An orderly charging scheduling method for electric vehicles based on improved Kepler algorithm

Abstract

A method for orderly charging scheduling of electric vehicles based on an improved Kepler algorithm includes the following steps: Step 1: Establishing a Prophet model to predict photovoltaic power generation in future time periods; Step 2: Constructing an objective function that minimizes battery loss cost, agent scheduling costs, and grid load variance. The agent scheduling costs are calculated based on the photovoltaic power generation in the future time period obtained in Step 1; Step 3: Setting constraints on the electric vehicle charging and discharging power, state of charge, charge capacity, and actual transformer capacity; Step 4: Defining these objective functions and constraints based on the improved algorithm to solve the orderly charging scheduling problem for electric vehicles. This method implements the scheduling model's "peak shaving" effect on electricity load, improving the algorithm's convergence speed and accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and in particular to an orderly charging scheduling strategy for electric vehicles based on an improved Kepler algorithm. Background Art

[0002] As more and more countries promote the development of new energy technologies to protect the environment, electric vehicles (EVs) continue to see increasing annual sales in the automotive industry due to their significant clean and environmental benefits. Currently, there are over 1.2 billion EVs on the planet, and this number is projected to reach a staggering 2 billion by 2035. The increasing per capita ownership of EVs means greater demand for charging. If EVs are charged in a disorderly manner, this will significantly impact the grid's supply and demand balance, leading to problems such as degraded power quality, increased peak loads, excessive harmonic injection, and power loss.

[0003] Patent publication number CN117507867A discloses a method for orderly charging of community electric vehicles. This method constructs a two-layer multi-objective optimization model based on minimizing the peak-to-valley difference of community load and user charging costs, and uses the rat swarm optimization algorithm to solve the optimization model to obtain the optimal charging plan; patent publication number CN116127828A discloses an electric vehicle charging method, device and related equipment based on time-of-use electricity prices. This patent uses the minimization of the peak-to-valley difference of the total power grid load as the objective function, and solves the electric vehicle orderly charging model through the artificial bee colony algorithm.

[0004] None of the above inventions consider the real-time changes in electric vehicle battery capacity, nor do they consider the interests of electric vehicles, charging station distributors, and the power grid to construct an objective function that constrains the interests of these three parties. Furthermore, existing solutions also suffer from slow convergence speeds and insufficient convergence accuracy. Summary of the Invention

[0005] The purpose of the present invention is to solve the technical problems existing in the existing electric vehicle orderly charging scheduling technology, which are that the real-time changes of the electric vehicle battery capacity are ignored, the scheduling model convergence speed is slow, and the convergence accuracy is insufficient, resulting in the scheduling efficiency and accuracy failing to meet the development needs of electric vehicles. A method for orderly charging scheduling of electric vehicles based on an improved Kepler algorithm is proposed.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0007] An electric vehicle orderly charging scheduling method based on an improved Kepler algorithm comprises the following steps:

[0008] Step 1: Establish a Prophet model to predict photovoltaic power generation in the future;

[0009] Step 2: Construct an objective function: Minimize the battery loss cost, the agent's dispatch cost, and the grid load variance. Calculate the agent's dispatch cost based on the future PV power generation obtained in Step 1.

[0010] Step 3: Set constraints: constrain the electric vehicle charging and discharging power, state of charge, charge capacity, and actual capacity of the Transformer;

[0011] Step 4: Optimize and process the above objective function and constraints based on the chaos mapping-lateral crossing-Lévy flight-Kepler algorithm to solve the scheduling problem of orderly charging of electric vehicles.

[0012] In Step 1, the Prophet model consists of the following four functions:

[0013]

[0014] Where, represents the photovoltaic power generation power predicted in the future i-th time period, g (i) Represents the overall growth trend of photovoltaic power generation, which is a trend function, s (i) Indicates a periodic trend, for example, a periodic function, h (i) Indicates the impact of non-fixed photovoltaic power generation failure on the prediction results, which is the influence function, ε (i) It represents abnormal fluctuations that cannot be predicted by the prediction model and is a random error function.

[0015] The battery loss cost minimization calculation in Step 2 includes:

[0016] First, one hour is a time period, so there are 24 time periods in a day; starting from 0:00 in a day and ending at 24:00, the corresponding time period is recorded as i = 1, 2, ..., 24; within a day, the electric vehicles are numbered as j = 1, 2, ..., m in the order of the time they are connected to the power grid;

[0017] Next, introduce the 0-1 variable n ij , which represents the alternating state of charging and discharging of electric vehicles:

[0018]

[0019] Where, P i,j is the charging and discharging power of the jth electric vehicle in the i-th time period; when P i,j >0, it means the jth electric car is in charging state; when P i,j <0, it means the electric vehicle is in the discharge state; when P i,j=0, it means the electric vehicle is neither charging nor discharging;

[0020] Then, the Gauss-Newton iteration method was used to perform nonlinear fitting on the relationship between the number of charge and discharge alternations and the remaining capacity of various batteries. The fitting curve is g j (n); furthermore, the sum of the battery loss costs of each electric vehicle h1 is minimized as follows:

[0021]

[0022] Where, is the cumulative number of charge and discharge alternations when the jth electric vehicle is connected to the grid, c j is the battery cost of the jth electric car, n jmax is the maximum number of charge and discharge cycles that the j-th electric vehicle battery can withstand.

