A Method for Configuring Charging Piles in Distribution Transformers Based on Double-Layer Optimization

Abstract

The present invention discloses a method for configuring charging piles in a transformer area based on double-layer optimization, which comprises the following steps: S1, establishing a corresponding orderly charging simulation model and an orderly charging strategy according to the scheduled orderly charging system; S2, establishing a double-layer optimization model for the orderly configuration of electric vehicle charging piles; S3, initializing the parameters of the double-layer optimization model; S4, in the double-layer optimization model, the upper-layer optimization model adopts an iterative algorithm, and the lower-layer optimization model adopts an improved BPSO algorithm and a Monte Carlo algorithm to solve the orderly charging load curve and the disorderly charging load curve respectively, solving the objective functions corresponding to the upper and lower layer optimization models, and obtaining the optimal ratio of orderly charging piles and the electric vehicle orderly charging plan. This method obtains the transformation plan of orderly charging piles in the transformer area through the double-layer optimization model, including the optimal configuration of the transformation capacity of orderly charging piles in the transformer area and the electric vehicle charging plan.

Description

Technical Field

[0001] The invention relates to the field of orderly charging strategies for electric vehicles, and in particular to a method for configuring charging piles in a charging area based on double-layer optimization. Background Art

[0002] In recent years, with the rapid development of electric vehicles and the support of national policies, the number of electric vehicles in my country has increased rapidly. At the same time, the problem of electric vehicle charging has become increasingly prominent and has become a livelihood issue of concern to a large number of electric vehicle owners. Charging piles are still the most important charging facilities for electric vehicles, so the construction of charging facilities has also developed rapidly in recent years, and the number of domestic charging piles has also increased rapidly. However, the disorderly charging of a large number of electric vehicles will have adverse effects on the power grid, such as increasing the burden on the power supply system of the power grid and affecting the normal power supply of the power grid; bringing harmonic pollution to the power grid system, thereby affecting the power quality of the power grid and users. One of the effective measures to solve this problem is to carry out intelligent transformation of charging piles, using orderly charging piles with "orderly charging" function to replace disordered charging piles without "orderly charging" function, reducing the burden of the power grid during peak power consumption and the impact of harmonic pollution.

[0003] As mentioned above, charging piles are the foundation for the development of the electric vehicle industry. In order to reduce the impact of disorderly charging on the power grid, some cities in my country stipulate that the direct construction ratio of charging piles in newly built residential areas is 20%-100%. Electric vehicles participate in orderly charging in a fully responsive manner. In some old residential areas, due to the development of charging pile technology, charging piles are generally not built in parking spaces or equipped with ordinary charging piles installed privately by car owners and publicly built by the area management party. This type of charging pile generally does not have monitoring, communication and dispatching functions. After the electric vehicle is connected to the charging pile, it is charged in a on-demand manner. With the advancement of the intelligent transformation of the area, Shanghai, Hangzhou and other cities have issued relevant policies to support the construction of orderly charging piles, encourage the intelligent transformation of charging piles in the area, and use orderly charging piles with "orderly charging" functions to replace disorderly charging piles without "orderly charging" functions. However, the construction of orderly charging piles requires higher construction costs than disorderly charging piles, and some electric vehicles may also have travel time restrictions and must adopt disorderly charging to replenish battery power. If orderly charging piles are built in all the substations, the investment cost will increase and cause waste of resources. Therefore, analyzing the configuration of orderly charging piles will provide certain guidance for the intelligent transformation and transition of charging piles in substations. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a method for configuring charging piles in a distribution area based on double-layer optimization. The method obtains an orderly charging pile transformation plan in the distribution area through a double-layer optimization model, which includes the optimal configuration of the orderly charging pile transformation capacity in the distribution area and an electric vehicle charging plan.

[0005] A method for configuring charging piles in a power distribution area based on double - layer optimization, comprising the following steps:

[0006] S1. Establish a double - layer optimization model for the orderly configuration of electric vehicle charging piles. The double - layer optimization model includes an upper - layer optimization model, a lower - layer optimization model, namely, an orderly charging optimization model and a strategy;

[0007] S2. Initialize the parameters of the double - layer optimization model. The parameters of the double - layer optimization model include the basic load p of the distribution transformer o,t and the rated capacity p lim , charging price, number of time periods M, number of electric vehicles N, rated capacity C of the electric vehicle battery, travel characteristics of electric vehicles, rated power P of orderly charging piles and disorderly charging piles r , charging efficiency η, operation years α, fixed cost C sta , conversion factor ε, discount rate r, annual income γ for reducing the unit peak - valley difference rate pvd , expansion cost γ per unit capacity of the distribution transformer del ;

[0008] S3. In the double - layer optimization model, the upper - layer optimization model uses an iterative algorithm, and the lower - layer optimization model uses an improved BPSO algorithm and a Monte Carlo algorithm to solve the orderly charging load curve and the disorderly charging load curve respectively, solve the objective functions corresponding to the upper - layer and lower - layer optimization models, and obtain the optimal ratio of orderly charging piles and the orderly charging plan for electric vehicles.

[0009] Preferably, the objective function of the upper - layer optimization model is:

[0010] Taking the return on investment R as the optimization goal, its objective function is:

[0011]

[0012] Among them, C inv is the annual investment cost of orderly charging piles, and I is to configure a certain capacity of orderly charging piles.

[0013] Preferably, the objective function of the lower - layer optimization model is:

[0014] The lower - layer optimization model takes the orderly charging objective function F as the optimization goal, and its objective function is:

[0015] minF = λ1F1′+λ2F′2+λ3F3′

[0016]

[0017] Among them, F1′, F′2, and F3′ are the normalized objective functions of the charging completion rate F1, charging cost F2, and load peak-valley difference F3 respectively, λ1, λ2, and λ3 are the weight coefficients of the charging completion rate F1, charging cost F2, and load peak-valley difference F3 respectively, and λ1 + λ2 + λ3 = 1, λ1 > 0, λ2 > 0, λ3 > 0; soc b,i , is the remaining power percentage at the start of charging for the EV i , and J b,i is the end time of the EV i journey, and J s,i is the start time of the EV i journey, and soc e,i is the expected remaining power percentage at the time of taking the vehicle for the EV i ; η is the charging efficiency, Δt is the time period length, and ρ t is the unit charging price at time period t, and ρ max , ρ min are the maximum and minimum unit charging prices respectively, max(p t ), min(p t ) are the maximum and minimum values of the total distribution transformer load respectively.

