Micro-grid multi-objective optimization scheduling method considering orderly charging and discharging of electric vehicle

Abstract

The invention discloses a micro-grid multi-objective optimization scheduling method considering orderly charging and discharging of an electric vehicle, which comprises the following steps: firstly, constructing a micro-grid mathematical model containing access of the electric vehicle by taking the lowest operation energy consumption of a micro-grid system and the lowest environmental governance cost as objectives, and fitting vehicle driving data; and obtaining a probability density function of the daily driving distance S of the electric vehicle and a probability density function of the final return moment t of the electric vehicle, and carrying out simulation prediction on the charging and discharging demands of the electric vehicle by utilizing a Monte Carlo method. Constructing a power constraint; the method comprises the following steps: improving a reference vector guided evolutionary algorithm, solving a microgrid mathematical model by using the improved reference vector guided evolutionary algorithm, solving a Pareto optimal leading edge of environmental economic dispatching, and obtaining an optimal dispatching method. According to the method, the multi-objective optimization problem of the micro-grid can be effectively solved, and the operation efficiency and the optimization capacity can be effectively improved by utilizing the improved reference vector guide evolutionary algorithm to solve.

Description

Technical Field

[0001] The present invention belongs to the technical field of multi-objective optimization scheduling of power systems, and in particular relates to a multi-objective optimization scheduling method for a microgrid taking into account orderly charging and discharging of electric vehicles. Background Art

[0002] The world today is facing a major energy crisis and environmental challenges. The rapid growth in energy demand has led to excessive greenhouse gas emissions and caused global climate change. In order to reduce the continued dependence on fossil fuels, mitigate the negative impact of climate change, and maintain people's quality of life, renewable energy as a low-carbon and clean energy form has received more and more attention. At the same time, with the rapid consumption of traditional fossil energy around the world, energy depletion and environmental pollution have occurred. Compared with traditional cars, electric vehicles, as a new type of transportation, have advantages such as environmental protection and energy saving. They can achieve zero-emission and low-carbon travel, and have become the focus of attention from all walks of life. The addition of a large number of electric vehicle loads to the power grid will have an impact on the power grid, so it is necessary to incorporate electric vehicle loads into the entire scheduling strategy.

[0003] The increasing popularity of renewable energy and the development of smart devices have also accelerated the pace of microgrid construction. Although the real-time dispatching methods of microgrids are constantly developing, there are still some challenges. On the one hand, electric vehicles can be regarded as a distributed energy storage device in microgrids, and the charging demand of electric vehicles can be adjusted according to the available power supply. Through the intelligent charging system, the charging time of electric vehicles can be scheduled during periods of low power demand or low electricity prices, thereby providing load flexibility for microgrids and reducing the operating costs of microgrids; on the other hand, most of the existing microgrid dispatching adopts single-objective intelligent optimization algorithms. Although single-objective intelligent optimization algorithms have the advantages of simple implementation and fast technical speed, single-objective optimization methods can only obtain one optimal solution in one calculation. Multi-objective intelligent optimization algorithms can optimize two conflicting objective functions at the same time, and multiple Pareto optimal solutions can be obtained with only one calculation. However, the existing multi-objective algorithms still have the problem of premature convergence. At the same time, due to the backwardness of Pareto optimal solution selection and storage strategies, they also have the defects of poor diversity and distribution of Pareto optimal solutions. Developing and studying multi-objective evolutionary algorithms with better performance is still a key technology for efficiently solving multi-objective microgrid dispatching problems, which can optimize economic benefits, environmental benefits and reliability.

[0004] The reference vector guided evolutionary algorithm is guided by a set of predefined reference vectors and targets the preferred subset of the entire Pareto front. The reference vector guided evolutionary algorithm uses the angle penalty distance (APD) as a selection strategy to guide the selection of reference vectors, measuring the distance from the solution to the ideal point and the distance from the solution to the reference vector, so that the reference vectors evenly cover the entire Pareto plane and find a set of Pareto optimal sets. The population is divided by dividing the subspace according to the spatial position relationship between the candidate solution and the reference vector. The effectiveness of the proposed model, strategy and algorithm is verified by examples through the scheduling results of different optimization objectives. The results show that the improved algorithm proposed in this paper can improve the economy and environmental protection of the microgrid system and promote the optimal operation of the microgrid. It provides guidance on how to prioritize and implement microgrid systems, thereby bringing more sustainable and cost-effective energy solutions. Summary of the invention

[0005] Purpose of the invention: In order to solve the problem of microgrid optimization scheduling, the present invention provides a microgrid multi-objective optimization scheduling method taking into account the orderly charging and discharging of electric vehicles. Considering the situation of electric vehicles connected to the microgrid, their charging and discharging needs need to be coordinated and managed with other loads and power sources in the microgrid. For the problem of microgrid optimization scheduling, an improved reference vector guided evolutionary algorithm is proposed. In the position update stage, the Chebyshev mapping method is used to generate an initialization population strategy to enhance the local search capability of the algorithm and have stronger optimization performance.

