Charging station reconstruction and extension method driven by user travel behavior and intelligent charging strategy

Abstract

According to the charging station reconstruction and extension method driven by the user travel behavior and the intelligent charging strategy, the travel behavior of the user is deeply analyzed, the travel behavior and the charging behavior of the user in the workday are predicted, the schedulable time window is dynamically generated for each user based on the prediction result, and charging is performed in the schedulable time window. The minimum user charging cost, the minimum power grid load mean square error and the minimum carbon emission serve as optimization objectives, and an optimal solution set of a user charging strategy is solved through multi-objective optimization; and taking the minimum total building cost, the minimum user total distance cost and the minimum penalty cost of uncovered users as optimization objectives, bringing the charging strategies after user scheduling into the charging station intelligent reconstruction and extension optimization model for multi-objective optimization solution, and finally obtaining charging station reconstruction and extension schemes under different charging strategies. According to the characteristics of different charging strategies, the construction layout of the charging station is dynamically adjusted, scientific guidance is provided for reconstruction and extension of the charging station, and the utilization efficiency and service quality of charging facilities are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of charging station construction, and specifically to a charging station renovation and expansion method driven by user travel behavior and intelligent charging strategy. Background Art

[0002] In recent years, with the continuous growth of electric vehicle ownership, many cities have adopted time-of-use electricity price strategies to guide electric vehicle charging behavior. However, relying solely on user-side electricity price adjustment is still not enough to solve the problem. The electricity required for electric vehicles to connect to the grid mainly comes from the combustion of fossil energy in the power grid, which will lead to increased carbon dioxide emissions. Therefore, it is necessary to comprehensively consider the impact of user charging costs, grid load mean square deviation, and carbon emissions to effectively guide user charging behavior.

[0003] In addition, existing electric vehicle charging facilities generally have a supply-demand imbalance problem, and the unreasonable layout of charging facilities is one of the important factors restricting the promotion of electric vehicles. Based on the trend that the charging behavior of electric vehicles will be affected by the intelligent charging scheduling method in the future, the present invention proposes a site selection and renovation and expansion method, which comprehensively considers the travel behavior habits of users and the charging distribution after intelligent charging scheduling, expands the areas with concentrated charging demand, and reasonably removes the areas with excess charging facilities, so as to achieve optimal allocation of resources.

[0004] To achieve the optimization of charging station renovation and expansion driven by user travel behavior and intelligent charging strategy, the following needs to be achieved: 1. Travel mode identification and charging demand prediction: Based on user travel behavior data, predict the dispatchable time window of each user to provide range constraints for intelligent charging scheduling; 2. Multi-factor driven intelligent charging scheduling optimization model: Taking the minimum user charging cost, the minimum grid load mean square error and the minimum carbon emission as the optimization goals, use multi-objective optimization to solve the optimal solution set of user charging strategy; 3. Charging station intelligent renovation and expansion optimization model based on multi-dimensional needs: Taking the minimum total construction cost, the minimum total distance cost of users, and the minimum penalty cost of uncovered users as the optimization goals, and incorporating the charging strategy after user scheduling into the charging station intelligent renovation and expansion optimization model for multi-objective optimization solution, finally obtain the charging station renovation and expansion plan under different charging strategies. Summary of the invention

[0005] In view of the shortcomings of the prior art, the technical problem to be solved by the present invention is to provide a method for the renovation and expansion of charging stations driven by user travel behavior and intelligent charging strategies, which determines the user charging strategy and realizes the efficient allocation of charging resources on the basis of taking into account the interests of the user's charging cost, grid load fluctuation and carbon emissions. At the same time, according to the characteristics of different charging strategies, the construction layout of the charging station is dynamically adjusted to provide scientific guidance for the renovation and expansion of the charging station, and improve the utilization efficiency and service quality of the charging facilities.

[0006] The technical solution of the present invention to solve the technical problem is: A charging station renovation and expansion method driven by user travel behavior and intelligent charging strategy, the charging station renovation and expansion method comprising the following contents: The functional areas reached by users are divided into residential areas, work areas, and other areas except residential areas and work areas, and the OD data of the functional areas reached by users are collected; the first departure time K, the stay time S in each area, the distance J between areas, and the power state SOC0 at the first departure of users are obtained every day in the past year, and the probability density functions of these four parameters are fitted, and their respective values ​​are assigned to each user through probability extraction; The user travel plans are divided into three types: the first type is from the residential area to the work area and then back to the residential area; the second type is from the residential area to the work area and then to other areas and finally back to the residential area; the third type is from the residential area to the work area and then to other areas, from other areas back to the work area and finally back to the residential area; For each user, a travel plan is randomly generated from the three types of user travel plans, and the battery status of the user when arriving at the next area is repeatedly judged based on the number of ferries in different areas on the same day in the travel plan. When the battery status at the area is lower than 30%, charging is required, and the schedulable time window is: the arrival time at the area ~ the departure time from the area; if the user does not depart on the day of arriving at the area, the schedulable time window is: the arrival time at the area ~ the first departure time of the next day; Obtain the predicted dispatchable time windows of all users and the total number of users N who need to charge in the planning area; The optimization objectives are to minimize user charging cost, grid load mean square error and carbon emission, and to use each user’s charging time z as the n It is necessary to build an intelligent charging scheduling optimization model within the schedulable time window and the power state of the user before and after charging must be between the upper and lower limits of the power state. The optimal solution set of the user's charging strategy is solved. The optimal solution set includes the charging time and charging power of each user after scheduling. Combined with the optimal solution set of user charging strategies obtained by the intelligent charging scheduling optimization model, and taking the minimum total construction cost, the minimum total distance cost of users, and the minimum penalty cost of uncovered users as the optimization objectives, an optimization model for intelligent renovation and expansion of charging stations is constructed; Solve the optimization model of intelligent renovation and expansion of charging stations, and output renovation and expansion plans for charging stations under different charging strategies. Each renovation and expansion plan includes the optimal location and optimal capacity of the charging station. Compare the optimal capacity of the charging station obtained with the number of charging piles in the charging station mThe size of the charging station determines the supply and demand relationship, and expansion is carried out when supply is less than demand. When supply exceeds demand, the use probability and maintenance cost of the charging station are used as input variables, and the Mamdani fuzzy reasoner is used to obtain the possibility of dismantling the charging station.