[0018] Preferably, the two-layer optimization model established in step S1 is set with constraint conditions, and the constraint conditions include charging pile capacity constraint, distribution transformer load constraint, charging time period constraint, and remaining power percentage constraint.

[0019] Preferably, step S3 includes the following sub-steps:

[0020] S3-1. Initialize the initial value of the ordered charging pile ratio (i.e., the ratio of the ordered charging pile capacity to the number of electric vehicles, and the initial value is set according to the actual situation) and the boundary of the ordered charging pile configuration ratio (i.e., the ratio of the maximum number of ordered charging piles that can be accommodated in this area to the number of electric vehicles);

[0021] S3-2. Calculate the ordered charging pile configuration capacity and the remaining capacity of the unordered charging piles from the ordered charging pile configuration ratio, which are OCP and DCP respectively, and pass them into the lower-layer model;

[0022] S3-3. The lower-layer optimization model calculates the unordered charging load of electric vehicles when the ordered charging piles are not transformed. According to the ordered charging pile configuration capacity, extract the electric vehicles that meet the ordered charging pile capacity, and calculate the ordered charging load curve of electric vehicles based on the improved BPSO algorithm; according to the remaining capacity of the unordered charging piles, extract the remaining electric vehicles that meet the remaining capacity of the unordered charging piles, and calculate the unordered charging load curve of electric vehicles based on the Monte Carlo algorithm, and pass it into the upper-layer optimization model;

[0023] S3-4. The upper-layer optimization model calculates the return on investment rate R of ordered charging;

[0024] S3-5. Determine whether the return on investment of orderly charging no longer rises. If it no longer rises, output the optimal ratio of orderly charging piles and the electric vehicle charging plan; otherwise, continue with step S3-6.

[0025] S3-6. Determine whether the ratio of orderly charging piles reaches the ratio boundary condition. If it does not reach the ratio boundary, adjust the ratio of orderly charging piles and go to step S3-2. If it reaches the ratio boundary, the algorithm terminates and outputs the optimal ratio of orderly charging piles and the electric vehicle orderly charging plan.

[0026] Preferably, step S3-3 includes the following sub-steps:

[0027] S3-3-1. Initialize the simulation parameters of the double-layer optimization model in S1. The simulation parameters include the basic load p of the distribution transformer o,t and the rated capacity p lim , the number of time periods M, the number of electric vehicles N, the rated capacity C of the electric vehicle battery, the expected remaining power percentage soc e , the threshold SOC of the remaining charge percentage for charging ls , the rated charging power P of the electric vehicle r and the charging efficiency η, the Monte Carlo convergence accuracy ε, the number of particles n in the BPSO particle swarm PSO , and the maximum number of iterations K; μ

[0028] S3-3-2. Randomly sample to obtain the end time t of the EV i trip, and the expression is as follows: b,i

[0029]

[0030] Randomly sample to obtain the start time t of the EV i trip, and the expression is as follows: s,i

[0031]

[0032] Randomly sample the residence duration T w,i of the trip, and the expression is as follows:

[0033]

[0034] Randomly sample the remaining charge soc at the start of charging b,i of the trip, and the expression is as follows:

[0035]

[0036] And calculate the end period J b,i of the trip, and the start period J s,i of the trip, and the expression is as follows:

[0037]

[0038]

[0039] Furthermore, calculate the charging selection flag bit s c,i , and the expression is as follows:

[0040]

[0041] where EV i represents the i-th electric vehicle, x represents a random variable, and the subscripts b, s, and w respectively represent the end of the trip, the start of the trip, and the stay duration. J b,i , J s,i are respectively the end period of the trip, the start period of the trip, t b,i is the end time of the trip, t s,i is the start time of the trip; s c,i = 1 indicates that EV i selects to charge the electric vehicle upon arrival, and s c,i = 0 indicates that EV i does not select to charge the electric vehicle upon arrival. SOC ls is the threshold percentage of the remaining charge for charging, and soc b,i is the percentage of the remaining charge at the start of charging for EV i sampled from the probability distribution model of the remaining charge at the start of charging;

[0042] S3-3-3. Determine whether the charging selection flag bit is 1. If it is 0, then return to S3-3-2 to simulate the next electric vehicle;

[0043] S3-3-4. Calculate the shortest charging duration T i of EV sm,i . During this duration, EV i charges continuously at the rated charging power P r and is superimposed on the unordered charging load curve of the previous i - 1 vehicles;

[0044] S3-3-5. Determine whether all electric vehicles have been calculated. If not, then return to step S3-3-2. Otherwise, go to S3-3-6;

[0045] S3-3-6. Calculate the mean value of the unordered charging load for each period under the current Monte Carlo simulation times, and calculate the coefficient of variation τ of the unordered charging load. The expression is as follows:

[0046]

[0047] where is the sample mean, and N xis the total number of samples.

[0048] S3-3-7. Determine whether the variance coefficient τ of the unordered charging load reaches the Monte Carlo convergence accuracy ε. If it is greater, increase the Monte Carlo simulation times by 1, return to S3-3-1 to continue the simulation; otherwise, output the total unordered charging load of electric vehicles.

[0049] S3-3-8. Randomly sample to obtain the end time t of the trips of [number of electric vehicles] electric vehicles, the start time t, the remaining charge soc at the start of charging, and the expected remaining charge soc of the electric vehicles, and calculate the charging selection flag bit to determine the charging demand of the electric vehicles. b,i , the start time t s,i , the remaining charge soc at the start of charging b,i , the expected remaining charge soc of the electric vehicle e,i , calculate the charging selection flag bit to determine the charging demand of the electric vehicle.