[0006] Technical solution: The present invention discloses a multi-objective optimization scheduling method for a microgrid considering orderly charging and discharging of electric vehicles, comprising the following steps:

[0007] Step 1: Establish a mathematical model of a microgrid including photovoltaic arrays, wind turbines, fuel cells, and micro turbines to obtain the forecast data of photovoltaic output, wind turbine output, and load power in the next day; then add the charging and discharging load caused by electric vehicles connected to the microgrid, and optimize the scheduling of the adjustable power output and energy storage charging and discharging within the microgrid with the goal of minimizing the microgrid operation cost and the microgrid environmental protection cost;

[0008] Step 2: By fitting the vehicle driving data, the probability density function of the daily driving distance S of the electric vehicle and the probability density function of the final return time t of the electric vehicle are obtained, and the charging and discharging demand of the electric vehicle is simulated and predicted using the Monte Carlo method to obtain the charging load curve of the electric vehicle;

[0009] Step 3: In order to ensure the stable operation of the microgrid system, construct power balance constraints, distributed power output constraints, exchange power constraints between the microgrid and the main grid, and power constraints of the energy storage device;

[0010] Step 4: Improve the reference vector guided evolutionary algorithm: use Chebyshev mapping to initialize the population distribution, and use the improved angle penalty distance formula to balance convergence and diversity;

[0011] Step 5: Use the improved reference vector to guide the evolutionary algorithm to solve the mathematical model of the microgrid, solve the Pareto optimal frontier of environmental economic dispatch, and obtain the optimal dispatch method.

[0012] Furthermore, the objective function of the microgrid mathematical model in step 1 is as follows:

[0013] The operation cost of the microgrid is the lowest for the first objective function:

[0014]

[0015] In the formula, C 1 represents the total operating cost of the microgrid, and T is the scheduling time period. fule (t) represents the fuel consumption cost of each distributed power source at time t; C CM (t) represents the equipment maintenance cost of each distributed generation at time t; C grid (t) represents the total cost of interaction between the microgrid and the main grid at time t;

[0016] The minimum environmental protection cost is the second objective function:

[0017]

[0018] In the formula, C CR.MT (t) represents the pollutant treatment fee when the micro gas turbine generates electricity; C CR.FC (t) represents the pollutant treatment fee when the fuel cell generates electricity; C CR.grid (t) represents the pollutant treatment fee of the main grid when the microgrid buys electricity from the main grid.

[0019] Furthermore, the probability density function of the Monte Carlo method for simulating and predicting the charging and discharging demand of electric vehicles in step 2 is as follows:

[0020] By fitting the vehicle driving data, the probability density function of the daily driving distance S of the electric vehicle is obtained:

[0021]

[0022] Where: Standard deviation δ s Take 0.89; expected value μ s Take 3.21;

[0023] The final return time t of the electric vehicle also obeys the normal distribution, and its probability density function is:

[0024]

[0025] Where: Standard deviation δ d Take 17.42; expected value μ d Take 3.43.

[0026] Furthermore, the charging and discharging requirements of the electric vehicle in step 2 are specifically as follows:

[0027] 1) Disordered charging model of electric vehicles: that is, the charging mode is based on the user's personal habits and random needs without taking any management measures; the charging load of electric vehicles is calculated:

[0028]

[0029] Where, T c Indicates charging time; W is power consumption per 100 kilometers; P c is the charging power; c For charging efficiency;

[0030] According to the final return time t of the electric vehicle and the charging duration T c , the Monte Carlo simulation method is used to fit the charging load demand of a single electric vehicle within a day, and the total charging load curve of the electric vehicle is obtained. The total charging load of the electric vehicle is added to the total load of the microgrid optimization scheduling;

[0031] 2) Orderly charging and discharging model of electric vehicles: The energy storage system of electric vehicles is regarded as a dispatchable resource, and electric vehicles are charged or discharged to the grid in an orderly manner according to demand; when the grid demand is peak, the pressure on the grid is relieved by releasing energy storage power to the grid; and when the grid demand is low, the idle electric vehicle batteries are used for charging;

[0032] The maximum discharge time of a single electric vehicle is calculated by the following formula:

[0033]

[0034] Where: Indicates the maximum discharge time of an electric vehicle; SOC max , SCO min Respectively represent the upper and lower limits of the state of charge of the electric vehicle; P d Indicates discharge power;

[0035] Since the scheduling period T is 0 to 24 hours, according to the final return time t of the electric vehicle and the maximum discharge time That is, the actual discharge time T d for:

[0036]

[0037] Sustainable charging time of a single electric vehicle T c It can be expressed as:

[0038]

[0039] T is the return time, T d is the actual discharge time of the electric vehicle, T c How long does it take to charge an electric vehicle?

