Micro-grid system multi-objective optimization method based on NSGA-GMO algorithm

Abstract

The invention relates to a micro-grid system multi-objective optimization method based on an NSGA-GMO algorithm. The micro-grid system comprising a photovoltaic cell, a wind driven generator, a gas turbine and an energy storage cell is constructed; analyzing the fuel cost of the micro gas turbine; constructing a multi-target optimal scheduling model of the micro-grid system; an NSGA-GMO algorithm is adopted to solve the multi-target optimal scheduling model, and an optimal solution is obtained; and according to the optimal solution, guiding the micro-grid system to dispatch and operate. According to the method, the multi-target optimal scheduling model of the micro-grid system containing the distributed power supply is constructed, and the optimal solution of the multi-target optimal scheduling model is solved by using the NSGA-GMO algorithm, so that the power generation cost and the pollution abatement cost of the micro-grid system can be reduced; and the NSGA-GMO algorithm has better optimization capability and convergence.

Description

Technical Field

[0001] The present invention belongs to the field of power grid optimal dispatching, and particularly relates to a multi-objective optimization method for a microgrid system based on the NSGA-GMO algorithm. Background Technique

[0002] With the continuous increase in energy demand and the increasingly serious environmental problems, the microgrid, as a new type of energy supply method, shows great potential in improving power reliability and reducing environmental impact. The microgrid has advantages such as high flexibility, strong reliability, and environmental protection, and can effectively integrate distributed generation resources such as wind energy and solar energy. However, the operation optimization of the microgrid faces the problem of multi-objective optimization, and multiple conflicting objectives such as economic benefits, environmental sustainability, and system reliability need to be considered simultaneously.

[0003] The paper "Handling Multi-Objective Problems Using Particle Swarm Optimization" by Coello Coello and Lechuga et al. published in the 3rd issue of the IEEE Transactions on Evolutionary Computation in 2004 disclosed the multi-objective particle swarm optimization algorithm (MOPSO). This algorithm uses Pareto dominance to guide the flight direction of particles and uses mutation operations to enhance randomness and solution diversity. However, the fast convergence solution of the multi-objective particle swarm optimization algorithm sometimes leads to the premature termination of the solution search process, resulting in an inaccurate Pareto front.

[0004] The paper "Research on Economic Optimal Dispatching of Microgrid Clusters Based on Improved Vulture Algorithm" by Zhou Hui et al. published in the 2nd issue of Acta Energiae Solaris Sinica in 2024 proposed an improved vulture algorithm, which can effectively reduce the economic cost of the microgrid system and improve the overall benefit of the system through an optimized dispatching strategy. However, while reducing the economic cost of the microgrid system, pollutant emissions and environmental impacts are also important factors that cannot be ignored. Therefore, how to balance economy and environmental protection has become an urgent problem to be solved in the microgrid dispatching optimization. Summary of the Invention

[0005] The object of the present invention is to address the above problems and provide a multi-objective optimization method for a microgrid system based on the NSGA-GMO algorithm. A model of a microgrid system including a photovoltaic cell (PV), a wind turbine (WT), a micro-turbine (MT), and a storage battery (SB) is established, and the optimization objectives of the microgrid system are determined. The elite non-dominated sorting method is combined with the geometric mean optimization algorithm using a feedback iteration mechanism. The geometric mean optimization algorithm improves the convergence accuracy by generating and selecting offspring using the feedback iteration mechanism. Combining the elite non-dominated method and the crowding distance selection strategy enhances the convergence and diversity of potential solutions and precisely locates the Pareto optimal solutions. The multi-objective optimal scheduling of the microgrid system is realized, and the power generation cost and environmental governance cost of the microgrid system are reduced.

[0006] To achieve the above object, the technical solution provided by the present invention is as follows: A multi-objective optimization method for a microgrid system based on the NSGA-GMO algorithm, comprising the following steps: Step 1: Construct a microgrid system including a photovoltaic cell, a wind turbine, a gas turbine, and a storage battery; Step 2: Analyze the fuel cost of the micro-turbine; Step 3: Construct a multi-objective optimal scheduling model for the microgrid system. The optimization objectives of the multi-objective optimal scheduling model include minimizing the power generation cost of the microgrid and minimizing the governance cost of the polluting gases generated by the microgrid power generation; Step 4: Use a solution method combining the non-dominated sorting genetic algorithm and the geometric mean optimizer to solve the multi-objective optimal scheduling model in Step 3 to obtain the optimal solution, i.e., the optimal scheduling plan; Step 5: According to the optimal scheduling plan obtained in Step 4, perform operation scheduling on the microgrid system.