[0023] The scheduling cost minimization calculation in Step 2 includes:

[0024] The dispatch cost is composed of the electricity purchase cost paid by the charging pile agent to the grid and the discharge cost paid by the charging pile agent to the electric vehicle;

[0025] 1) Calculation of electricity purchase cost

[0026] Introducing the concept of energy storage capacity to measure the real-time storage energy of the charging pile;

[0027] 0≤C i ≤C max ,i=1,2,…,24

[0028] Where C i is the energy stored in the charging pile in the i-th time period, C max The energy storage capacity of the charging pile;

[0029] If the sum of the photovoltaic power generation energy, the energy obtained from the grid and the energy stored in the charging pile in the previous period is greater than the charging demand of the electric vehicle in this period, the excess energy will be stored in the next period. Otherwise, there will be no energy storage in this period, that is:

[0030]

[0031] Where, T is the time of each period, that is, T = 1h, is the photovoltaic power generation power in the i-th time period, and is the power obtained by the charging pile from the grid in the i-th time period. When , it means the charging pile obtains power from the grid; when When , it means the grid obtains power from the charging pile;

[0032] The electricity purchase cost paid by the charging pile agent to the grid in the i-th time period is price i ,Right now:

[0033]

[0034] Where, is the price of electricity purchased from the power grid by the charging pile agent in the i-th time period, The price of electricity sold by the charging pile agent to the grid in the i-th time period;

[0035] Therefore, the electricity purchase cost paid by the charging pile agent to the grid in one day is h 2,1 It can be expressed as:

[0036]

[0037] 2) Calculation of discharge cost

[0038] When p i When <0, it means that the electric vehicle is in the discharge state, so the discharge cost of each model of electric vehicle within one day is h 2,2 It can be expressed as:

[0039]

[0040] Where p i,j is the charging and discharging power of the jth electric vehicle in the i-th time period, is the discharge compensation price of electric vehicles in the i-th time period;

[0041] Therefore, the scheduling cost h2 is minimized as follows:

[0042] minh2=h 2,1 +h 2,2 .

[0043] The grid load variance minimization calculation in Step 2 includes:

[0044] The average value of the grid load μ is expressed as:

[0045]

[0046] Where, is the power obtained by the charging pile from the grid in the i-th period, n is the number of periods in a day, that is, n = 24;

[0047] Then, the power load variance is calculated. The smaller the variance, the smaller the grid load change, the more stable the load, and the higher the grid stability. The grid load variance h3 is minimized as follows:

[0048]

[0049] In Step 3, the charge and discharge power are constrained as follows:

[0050] Whether the electric vehicle is in the charging or discharging state, its power should be less than or equal to the maximum power. When it is not connected to the grid, the power is 0, that is:

[0051]

[0052] Where, P i,j is the charging and discharging power of the jth electric vehicle in the i-th time period, P jmax is the maximum charging and discharging power of the jth electric vehicle, is the time when the jth electric vehicle is connected to the grid, For the user of the jth electric vehicle, input the time when the charging pile leaves the grid;

[0053] In Step 3, the state of charge is constrained as follows:

[0054] To prevent rapid battery degradation, the residual power of an electric vehicle after discharge should not be less than 10% of the rated capacity, and the residual power of an electric vehicle after charging should not exceed 95% of the rated capacity; that is:

[0055]

[0056] Where, SOC i,j is the state of charge of the j-th electric vehicle after the i-th time period, is the state of charge of the jth electric vehicle when it is connected to the grid, B j The battery capacity of the jth electric vehicle, η is the charging and discharging efficiency;

[0057] In Step 3, the charge capacity is constrained as follows:

[0058] From the time an electric vehicle is connected to the grid until it leaves the grid, the amount of electricity supplied by the charging pile to the electric vehicle should meet the user's established needs. Therefore:

[0059]

[0060] Where, For the jth electric vehicle user, the expected state of charge of the charging pile when leaving the grid is input, g j (n) is a nonlinear fitting function of the number of charge and discharge cycles of the j-th electric vehicle and the remaining capacity of various batteries;

[0061] In Step 3, the actual capacity of the Transformer is constrained as follows:

[0062] The total power of the entire dispatching system in any time period should not exceed the distribution capacity of the area, so:

[0063]

[0064] Where, is the basic load of the area where electric vehicles are connected to the grid in the i-th time period, S max is the distribution transformer capacity of the area, is the power factor.