[0040] According to the final return time t of the electric vehicle and the charging duration T c ,The Monte Carlo simulation method is used to fit the charging and discharging load demand of a single electric vehicle within a day, and the orderly charging and discharging load curve of the electric vehicle is obtained. The orderly charging and discharging load of the electric vehicle is added to the total load of the microgrid optimal scheduling.

[0041] Furthermore, the power balance constraint conditions constructed in step 3 are as follows:

[0042] Power balance constraints:

[0043]

[0044] Where: Indicates that each distributed generation unit provides useful power in the period t; P EV (t) represents the power of the electric vehicle; P grid (t) represents the active power generated by the microgrid and the main grid during the period t; P load (t) represents the system load power during the period t.

[0045] Furthermore, the improvement of the reference vector guided evolutionary algorithm in step 4 is as follows:

[0046] First, the population distribution is initialized using the Chebyshev mapping method. The formula is as follows:

[0047] x n+1 = cos(k arc cos x n ),x n ∈[-1,1]

[0048] The improved angle penalty distance formula is used to balance convergence and diversity. Its mathematical expression is:

[0049]

[0050] In the formula, ||f′ t,i || represents the modulus of the target vector of individual i in the t-th generation subpopulation, represents the mean value of the modulus of all individual target vectors associated with the t-th generation reference vector v in the subpopulation, m represents the number of targets in the multi-target problem, θ t,i,j Represents the individual i in the t-th generation subpopulation and the reference vector v j The acute angle between is the reference vector v t,j The minimum angle between the vector and other reference vectors in the current generation, Represents the reference vector ν associated with all individuals in the t-th generation subpopulation i The mean of the angles between .

[0051] Furthermore, the step 5 uses the improved reference vector to guide the evolutionary algorithm to solve the microgrid mathematical model, which specifically includes the following steps:

[0052] Step 5.1: Initialize the population distribution using Chebyshev mapping;

[0053] Step 5.2: Generate a reference vector set, which is a set of uniformly distributed vectors representing some important directions in the target space;

[0054] Step 5.3: Evaluate fitness: For each individual, calculate its fitness value in the target space by mapping the decision variables to the target space;

[0055] Step 5.4: Parent selection, selecting parent individuals from the population to produce new offspring;

[0056] Step 5.5: Generate offspring, use mutation and crossover operations to process parent individuals to generate new offspring individuals;

[0057] Step 5.6: Elite selection: select some individuals with higher fitness values ​​from the population as elite individuals;

[0058] Step 5.7: Use the reference vector to divide the target space into multiple subspaces, and perform selection in each subspace separately;

[0059] Step 5.8: Target value conversion: Make sure the converted target values ​​are all in the first quadrant;

[0060] Step 5.9: Population division: After translating the target value, divide the population P t Divide into N sub-populations;

[0061] Step 5.10: Angle penalty distance calculation, using the improved angle penalty distance method to balance convergence and diversity;

[0062] Step 5.11: Reference vector adaptation, adjust and update the reference vector set according to the distribution of individuals in the population. This helps guide the search direction of the algorithm and promotes a better distribution of the population in the target space;

[0063] Step 5.12: Select the next generation population from the parent generation and the offspring generation;

[0064] Step 5.13: Check the termination condition: Determine whether the termination condition is met. If the number of iterations reaches the maximum, the loop ends. If not, continue iterating until the termination condition is met.

[0065] Beneficial effects:

[0066] 1. The present invention constructs a complete microgrid system, including a power generation system, a renewable energy system and an energy storage device, and considers the situation where electric vehicles are connected to the microgrid. Their charging and discharging needs need to be coordinated and managed with other loads and power sources in the microgrid.

[0067] 2. The present invention establishes a microgrid system involving electric vehicles, takes into account the disorderly charging and orderly charging and discharging of electric vehicles, uses real-time electricity prices to guide the orderly charging and discharging of electric vehicles, and further demonstrates the benefits of applying these strategies from the perspective of microgrid operation.

