following steps:

[0043] (1) A microgrid system model under the grid-connected operation mode is constructed. Taking the operation and maintenance cost f1(x) and the environmental protection cost f2(x) as the optimization targets, a mathematical model is established. The objective function is:

[0044] Min C=Min(f1(x),f2(x))

[0045] Among them, the operation and maintenance cost f1(x) includes the transaction cost between the microgrid and the main grid, the fuel cost and maintenance cost of the distributed power source, and the environmental protection cost f2(x) includes the cost of pollutant treatment;

[0046] (2) Optimizing scheduling based on adaptive parameter adjustment and mutation particle swarm algorithm, the particle swarm algorithm includes adaptive parameter adjustment and mutation operation:

[0047] The particle velocity and position update formula is:

[0048]

[0049] Among them, ω is the inertia weight, c1 and c2 are the individual learning factor and the group learning factor respectively, r1 and r2 are random numbers between [0,1], and p i,best is the historical optimal position of the ith particle, g best is the global optimal position, x i (t) is the position information of the particle in the tth generation, v i (t+1) is the velocity information of the particle in the t+1 generation;

[0050] Adaptive parameter adjustment includes dynamic adjustment of inertia weight ω, individual learning factor c1 and group learning factor c2, and the formula is:

[0051]

[0052] Among them, IT is the current iteration number, MI is the maximum iteration number, ω s ,ω e are the initial and final values ​​of the inertia weight, c 1s 、c 1e 、c 2s 、c 2e are the initial and final values ​​of the individual learning factor and group learning factor respectively;

[0053] Mutation operation: In each iteration, 60 dimensions of particles are randomly selected for mutation operation. The mutation probability pm is adaptively adjusted with the number of iterations. The formula is:

[0054]

[0055] The mutation probability is higher in the initial iteration, which increases the population diversity and helps the global search; it gradually decreases in the later iteration to avoid destroying the existing optimization results.

[0056] Furthermore, the operation and maintenance cost f1(x) of the microgrid system is determined by the following formula:

[0057]

[0058] Among them, C GRID (t), C MT (t), C DE (t), C PV (t), CWT (t), C BT (t) are the operating costs of transactions with the main grid, micro gas turbines, diesel generators, photovoltaic power generation, wind turbines and batteries, respectively; c buy (t), c sell (t) are the unit prices of electricity sold and purchased from the main grid at time t, C MT.F (t), C DE.F (t) are the fuel costs of the micro gas turbine and diesel generator, respectively, C MT.M (t), C DE.M (t) are the maintenance costs of the micro gas turbine and diesel generator respectively.

[0059] Furthermore, the environmental protection cost f2(x) is determined by the following formula:

[0060]

[0061] Among them, C GRID.EN (t), C MT.EN (t), C DE.EN (t) are the environmental protection costs of the main grid, micro gas turbine and diesel generator respectively. k is the unit price of treatment of the kth type of pollutant, γ grid,k is the emission coefficient of the kth type of pollutant.

[0062] Furthermore, the optimization scheduling step based on the adaptive scheduling and mutation particle swarm algorithm includes the following constraints:

[0063] (1) Power balance constraints:

[0064] P GRID (t)+P PV (t)+P WT (t)+P MT (t)+P DE (t)+P BT (t) = P L (t)

[0065] Among them, P L (t) is the power of the load at time t;

[0066] (2) DG output constraints:

[0067] P i.min ≤P i ≤P i.max

[0068] Among them, P i.min , P i.max They are the upper and lower limits of each micro power source output respectively;

[0069] (3) Battery capacity constraints:

[0070] SOC MIN ≤SOC(t)≤SOC MAX

[0071] Among them, SOC MIN ,SOC MAX are the minimum and maximum capacities of the battery respectively;

[0072] (4) DE and MT climbing constraints:

[0073]

[0074] Among them, R DE ,R MT They are the climbing power limits of DE and MT respectively.