[0050] S3-3-9. Determine whether there are electric vehicles connected in the current time period. If there are electric vehicles connected, obtain the data of all electric vehicles that need to charge at the charging piles connected in the current time period, including the charging selection flag bit s, the end time t of the trip, the departure time t, the remaining charge soc at the start of charging, and the expected remaining charge soc, and calculate the end time period J of the trip and the start time period J of the trip. If there are no electric vehicles connected, go to S3-3-14. c,i , the end time t of the trip b,i , the departure time t s,i , the remaining charge soc at the start of charging b,i , the expected remaining charge soc e,i , calculate the end time period J of the trip b,i , the start time period J of the trip s,i , if there are no electric vehicles connected, go to S3-3-14.

[0051] S3-3-10. Adopt the method of patching the solution

[0052] The charging time constraint is as follows:

[0053]

[0054] Initialize the particle swarm Ensure that the end time period and start time period of each initialized particle meet the time period constraints of the electric vehicle, and at the same time, constrain the number of effective charging time periods of each particle to be [expression for rounding up the minimum charging duration divided by the interval of each charging time period], as follows: It means rounding up the minimum charging duration divided by the interval of each charging time period, and the expression is as follows:

[0055]

[0056] S3-3-11. Start iteration using the improved BPSO optimization algorithm, and output the optimal fitness value and the global extreme value gbest k ;

[0057] S3-3-12. Determine whether the current iteration number reaches the maximum iteration number K. If it does not reach, return to S3-3-11.

[0058] S3-3-13. Output the optimal fitness value of the particles and the global extreme value gbest k , and obtain the orderly charging plan for electric vehicles according to the global extreme value to schedule the orderly charging of electric vehicles;

[0059] S3-3-14. Determine whether the current time period has reached 96 time periods. If not, enter S3-3-19. If it has reached, respectively count the total distribution transformer load p t , the peak-valley difference ΔP L , the peak-valley difference rate standard deviation σ L , load rate ζ t , load factor K L , the electric vehicle charging completion rate F1, the charging cost and other result data F2, and draw the total load curve of the orderly charging of electric vehicles. The expression is as follows:

[0060]

[0061] where p ave is the average load of the distribution transformer, and the calculation formula is as follows:

[0062]

[0063]

[0064] where p lim is the rated capacity of the distribution transformer, and cosθ is the power factor of the distribution transformer.

[0065]

[0066]

[0067]

[0068] where in the above formula N = OCP.

[0069] Preferably, the improved BPSO optimization algorithm is as follows:

[0070] Calculate the fitness value of each particle, select the particle with the smallest fitness to update the individual extreme value and the global extreme value of the population, and update the particle velocity and the inertia weight coefficient ω. The expression is as follows:

[0071]

[0072]

[0073] Update the particle position At the same time, according to the particle velocity The magnitude of the absolute value preferentially changes the particle positions in the dimension with a larger absolute value of the speed, and ensures that the number of effective charging periods for each particle remains Calculate the fitness value of the updated particle, and update the individual extreme values of the population again and the global extreme value gbest k ,

[0074]

[0075]

[0076] where k is the current iteration number, is the speed of particle i at the (k + 1)-th iteration, ω is the inertia weight coefficient, is the speed of particle i at the k-th iteration, c1 and c2 are acceleration factors, usually taking the value of 2, r1 and r2 are random numbers uniformly distributed between [0, 1], is the individual extreme value of particle i at the k-th iteration, gbest k The global extreme value of the particle population at the k-th iteration, is the position of particle i at the k-th iteration, is the position of particle i at the (k + 1)-th iteration; ω max , ω min are the maximum and minimum values of the inertia weight coefficient respectively, K is the maximum number of iterations; v i,j is the speed value of particle i in dimension j, x i,j is the position of particle i in dimension j, rand is a random number uniformly distributed between [0, 1].

[0077] Advantages of the present invention:

[0078] The use of the double-layer optimization model is suitable for studying the planning and optimization problems among multiple decision-makers, thereby realizing the control of the orderly charging strategy of electric vehicles, achieving the goal of maximizing the return on investment of orderly charging facilities and realizing the optimal orderly charging pile configuration ratio based on "peak shaving and valley filling" of the distribution network. The charging pile configuration method for this substation area can better meet the needs of the actual orderly charging pile configuration problem, which can not only improve the income of orderly charging piles, play a role of "peak shaving and valley filling" for the total grid load, and reduce the burden on the power grid; but also reduce the investment and transformation cost of the substation area for charging piles, and prevent waste of resources. Description of the drawings

[0079] Figure 1 Double-layer optimization process for electric vehicle charging pile configuration.

[0080] Figure 2 Return on investment of orderly charging under double-layer optimization.

[0081] Figure 3 Disorderly and orderly charging loads of electric vehicles. Detailed implementation manners

[0082] In order to make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0083] The present invention relates to an optimization method for the configuration problem of orderly charging piles, and uses a two-layer optimization model to generate an orderly charging pile configuration ratio with the highest return on investment. The upper and lower layer optimization objectives of the two-layer optimization model are the return rate and the orderly charging strategy objective respectively, as follows Figure 1 shown. The objective of the present invention is to propose a method for configuring orderly charging in a substation area based on two-layer optimization, so as to realize the control of the orderly charging strategy of electric vehicles, and achieve the objectives of the maximum return on investment of orderly charging facilities and the optimal orderly charging pile configuration ratio based on "peak shaving and valley filling" of the distribution network. To illustrate the effect of the present invention, the following takes the orderly charging system of electric vehicles in a certain substation area as the implementation object of the present invention to describe the present invention in detail:

[0084] S1. Establish a two-layer optimization model for the configuration of orderly charging piles for electric vehicles, and the two-layer optimization model includes an upper-layer optimization model and a lower-layer optimization model.

[0085] The upper and lower layer optimization models and the constraint conditions are as follows Figure 1 shown.