[0068] 3. Aiming at the optimization scheduling problem of microgrids, the present invention proposes an improved reference vector guided evolutionary algorithm, which uses Chebyshev mapping method to generate the initialization population strategy in the position update phase to enhance the local search ability of the algorithm. At the same time, in order to successfully achieve the balance between convergence and diversity in high-dimensional multi-objective evolutionary algorithms, an improved angular penalty distance method is proposed. The results show that the improved reference vector guided evolutionary algorithm has stronger optimization performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 System flow chart for optimizing dispatch of microgrids;

[0070] Figure 2 Flowchart of the guided evolutionary algorithm for improved reference vectors. DETAILED DESCRIPTION

[0071] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0072] The present invention discloses a multi-objective optimization scheduling method for a microgrid considering orderly charging and discharging of electric vehicles, which mainly includes the following steps:

[0073] 1. Establishment of Microgrid Mathematical Model

[0074] A mathematical model of a microgrid including photovoltaic arrays, wind turbines, fuel cells, and micro turbines is established. With the goal of minimizing the microgrid operating cost and the microgrid environmental protection cost, the adjustable power output and energy storage charging and discharging within the microgrid are optimized.

[0075] The lowest power generation cost of the microgrid is the first objective function:

[0076]

[0077] In the formula, C 1 represents the total operating cost of the microgrid, and T is the scheduling time period. fule (t) represents the fuel consumption cost of each distributed power source at time t, and its mathematical model is shown in formula (2). CM (t) represents the equipment maintenance cost of each distributed power source at time t, and its mathematical model is shown in formula (3). grid (t) represents the total cost of interaction between the microgrid and the main grid at time t, and its mathematical model is shown in formula (4).

[0078] C fule (t) = C MT (t)+C FC (t) (2)

[0079] In the formula, C MT (t) represents the fuel cost consumed by the micro gas turbine when generating electricity; C FC (t) represents the fuel cost consumed when the fuel cell generates electricity.

[0080] C CM (t) = C CM.MT (t)+C CM.FC (t) (3)

[0081] In the formula, C CM.MT (t) represents the maintenance cost of the micro gas turbine when generating electricity; C CM.FC (t) represents the maintenance cost when the fuel cell is generating electricity.

[0082]

[0083] In formula (4), C bp (t) represents the electricity purchase price between the microgrid and the large grid at time t. sp (t) represents the electricity price between the microgrid and the large grid at time t. grid (t) represents the interaction power between the microgrid and the large grid at time t.

[0084] The minimum environmental protection cost is the second objective function:

[0085]

[0086] in

[0087]

[0088] In the formula, C CR.MT (t) represents the pollutant treatment fee when the micro gas turbine generates electricity, P MT (t) represents the power of the micro gas turbine when generating electricity, K n.MT The penalty fee rate for producing the nth type of polluting gas when the gas turbine generates electricity; C CR.FC (t) represents the pollutant treatment fee when the fuel cell generates electricity, P FC (t) represents the power of the fuel cell, K n.FC The penalty fee rate for generating the nth pollutant gas when the fuel cell generates electricity; C CR.grid (t) represents the pollutant treatment fee of the main grid when the microgrid buys electricity from the main grid, where C n K is the cost of treating each n pollutant gases. n.grid The penalty fee rate for producing the nth type of polluting gas for the main grid electricity production.

[0089] 2. Use the Monte Carlo method to simulate and predict the charging and discharging demand of electric vehicles

[0090] By fitting the vehicle driving data, the probability density function of the daily driving distance S of the electric vehicle is obtained:

[0091]

[0092] Where: Standard deviation δ s Take 0.89; expected value μ s Take 3.21. Randomly generate the daily mileage S of the mth electric vehicle according to the probability density function of the daily mileage of electric vehicles.

[0093] The final return time t of the electric vehicle also obeys the normal distribution, and its probability density function is:

[0094]

[0095] Where: Standard deviation δ d Take 17.42; expected value μ d Take 3.43. Randomly generate the daily charging start time t of the mth electric vehicle according to the probability density function of the daily charging start time of the electric vehicle.

[0096] Disordered charging model of electric vehicles:

[0097] Disordered charging of electric vehicles refers to a charging mode in which electric vehicles are charged according to the user's personal habits and random needs without taking any management measures.

[0098] The electric vehicle charging load can be calculated from equations (7) and (8):

[0099]

[0100] Where, T c Indicates charging time; W is power consumption per 100 kilometers; P c is the charging power; c For charging efficiency.

[0101] According to the final return time t of the electric vehicle and the charging duration T c ,The Monte Carlo simulation method is used to fit the charging load demand of a single electric vehicle within a day, and the total charging load curve of the electric vehicle can be obtained. The total charging load of the electric vehicle is added to the total load of the microgrid optimal scheduling.