[0065] The specific steps of solving the scheduling model using the chaos map-lateral cross-Lévy flight-Kepler algorithm in Step 4 are as follows:

[0066] 1) Use the fourth-order Chebyshev chaotic map to construct the original solar system with M planets and N dimensions, and initialize the orbital eccentricity, revolution period and position of the planets; position X r,s The initialization formula is as follows:

[0067] chaos(r+1)=cos(4×cos -1 (chaos(r)))、

[0068]

[0069] Where r is the number of planets, one planet represents a set of candidate solutions, and s is the decision variable X in the optimization model. r,s The number of decision variables in this optimization model is 24×m P i,j , 24 and λ1, λ2, λ3, chaos(r) is the chaos parameter of the rth planet, The decision variables X are r,s The upper and lower limits of

[0070] 2) Calculate the fitness function value of each planet, and use the one with the minimum fitness value as the fit best Planet X as the Sun in Galaxy best , and calculate the mass of all celestial bodies in the galaxy and the gravitational force between the sun and other planets;

[0071] 3) Using the horizontal crossover strategy, each of the M-1 planets orbiting the sun generates a new satellite to strengthen the competition between celestial bodies. The specific formula is as follows:

[0072]

[0073] Where t is the number of iterations, rand is a random number between 0 and 1, c is a random number between -1 and 1, and X a (t) and X b(t) is the position of two random planets, Sate a (t), Sate b (t) are X a (t) and X b (t) satellite (candidate solution); whether the satellite can replace the planet depends on their respective fitness values. Only when the best fitness value is better than the planet can it be retained in the subsequent position update operation. The specific formula is as follows:

[0074]

[0075] 4) Calculate the speed of the planet orbiting the sun and the normalized Euclidean distance R between the planet and the sun r (t), and update the position of the planet using the following formula:

[0076] When R r (t)<0.5:

[0077]

[0078] When R r (t)≥0.5:

[0079]

[0080] Where, is the position of the rth planet, is the position of the sun, and are the positions of two random planets; h is the adaptive factor that controls the distance between the sun and the current planet in the current iteration number t, is a 0-1 operator, η is a ±1 operator, The revolution speed of the rth planet, F gr (t) the gravitational force between the rth planet and the sun, rand3 is a normally distributed random number;

[0081] 5) Using the Levy flight strategy, planets closer to the sun conduct local searches near the sun using small jumps, while planets farther from the sun conduct searches in unknown regions of space using large jumps. The specific formula is as follows:

[0082]

[0083] Where γ is the transmission parameter, χ is the search step size, u and v are random operators following the normal distribution, and β∈(0,2) is the parameter that controls the shape of the distribution.

[0084] 6) Determine whether the same result is obtained in consecutive iterations. If so, exit the iteration loop; if not, update and archive the new non-inferior solution and continue iterating.

[0085] A chaotic map-lateral cross-Lévy flight-Kepler algorithm is used to solve a scheduling model and includes the following steps:

[0086] 1) Use the fourth-order Chebyshev chaotic map to construct the original solar system with M planets and N dimensions, and initialize the orbital eccentricity, revolution period and position X of the planets. r,s ;

[0087] 2) Calculate the fitness function value of each planet, and use the one with the minimum fitness value as the fit best Planet X as the Sun in Galaxy best , and calculate the mass of all celestial bodies in the galaxy and the gravitational force between the sun and other planets;

[0088] 3) Using a horizontal crossover strategy, each of the M-1 planets orbiting the sun generates a new satellite to intensify competition between celestial bodies;

[0089] 4) Calculate the speed of the planet orbiting the sun and the normalized Euclidean distance R between the planet and the sun r (t), and update the position of the planet;

[0090] 5) Using the Levy flight strategy, planets closer to the sun conduct local searches near the sun in small jumps, while planets farther from the sun conduct large jumps to search unknown areas of space;

[0091] 6) Determine whether the same result is obtained in consecutive iterations. If so, exit the iteration loop; if not, update and archive the new non-inferior solution and continue iterating.

[0092] In step 1), position X r,s The initialization formula is as follows:

[0093] chaos(r+1)=cos(4×cos -1 (chaos(r)))、

[0094]

[0095] Where r is the number of planets, one planet represents a set of candidate solutions, and s is the number of decision variables in the optimization model. There are 24×m decision variables in the optimization model. i,j , 24 and λ1, λ2, λ3, chaos(r) is the chaos parameter of the rth planet, The decision variables X are r,s The upper and lower limits of

[0096] In step 3), the specific formula used is as follows:

[0097]

[0098] Where t is the number of iterations, rand is a random number between 0 and 1, c is a random number between -1 and 1, and X a (t) and X b (t) is the position of two random planets, Sate a (t), Sate b (t) are X a (t) and X b (t) satellite (candidate solution); whether the satellite can replace the planet depends on their respective fitness values. Only when the best fitness value is better than the planet can it be retained in the subsequent position update operation. The specific formula is as follows:

[0099]

[0100] In step 4), the planet's position is updated using the following formula:

[0101] When R r (t)<0.5:

[0102]

[0103] When R r (t)≥0.5:

[0104]

[0105] Where, is the position of the rth planet, is the position of the sun, and are the positions of two random planets; h is the adaptive factor that controls the distance between the sun and the current planet in the current iteration number t, is a 0-1 operator, η is a ±1 operator, The revolution speed of the rth planet, F gr (t) the gravitational force between the rth planet and the sun, rand3 is a normally distributed random number;

[0106] In step 5), the specific formula is as follows:

[0107]

[0108]

[0109] Where γ is the transmission parameter, χ is the search step size, u and v are random operators following the normal distribution, and β∈(0,2) is the parameter that controls the shape of the distribution.