[0086] Ⅰ) The upper-layer optimization model takes the return on investment R as the optimization objective, and its objective function is:

[0087]

[0088] Where C inv is the annual investment cost of the orderly charging pile, as shown in Equation (2), and I is the annual investment income obtained after the orderly charging strategy is optimized and solved by introducing the orderly charging pile with a certain capacity into the lower-layer model, as shown in Equation (5).

[0089] The annual investment cost C inv of the orderly charging pile on the grid side generally includes the construction cost and the operation and maintenance cost.

[0090] C inv = C con + C ope (2)

[0091] Where C con is the annual construction cost, Cope is the annual operation and maintenance cost.

[0092] Specifically, the annual construction cost is calculated as follows:

[0093] The construction cost of the charging pile includes fixed costs and variable costs. The fixed costs include construction engineering costs, installation engineering costs, site rental fees, and other fixed expenses. The variable costs include the equipment purchase cost of the charging pile. Let n be the configured capacity of the orderly charging pile, then the annual construction cost of the orderly charging pile is shown in Equation (3).

[0094]

[0095] In the formula, C sta , C var , β, and α are the fixed cost, variable cost, unit price, and depreciation life of the orderly charging pile respectively, and r is the discount rate.

[0096] The annual operation and maintenance cost is calculated as follows:

[0097] The operation and maintenance cost mainly includes the human operation cost and maintenance cost invested after the project is completed. The operation mode of the orderly charging pile adopts the "unified construction and unified operation" mode. Therefore, a certain amount of operation and maintenance cost needs to be invested every year during the operation to maintain the normal operation of the entire orderly charging system. Since the operation and maintenance cost has great uncertainty, it is usually determined by converting the fixed cost according to a certain conversion coefficient. Therefore, the annual operation and maintenance cost C ope of the electric vehicle charging pile is shown in Equation (1.4).

[0098] C ope = εC sta (4)

[0099] In the formula, ε is the conversion coefficient.

[0100] The annual investment income I includes the income brought by reducing the peak-valley difference rate of the total load and the income brought by delaying the expansion of the distribution transformer. The calculation formula is shown in Equation (5).

[0101] I = I pvd + I del (5)

[0102] In the formula, I pvd is the income from reducing the peak-valley difference rate, and I del is the income from delaying the expansion of the distribution transformer.

[0103] It can be understood that it also involves the income from reducing the peak-valley difference rate:

[0104] Peak-valley difference rate It can reflect the degree of "peak shaving and valley filling" of the distribution transformer load by configuring orderly charging piles. The smaller the peak-valley difference rate, the greater the positive effect of orderly charging on the distribution transformer network, which can reduce the distribution transformer network loss, peak regulation and frequency modulation, and the cost of spinning reserve. Define the annual peak-valley difference rate reduction benefit I pvd is the benefit brought by configuring orderly charging piles to reduce the peak-valley difference rate of the load, as shown in Equation (6).

[0105]

[0106] In the formula, γ pvd is the annual benefit coefficient for reducing the unit peak-valley difference rate, is the peak-valley difference rate of the total distribution transformer load when all electric vehicles charge disorderly, is the peak-valley difference rate of the total distribution transformer load when configuring a certain capacity of orderly charging piles, both of which can be calculated by Equation (7).

[0107]

[0108] In the formula, ΔP L is the peak-valley difference, which can be calculated by Equation (8), and p t is the total load in period t.

[0109] ΔP L =max(p t )-min(p t ), t = 1, 2,..., M (8)

[0110] In the formula, p t is the total load in period t.

[0111] It can be understood that it also involves the benefit of delaying the expansion of the distribution transformer:

[0112] If the disorderly charging load is allowed to develop, the power grid side will inevitably expand the distribution transformer for the safe operation of the distribution transformer. After configuring a certain capacity of orderly charging piles for electric vehicles to participate in orderly charging scheduling, it can reduce the demand for the distribution transformer capacity at the peak moment of load power consumption, reduce the harm caused by large-scale disorderly charging load to the power grid, delay the expansion pressure of the distribution transformer, and extend the service life of the distribution transformer. Therefore, this chapter takes into account the benefit of configuring orderly charging piles for delaying the expansion and upgrade of the distribution transformer, as shown in Equation (9).

[0113]

[0114] In the formula, γ del is the expansion cost per unit capacity of the distribution transformer, is the maximum value of the disorderly charging load when all electric vehicles charge disorderly, and p dis,t is the disorderly charging load of electric vehicles in period t after configuring a certain capacity of orderly charging piles.

[0115] Combining formulas (1)-(9), the maximum return on investment of the upper-layer model can be obtained as follows:

[0116]

[0117] Ⅱ) The lower layer takes the ordered charging objective function F as the optimization objective, and its objective function is:

[0118] minF = λ1F1′ + λ2F′2 + λ3F3′ (11)

[0119]

[0120] Among them, F1′, F′2, and F3′ are the normalized objective functions of the charging completion rate F1, charging cost F2, and load peak-valley difference F3 respectively, and λ1, λ2, and λ3 are the weight coefficients of the charging completion rate F1, charging cost F2, and load peak-valley difference F3 respectively, and λ1 + λ2 + λ3 = 1, λ1 > 0, λ2 > 0, λ3 > 0; soc b,i , is the percentage of the remaining battery charge at the start of charging for the EV i , J b,i is the end time of the EV i journey, J s,i is the start time of the EV i journey, soc e,i is the expected remaining battery charge percentage when picking up the vehicle for the EV i ; η is the charging efficiency, Δt is the time period length, ρ t is the unit charging price at time period t, ρ max , ρ min are the maximum and minimum unit charging prices respectively. max(p t ), min(p t ) are the maximum and minimum total loads of the distribution transformer respectively.

[0121] Specifically, the constraint conditions are as follows:

[0122] (1) Charging pile capacity constraint

[0123] Since the number of electric vehicles in the distribution area is N, when the capacity of the ordered charging piles configured is n, the remaining capacity of the disordered charging piles is N - n, and the configured capacity of the ordered charging piles cannot exceed the number of electric vehicles.