[0075] Furthermore, the specific process of the adaptive adjustment and mutation particle swarm algorithm is as follows:

[0076] Step 1: Initialize the population and external archive library. Initialize particle positions, particle velocities, p i,best and g best ;

[0077] Step 2: Calculate the fitness value;

[0078] Step 3: Update the population;

[0079] Update particle position, velocity, p i,best , g best , mesh division and ω, c1, c2, pm parameters;

[0080] Step 4: Determine dominance and update external archives;

[0081] Step 5: Determine whether the archive space is exceeded;

[0082] Step 6: Determine whether the maximum number of iterations is exceeded. If so, output the Pareto frontier solution set; if not, repeat steps 2-5.

[0083] Beneficial effects:

[0084] (1) Through the adaptive parameter adjustment mechanism and mutation operation, the algorithm of the present invention can conduct a wider search in the solution space, effectively avoiding the premature convergence problem in the traditional PSO algorithm and enhancing the global optimization capability.

[0085] (2) The present invention introduces mutation operations to maintain the diversity of the population, ensuring that particles can continuously explore new solution spaces during the iteration process, thereby improving the quality of the solution.

[0086] (3) The present invention can simultaneously optimize multiple conflicting objectives of the microgrid through the Pareto dominance mechanism and form a Pareto frontier solution set, providing a set of optional optimal scheduling schemes. Users can choose the best scheme according to actual needs.

[0087] (4) The present invention is applicable to microgrid systems of different scales and structures, especially multi-objective optimization scenarios that require consideration of both economic and environmental benefits, and has broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] Figure 1 It is a basic structure diagram of the microgrid adopted by the present invention;

[0089] Figure 2 This is a flow chart of the particle swarm algorithm for adaptive tuning and mutation proposed by the present invention;

[0090] Figure 3 This is a variation trend diagram of w, c1, and c2 proposed by the present invention;

[0091] Figure 4 This is the PM variation trend diagram proposed by the present invention;

[0092] Figure 5 It is a microgrid load change, PV and WT power generation prediction diagram of an embodiment of the present invention;

[0093] FIG6( a ) shows the DG output in the economic mode of an embodiment of the present invention;

[0094] FIG6( b ) is a diagram showing the microgrid load and DG output results in the economic mode of an embodiment of the present invention;

[0095] FIG. 7( a ) shows the DG output in the environmental protection mode of an embodiment of the present invention;

[0096] FIG7( b ) is a diagram showing the microgrid load and DG output results in the environmental protection mode of an embodiment of the present invention;

[0097] Figure 8 A Pareto frontier solution set distribution diagram of an embodiment of the present invention;

[0098] Fig. 9 It is a comparison of the iterative convergence curves of the two algorithms in the embodiment of the present invention;

[0099] Fig.10 This is a real-time electricity price map according to an embodiment of the present invention. DETAILED DESCRIPTION

[0100] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments:

[0101] like Figure 1As shown in the figure, the microgrid system consists of multiple distributed power sources and energy storage devices, including photovoltaic power generation (PV), wind turbine (WT), diesel generator (DE), micro gas turbine (MT) and battery (BT), as well as an interface connected to the main grid (GRID). The microgrid meets the load demand through these distributed energy sources and energy storage systems, while balancing economic benefits and environmental benefits. In the actual operation of the microgrid, each distributed power source has different power generation characteristics and operating limitations, so detailed scheduling planning is required. The specific system composition and model description are as follows:

[0102] (1) Photovoltaic power generation system (PV): Photovoltaic power generation converts solar energy into electrical energy. Its power output is greatly affected by sunlight intensity and temperature changes. The output power of photovoltaic power generation P PV (t) is determined by the following formula:

[0103]

[0104] Among them, P PV (t) is the rated power of the PV module, P S is the maximum PV output power under standard conditions, S A is the current sunshine intensity, So is the standard sunshine intensity. k is the temperature coefficient, T c is the current operating temperature of PV; T r is the reference temperature.