[0102] Electric vehicle orderly charging and discharging model:

[0103] Orderly charging and discharging of electric vehicles is to use the energy storage system of electric vehicles as a dispatchable resource, and electric vehicles discharge or discharge to the grid in an orderly manner according to demand. When the grid demand is peak, the pressure on the grid can be relieved by releasing the energy storage power to the grid; and when the grid demand is low, the idle electric vehicle batteries can be used for charging, thereby playing a role in balancing the load and ensuring the stability of the microgrid system.

[0104] The maximum discharge time of a single electric vehicle can be calculated by the following formula:

[0105]

[0106] Where: Indicates the maximum discharge time of an electric vehicle; SOC max , SCO min Respectively represent the upper and lower limits of the state of charge of the electric vehicle; P d Indicates the discharge power.

[0107] Since the scheduling period T is 0 to 24 hours, according to the final return time t of the electric vehicle and the maximum discharge time The actual discharge time T can be obtained d for:

[0108]

[0109] Sustainable charging time of a single electric vehicle T c It can be expressed as:

[0110]

[0111] T is the return time, T d is the actual discharge time of the electric vehicle, T c How long does it take to charge an electric car?

[0112] According to the final return time t of the electric vehicle and the charging duration T c ,The Monte Carlo simulation method is used to fit the charging load demand of a single electric vehicle within a day, and the sequential charging and discharging load curve of the electric vehicle can be obtained. The sequential charging and discharging load of the electric vehicle is added to the total load of the microgrid optimization scheduling.

[0113] 3. Construct power balance constraints, distributed power output constraints, exchange power constraints between microgrids and main grids, and power constraint formulas for energy storage devices.

[0114] Power balance constraints:

[0115]

[0116] Where: Indicates that each distributed generation unit provides useful power in the period t; P EV (t) represents the power of the electric vehicle; P grid (t) represents the active power generated by the microgrid and the main grid during the period t; P load (t) represents the system load power during the period t.

[0117] Micro gas turbine output constraints:

[0118]

[0119] Where: are the upper and lower limits of the micro gas turbine output respectively; MT is the upper limit of the climbing power of the micro gas turbine.

[0120] Fuel cell output constraints:

[0121]

[0122] Where: are the upper and lower limits of the fuel cell output respectively; FC is the upper limit of the fuel cell's climbing power.

[0123] The exchange power constraints between the microgrid and the main grid are:

[0124]

[0125] Where: They represent the lower and upper limits of the interaction power between the microgrid and the large grid respectively;

[0126] Power constraints of energy storage devices:

[0127]

[0128] Where: It is the upper and lower limits of the energy storage device output. A positive value indicates power input, and a negative value indicates power output. SOC max (t), SOC min (t) are the upper and lower limits of the energy storage capacity at time t.

[0129] 4. Improvement of reference vector guided evolutionary algorithm

[0130] First, in order to generate an initial population with a wide range, uniform distribution, and strong global search capability, the Chebyshev mapping method is used to initialize the population distribution. The formula is shown in formula (18):

[0131] x n+1 = cos(k arc cos x n ),x n ∈[-1,1] (18)

[0132] Where k is the order of the Chebyshev polynomial, which can generate unrelated chaotic sequences with ergodic properties. The steps of the Chebyshev mapping population are as follows:

[0133] (1) Assuming the population size is N, a D-dimensional vector is randomly generated in the D-dimensional space as the initial individual, Y = (y 1 ,y 2 ,...,y d ),v i ∈[-1,1];

[0134] (2) Use formula (18) to iterate each dimension of Y N-1 times to generate the remaining N-1 individuals;

[0135] (3) The mapping formula is shown in formula (19). Formula (19) is used to map the generated N individuals to the solution search space in sequence:

[0136] x id = l d +(1+y id )×(u d -l d ) / 2 (19)

[0137] Among them, u d , l dare the upper and lower bounds of the dth dimension of the search space, respectively, id is the dth dimension of the ith individual, x id is the d-dimensional coordinate value of the i-th individual in the search space.

[0138] The specific solution process of the reference vector guided evolutionary algorithm is:

[0139] (1) Set the reference vector and generate n evenly distributed points on the hyperplane, where M is the number of targets. The reference vector generation formula is:

[0140]

[0141] The unit reference vector is set to:

[0142]

[0143] ν i The normalized vector generated by the reference point and the origin. Given two vectors v 1 and v 2 , the cosine of the acute angle θ between two vectors can be used to measure the spatial relationship between them, calculated as

[0144]

[0145] Where ||·|| is the calculation norm, that is, the length of the vector. It is used to measure the spatial relationship between the target vector and the reference vector in the objective space. Since the reference vector has been normalized when it is generated, it is no longer necessary to normalize it when calculating the cosine value.