[0124] 0 ≤ n ≤ N (13)

[0125] (2) Distribution transformer load constraint

[0126] The total load of the distribution transformer at time period t consists of the basic load, the disordered charging load of electric vehicles, and the ordered charging load of electric vehicles.

[0127] pt = p o,t + p dis,t + p ord,t (14)

[0128] Wherein, p o,t is the basic load of the distribution transformer in period t, p dis,t is the unordered charging load of electric vehicles in period t, p ord,t is the ordered charging load of electric vehicles in period t.

[0129] The total load of the distribution transformer at different times should be within its rated capacity.

[0130] p t ≤ p lim (15)

[0131] (3) Charging period constraint

[0132] The charging end time of electric vehicles cannot exceed the going-out time.

[0133]

[0134] (4) Remaining battery percentage constraint

[0135] During a complete charging cycle of an electric vehicle, the remaining battery percentage when picking up the vehicle cannot exceed 1.

[0136] soc b,i ≤ soc a,i ≤ 1 (17)

[0137] S2. Initialize the simulation parameters of the distribution transformer area, electric vehicles, and charging piles, including the basic load p 0，t of the distribution transformer and its rated capacity p lim , charging price, number of time periods M, number of electric vehicles N, rated capacity C of the electric vehicle battery, travel characteristics of electric vehicles, rated power P of ordered and unordered charging piles, charging efficiency η, operation years α, fixed cost C sta , conversion factor ε, discount rate r, annual income γ for reducing the unit peak-valley difference rate pvd , expansion cost γ per unit capacity of the distribution transformer del and other parameters.

[0138] S3. In the two-layer optimization model, the upper layer uses an iterative algorithm, and the lower layer uses an improved BPSO algorithm and a Monte Carlo algorithm to solve the ordered charging load curve and the unordered charging load curve respectively, solve the corresponding objective functions of the upper and lower layers, and obtain the optimal ratio of ordered charging piles and the ordered charging plan for electric vehicles. Specifically, it includes the following sub-steps:

[0139] S3-1. Initialize the initial value of the ratio of ordered charging piles and the boundary of the configuration ratio of ordered charging piles.

[0140] S3-2. Calculate the configured capacity of the orderly charging piles and the remaining capacity of the disorderly charging piles as OCP and DCP respectively from the orderly charging pile configuration ratio, and input them into the lower-layer model.

[0141] S3-3. The lower-layer model calculates the disorderly charging load of electric vehicles when the orderly charging piles are not modified. According to the configured capacity of the orderly charging piles, extract the electric vehicles that meet the capacity of the orderly charging piles, and calculate the orderly charging load curve of electric vehicles based on the improved BPSO algorithm. According to the remaining capacity of the disorderly charging piles, extract the remaining electric vehicles that meet the remaining capacity of the disorderly charging piles, and calculate the disorderly charging load curve of electric vehicles based on the Monte Carlo algorithm, and input them into the upper-layer model. Specifically, it is the following sub-steps:

[0142] S3-3-1. Initialize the simulation parameters of the distribution transformer area, electric vehicles, charging piles, etc., including the basic load p o,t and the rated capacity p lim , the number of time periods M, the number of electric vehicles N, the rated capacity C of the electric vehicle battery, the expected remaining power percentage soc e , the remaining power percentage threshold SOC of the charging ls , the rated charging power P of the electric vehicle r and the charging efficiency η, the Monte Carlo convergence accuracy ε, the number of BPSO particle populations n PSO , parameters such as the maximum number of iterations K, etc.

[0143] S3-3-2. Randomly sample to obtain the end time t i of the EV trip, the start time t b,i , the stay duration T s,i , the remaining power soc at the start of charging according to equations (18), (19), (20), (21) respectively, and calculate the end period J w,i of the trip, the start period J b,i , and calculate the charging selection flag bit s according to equation (24) b,i , the start period J s,i , and calculate the charging selection flag bit s according to equation (24) c,i

[0144]

[0145]

[0146]

[0147]

[0148]

[0149]

[0150]

[0151] Among them, in Formulas (22) and (23), J b,i , J s,i are respectively the end period of the trip, the start period of the trip, t b,i is the end time of the trip, t s,i is the start time of the trip; in Formula (24), s c,i =1 indicates that the EV i chooses to charge the electric vehicle upon arrival, s c,i =0 indicates that the EV i does not choose to charge the electric vehicle upon arrival, SOC ls is the remaining charge percentage threshold for charging, soc b,i is the remaining charge percentage at the start of charging of the EV i sampled according to the probability distribution model of the remaining charge percentage at the start of charging in Formula (21).

[0152] S3-3-3. Determine whether the charging selection flag bit is 1. If it is 0, then return to S4-3-2 to simulate the next electric vehicle.

[0153] S3-3-4. Calculate the shortest charging duration T i of the EV sm,i . During this duration, the EV i charges continuously at the rated charging power P r , and it is superimposed on the unordered charging load curve of the previous i-1 vehicles.

[0154] S3-3-5. Determine whether all electric vehicles have been calculated. If not, then return to step S3-3-2, otherwise go to S3-3-6.

[0155] S3-3-6. Calculate the mean value of the unordered charging load in each period under the current Monte Carlo simulation times, and calculate the variance coefficient τ of the unordered charging load according to Formula (25).

[0156]

[0157] S3-3-7. Determine whether the variance coefficient τ of the unordered charging load reaches the Monte Carlo convergence accuracy ε. If it is greater, then the Monte Carlo simulation times +1, return to S3-3-1 to continue the simulation, otherwise output the total unordered charging load of the electric vehicles.