[0007] Preferably, in Step 2, the fuel cost of the micro-turbine is: C MT = ; In the formula, C MT is the fuel cost, C FUEL is the unit price of the fuel, represents the output power of the k-th micro-turbine in the m-th period, η MT is the power generation efficiency of the micro-turbine, C LHV is the lower calorific value of the fuel, m is the period serial number, M represents the number of periods; k is the serial number of the micro-turbine, and K represents the number of micro-turbines.

[0008] Preferably, in step 3, the optimization function for minimizing the power generation cost of the microgrid is: min F 1 = ; = + k MT · ; = k SB · ; = k Grid ·P Grid,m +E Grid ·P Grid,m ; Where: F 1 is the power generation cost of the microgrid, for m Time period k The operation and maintenance cost of a micro gas turbine is for m Time period n The operation and maintenance cost of each energy storage battery, is the operation and maintenance cost of the interaction between the microgrid and the large grid during period m; P Grid,m express m The interactive power between the microgrid and the large grid during the time period; express m Time period n The output power of each energy storage battery; k MT , k SB , k Grid are the operation and maintenance management coefficients of the micro gas turbine, energy storage battery, microgrid and large grid interaction; E Grid is the cost of purchasing or selling electricity at time m; n is the serial number of the energy storage battery, and N is the number of energy storage batteries.

[0009] Preferably, in step 3, the objective function for minimizing the cost of treating polluted gases generated by the microgrid is: Min F 2 = ; In the formula, F 2 represents the cost of treating polluted gases generated by microgrids, is the emission coefficient of the i-th pollutant of the j-th distributed generation, is the emission control cost of the i-th pollutant of the j-th distributed power source, j is the serial number of the distributed power source; i is the type of pollutant; J represents the number of types of micro-power sources in the microgrid; I represents the number of types of pollutant gases generated by the microgrid.

[0010] Preferably, the constraint conditions of the optimization scheduling model include total power balance constraint, output power balance constraint of each micro-power source, charge and discharge boundary value constraint of the energy storage battery, and interactive power constraint between the microgrid and the large grid.

[0011] Further, the total power balance constraint of the microgrid is: = + + ; In the formula, P total is the total power demand of the microgrid system; P j is the output power of the j-th distributed power source, P SB is the charging or discharging power of the energy storage battery; P Grid represents the interactive power between the microgrid and the large grid; Further, the output power balance constraint of each distributed power source is: P j,min ≤P j ≤P j,max ; In the formula, P j,max and P j,min are the maximum and minimum output powers of the j-th distributed power source respectively.

[0012] Further, the charge and discharge boundary value constraint of the energy storage battery is: SOC min ≤ SOC ≤ SOC max ; SOC start = SOC end ; P SB,min ≤P SB ≤P SB,max ; In the formula, SOC is the current battery level of the energy storage battery; SOC max and SOC min are the maximum and minimum battery levels that the energy storage battery can reach respectively, SOC start and SOC end are the battery levels at the initial and end states of a cycle of the energy storage battery respectively, P SB,max and P SB,min are the maximum and minimum values of the charging and discharging powers of the energy storage battery respectively.

[0013] Further, the interactive power constraint between the microgrid and the large grid is: P Grid,min ≤P Grid ≤P Grid,max ; In the formula, PGrid,max , P Grid,min are the maximum and minimum powers of the interaction between the microgrid and the large grid, respectively.