[0110] Compared with the prior art, the present invention has the following technical effects:

[0111] 1) The present invention develops a scheduling strategy for orderly charging of electric vehicles by energy storage charging piles to reduce the adverse effects of electric vehicles connected to the grid;

[0112] 2) This invention builds a model and structure for a safe, reliable, and sustainable dispatching system from the perspectives of users, charging pile agents, and power grid companies.

[0113] 3) Based on the original Kepler algorithm, this paper combines chaotic mapping, lateral crossing and Levy flight strategy to propose an improved Kepler optimization algorithm to overcome its shortcomings such as imbalance between exploration and development, slow convergence speed and insufficient convergence accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0114] The present invention will be further described below with reference to the accompanying drawings and embodiments:

[0115] Figure 1 It is a structural diagram of the scheduling system of the present invention;

[0116] Figure 2 This is a flow chart of the improved Kepler algorithm of the present invention;

[0117] Figure 3 A comparison diagram of power grid loads under the dispatching strategy of the present invention and under the disordered dispatching strategy;

[0118] Figure 4 A comparison chart of the optimization iterations of the improved Kepler algorithm of the present invention and the classic Kepler algorithm;

[0119] Figure 5 This is a trend chart of battery loss cost changes in the present invention. DETAILED DESCRIPTION

[0120] Figure 1 As shown, this paper provides an orderly charging scheduling method for electric vehicles based on an improved Kepler algorithm, which includes the following steps:

[0121] Step 1: Establish a Prophet model to predict photovoltaic power generation in the future;

[0122] Step 2: Construct an objective function: Minimize the battery loss cost, the agent's dispatch cost, and the grid load variance. Calculate the agent's dispatch cost based on the future PV power generation obtained in Step 1.

[0123] Step 3: Set constraints: constrain the electric vehicle charging and discharging power, state of charge, charge capacity, and actual capacity of the Transformer;

[0124] Step 4: Optimize and process the above objective function and constraints based on the chaos mapping-lateral crossing-Lévy flight-Kepler algorithm to solve the scheduling problem of orderly charging of electric vehicles.

[0125] The Prophet model in Step 1 is a decomposable time series model, consisting of the following four functions:

[0126]

[0127] Where, represents the photovoltaic power generation power predicted in the future i-th time period, g (i) Represents the overall growth trend of photovoltaic power generation, which is a trend function, s (i) Indicates a periodic trend, for example, a periodic function, h (i) Indicates the impact of non-fixed photovoltaic power generation failure on the prediction results, which is the influence function, ε (i) It represents abnormal fluctuations that cannot be predicted by the prediction model and is a random error function.

[0128] The battery loss cost minimization calculation in Step 2 includes:

[0129] First, consider one hour as a time period, so there are 24 time periods in a day. Starting at 0:00 AM and ending at 12:00 PM, the corresponding time periods are denoted as i = 1, 2, ..., 24. Within a day, electric vehicles are numbered j = 1, 2, ..., m, in the order in which they were connected to the grid.

[0130] Next, introduce the 0-1 variable n ij , which represents the alternating state of charging and discharging of electric vehicles:

[0131]

[0132] Where, P i,j is the charging and discharging power of the jth electric vehicle in the i-th time period. i,j >0, it means the jth electric car is in charging state; when P i,j <0, it means the electric vehicle is in the discharge state; when P i,j =0, it means that the electric vehicle is neither charging nor discharging.

[0133] Then, the Gauss-Newton iteration method was used to perform nonlinear fitting on the relationship between the number of charge and discharge alternations and the remaining capacity of various batteries. The fitting curve is g j(n). Then, the sum of the battery loss costs of each electric vehicle h1 is minimized as follows:

[0134]

[0135] Where, is the cumulative number of charge and discharge alternations when the jth electric vehicle is connected to the grid, c j is the battery cost of the jth electric car, n jmax is the maximum number of charge and discharge cycles that the j-th electric vehicle battery can withstand.

[0136] The scheduling cost minimization calculation in Step 2 includes:

[0137] The dispatch cost consists of the electricity purchase cost paid by the charging pile agent to the grid and the discharge cost paid by the charging pile agent to the electric vehicle.

[0138] 1) Calculation of electricity purchase cost

[0139] The concept of energy storage capacity is introduced to measure the real-time storage energy of the charging pile.

[0140] 0≤C i ≤C max ,i=1,2,…,24

[0141] Where C i is the energy stored in the charging pile in the i-th time period, C max The energy storage capacity of the charging pile.

[0142] If the sum of the photovoltaic power generation energy, the energy obtained from the grid and the energy stored in the charging pile in the previous period is greater than the charging demand of the electric vehicle in this period, the excess energy will be stored in the next period. Otherwise, there will be no energy storage in this period, that is:

[0143]

[0144] Where, T is the time of each period, that is, T = 1h, is the photovoltaic power generation power in the i-th time period, and is the power obtained by the charging pile from the grid in the i-th time period. When , it means the charging pile obtains power from the grid; when When , it means that the grid obtains power from the charging pile.