[0158] S3-3-8. Randomly sample to obtain the end time t b,i of the trip, the start time t s,i of the trip, and the remaining charge soc at the start of charging of the electric vehicle respectively according to Formulas (18), (19), and (21)b,i The expected remaining battery level SOC of the electric vehicle e,i and calculate the charging selection flag according to Equation (24) to determine the charging demand of the electric vehicle

[0159] S3-3-9. Determine whether there is an electric vehicle connected in the current period. If there is an electric vehicle connected, obtain the data of all electric vehicles that need to be charged at the charging piles connected in the current period, including the charging selection flag as s c,i the end time t of the trip b,i the departure time t s,i the initial remaining battery level SOC for charging b,i the expected remaining battery level SOC e,i and calculate the end-of-trip period J according to Equations (22) and (23) b,i the start period J s,i If there is no electric vehicle connected, go to S3-3-14

[0160] S3-3-10. Adopt the method of patching the solution to initialize the particle swarm according to the charging time constraint of Equation (26) in the way of Equation (27) Ensure that the end-of-trip period and the start period of each initialized particle meet the time period constraints of the electric vehicle, and at the same time, the number of effective charging time periods of each particle is which means rounding up the minimum charging duration divided by the interval of each charging time period

[0161]

[0162]

[0163] S3-3-11. Start iteration using the improved BPSO optimization algorithm. Calculate the fitness value of each particle according to Equation (11), and select the particle with the minimum fitness to update the individual extreme value and the global extreme value of the population. Update the particle velocity and the inertia weight coefficient ω according to Equations (28) and (29). Update the particle position according to Equations (30) and (31) At the same time, according to the particle velocity preferably change the particle position in the dimension with the larger absolute value of the velocity according to the magnitude of the absolute value of the velocity, and ensure that the number of effective charging time periods of each particle is still Calculate the fitness value of the updated particle and update the individual extreme value of the population again and the global extreme value gbest k .

[0164]

[0165]

[0166]

[0167]

[0168] Among them, in formula (28), k is the current iteration number, is the velocity of particle i at the (k + 1)-th iteration, ω is the inertia weight coefficient, is the velocity of particle i at the k-th iteration, c1 and c2 are acceleration factors respectively, usually taking the value of 2, r1 and r2 are random numbers uniformly distributed between [0, 1], is the personal best value of particle i at the k-th iteration, gbest k is the global best value of the particle population at the k-th iteration, is the position of particle i at the k-th iteration, is the position of particle i at the (k + 1)-th iteration; in formula (29), ω max and ω min are the maximum and minimum values of the inertia weight coefficient respectively,

[0169] In this embodiment, ω max is set to 0.9, ω min is set to 0.4, K is the maximum iteration number; in formulas (30) and (31), v i,j is the velocity value of particle i in dimension j, x i,j is the position of particle i in dimension j, rand is a random number uniformly distributed between [0, 1].

[0170] S3-3-12. Determine whether the current iteration number has reached the maximum iteration number K. If not, return to S3-3-11.

[0171] S3-3-13. Output the optimal fitness value of the particle and the global best value gbest k , and obtain the orderly charging plan for electric vehicles according to the global best value to schedule the orderly charging of electric vehicles.

[0172] S3-3-14. Determine whether the current time period has reached 96 time periods. If not, enter S3-3-9. If it has reached, then respectively count the total distribution transformer load p t , peak-valley difference ΔP L , peak-valley difference rate , standard deviation σ L , load rate ζ t , load factor K L , electric vehicle charging completion rate F1, charging cost and other result data F2 according to formulas (7), (8), (32), (34), (35), (36), (37), and draw the total load curve of electric vehicle orderly charging.

[0173]

[0174] Among them, p ave is the average load of the distribution transformer, and the calculation formula is shown in Equation (33).

[0175]

[0176]

[0177] Among them, p lim is the rated capacity of the distribution transformer, and cosθ is the power factor of the distribution transformer.

[0178]

[0179]

[0180]

[0181] Among them, in Equations (36) and (37), N = OCP.

[0182] S3-5. The upper-layer model calculates the orderly charging investment return rate R according to Equation (1-10).

[0183] S3-6. Judge whether the orderly charging investment return rate no longer rises. If it no longer rises, output the optimal ratio of orderly charging piles and the electric vehicle charging plan. Otherwise, go to step S3-7.

[0184] S3-7. Judge whether the ratio of orderly charging piles reaches the ratio boundary condition. If it does not reach the ratio boundary, adjust the ratio of orderly charging piles and go to step S3-2. If it reaches the ratio boundary, the algorithm terminates and outputs the optimal ratio of orderly charging piles and the electric vehicle orderly charging plan.

[0185] During the optimization process, the orderly charging investment return rate under double-layer optimization is as Figure 2 shown; when the ratio of orderly charging piles is small, the investment return rate increases rapidly. When it reaches 0.71, the investment return rate shows an inflection point and reaches a maximum value of 97.09%. Subsequently, it slowly decreases. At this time, continuing to increase the ratio of orderly charging piles cannot improve the investment return rate. The corresponding configuration ratio of orderly charging piles under a high investment return rate is about 0.65 - 0.72. Therefore, for this example, when the ratio of orderly charging piles to disorderly charging piles is 0.71:0.29, the orderly charging investment return rate can be maximized. At the same time, under the optimal ratio of orderly charging piles, the total load of the distribution transformer is as Figure 3As shown, the configured orderly charging piles after optimized calculation provide orderly charging services for electric vehicles, reasonably arrange the charging time of electric vehicles, greatly increase the load during the valley period, and significantly improve the peak-valley difference of the load. Compared with disorderly charging, the peak-valley difference is reduced from 1584.30 kW to 719.55 kW, and the peak-valley difference rate is reduced from 78.55% to 44.30%. In addition, after the remaining untransformed disorderly charging piles are used for disorderly charging of electric vehicles, the peak value of the total load of the distribution transformer is almost the same as the rated capacity limit of the distribution transformer. However, the disorderly charging behavior still has a negative impact on the distribution transformer, forming a state of peak-on-peak with the basic load, posing a threat to the safe operation of the distribution transformer. With the gradual progress of the intelligent transformation project of orderly charging piles, it is necessary to gradually transform disorderly charging piles with orderly charging piles to achieve a better peak shaving and valley filling effect. Therefore, according to the above pictures and text descriptions, it can be seen that the present invention not only achieves the purpose of "peak shaving and valley filling" and reducing the load of the distribution network, but also proposes a method for maximizing the investment return by reasonably configuring the capacity of orderly charging piles at present. At the same time, we also investigated the influence of different electric vehicle penetration rates on the configuration results. Based on the above double-layer model, the optimal ratio of orderly charging piles, investment return, investment return rate, and relevant grid-side indicators under different penetration rates are shown in Table 1. As shown in the above charts, it can be seen that the transformation and construction of orderly charging piles will greatly improve the charging completion rate of electric vehicles, meet the charging needs of electric vehicles, and at the same time smooth the load fluctuation of the distribution transformer.