[0014] Furthermore, in step 4, the geometric mean optimizer makes all individual best agents act together by generating a unique global guiding agent to improve the diversity among the individual best agents; the geometric mean optimizer enhances the exploration ability of the search individuals in the initial iteration and also enhances their exploitation ability in the later iteration by setting the scaling parameter vector μ and decreasing its range, helping to achieve a good balance between exploration and exploitation; the fitness of the search individuals is measured using the fuzzy membership function value, and the fitness and diversity of the search individuals in the search space are evaluated through the double fitness index DFI; = + ; = ; = = ; where represents the position vector of the global guiding agent calculated by the i-th search individual in the t-th iteration, represents mutation for guiding the search individuals; w is a control parameter, and randn is a random vector generated from the standard normal distribution, represents the standard deviation vector of the individual optimal solutions so far, represents the maximum standard deviation value of the individual optimal solutions so far; represents the fuzzy membership function value of the L-th individual best agent so far, and represent the average value and the standard deviation of the objective function value of the current individual optimal agent at the t-th iteration, respectively, represents the objective function value of the L-th individual optimal agent so far at the t-th iteration; represents the double fitness index of the i-th search individual at the t-th iteration; represents the individual optimal position vector of the L-th search individual so far.

[0015] The update formulas for the velocity and position of the search individuals are: = ; ; ; ; In the formula and respectively represent the speeds of the i-th search individual at the t-th and (t + 1)-th iterations; and respectively represent the positions of the i-th search individual at the t-th and (t + 1)-th iterations; t represents the number of iterations, represents the maximum number of iterations; is the scaling parameter vector, representing the step size of the search individual towards its guidance; rand is a random number generated within the range of [0, 1].

[0016] Preferably, in step 4, the solution method combining the non-dominated sorting genetic algorithm and the geometric mean optimizer includes the following steps: (1) Input the predicted data of wind power, photovoltaic power and load, and set the algorithm parameters; (2) Generate a random population P t with a size of N, and use the geometric mean optimizer to generate a new population based on the population P t , and the position of the new population is S t , and calculate the fitness values of the two populations respectively on the basis of considering the constraint conditions of each distributed power source; (3) Use the feedback iteration mechanism to generate as the state of the offspring population Q t ; (4) Combine P t and Q t into the population A t , perform fast non-dominated sorting and crowding distance calculation on it, and select the dominant solution set as the new parent P t+1 ; (5) Determine whether the iteration end condition is reached. If so, end the operation and output the load distribution result of the distributed power source; if the judgment result is no, return to step (2) to continue the iteration.

[0017] Compared with the prior art, the beneficial effects of the present invention include: 1) By constructing a multi-objective optimization scheduling model of a microgrid system including distributed power sources, and taking the minimum power generation cost and minimum pollution control cost of the microgrid as the optimization objectives, and using the NSGA-GMO algorithm to solve the optimal solution of the multi-objective optimization scheduling model, the present invention can reduce the power generation cost of the microgrid system, improve the utilization rate and penetration rate of renewable energy, and improve the environmental protection benefits.

[0018] 2) The geometric mean optimization algorithm of the present invention uses a feedback iteration mechanism to generate and select offspring to improve the convergence accuracy, and maintains and enhances the convergence and robustness of potential solutions.

[0019] 3) The present invention combines the elite non-dominated sorting method with the geometric mean optimization algorithm using a feedback iteration mechanism, which has strong robustness and flexibility in dealing with diverse optimization objectives, and can always achieve good convergence, obtaining a diverse Pareto front, and can serve a wider multi-objective optimal scheduling system.

[0020] 4) The microgrid scheduling system provided by the present invention uses the NSGA-GMO algorithm to solve the multi-objective optimal scheduling model of the microgrid system, obtains the optimal scheduling scheme with the minimum power generation cost and the minimum pollution control cost of the microgrid, issues scheduling instructions to each distributed power source, and realizes the intelligent optimal scheduling of the microgrid scheduling system; the present invention can adapt to different operation scenarios and requirements, flexibly respond to various changes in the operation of the microgrid by adjusting the calculation parameters and algorithm settings of each module, and realize the economic operation of the microgrid; by optimizing the scheduling strategy and reasonably allocating the output of each distributed power source, the overall power supply reliability of the microgrid is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The present invention will be further described below in conjunction with the drawings and embodiments.

[0022] Figure 1 It is a schematic structural diagram of the microgrid system according to the embodiment of the present invention.

[0023] Figure 2 It is a flowchart of the NSGA-GMO algorithm according to the embodiment of the present invention.

[0024] Figure 3 It is a predicted power curve diagram of the photovoltaic cell, wind turbine and electrical load according to the embodiment of the present invention.