[0145] The electricity purchase cost paid by the charging pile agent to the grid in the i-th time period is price i ,Right now:

[0146]

[0147] Where, is the price of electricity purchased from the power grid by the charging pile agent in the i-th time period, The price of electricity sold by the charging pile agent to the grid in the i-th time period.

[0148] Therefore, the electricity purchase cost paid by the charging pile agent to the grid in one day is h 2,1 It can be expressed as:

[0149]

[0150] 2) Calculation of discharge cost

[0151] When p i When <0, it means that the electric vehicle is in the discharge state, so the discharge cost of each model of electric vehicle within one day is h 2,2 It can be expressed as:

[0152]

[0153] Where p i,j is the charging and discharging power of the jth electric vehicle in the i-th time period, is the discharge compensation price of the electric vehicle in the i-th time period.

[0154] Therefore, the scheduling cost h2 is minimized as follows:

[0155] minh2=h 2,1 +h 2,2

[0156] The grid load variance minimization calculation in Step 2 includes:

[0157] The average value of the grid load μ is expressed as:

[0158]

[0159] Where, is the power obtained by the charging pile from the grid during the i-th period, and n is the number of periods in a day, that is, n=24.

[0160] Then, the power load variance is calculated. The smaller the variance, the smaller the grid load change, the more stable the load, and the higher the grid stability. Minimizing the grid load variance h3 is expressed as:

[0161]

[0162] Since the optimization objectives of the three objective functions are all minimized, this paper introduces a positive number λ before each objective function and then sums them up, transforming the multi-objective function into a single objective function:

[0163] minh=λ1h1+λ2h2+λ3h3

[0164]

[0165] In Step 3, the charge and discharge power are constrained as follows:

[0166] Whether the electric vehicle is in the charging or discharging state, its power should be less than or equal to the maximum power. When it is not connected to the grid, the power is 0, that is:

[0167]

[0168] Where, P i,j is the charging and discharging power of the jth electric vehicle in the i-th time period, P jmax is the maximum charging and discharging power of the jth electric vehicle, is the time when the jth electric vehicle is connected to the grid, The user of the jth electric vehicle inputs the time when the charging pile leaves the grid.

[0169] In Step 3, the state of charge is constrained as follows:

[0170] To prevent rapid battery degradation, the residual power of an electric vehicle after discharge should not be less than 10% of the rated capacity, and the residual power of an electric vehicle after charging should not exceed 95% of the rated capacity. That is:

[0171]

[0172] Where, SOC i,j is the state of charge of the j-th electric vehicle after the i-th time period, is the state of charge of the jth electric vehicle when it is connected to the grid, B j The battery capacity of the jth electric vehicle, η is the charging and discharging efficiency.

[0173] In Step 3, the charge capacity is constrained as follows:

[0174] From the time an electric vehicle is connected to the grid until it leaves the grid, the amount of electricity supplied by the charging pile to the electric vehicle should meet the user's established needs. Therefore:

[0175]

[0176] Where, For the jth electric vehicle user, the expected state of charge of the charging pile when leaving the grid is input, g j (n) is a nonlinear fitting function of the number of charge and discharge alternations of the j-th electric vehicle and the remaining capacity of various batteries.

[0177] In Step 3, the actual capacity of the Transformer is constrained as follows:

[0178] The total power of the entire dispatching system in any time period should not exceed the distribution capacity of the area, so:

[0179]

[0180] Where, is the basic load of the area where electric vehicles are connected to the grid in the i-th time period, S max is the distribution transformer capacity of the area, is the power factor.

[0181] like Figure 2 As shown, the specific steps of solving the scheduling model by using the chaos map-lateral cross-Lévy flight-Kepler algorithm in Step 4 are as follows:

[0182] 1) Use the fourth-order Chebyshev chaotic map to construct the original solar system with M planets and N dimensions, and initialize the orbital eccentricity, revolution period and position of the planets; position X r,s The initialization formula is as follows:

[0183] chaos(r+1)=cos(4×cos -1 (chaos(r)))、

[0184]

[0185] Where r is the number of planets, one planet represents a set of candidate solutions, and s is the decision variable X in the optimization model. r,s The number of decision variables in this optimization model is 24×m P i,j , 24 and λ1, λ2, λ3, chaos(r) is the chaos parameter of the rth planet, The decision variables X are r,s The upper and lower limits of

[0186] 2) Calculate the fitness function value of each planet, and use the one with the minimum fitness value as the fit best Planet X as the Sun in Galaxy best , and calculate the mass of all celestial bodies in the galaxy and the gravitational force between the sun and other planets;

[0187] 3) Using the horizontal crossover strategy, each of the M-1 planets orbiting the sun generates a new satellite to strengthen the competition between celestial bodies. The specific formula is as follows:

[0188]