[0105] (2) Wind turbine (WT): A wind turbine generates electricity through wind power, and its power generation depends on the wind speed. The output power of the generator is P WT (t) Due to the limitation of wind speed, the power generation characteristics are different when the wind speed is in different ranges. The model is as follows:

[0106]

[0107] v w (t) is the actual wind speed at time t. V ci is the cut-in wind speed, V r is the rated working wind speed of WT, V co When the wind speed is between the cut-in wind speed and the rated wind speed, the WT operates at rated power, and the output power is proportional to the square of the wind speed.

[0108] (3) Diesel generator (DE): Diesel generators are often used to provide electricity during peak load periods. They have high operating costs and are accompanied by certain pollutant emissions. The output power P of a diesel generator DEThe relationship between it and its operating cost can be described by a quadratic function model:

[0109]

[0110] C DE.F is the fuel cost of DE; P DE is the output active power of DE; α, β, γ are the fuel cost coefficients of DE. In this model, α=6, β=0.1801, γ=1.1×10 -4 .

[0111] (4) Micro gas turbine (MT): Micro gas turbine has the characteristics of high efficiency and flexibility, and is suitable for flexible response to load changes. Its power generation efficiency changes with the output power, and the efficiency model is:

[0112]

[0113] Among them, η MT Indicates power generation efficiency, P MT Represents output power. The fuel cost of MT is closely related to power generation efficiency, as shown in the following formula:

[0114]

[0115] Among them, C MT.F is the fuel cost of DE; C and LHV are the natural gas price and lower heating value, respectively.

[0116] (5) Battery (BT): The output form of the battery is direct current. The charge and discharge state is determined by the total load, DG power generation, and the power purchase and sales situation between the microgrid and the main grid. The battery capacity (SOC) is closely related to the charge and discharge power. The expression is as follows:

[0117]

[0118] SOC(t) and SOC(t-1) are the capacities at time t and time t-1 respectively; η ch , η dis is the charging efficiency and discharging efficiency, both of which are taken as 0.9 in this experiment. BT (t)≥0, the battery is charged, P BT (t)<0, battery discharge, P BT (t) is the charging and discharging power of the battery.

[0119] (6) Main grid interface (GRID): The microgrid can trade electricity through the interface with the main grid. When the microgrid is short of internal power generation, it can purchase electricity from the main grid. buy (t), when there is excess power generation, electricity is sold to the main grid C sell (t).

[0120] The system architecture of the microgrid provides a variety of feasible solutions for dispatch optimization, which requires considering the optimal combination of the output and cost of each distributed energy source. In actual engineering applications, the conditional constraints must be taken into account, including line transmission power constraints, power balance constraints, DG output constraints, battery capacity constraints, DE, MT climbing constraints. The specific constraints are as follows:

[0121] (1) Line transmission power constraints:

[0122] P li.min ≤P li ≤P li.max

[0123] (2) Power balance constraints

[0124] P GRID (t)+P PV (t)+P WT (t)+P MT (t)+P DE (t)+P BT (t) = P L (t)

[0125] (3) DG output constraints

[0126] P i.min ≤P i ≤P i.max

[0127] (4) Battery capacity constraints

[0128] SOC MIN ≤SOC(t)≤SOC MAX

[0129] (5) DE and MT climbing constraints

[0130]

[0131] According to the basic structure of the microgrid, the present invention obtains real data, comprehensively considers the above constraints, and performs multi-objective optimization scheduling on the microgrid in the grid-connected operation mode based on the AP-MPSO algorithm. The algorithm flow chart is as follows: Figure 2 As shown. According to different needs, the dispatch mode is divided into economic mode and environmental protection mode. Specifically, it includes the following steps:

[0132] (1) Initialize the population and external archive. Initialize particle positions, particle velocities, and p i,best and g best In this study, the number of particles is 100, each particle is 120-dimensional, corresponding to one scheduling scheme; the maximum capacity of the external archive library is 100 (a maximum of 100 non-dominated solutions can be stored).