[0025] Figure 4 It is a comparison diagram of the Pareto front solutions solved by the method of the present invention and the MOPSO algorithm.

[0026] Figure 5 It is an output curve diagram of each distributed power source of the optimal scheduling scheme solved by the present invention in the embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0027] As Figure 1 shown, the multi-objective optimization method of the microgrid system based on the NSGA-GMO algorithm combination includes the following steps: Step 1: Construct a microgrid system including a solar photovoltaic power station, a wind turbine, a gas turbine and a energy storage battery.

[0028] The energy storage battery shows a charging or discharging state at time m, which are respectively expressed as: SOC(m)= SOC(m-1) – P SB (m)·△m·η ch(1) SOC(m)= SOC(m - 1) – P SB (m)·△m / η dis (2) In the formula, SOC(m) and SOC(m - 1) are the remaining power of the energy storage battery at moments m and m - 1 respectively. P SB (m) is the power of the energy storage battery at moment m. When P SB (m) is positive, the energy storage battery is in the discharging state. When P SB (m) is negative, the energy storage battery is in the charging state. η ch and η dis are the charging and discharging efficiencies respectively, and △m is the time interval.

[0029] Step 2: Analyze the fuel cost of the micro gas turbine.

[0030] The fuel cost of the micro gas turbine is: C MT = ; In the formula, C MT is the fuel cost, C FUEL is the unit price of the fuel, represents the output power of the k - th micro gas turbine in the m - th period, η MT is the power generation efficiency of the micro gas turbine, C LHV is the lower calorific value of the fuel, m is the period serial number, M represents the number of periods; k is the serial number of the micro gas turbine, and K represents the number of micro gas turbines.

[0031] Step 3: Construct a multi - objective optimization scheduling model for the micro - grid system. The optimization objectives of this optimization scheduling model include minimizing the power generation cost of the micro - grid and minimizing the pollution control cost.

[0032] The optimization function for minimizing the power generation cost of the micro - grid is: min F 1 = ; = + k MT · ; = k SB · ; = k Grid ·P Grid,m +E Grid ·P Grid,m ; In the formula: F 1 is the power generation cost of the micro - grid, For m the operation and maintenance cost of the k nth micro gas turbine in the mth period, For m the operation and maintenance cost of the n kth energy storage battery in the mth period, is the operation and maintenance cost of the microgrid interacting with the large grid in the mth period; P Grid,m represents m the interaction power between the microgrid and the large grid in the mth period; represents m the output power of the n kth energy storage battery in the mth period; k MT , k SB , k Grid are respectively the operation and maintenance management coefficients of the micro gas turbine, energy storage battery, and microgrid interacting with the large grid; E Grid is the power purchase or sale cost at the mth moment; n is the serial number of the energy storage battery, and N is the number of energy storage batteries.

[0033] The optimization objective function for minimizing the treatment cost of polluting gases generated by the microgrid is: Min F 2 = ; In the formula, F 2 is the polluting gas treatment cost, is the emission coefficient of the ith pollutant of the jth micro power source, is the emission treatment cost of the ith pollutant of the jth micro power source, j is the type of micro power source, and i is the type of pollutant.

[0034] The constraint conditions of the multi-objective optimization scheduling model include: 1) Total power balance constraint of the microgrid: = + + ; In the formula, P total is the total power demand of the microgrid; P j is the output power of the jth distributed power source, P SB is the power during charging or discharging of the energy storage battery; P Grid represents the interaction power between the microgrid and the large grid.

[0035] 2) Output power balance constraint of each micro power source: P j,min ≤P j ≤P j,max ; In the formula, P j,max , P j,minThey are the maximum and minimum output powers of the j-th distributed power source respectively.

[0036] 3) Charge and discharge boundary value constraints of the energy storage battery: SOC min ≤ SOC ≤ SOC max ; SOC start = SOC end ; P SB,min ≤P SB ≤P SB,max ; In the formula, SOC is the current power level of the energy storage battery; SOC max , SOC min are the maximum and minimum power levels that the energy storage battery can reach respectively, SOC start , SOC end are the initial power level and the power level at the end of a cycle respectively, P SB,max , P SB,min are the maximum and minimum values of the charging and discharging powers of the energy storage battery respectively.