[0146] (2) Population initialization: Generate NP randomly distributed decision vectors in the decision space, where NP represents the size of the population.

[0147] (3) Generation of offspring. In the reference vector guided evolution algorithm, the operation of generating offspring is the differential operation, that is, simulating binary crossover (SBX) and polynomial mutation to create the offspring population.

[0148] (4) Reference vector-based selection strategy: Use the reference vector to divide the target space into multiple subspaces, and perform selection in each subspace separately. The steps of the reference vector-based selection strategy are as follows:

[0149] a. Target value conversion

[0150] The starting point of the reference vector in this paper is always the origin of the coordinate system. t The target value of individuals in Convert to F t ', transformed by the following method:

[0151]

[0152] where i=1,...,|P t |, f t,i and f′ t,i are the target vectors of individual i before and after translation, Represents the minimum objective value calculated by Ft. The role of the conversion operation is to ensure that the converted objective values ​​are all in the first quadrant, where the poles of each objective function are on the corresponding coordinate axis, thereby maximally covering the reference vector.

[0153] b. Population division

[0154] After translating the target value, the population P t Divide into N subpopulations By associating each individual with its closest reference vector, where N is the number of reference vectors, the spatial relationship of two vectors is calculated with reference to the acute angle between them, i.e. the cosine value between the target vector and the reference vector:

[0155]

[0156] Among them, θ t,i,j Denotes the target vector f′ t,i and the reference vector v t,j Thus, individual I t,i Assigned to subpopulation If and only if f′ t,i and v t,j The smallest among all reference vectors (i.e. the largest cosine value):

[0157]

[0158] Among them, I t,i Indicates P t The i-th individual in, i=1,...,|P t |.

[0159] c. Angle Penalty Distance Calculation

[0160] The angle penalty distance is used to balance convergence and diversity in multidimensional optimization problems. Its mathematical expression is:

[0161] d t,i,j =(1+P(θ t,i,j ))·||f′ t,i || (26)

[0162] The mathematical description of the IAPD method is given below, as shown in formula (27):

[0163]

[0164] In the formula, ||f′ t,i || represents the modulus of the target vector of individual i in the t-th generation subpopulation, represents the mean value of the modulus of all individual target vectors associated with the t-th generation reference vector v in the subpopulation, m represents the number of targets in the multi-target problem, θ t,i,j Represents the individual i in the t-th generation subpopulation and the reference vector v j The acute angle between is the reference vector v t,j The minimum angle between the vector and other reference vectors in the current generation, Represents the reference vector ν associated with all individuals in the t-th generation subpopulation i The improved APD method can overcome the shortcomings of the original APD method in balancing the convergence and diversity of the algorithm.

[0165] (5) Reference vector adaptation: In order to obtain uniformly distributed solutions on the unevenly distributed Pareto plane, the reference vector needs to be adaptive. The reference vector is dynamically adjusted according to the distribution of candidate solutions, which improves the efficiency of processing the objective function. Its mathematical expression is:

[0166]

[0167] where i = 1, ..., N, ν t+1,i represents the i-th adaptive reference vector of the next generation t+1, ν 0,i represents the i-th uniformly distributed reference vector generated in the initialization phase, and They represent the maximum and minimum values ​​of each objective function in the t+1th generation respectively. The operator represents the Hadamard product of the element multiplied by two vectors (or matrices) of the same size. Using the above reference vector adaptation strategy, a uniformly distributed solution can be obtained even if the objective function is not normalized to the same range.

[0168] 5. Use the improved reference vector to guide the evolutionary algorithm to solve the mathematical model of the microgrid, solve the Pareto optimal frontier of environmental economic dispatch, and obtain the optimal dispatch method

[0169] Step 5.1: Initialize the population distribution using the Chebyshev mapping method.

[0170] Step 5.2: Generate a set of reference vectors. Reference vectors are a set of uniformly distributed vectors that represent some important directions in the target space.

[0171] Step 5.3: Evaluate fitness: For each individual, calculate its fitness value in the target space. This can be achieved by mapping the decision variables to the target space.

[0172] Step 5.4: Parent selection, selecting parent individuals from the population to produce new offspring

[0173] Step 5.5: Generate offspring by using mutation and crossover operations to process parent individuals and generate new offspring individuals.

[0174] Step 5.6: Elite selection: Select some individuals with higher fitness values ​​from the population as elite individuals.