[0133] (2) Calculate the fitness value. According to the system model of the microgrid, the scheduling scheme of each particle is evaluated, and the quality of the particle is judged by the Pareto dominance concept. The fitness function of each particle includes two optimization objectives: operation and maintenance cost f1 and environmental protection cost f2. The specific calculation formula is as follows:

[0134]

[0135] Among them, C GRID (t), C MT (t), C DE (t), C PV (t), C WT (t), C BT (t) are the operating costs of transactions with the main grid, micro gas turbines, diesel generators, photovoltaic power generation, wind turbines and batteries. buy (t), c sell (t) are the unit prices of electricity sold and purchased from the main grid at time t. MT.F (t), C DE.F (t) are the fuel costs of the micro gas turbine and diesel generator, respectively, C MT.M (t), C DE.M (t) are the maintenance costs of the micro gas turbine and diesel generator respectively.

[0136]

[0137] Among them, C GRID.EN (t), C MT.EN (t), C DE.EN (t) are the environmental protection costs of the main grid, micro gas turbine and diesel generator respectively. k is the unit price of treatment of the kth type of pollutant, γ grid,k is the emission coefficient of the kth type of pollutant.

[0138] (3) Update the population. In each iteration, the speed and position of each particle are updated according to the basic formula of particle swarm optimization:

[0139]

[0140] Among them, the inertia weight ω controls the particle's ability to maintain the current speed, the learning factors c1 and c2 represent the particle's ability to approach the individual's historical optimal position and the group's global optimal position, respectively, and r1 and r2 are random numbers. In order to balance the global search and local search capabilities, the present invention introduces an adaptive parameter adjustment mechanism. As the number of iterations increases, the inertia weights ω, c1, and c2 will be dynamically adjusted, and the formula is as follows:

[0141]

[0142] Among them, ω s ,ω e are the initial and final values ​​of ω respectively; c 1s 、c 1e are the initial value and final value of c1 respectively; c 2s 、c 2e are the initial value and the end value of c2 respectively; IT and MI are the current number of iterations and the maximum number of iterations respectively. In the adaptive parameter adjustment method, w, c1, and c2 are dynamically adjusted with the iteration process. In the early stage of iteration, w, c1 is larger and c2 is smaller. The current moving direction and the individual historical optimal position have a larger weight, which is conducive to the global search of the particle swarm and avoids converging to the local optimal value; in the later stage of iteration, w, c1 is smaller and c2 is larger, which is more conducive to local refinement search and accelerates the convergence speed. In addition, the ω, c1 change curve is a convex function in the early stage, and the c2 change curve is a concave function in the early stage and a convex function in the later stage. This setting further improves the global search ability and convergence speed of the particles. The specific change trends are as follows Figure 3 shown.

[0143] In order to further enhance the global search capability of particles and overcome the adverse effects of random initialization, a mutation operation is introduced to randomly select 60 dimensions of particles for mutation. The mutation probability is adaptively adjusted according to the number of iterations. The mutation probability formula is as follows:

[0144]

[0145] Among them, α is used to adjust the concavity of the curve. A larger α value will cause pm to decrease faster. During the adaptive mutation process, pm is dynamically adjusted with the iteration process. In the early stage of the iteration, the mutation probability is large, which increases the population diversity and global search ability, and is also conducive to overcoming the adverse effects of random initialization; in the later stage of the iteration, in order not to destroy the learning results of the particles and the local refinement search, the mutation probability is 0. The specific change trend of pm is as follows Figure 4 shown.

[0146] (4) Determine dominance and update the external archive. Determine the dominance of the original archive Pareto frontier solution set and the newly obtained Pareto frontier solution set, and use all non-dominated solutions as the new Pareto frontier solution set.

[0147] (5) Determine whether the archive space is exceeded. If the archive space is not exceeded, proceed to the next step; if the archive space is exceeded, calculate the number of particles in each grid according to the divided grids, select the grid according to the roulette selection method, and randomly delete a non-dominated solution in the selected grid.

[0148] (6) Determine whether the maximum number of iterations has been exceeded. If so, output the Pareto frontier solution set; if not, loop through (2)-(5).