[0037] 4) Interactive power constraints between the microgrid and the main grid: P Grid,min ≤P Grid ≤P Grid,max ; In the formula, P Grid,max , P Grid,min are the maximum and minimum interactive powers between the microgrid and the main grid respectively.

[0038] Step 4: Solve the multi-objective optimization scheduling model using the NSGA-GMO algorithm to obtain the optimal solution; The Geometric Mean Optimizer (GMO) is a metaheuristic technique proposed by Farshad Rezaei et al. in 2023. It has advantages in the exploration and exploitation phases, can create a search population using potential solutions, and shows high efficiency in solving objective optimization problems. It uses the generation of unique global guiding agents to make all individual best agents act together, thus maintaining a high diversity among the best agents. A scaling parameter vector μ is set and its range is decreased, which can enhance the exploration ability of the agents in the initial iterations and at the same time enhance the exploitation ability in the later iterations, helping to achieve a good balance between exploration and exploitation. The fuzzy membership function value MF is used to measure the fitness of an agent, and the search agent's fitness and diversity in the search space are evaluated simultaneously through the dual fitness index DFI. It is defined as follows: = + ; = ; = = ; is the mutated used to guide the search agent, w is the control parameter, and randn is a random vector generated from the standard normal distribution, is the standard deviation vector calculated for the individual optimal solution so far, is the maximum standard deviation value on the dimension of the individual optimal solution so far; is the fuzzy membership function value of the L-th individual optimal agent so far, and represent the mean and standard deviation of the objective function values of all current individual optimal agents at the t-th iteration, respectively, is the objective function value of the L-th individual optimal agent so far at the t-th iteration; is the bi-fitness index of the i-th agent at the t-th iteration; is the position vector of the unique global guiding agent calculated by the i-th agent in the t-th iteration, represents the individual optimal position vector of the L-th search agent so far; e is the base of the natural logarithm.

[0039] The update formulas for the agent velocity and position are as follows: = ; ; ; ; where and represent the velocities of the i-th agent at the t-th and t+1-th iterations, respectively; and represent the positions of the i-th agent at the t-th and t+1-th iterations, respectively; is the control parameter, t is the current iteration number, is the maximum iteration number; is a scaling parameter vector representing the step size of the search agent towards its guidance; rand is a random number generated in the range [0, 1].

[0040] The elitist non-dominated sorting method is a multi-objective optimization technique that has been widely used due to its simplicity, efficiency, and significant advantages. During the optimization process, the fast non-dominated sorting algorithm is used to hierarchically partition individuals: First, individuals in the population that are not dominated by any other individuals are identified and stored in a set to form the pareto = 1 layer set; Then, for each individual in the set, the individuals it dominates are checked, and the number of these dominated individuals is decreased by 1, and then these individuals are moved to another set. This process is repeated until all individuals are graded and assigned corresponding pareto ranks.

[0041] To maintain the diversity of the population, crowding distance is used to estimate the density around a certain solution in the population. The method for calculating crowding distance is: First, the population is sorted in ascending order according to the values of each objective function; For each objective function, the boundary solutions are assigned an infinite distance value, while the solutions in the middle are assigned distance values according to the normalized absolute value of the difference in the objective function values between them and their adjacent solutions.

[0042] The solution method combining the non-dominated sorting genetic algorithm and the geometric mean optimizer (NSGA-GMO) generates the population S by using the geometric mean optimizer t , and using the feedback iteration mechanism, the r-th individual is selected from the population state X t , the i-th individual is selected from S t , and a new individual is generated according to the weights λ 1 , λ 2 to form the solution set Q , the initial population P t is combined with Q t to form the solution set A t , and elitist non-dominated sorting is performed on it. Using the crowding selection mechanism, the greater the crowding degree of the solution, the higher the selection probability, which can effectively guide the search process, ensure the balance between exploitation and exploration, and improve the convergence, diversity, and comprehensiveness of the obtained solutions. t In the embodiment, the NSGA-GMO algorithm is used to solve the multi-objective optimal scheduling model of the microgrid system. As