[0189] Where t is the number of iterations, rand is a random number between 0 and 1, c is a random number between -1 and 1, and X a (t) and X b (t) is the position of two random planets, Sate a (t), Sate b (t) are X a (t) and X b (t) satellite (candidate solution); whether the satellite can replace the planet depends on their respective fitness values. Only when the best fitness value is better than the planet can it be retained in the subsequent position update operation. The specific formula is as follows:

[0190]

[0191] 4) Calculate the speed of the planet orbiting the sun and the normalized Euclidean distance R between the planet and the sun r (t), and update the position of the planet using the following formula:

[0192] When R r (t)<0.5:

[0193]

[0194] When R r (t)≥0.5:

[0195]

[0196] Where, is the position of the rth planet, is the position of the sun, and are the positions of two random planets; h is the adaptive factor that controls the distance between the sun and the current planet in the current iteration number t, is a 0-1 operator, η is a ±1 operator, The revolution speed of the rth planet, F gr (t) the gravitational force between the rth planet and the sun, rand3 is a normally distributed random number;

[0197] 5) Using the Levy flight strategy, planets closer to the sun conduct local searches near the sun using small jumps, while planets farther from the sun conduct searches in unknown regions of space using large jumps. The specific formula is as follows:

[0198]

[0199] Where γ is the transmission parameter, χ is the search step size, u and v are random operators following the normal distribution, and β∈(0,2) is the parameter that controls the shape of the distribution.

[0200] 6) Determine whether the same result is obtained in consecutive iterations. If so, exit the iteration loop; if not, update and archive the new non-inferior solution and continue iterating.

[0201] The present invention also includes a chaos map-lateral cross-Lévy flight-Kepler algorithm, which is used to solve the scheduling model and includes the following steps:

[0202] 1) Use the fourth-order Chebyshev chaotic map to construct the original solar system with M planets and N dimensions, and initialize the orbital eccentricity, revolution period and position X of the planets. r,s ;

[0203] 2) Calculate the fitness function value of each planet, and use the one with the minimum fitness value as the fit best Planet X as the Sun in Galaxy best , and calculate the mass of all celestial bodies in the galaxy and the gravitational force between the sun and other planets;

[0204] 3) Using a horizontal crossover strategy, each of the M-1 planets orbiting the sun generates a new satellite to intensify competition between celestial bodies;

[0205] 4) Calculate the speed of the planet orbiting the sun and the normalized Euclidean distance R between the planet and the sun r (t), and update the position of the planet;

[0206] 5) Using the Levy flight strategy, planets closer to the sun conduct local searches near the sun in small jumps, while planets farther from the sun conduct large jumps to search unknown areas of space;

[0207] 6) Determine whether the same result is obtained in consecutive iterations. If so, exit the iteration loop; if not, update and archive the new non-inferior solution and continue iterating.

[0208] In step 1), position X r,s The initialization formula is as follows:

[0209] chaos(r+1)=cos(4×cos -1 (chaos(r)))、

[0210]

[0211] Where r is the number of planets, one planet represents a set of candidate solutions, and s is the number of decision variables in the optimization model. There are 24×m decision variables in the optimization model. i,j , 24 and λ1, λ2, λ3, chaos(r) is the chaos parameter of the rth planet, The decision variables X are r,s The upper and lower limits of

[0212] In step 3), the specific formula used is as follows:

[0213]

[0214] Where t is the number of iterations, rand is a random number between 0 and 1, c is a random number between -1 and 1, and X a (t) and X b (t) is the position of two random planets, Sate a (t), Sate b (t) are X a (t) and X b (t) satellite (candidate solution); whether the satellite can replace the planet depends on their respective fitness values. Only when the best fitness value is better than the planet can it be retained in the subsequent position update operation. The specific formula is as follows:

[0215]

[0216] In step 4), the planet's position is updated using the following formula:

[0217] When R r (t)<0.5:

[0218]

[0219] When R r (t)≥0.5:

[0220]

[0221] Where, is the position of the rth planet, is the position of the sun, and are the positions of two random planets; h is the adaptive factor that controls the distance between the sun and the current planet in the current iteration number t, is a 0-1 operator, η is a ±1 operator, The revolution speed of the rth planet, F gr (t) the gravitational force between the rth planet and the sun, rand3 is a normally distributed random number;

[0222] In step 5), the specific formula is as follows:

[0223]

[0224] Where γ is the transmission parameter, χ is the search step size, u and v are random operators following the normal distribution, and β∈(0,2) is the parameter that controls the shape of the distribution.

[0225] Example:

[0226] So far, the task of calculating the orderly charging scheduling strategy for electric vehicles has been successfully completed. The following example illustrates this:

[0227] This article collects data from a charging station in a distribution area in Jinan, Shandong Province, on July 19, 2024. A total of 36 electric vehicles were involved in charging and discharging. It is assumed that these 36 electric vehicles are of the same model and that the users agree to the scheduling agreement. The key parameters of the distribution area, charging station, and electric vehicles are shown in the following table:

[0228]

[0229] The latest time-of-use electricity prices for 2024 announced by State Grid Shandong Electric Power Company are as follows:

[0230]

[0231]

[0232] Compare the grid load conditions under the present invention and disorderly charging modes, such as Figure 3 shown.