[0149] In order to verify the effectiveness of the present invention, multiple simulation experiments were conducted. The simulation results show that the AP-MPSO algorithm of the present invention can effectively solve the microgrid scheduling problem, especially in balancing the operation and maintenance costs and environmental protection costs. The load change curve of the microgrid and the power generation prediction curve of PV and WT are shown in Figure 2. Figure 5 The output of each distributed power source under economic (lowest total cost) and environmental protection mode (lowest environmental protection cost) is shown in Figure 6 and Figure 7 respectively. The Pareto frontier solution set is as follows: Figure 8 As shown in the figure. The difference between the two different operating modes lies in the changes in the output of DE and MT. In the economic mode, the output of DE is significantly greater than that of MT, because the operating cost of DE is lower than that of MT. In the environmental mode, the output of MT is significantly greater than that of DE, because the environmental protection cost of DE is higher than that of MT. There is no power outage in both dispatching modes, which ensures the reliability of power supply.

[0150] In addition, for this example, the AP-MPSO algorithm is compared with the PSO algorithm. The convergence curve is as follows: Fig. 9 As shown. When the PSO algorithm is used to optimize the microgrid, the convergence speed is fast (about 50 generations), and it is easy to fall into the local optimal solution; the AP-MPSO algorithm is used to optimize the microgrid, which overcomes the disadvantage of the PSO algorithm being easy to mature prematurely, and the operation results are significantly better than the PSO algorithm. The optimal solution obtained by the PSO algorithm is 1334.7 yuan; the optimal solution obtained by the AP-MPSO algorithm is 1241.64 yuan, a decrease of 6.97%. Combined with the real-time electricity price ( Fig.10 ), and further analyze the operation results. When the real-time electricity price is low, BT operates in a charging state, and the microgrid purchases more electricity from the main grid; when the real-time electricity price is high, BT operates in a discharging state, and the microgrid purchases less electricity from the main grid. As high-quality power supply equipment with low operating costs and no pollution, 96.4% of the power generated by PV and WT is used by the microgrid during operation.

[0151] The AP-MPSO algorithm proposed in this application shows superior performance in solving the multi-objective optimization scheduling problem of microgrids. By introducing an adaptive parameter adjustment mechanism, the ω, c1, c2, and pm parameters are dynamically adjusted according to the characteristics of the optimization process, overcoming the shortcomings of the traditional PSO being prone to premature maturity, and effectively improving the global search capability and convergence accuracy of the algorithm; the addition of an adaptive mutation mechanism increases the diversity of solutions, overcomes the shortcomings of the unreasonable initialization of the PSO algorithm, and ensures the adaptability of the algorithm under different load demands and resource distribution conditions.

[0152] The above description is only a preferred embodiment of the present invention and does not constitute any other form of limitation to the present invention. Any modification or equivalent change made based on the technical essence of the present invention still falls within the scope of protection required by the present invention.