[0043] shown, the NSGA-GMO algorithm includes: Figure 2 (1) Input the predicted data of wind power, photovoltaic, and load, and set the algorithm parameters; (2) Generate a random population P of size N , and use the GMO algorithm to generate a new population S from P t , and calculate the fitness values of the two populations respectively on the basis of considering the constraint conditions of each distributed power source model; t (3) Use the feedback iteration mechanism to generate t (44) (3) Use the feedback iteration mechanism to generate As the offspring population Q t ; (4) Combine P t and Q t to form population A t , perform fast non-dominated sorting and crowding distance calculation on it, and select the dominant solution set as the new parent P t+1 ; (5) Determine whether the iteration end condition is reached. If so, end the operation and output the load distribution result of the distributed power source; if the judgment result is no, return to step (2) to continue the iteration.

[0044] Step 5: According to the optimal solution obtained in step 4, perform operation scheduling on the microgrid system.

[0045] Case study: The scheduling period of the microgrid is 1 day, divided into 24 time periods, i.e., M = 24. The operating parameters of the microgrid system are shown in Table 1, and the predicted powers of photovoltaic, wind power, and load are as Figure 3 shown. Among them, the rated capacity of the energy storage battery is 300 kw, the charge and discharge efficiency is 75%, the initial state of SOC is 50%, the upper limit is 80%, and the lower limit is 20%. The parameter settings of the MOPSO algorithm for comparative analysis include: inertia weight w init = 0.9, w end = 0.4, learning factor c 1 = c 2 = 2.0. The population size N = 60 and the number of iterations T = 200 are set for both the MOPSO algorithm and the NSGA-GMO algorithm of the present invention.

[0046] Table 1 Operating parameters of each distributed power source

[0047] The trading electricity prices between the microgrid system and the large power grid during normal periods, peak periods, and valley periods are shown in Table 2.

[0048] Table 2 Trading electricity prices between the microgrid and the large power grid at different times

[0049] To verify the feasibility and superiority of the NSGA-GMO algorithm, a comparative experiment is conducted with the MOPSO algorithm. Substitute the above micro-power operating parameters and algorithm parameters into the case study, use a common initial population, and solve in the same simulation environment to obtain the Figure 4 Pareto solution set shown. The solution results show that the optimization result using the NSGA-GMO algorithm is significantly better than the MOPSO algorithm, with a faster convergence speed and a more comprehensive and uniform distribution of the solution set.

[0050] The NSGA-GMO algorithm and the MOPSO algorithm are used to solve the comprehensive optimization problem of minimizing the power generation cost and the pollution control cost respectively, and the corresponding objective values are shown in Table 3. The power generation cost obtained by the NSGA-GMO algorithm is 4.5% lower than the optimization result of the MOPSO algorithm, and the pollution control cost is reduced by 7.2%.

[0051] Table 3 Objective function values of different algorithms

[0052] The scheduling schemes obtained based on the solutions of the two algorithms show that within the scheduling period, the optimal output schemes of each micro-power source are as follows: during the valley periods with low load demand, i.e., 1:00 - 6:00 and 23:00 - 24:00, the microgrid system meets the load demand by dispatching wind power and gas turbines. If there is a shortage, it is supplemented by purchasing electricity from the main grid; in addition, the energy storage battery is charged during this period. For the peak periods with high load demand, i.e., 18:00 - 21:00, the energy storage battery discharges to balance the output of each micro-power source and sells the excess electricity to the main grid to reduce costs. During other periods, the output of each distributed power source is as Figure 5 shown, where Grid represents the interaction power between the microgrid and the main grid.

[0053] The simulation results show that the NSGA-GMO algorithm provided by the present invention has better optimization ability and convergence compared with the solution results of the MOPSO algorithm.

[0054] The microgrid scheduling system of the above microgrid system multi-objective optimization method includes:[[]] A gas turbine cost calculation module for calculating the fuel cost of the micro gas turbine; An energy storage battery charge and discharge calculation module for calculating the charge and discharge amount of the energy storage battery; A microgrid power generation cost calculation module for calculating the power generation cost of the microgrid system; A pollution control cost calculation module for calculating the pollution control cost of the polluting gases generated by the power generation of each distributed power source in the microgrid system; A microgrid power purchase and sale calculation module for calculating the cost or revenue of the microgrid system's power purchase and sale with the main grid; An NSGA-GMO algorithm module that combines the elitist non-dominated sorting method with a geometric mean optimizer using a feedback iteration mechanism for solving the multi-objective optimal scheduling model of the microgrid system; An optimal scheduling scheme solving module for constructing a multi-objective optimal scheduling model of the microgrid system, calling the NSGA-GMO algorithm module, obtaining the optimal solution, and generating the optimal scheduling scheme; The command and dispatch module issues dispatch instructions for the output of each distributed power source to the operation units of photovoltaic cells, wind turbines, gas turbines, and energy storage batteries respectively according to the optimal dispatch plan obtained by the optimal dispatch plan solving module.