[0186] Table 1 Grid-side and user-side related indicators at the optimal ratio of orderly charging piles under different penetration rates

[0187]

Claims

1. A method for configuring charging piles in a transformer substation area based on double-layer optimization, characterized in that, It includes the following steps: S1. Establish a two-layer optimization model for the orderly charging pile configuration of electric vehicles. The two-layer optimization model includes an upper-layer optimization model, a lower-layer optimization model, namely an orderly charging optimization model and a strategy; The objective function of the upper-layer optimization model is: Taking the R return on investment as the optimization objective, its objective function is: Among them, C inv is the annual investment cost of the orderly charging piles, and I is to configure orderly charging piles with a certain capacity; The objective function of the lower-layer optimization model is: The lower-layer optimization model takes the orderly charging objective function F as the optimization objective, and its objective function is: minF = λ1F1′ + λ2F2′ + λ3F3′ Among them, F1′, F2′, and F3′ are the normalized objective functions of the charging completion rate F1, charging cost F2, and load peak-valley difference F3, respectively. λ1, λ2, and λ3 are the weight coefficients of the charging completion rate F1, charging cost F2, and load peak-valley difference F3, respectively, and λ1 + λ2 + λ3 = 1, λ1 > 0, λ2 > 0, λ3 > 0; soc b,i is the i percentage of the remaining battery level at the start of charging for the EV, J b,i is the i end time of the trip for the EV, J s,i is the i start time of the trip, soc e,i is the i desired remaining battery level percentage when picking up the vehicle; η is the charging efficiency, Δt is the time period length, ρ t is the unit charging price at time period t, ρ max and ρ min are the maximum and minimum unit charging prices, respectively. max(p t ) and min(p t ) are the maximum and minimum total transformer loads; The two-layer optimization model established in step S1 is set with constraint conditions, and the constraint conditions include charging pile capacity constraint, distribution transformer load constraint, charging time period constraint, and remaining power percentage constraint; S2. Initialize the two-layer optimization model established in S1. The parameters of the two-layer optimization model include the basic load p of the distribution transformer o,t and the rated capacity p lim , charging price, number of time periods M, number of electric vehicles N, rated capacity C of the electric vehicle battery, travel characteristic quantity of the electric vehicle, rated power P of the ordered charging pile and the unordered charging pile r , charging efficiency η, operation years α, fixed cost C sta , conversion coefficient ε, discount rate r, annual income γ for reducing the unit peak-valley difference rate pvd , expansion cost γ per unit capacity of the distribution transformer del ; S3. In the two-layer optimization model, the upper-layer optimization model uses an iterative algorithm, and the lower-layer optimization model uses an improved BPSO algorithm and a Monte Carlo algorithm to solve the orderly charging load curve and the disorderly charging load curve respectively, solve the objective functions corresponding to the upper and lower layer optimization models, and obtain the optimal ratio of orderly charging piles and the orderly charging plan for electric vehicles. Step S3 includes the following sub-steps: S3-1. Initialize the initial value of the ratio of orderly charging piles and the boundary of the ratio of orderly charging pile configuration. The initial value of the ratio of orderly charging piles is the ratio of the capacity of orderly charging piles to the number of electric vehicles, and the boundary of the ratio of orderly charging pile configuration is the ratio of the maximum number of orderly charging piles that can be accommodated in this area to the number of electric vehicles; S3-2. Calculate the configured capacity of orderly charging piles and the remaining capacity of disorderly charging piles from the ratio of orderly charging pile configuration, which are OCP and DCP respectively, and pass them into the lower-layer model; S3-3. The lower-layer optimization model calculates the disorderly charging load of electric vehicles when the orderly charging piles are not transformed. According to the configured capacity of the orderly charging piles, electric vehicles that meet the capacity of the orderly charging piles are selected, and the orderly charging load curve of electric vehicles is calculated based on the improved BPSO algorithm; according to the remaining capacity of the disorderly charging piles, the remaining electric vehicles that meet the remaining capacity of the disorderly charging piles are selected, and the disorderly charging load curve of electric vehicles is calculated based on the Monte Carlo algorithm, and then passed into the upper-layer optimization model; S3-4. The upper-layer optimization model calculates the return on investment R of orderly charging; S3-5. Judge whether the return on investment of orderly charging no longer rises. If it no longer rises, output the optimal ratio of orderly charging piles and the electric vehicle charging plan. Otherwise, continue with step S3-6; S3-6. Judge whether the ratio of orderly charging piles reaches the ratio boundary condition. If it does not reach the ratio boundary, adjust the ratio of orderly charging piles and go to step S3-2. If it reaches the ratio boundary, the algorithm terminates and outputs the optimal ratio of orderly charging piles and the orderly charging plan for electric vehicles.