Claims

1. A microgrid optimization scheduling method based on adaptive scheduling and mutation particle swarm algorithm, characterized in that: The following steps are involved: (1) A microgrid system model under the grid-connected operation mode is constructed. Taking the operation and maintenance cost f1(x) and the environmental protection cost f2(x) as the optimization targets, a mathematical model is established. The objective function is: Min C=Min(f1(x),f2(x)) Among them, the operation and maintenance cost f1(x) includes the transaction cost between the microgrid and the main grid, the fuel cost and maintenance cost of the distributed power source, and the environmental protection cost f2(x) includes the cost of pollutant treatment, which must take into account the conditional constraints, including line transmission power constraints, power balance constraints, DG output constraints, battery capacity constraints, DE, and MT climbing constraints; (2) Based on the adaptive scheduling and mutation particle swarm algorithm, the scheduling scheme of each particle is evaluated according to the system model of the microgrid, and the quality of the particle is judged by the Pareto dominance concept. The fitness function of each particle includes two optimization objectives: operation and maintenance cost f1 and environmental protection cost f2. The particle swarm algorithm includes adaptive parameter adjustment and mutation operation: The particle velocity and position update formula is: Among them, ω is the inertia weight, c1 and c2 are the individual learning factor and the group learning factor respectively, r1 and r2 are random numbers between [0,1], and p i,best is the historical optimal position of the ith particle, g best is the global optimal position, x i (t) is the position information of the particle in the tth generation, v i (t+1) is the velocity information of the particle in the t+1 generation; Adaptive parameter adjustment includes dynamic adjustment of inertia weight ω, individual learning factor c1 and group learning factor c2, and the formula is: Among them, IT is the current iteration number, MI is the maximum iteration number, ω s ,ω e are the initial and final values ​​of the inertia weight, c 1s 、c 1e 、c 2s 、c 2e are the initial and final values ​​of the individual learning factor and group learning factor respectively; Mutation operation: In each iteration, 60 dimensions of particles are randomly selected for mutation operation. The mutation probability pm is adaptively adjusted with the number of iterations. The formula is: The mutation probability is higher in the initial iteration, which increases the population diversity and helps the global search; it gradually decreases in the later iteration to avoid destroying the existing optimization results.

2. The microgrid optimization scheduling method based on the adaptive modulation and mutation particle swarm algorithm according to claim 1 is characterized by: The operation and maintenance cost f1(x) of the microgrid system is determined by the following formula: Among them, C GRID (t), C MT (t), C DE (t), C PV (t), C WT (t), C BT (t) are the operating costs of transactions with the main grid, micro gas turbines, diesel generators, photovoltaic power generation, wind turbines and batteries, respectively; c buy (t), c sell (t) are the unit prices of electricity sold and purchased from the main grid at time t, C MT.F (t), C DE.F (t) are the fuel costs of the micro gas turbine and diesel generator, respectively, C MT.M (t), C DE.M (t) are the maintenance costs of the micro gas turbine and diesel generator respectively.

3. The microgrid optimization scheduling method based on the adaptive modulation and mutation particle swarm algorithm according to claim 1 is characterized in that: The environmental protection cost f2(x) is determined by the following formula: Among them, C GRID.EN (t), C MT.EN (t), C DE.EN (t) are the environmental protection costs of the main grid, micro gas turbine and diesel generator respectively, C k is the unit price of treatment of the kth type of pollutant, γ grid,k is the emission coefficient of the kth type of pollutant.

4. The microgrid optimization scheduling method based on the adaptive modulation and mutation particle swarm algorithm according to claim 1 is characterized in that: The optimization scheduling step based on the adaptive scheduling and mutation particle swarm algorithm includes the following constraints: (1) Power balance constraints: P GRID (t)+P PV (t)+P WT (t)+P MT (t)+P DE (t)+P BT (t)=P L (t) Among them, P L (t) is the power of the load at time t; (2) DG output constraints: P i.min ≤P i ≤P i.max Among them, P i.min , P i.max They are the upper and lower limits of each micro power source output respectively; (3) Battery capacity constraints: SOC MIN ≤SOC(t)≤SOC MAX Among them, SOC MIN ,SOC MAX are the minimum and maximum capacities of the battery respectively; (4) DE and MT climbing constraints: Among them, R DE ,R MT They are the climbing power limits of DE and MT respectively.

5. The microgrid optimization scheduling method based on the adaptive modulation and mutation particle swarm algorithm according to claim 1 is characterized in that: The specific process of the adaptive adjustment and mutation particle swarm algorithm is as follows: Step 1: Initialize the population and external archive library, initialize particle position, particle velocity, p i,best and g best ; Step 2: Calculate the fitness value; Step 3: Update the population; Update particle position, velocity, p i,best , g best , mesh division and ω, c1, c2, pm parameters; Step 4: Determine dominance and update external archives; Step 5: Determine whether the archive space is exceeded; Step 6: Determine whether the maximum number of iterations is exceeded. If so, output the Pareto frontier solution set; if not, repeat steps 2-5.

Images