Claims

1. A multi-objective optimization method for microgrid system based on NSGA-GMO algorithm, characterized in that: The following steps are involved: Step 1: Build a microgrid system including photovoltaic cells, wind turbines, gas turbines and energy storage batteries; Step 2: Analyze the fuel cost of the microturbine; Step 3: construct a multi-objective optimization scheduling model for the microgrid system, wherein the optimization objectives of the multi-objective optimization scheduling model include minimizing the microgrid power generation cost and minimizing the cost of treating polluted gases generated by the microgrid power generation; Step 4: Use a solution method combining a non-dominated sorting genetic algorithm and a geometric mean optimizer to solve the multi-objective optimization scheduling model in step 3 and obtain the optimal solution, i.e., the optimal scheduling plan; Step 5: According to the optimal scheduling plan obtained in step 4, the microgrid system is operated and scheduled.

2. The multi-objective optimization method for a microgrid system according to claim 1, characterized in that: In step 2, the fuel cost of the micro gas turbine is: C MT = ； In the formula, C MT is the fuel cost, C FUEL is the unit price of fuel, represents the output power of the kth micro gas turbine in time period m, η MT is the power generation efficiency of the micro gas turbine, C LHV is the lower calorific value of the fuel, m is the time period number, M represents the number of time periods; k is the number of the micro gas turbine, K represents the number of micro gas turbines.

3. The multi-objective optimization method for a microgrid system according to claim 2, characterized in that: In step 3, the optimization function for minimizing the power generation cost of the microgrid is: my F1= ; = + k MT · ； = k SB · ； = k Grid ·P Grid,m +E Grid ·P Grid,m ; Where: F1 is the power generation cost of the microgrid, for m Time period k The operation and maintenance cost of a micro gas turbine is for m Time period n The operation and maintenance cost of each energy storage battery, is the operation and maintenance cost of the interaction between the microgrid and the large grid during period m; P Grid,m express m The interactive power between the microgrid and the large grid during the time period; express m Time period n The output power of each energy storage battery; k MT , k SB , k Grid are the operation and maintenance management coefficients of the micro gas turbine, energy storage battery, microgrid and large grid interaction; E Grid is the cost of purchasing or selling electricity at time m; n is the serial number of the energy storage battery, and N is the number of energy storage batteries.

4. The multi-objective optimization method for a microgrid system according to claim 3, characterized in that: In step 3, the objective function for minimizing the cost of treating polluted gases generated by the microgrid is: <h2 style=";text-align:left;direction:ltr">Min F2=<h2 style=";text-align:left;direction:ltr"> <h2 style=";text-align:left;direction:ltr"> ; In the formula, F2 represents the cost of treating polluted gas generated by microgrid. is the emission coefficient of the i-th pollutant of the j-th distributed generation, is the emission control cost of the i-th pollutant of the j-th distributed power source, j is the serial number of the distributed power source; i is the type of pollutant; J represents the number of types of micro-power sources in the microgrid; I represents the number of types of pollutant gases generated by the microgrid.

5. The multi-objective optimization method for a microgrid system according to claim 4, characterized in that: In step 3, the constraints of the multi-objective optimization scheduling model of the microgrid system include the total power balance constraint of the microgrid and the output power balance constraint of each distributed power source. = + + ； P j,min ≤P j ≤P j,max ; Where P total is the total power demand of the microgrid system; P j is the output power of the jth distributed power source, P SB P is the charging or discharging power of the energy storage battery; Grid Represents the interaction power between the microgrid and the large grid; P j,max , P j,min are the maximum and minimum output powers of the j-th distributed power source respectively.