[0233] It can be clearly seen that in the disordered charging mode, the grid load fluctuates greatly, with the highest peak at around 7:00 pm, reaching 230.06 kW. In the mode of the present invention, although the maximum peak also occurs between 7:00 and 8:00 pm, it is significantly smaller than the peak in the wireless charging mode. This proves that the model has achieved the goal of "peak shaving" well. In addition, the period from 9:00 am to 5:00 pm is a low point for the grid load in the disordered charging mode, while the grid load in the ordered charging mode during the same period shows a smooth downward trend. Therefore, the model is still very effective in "valley filling".

[0234] The improved Kepler algorithm of the present invention is compared with the classic Kepler algorithm optimization iteration, as shown in the following example: Figure 4 shown.

[0235] As can be seen from the figure, the objective function value of the improved Kepler algorithm is lower in the convergence stage, which makes the convergence speed faster and the convergence accuracy higher when solving the optimization problem.

[0236] During the scheduling process of the present invention, the battery loss cost change trend is as follows: Figure 5 .

[0237] The figure shows that during periods of low electricity prices, battery depletion costs show a slow upward trend. This trend becomes increasingly steeper from normal electricity price periods to peak periods, and then to peak periods. Higher electricity prices during these periods indicate higher grid loads. The faster the rise in battery depletion costs, the more frequent regulation is. This result indicates that regulation actions often occur during periods of high grid load, effectively achieving peak load shaving and valley filling.

Claims

1. An orderly charging scheduling method for electric vehicles based on an improved Kepler algorithm, characterized in that: The following steps are involved: Step 1: Establish a Prophet model to predict photovoltaic power generation in the future; Step 2: Construct an objective function: Minimize the battery loss cost, the agent's dispatch cost, and the grid load variance. Calculate the agent's dispatch cost based on the future PV power generation obtained in Step 1. Step 3: Set constraints: constrain the electric vehicle charging and discharging power, state of charge, charge capacity, and actual capacity of the Transformer; Step 4: Optimize and process the above objective function and constraints based on the Chaos Mapping-Transverse Crossing-Lévy Flight-Kepler algorithm to solve the scheduling problem of orderly charging of electric vehicles; The specific steps of solving the scheduling model using the chaos map-lateral cross-Lévy flight-Kepler algorithm in Step 4 are as follows: 1) The number of planets constructed using the fourth-order Chebyshev chaos map is M , dimension is N The original solar system, and initialize the orbital eccentricity, revolution period and position of the planet; position The initialization formula is as follows: 、 ； Where, r is the number of planets, one planet represents a set of candidate solutions, s To optimize the decision variables in the model The number of decision variables in this optimization model is indivual , 24 as well as 、 、 , For the r The chaos parameters of the planet, 、 The decision variables The upper and lower limits of 2) Calculate the fitness function value of each planet, and take the one with the minimum fitness value The planet as the sun in the galaxy , and calculate the mass of all celestial bodies in the galaxy and the gravitational force between the sun and other planets; 3) Using horizontal crossover strategy, Each planet orbiting the sun generates a new satellite to intensify the competition between celestial bodies. The specific formula is as follows: Where, t is the number of iterations, is a random number between 0 and 1, is a random number between -1 and 1. and are two random planet positions, 、 They are and Satellites; whether a satellite can replace a planet depends on their respective fitness values. Only when the best fitness value is better than the planet can it be retained in the subsequent position update operation. The specific formula is as follows: ； 4) Calculate the speed of the planet orbiting the sun and the normalized Euclidean distance between the planet and the sun , and update the planet's position using the following formula: when hour: （Y2）； when hour: Where, For the r The position of the planets, is the position of the sun, and are the positions of two random planets; Is to control the current number of iterations Adaptive factor for the distance between the Sun and the current planet, is a 0-1 operator, yes Operator, No. r The planet's orbital speed, No. r The gravitational force between the planets and the sun, is a normally distributed random number; 5) Using the Levy flight strategy, planets closer to the sun conduct local searches near the sun using small jumps, while planets farther from the sun conduct searches in unknown regions of space using large jumps. The specific formula is as follows: ； ； ； Where, is the transmission parameter, is the search step length, 、 is a random operator following a normal distribution, is a parameter that controls the shape of the distribution; 6) Determine whether the same result is obtained in consecutive iterations. If so, exit the iteration loop; if not, update and archive the new non-inferior solution and continue iterating.

2. The method according to claim 1, characterized in that In Step 1, the Prophet model consists of the following four functions: ； Where, Indicates the future i The photovoltaic power generation power predicted for each time period is Represents the overall growth trend of photovoltaic power generation, which is a trend function. Indicates a periodic trend, for example, a periodic function, Indicates the impact of non-fixed photovoltaic power generation failure on the prediction results, which is the influence function, It represents abnormal fluctuations that cannot be predicted by the prediction model and is a random error function.