2. The method for configuring charging piles in a transformer substation area based on double-layer optimization according to claim 1, wherein, Step S3-3 includes the following sub-steps: S3-3-1. Initialize the simulation parameters of the orderly charging simulation model in S1. The simulation parameters include the basic load p of the distribution transformer o,t and the rated capacity p lim , the number of time periods M, the number of electric vehicles N, the rated capacity C of the electric vehicle battery, the expected remaining power percentage soc e , the threshold SOC of the remaining charge percentage for charging ls , the rated charging power P of the electric vehicle r and the charging efficiency η, the Monte Carlo convergence accuracy ε, the number n of the BPSO particle population PSO , the maximum number of iterations K; μ S3-3-2. Random sampling to obtain EV i The end time t of the journey b,i , and the expression is as follows: Random sampling to obtain EV i Starting time t s,i , and the expression is as follows: Random sampling residence time T w,i , and the expression is as follows: Randomly sample the remaining battery charge (SOC) at the start of charging b,i , and the expression is as follows: And calculate the end period J of the travel b,i , the start period J s,i , The expressions are as follows: Furthermore, calculate the charging selection flag bit s c,i , and the expression is as follows: Among them, EV i represents the i-th electric vehicle, x represents a random variable, and the subscripts b, s, w respectively represent the end of the trip, the start of the trip, and the dwell time. J b,i and J s,i are respectively the end time period of the trip and the start time period of the trip. t b,i is the end time of the trip, and t s,i is the start time of the trip; s c,i = 1 indicates that EV i chooses to charge the electric vehicle upon arrival, and s c,i = 0 indicates that EV i does not choose to charge the electric vehicle upon arrival. SOC ls is the percentage threshold of the remaining charge for charging, and soc b,i is the percentage of the remaining charge at the start of charging for EV obtained by sampling from the probability distribution model of the remaining charge at the start of charging. i ​ S3-3-3. Judge whether the charging selection flag bit is 1. If it is 0, return to S3-3-2 to simulate the next electric vehicle; S3-3-4. Calculate the EV i The shortest charging duration T sm,i During this duration, the EV i Charges continuously at the rated charging power P r And superimposes it on the unordered charging load curves of the previous i - 1 vehicles; S3-3-5. Judge whether all electric vehicles have been calculated. If not, return to step S3-3-2. Otherwise, go to S3-3-6; S3-3-6. Calculate the mean value of the unordered charging load for each time period under the current Monte Carlo simulation times, and calculate the variance coefficient τ of the unordered charging load. The expression is as follows: where is the sample mean, and N x is the total number of samples; S3-3-7. Determine whether the variance coefficient τ of the unordered charging load reaches the Monte Carlo convergence accuracy ε. If it is greater, then the Monte Carlo simulation times +1, return to S3-3-1 to continue the simulation. Otherwise, output the total unordered charging load of electric vehicles; S3-3-8. Randomly sample to obtain the end time t of the driving of an electric vehicle b,i , the start time t s,i , the remaining power soc at the start of charging b,i , the expected remaining power soc of the electric vehicle e,i , calculate the charging selection flag bit to determine the charging demand of the electric vehicle; S3-3-9. Determine whether there is an electric vehicle connected in the current period. If there is an electric vehicle connected, obtain the data of all electric vehicles that need to be charged at the charging piles in the current period, including the charging selection flag bit s c,i , the end time t of the trip b,i , the departure time t s,i , the remaining battery power soc at the start of charging b,i , the expected remaining battery power soc e,i , calculate the end period J of the trip b,i , the start period J s,i , if there is no electric vehicle connected, go to S3-3-14; S3-3-10. Adopt the method of patching the solution, The charging time constraint, the expression is as follows: Initialize the particle swarm Ensure that the end time period of each initialized particle's journey conforms to the time period constraints of the electric vehicle, and at the same time, constrain the number of effective charging time periods of each particle to be It means rounding up the minimum charging duration divided by the interval of each charging time period. The expression is as follows: S3-3-11. Start iteration using the improved BPSO optimization algorithm and output the optimal fitness value and the global extreme value gbest k ; S3-3-12. Determine whether the current iteration times reach the maximum iteration times K. If not, return to S3-3-11; S3-3-13. Output the optimal fitness value of the particles and the global extreme value gbest k , and obtain the orderly charging plan for electric vehicles according to the global extreme value to schedule the orderly charging of electric vehicles; S3-3-14. Determine whether the current period has reached 96 time periods. If not, go to S3-3-19. If so, respectively count the total distribution transformer load p t , the peak-valley difference ΔP L , the peak-valley difference rate standard deviation σ L , the load factor ζ t , the load rate K L , the electric vehicle charging completion rate F1, the charging cost and other result data F2, and draw the total load curve of the orderly charging of electric vehicles. The expression is as follows: Among them, p ave is the average load of the distribution transformer, and the calculation formula is as follows: where p lim is the rated capacity of the distribution transformer, and cosθ is the power factor of the distribution transformer; Among them, in equations (36) and (37), N = OCP.

3. The method for configuring charging piles in a transformer substation area based on double-layer optimization according to claim 2, wherein, The improved BPSO optimization algorithm is as follows: Calculate the fitness value of each particle, select the particle with the minimum fitness to update the individual extreme value and global extreme value of the population, and update the particle velocity V i k+1 and the inertia weight coefficient ω, the expression is as follows: Update the particle position Meanwhile, according to the particle velocity V i k+1 Give priority to changing the particle position in the dimension with a larger absolute value of velocity according to the magnitude of the absolute value of the velocity, and ensure that the number of effective charging periods of each particle remains Calculate the fitness value of the updated particle and update the individual extreme value of the population again And the global extreme value gbest k , where k is the current iteration number, V i k+1 is the velocity of particle i at the (k + 1)-th iteration, ω is the inertia weight coefficient, V i k is the velocity of particle i at the k-th iteration, c1 and c2 are acceleration factors, usually taking the value of 2, r1 and r2 are random numbers uniformly distributed between [0, 1], is the personal best of particle i at the k-th iteration, gbest k is the global best of the particle swarm at the k-th iteration, is the position of particle i at the k-th iteration, is the position of particle i at the (k + 1)-th iteration; ω max 、ω min are the maximum and minimum values of the inertia weight coefficient respectively, K is the maximum number of iterations; v i,j is the velocity value of particle i in dimension j, x i,j is the position of particle i in dimension j, rand is a random number uniformly distributed between [0, 1].

Images