[0175] Step 5.7: Use the reference vector to divide the target space into multiple subspaces, and perform selection separately in each subspace.

[0176] Step 5.8: Target value conversion: Make sure the converted target values ​​are all in the first quadrant.

[0177] Step 5.9: Population division: After translating the target value, divide the population P t Divide into N sub-populations.

[0178] Step 5.10: Angle penalty distance calculation, using the improved angle penalty distance method to balance convergence and diversity.

[0179] Step 5.11: Reference vector adaptation, adjust and update the reference vector set according to the distribution of individuals in the population. This helps guide the search direction of the algorithm and promotes a better distribution of the population in the target space.

[0180] Step 5.12: Select the next generation population from the parents and offspring.

[0181] Step 5.13: Check the termination condition: Determine whether the termination condition is met. If the number of iterations reaches the maximum, the loop ends. If not, continue iterating until the termination condition is met.

[0182] The technical means disclosed in the scheme of the present invention are not limited to the technical means disclosed in the above-mentioned implementation mode, but also include technical schemes composed of any combination of the above-mentioned technical features. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications are also regarded as the protection scope of the present invention.

Claims

1. A multi-objective optimization scheduling method for microgrids considering orderly charging and discharging of electric vehicles, characterized in that: The steps include: Step 1: Establish a mathematical model of a microgrid including photovoltaic arrays, wind turbines, fuel cells, and micro turbines to obtain the forecast data of photovoltaic output, wind turbine output, and load power in the next day; then add the charging and discharging load caused by electric vehicles connected to the microgrid, and optimize the scheduling of the adjustable power output and energy storage charging and discharging within the microgrid with the goal of minimizing the microgrid operation cost and the microgrid environmental protection cost; Step 2: By fitting the vehicle driving data, the probability density function of the daily driving distance S of the electric vehicle and the probability density function of the final return time t of the electric vehicle are obtained, and the charging and discharging demand of the electric vehicle is simulated and predicted using the Monte Carlo method to obtain the charging load curve of the electric vehicle; Step 3: In order to ensure the stable operation of the microgrid system, construct power balance constraints, distributed power output constraints, exchange power constraints between the microgrid and the main grid, and power constraints of the energy storage device; Step 4: Improve the reference vector guided evolutionary algorithm: use Chebyshev mapping to initialize the population distribution, and use the improved angle penalty distance formula to balance convergence and diversity; Step 5: Use the improved reference vector to guide the evolutionary algorithm to solve the mathematical model of the microgrid, solve the Pareto optimal frontier of environmental economic dispatch, and obtain the optimal dispatch method.

2. The multi-objective optimization scheduling method for microgrids considering orderly charging and discharging of electric vehicles according to claim 1 is characterized in that: The objective function of the microgrid mathematical model in step 1 is as follows: The operation cost of the microgrid is the lowest for the first objective function: In the formula, C1 represents the total operating cost of the microgrid, and T is the scheduling time period. fule (t) represents the fuel consumption cost of each distributed power source at time t; C CM (t) represents the equipment maintenance cost of each distributed generation at time t; C grid (t) represents the total cost of interaction between the microgrid and the main grid at time t; The minimum environmental protection cost is the second objective function: In the formula, C CR.MT (t) represents the pollutant treatment fee when the micro gas turbine generates electricity; C CR.FC (t) represents the pollutant treatment fee when the fuel cell generates electricity; C CR.grid (t) represents the pollutant treatment fee of the main grid when the microgrid buys electricity from the main grid.

3. The multi-objective optimization scheduling method for microgrids considering orderly charging and discharging of electric vehicles according to claim 1 is characterized in that: The probability density function of the Monte Carlo method for simulating and predicting the charging and discharging demand of electric vehicles in step 2 is as follows: By fitting the vehicle driving data, the probability density function of the daily driving distance S of the electric vehicle is obtained: Where: Standard deviation δ s Take 0.89; expected value μ s Take 3.21; The final return time t of the electric vehicle also obeys the normal distribution, and its probability density function is: Where: Standard deviation δ d Take 17.42; expected value μ d Take 3.

43.