3. The method according to claim 1, characterized in that The battery loss cost minimization calculation in Step 2 includes: First, one hour is a time period, so there are 24 time periods in a day; starting from 0:00 in a day and ending at 24:00, the corresponding time period is ; Electric vehicles are numbered in the order of their access to the grid within a day. ; Next, introduce the 0-1 variable , which represents the alternating state of charging and discharging of electric vehicles: ； Where, For the j Electric cars in the i The charging and discharging power in each time period; When j An electric vehicle is in charging state; when When , it means the electric vehicle is in the discharge state; when When , it means that the electric vehicle is neither charging nor discharging; Then, the Gauss-Newton iteration method was used to perform nonlinear fitting on the relationship between the number of charge and discharge alternations and the remaining capacity of various batteries. The fitting curve is: ; Furthermore, the sum of the battery loss costs of each electric vehicle The minimization is expressed as: ； Where, For the j The cumulative number of charge and discharge alternations when an electric vehicle is connected to the power grid, For the j The battery cost of an electric car, For the j The maximum number of charge and discharge cycles that an electric vehicle battery can withstand. Indicates the alternating state of charging and discharging of electric vehicles.

4. The method according to claim 1, wherein The scheduling cost minimization calculation in Step 2 includes: The dispatch cost is composed of the electricity purchase cost paid by the charging pile agent to the grid and the discharge cost paid by the charging pile agent to the electric vehicle; 1) Calculation of electricity purchase cost Introducing the concept of energy storage capacity to measure the real-time storage energy of the charging pile; ； Where, For charging pile i The energy stored in a period of time, Energy storage capacity of the charging pile; If the sum of the photovoltaic power generation energy, the energy obtained from the grid and the energy stored in the charging pile in the previous period is greater than the charging demand of the electric vehicle in this period, the excess energy will be stored in the next period. Otherwise, there will be no energy storage in this period, that is: ； Where, T is the time of each period, that is , For the i The photovoltaic power generation power in each time period, and ; For the i The power obtained by the charging pile from the grid during the time period, When , it means the charging pile obtains power from the grid; when When , it means the grid obtains power from the charging pile; For the j Electric cars in the i The charging and discharging power in each time period; In the i The electricity purchase cost paid by the charging pile agent to the grid during the time period is ,Right now: ； Where, For charging pile agents i The price of electricity purchased from the grid during a certain period of time, Charging pile agent in the first i The price of electricity sold to the grid during a certain time period; Therefore, the electricity purchase cost paid by the charging pile agent to the grid in one day is It can be expressed as: ； 2) Calculation of discharge cost when When the electric vehicle is in the discharge state, the discharge cost of each model of electric vehicle in one day is It can be expressed as: ； Where, For the j Electric cars in the i The charging and discharging power of each time period, For electric vehicles i Discharge compensation price for each time period; Therefore, the scheduling cost The minimization is expressed as: 。 5. The method according to claim 1, wherein The grid load variance minimization calculation in Step 2 includes: Average grid load Expressed as: ； Where, For the i The power that the charging pile obtains from the grid during the period, n is the number of hours in a day, i.e. ; Then, the power load variance is calculated; the smaller the variance, the smaller the grid load change, the more stable the load, and the higher the grid stability; grid load variance The minimization is expressed as: 。 6. The method according to any one of claims 1 to 5, characterized in that In Step 3, the charge and discharge power are constrained as follows: Whether the electric vehicle is in the charging or discharging state, its power should be less than or equal to the maximum power. When it is not connected to the grid, the power is 0, that is: ； Where, For the j Electric cars in the i The charging and discharging power of each time period, For the j The maximum charging and discharging power of an electric vehicle, For the j The time it takes for electric vehicles to be connected to the grid, For the j The user of an electric vehicle enters the time when the charging station leaves the grid; In Step 3, the state of charge is constrained as follows: To prevent rapid battery degradation, the residual power of an electric vehicle after discharge should not be less than 10% of the rated capacity, and the residual power of an electric vehicle after charging should not exceed 95% of the rated capacity; that is: ； Where, For the j Electric cars in the i The state of charge after a period of time, For the j The state of charge of an electric vehicle when it is connected to the grid. No. j The battery capacity of an electric vehicle, is the charge and discharge efficiency; For the j The cumulative number of charge and discharge alternations when an electric vehicle is connected to the power grid; Indicates the alternating state of charging and discharging of electric vehicles; T The time for each period; In Step 3, the charge capacity is constrained as follows: From the time an electric vehicle is connected to the grid until it leaves the grid, the amount of electricity supplied by the charging pile to the electric vehicle should meet the user's established needs. Therefore: ； Where, For the j The expected state of charge of the electric vehicle user when leaving the grid is input into the charging pile. For the j A nonlinear fitting function of the number of charge and discharge cycles of an electric vehicle and the remaining capacity of various batteries; In Step 3, the actual capacity of the Transformer is constrained as follows: The total power of the entire dispatching system in any time period should not exceed the distribution capacity of the area, so: ； Where, For charging pile i The energy stored in a period of time, For the i The basic load of the area where electric vehicles are connected to the grid during a time period, Power distribution for the region Transformer capacity, is the power factor.

Images