6. The multi-objective optimization method for a microgrid system according to claim 5, characterized in that: In step 3, the constraints of the multi-objective optimization scheduling model of the microgrid system also include the charging and discharging boundary value constraints of the energy storage battery: SOCIETY min ≤ SOC ≤ SOC max ; SOCIETY start = SOC end ; P SB,min ≤P SB ≤P SB,max ; In the formula, SOC is the current power of the energy storage battery; SOC max , SOC min They are the maximum and minimum power that the energy storage battery can reach, SOC start , SOC end are the initial and final state power of the energy storage battery in one cycle, P SB,max , P SB,min They are respectively the maximum and minimum values ​​of the charging and discharging power of the energy storage battery.

7. The multi-objective optimization method for a microgrid system according to claim 6, characterized in that: In step 3, the constraints of the multi-objective optimization scheduling model of the microgrid system also include the interactive power constraints between the microgrid and the large grid: P Grid,min ≤P Grid ≤P Grid,max ; Where P Grid,max , P Grid,min They are the maximum and minimum power of interaction between the microgrid and the large grid, respectively.

8. The multi-objective optimization method for a microgrid system according to claim 7, characterized in that: In step 4, the geometric mean optimizer generates a unique global bootstrap proxy To make all individual best agents work together and improve the diversity among individual best agents; The geometric mean optimizer enhances the exploration ability of the search individual in the initial iteration and enhances its exploitation ability in the later iteration by setting the scaling parameter vector μ and decreasing its range, thereby helping to achieve a good balance between exploration and exploitation. The fuzzy membership function value is used to measure the fitness of the search individual, and the dual fitness index DFI is used to evaluate the fitness and diversity of the search individual in the search space. = + ； = ； = ； in, represents the position vector of the global guidance agent calculated by the i-th search individual in the t-th iteration, express The variation of is used to guide the search individual; w is the control parameter, randn is a random vector generated from the standard normal distribution, represents the standard deviation vector of the individual optimal solution so far, Represents the maximum standard deviation of the individual optimal solution so far; represents the fuzzy membership function value of the Lth best individual agent so far, and They represent the mean and standard deviation of the objective function value of the current individual optimal agent at the tth iteration, represents the objective function value of the Lth individual optimal agent so far at the tth iteration; represents the dual fitness index of the i-th search individual at the t-th iteration; Represents the individual optimal position vector of the Lth search individual so far; The update formula for the speed and position of the search individual is: = ； ； ； ； In the formula , They represent the speed of the tth and t+1th iterations of searching for individual i respectively; , They represent the positions of the tth and t+1th iterations of the search individual i respectively; t represents the number of iterations, Indicates the maximum number of iterations; is the scaling parameter vector; rand is a random number.

9. The multi-objective optimization method for a microgrid system according to claim 8, characterized in that: In step 4, the solution method combining the non-dominated sorting genetic algorithm and the geometric mean optimizer adopts a feedback iteration mechanism to generate new solutions, effectively guide the search process, ensure the balance between exploration and utilization, and improve the convergence, coverage and diversity of the solution; The formula for generating a new solution using the feedback iteration mechanism is: =λ1 +λ2 ; λ1= ; λ2= ; λ1+λ2=1; In the formula is the state of the i-th search individual in the t+1 generation; is the state of the rth search individual in the tth generation; λ1 and λ2 are weight coefficients respectively; for The fitness value of for The fitness value of .

10. The multi-objective optimization method for a microgrid system according to claim 8 or 9, characterized in that: In step 4, the solution method combining the non-dominated sorting genetic algorithm with the geometric mean optimizer includes the following steps: (1) Input forecast data of wind power, photovoltaic power and load, and set algorithm parameters; (2) Generate a random population P of size N t , using the geometric mean optimizer in the population P t A new population is generated based on t , the fitness values ​​of the two populations are calculated respectively based on the constraints of each distributed power source; (3) Generate using feedback iteration mechanism As the offspring population Q t Status; (4) P t With Q t Combination into population A t , perform fast non-dominated sorting and crowding distance calculation, and select the dominant solution set as the new parent P t+1 ; (5) Determine whether the iteration end condition is met. If so, terminate the operation and output the load distribution result of the distributed generation; if the judgment result is no, return to step (2) to continue iteration.

Images