4. The multi-objective optimization scheduling method for microgrids considering orderly charging and discharging of electric vehicles according to claim 3 is characterized in that: The specific charging and discharging requirements of the electric vehicle in step 2 are as follows: 1) Disordered charging model of electric vehicles: that is, the charging mode is based on the user's personal habits and random needs without taking any management measures; the charging load of electric vehicles is calculated: Where, T c Indicates charging time; W is the power consumption per 100 kilometers; P c is the charging power; c For charging efficiency; According to the final return time t of the electric vehicle and the charging duration T c , the Monte Carlo simulation method is used to fit the charging load demand of a single electric vehicle within a day, and the total charging load curve of the electric vehicle is obtained. The total charging load of the electric vehicle is added to the total load of the microgrid optimization scheduling; 2) Orderly charging and discharging model of electric vehicles: The energy storage system of electric vehicles is regarded as a dispatchable resource, and electric vehicles are charged or discharged to the grid in an orderly manner according to demand; when the grid demand is peak, the pressure on the grid is relieved by releasing energy storage power to the grid; and when the grid demand is low, the idle electric vehicle batteries are used for charging; The maximum discharge time of a single electric vehicle is calculated by the following formula: Where: Indicates the maximum discharge time of an electric vehicle; SOC max , SCO min Respectively represent the upper and lower limits of the state of charge of the electric vehicle; P d Indicates discharge power; Since the scheduling period T is 0 to 24 hours, according to the final return time t of the electric vehicle and the maximum discharge time That is, the actual discharge time T d for: Sustainable charging time of a single electric vehicle T c It can be expressed as: T is the return time, T d is the actual discharge time of the electric vehicle, T c How long does it take to charge an electric vehicle? According to the final return time t of the electric vehicle and the charging duration T c ,The Monte Carlo simulation method is used to fit the charging and discharging load demand of a single electric vehicle within a day, and the orderly charging and discharging load curve of the electric vehicle is obtained. The orderly charging and discharging load of the electric vehicle is added to the total load of the microgrid optimal scheduling.

5. The multi-objective optimization scheduling method for microgrids considering orderly charging and discharging of electric vehicles according to claim 1 is characterized in that: The power balance constraints constructed in step 3 are as follows: Power balance constraints: Where: Indicates that each distributed generation unit provides useful power in the period t; P EV (t) represents the power of the electric vehicle; P grid (t) represents the active power generated by the microgrid and the main grid during the period t; P load (t) represents the system load power during the period t.

6. The multi-objective optimization scheduling method for microgrids considering orderly charging and discharging of electric vehicles according to claim 1 is characterized in that: The improvement of the reference vector guided evolutionary algorithm in step 4 is as follows: First, the population distribution is initialized using the Chebyshev mapping method. The formula is as follows: x n+1 =cos(k arc cos x n ),x n ∈[-1,1] The improved angle penalty distance formula is used to balance convergence and diversity. Its mathematical expression is: In the formula, ||f′ t,i || represents the modulus of the target vector of individual i in the t-th generation subpopulation, represents the mean value of the modulus of all individual target vectors associated with the t-th generation reference vector v in the subpopulation, m represents the number of targets in the multi-target problem, θ t,i,j Represents the individual i in the t-th generation subpopulation and the reference vector v j The acute angle between is the reference vector v t,j The minimum angle between the vector and other reference vectors in the current generation, Represents the reference vector ν associated with all individuals in the t-th generation subpopulation i The mean of the angles between .

7. The multi-objective optimization scheduling method for microgrids considering orderly charging and discharging of electric vehicles according to claim 6 is characterized in that: The step 5 uses the improved reference vector to guide the evolutionary algorithm to solve the microgrid mathematical model, which specifically includes the following steps: Step 5.1: Initialize the population distribution using Chebyshev mapping; Step 5.2: Generate a reference vector set, which is a set of uniformly distributed vectors representing some important directions in the target space; Step 5.3: Evaluate fitness: For each individual, calculate its fitness value in the target space by mapping the decision variables to the target space; Step 5.4: Parent selection, selecting parent individuals from the population to produce new offspring; Step 5.5: Generate offspring, use mutation and crossover operations to process parent individuals to generate new offspring individuals; Step 5.6: Elite selection: select some individuals with higher fitness values ​​from the population as elite individuals; Step 5.7: Use the reference vector to divide the target space into multiple subspaces, and perform selection in each subspace separately; Step 5.8: Target value conversion: Make sure the converted target values ​​are all in the first quadrant; Step 5.9: Population division: After translating the target value, divide the population P t Divide into N sub-populations; Step 5.10: Angle penalty distance calculation, using the improved angle penalty distance method to balance convergence and diversity; Step 5.11: Reference vector adaptation, adjust and update the reference vector set according to the distribution of individuals in the population. This helps guide the search direction of the algorithm and promotes a better distribution of the population in the target space; Step 5.12: Select the next generation population from the parent generation and the offspring generation; Step 5.13: Check the termination condition: Determine whether the termination condition is met. If the number of iterations reaches the maximum, the loop ends. If not, continue iterating until the termination condition is met.

Images