[0007] Furthermore, the intelligent charging scheduling optimization model and the charging station intelligent renovation and expansion optimization model are solved by particle swarm optimization algorithm and the third-generation non-dominated sorting genetic algorithm; For the intelligent charging scheduling optimization model, the user charging time is encoded as an independent variable, using a two-layer encoding form. The first layer is the start time of charging, and the second layer is the charging duration. The chromosome length is the total number of users who need to charge in the planning area. For the optimization model of intelligent renovation and expansion of charging stations, the location of the charging stations is encoded as an independent variable, and the chromosome length is the number of charging stations.

[0008] In the present invention, when performing multi-objective optimization solutions, existing commonly used multi-objective optimization methods, such as NSGA-II algorithm, PSO-NSGA-II algorithm, multi-objective genetic optimization algorithm, etc., can also be used.

[0009] Furthermore, the charging station intelligent renovation and expansion optimization model is constrained by the distance to the charging station not exceeding the service radius of the charging station and the number of charging piles in the charging station not exceeding the maximum capacity specified in the design of the charging station.

[0010] Furthermore, the optimization objective functions f1-f3 of the intelligent charging scheduling optimization model are formulas (12)-(14), and the constraints are formulas (15)-(16): (12) (13) (14) (15) (16) Among them, the objective function f1 represents the minimization of user charging cost, f2 represents the minimization of grid load mean square error, and f3 represents the minimization of carbon emissions. Formula (15) constrains the charging time z of the nth user n Must meet its schedulable time window range, A n is the left time window of the schedulable time window, B n is the right time window of the schedulable time window; Formula (16) constrains the state of charge SOC of user n before and after charging n Both are between the upper and lower limits of the state of charge, SOC min is the minimum value of the state of charge, SOC maxis the maximum value of the power state; variable is a binary variable, where =1 means user n charges in time period t, =0 means no charging; N is the total number of users who need to charge in the planning area; W t The electricity price in different time periods; It is the basic load of the power grid in different time periods; is the charging power of user n in time period t; P avr is the average load of the power grid; T is the total number of time periods; Q n Charge amount for the nth user; CEF t is the carbon emission factor in time period t.

[0011] Furthermore, the optimization objective functions f4-f6 of the charging station intelligent renovation and expansion optimization model are as follows: Formulas (17)-(19), and the constraints are as follows: Formulas (22)-(23): (17) (18) (19) (20) (twenty one) (twenty two) (twenty three) Among them, the objective function f4 represents the minimization of the total travel cost of users, f5 represents the minimization of the total construction cost, and f6 represents the minimization of the penalty cost of uncovered users; represents the distance from the starting point O to the charging station m for the nth user, represents the distance from the charging station m to the end point D of the nth user, is a binary variable, =1 means the nth user is charging at the mth charging station, =0 means no charging; r is the discount rate, year is the service years of the charging pile; N is the total number of users who need charging in the planning area; C u is the unit penalty cost for failing to cover users; M is the number of charging stations in the planned area; Represents the planning area, R ser Represents the service radius of the charging station; C c,m represents the construction cost of the mth charging station; ARER m Represents the site construction area; y m is a binary variable, y m=0 means that charging station m originally has a charging station, y m =1 means there is no charging station; AM m represents the demand for charging station m, which changes dynamically due to the influence of different user charging strategies given by the intelligent charging scheduling optimization model; is the land price per unit area, O m is the number of original charging piles at the charging station, is the construction cost of the charging station, C cp is the price of a single charging pile, C m is the maintenance cost of a single charging pile; Formula (22) constrains that the distance to the charging station cannot exceed the service radius of the charging station; Formula (23) constrains that the number of charging piles in the charging station m cannot exceed the maximum capacity H specified in the design of the charging station max .

[0012] The present invention also protects a computer-readable storage medium having a computer program stored thereon, which can implement the steps of the charging station renovation and expansion method when executed by a processor.

[0013] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention first identifies travel patterns and predicts charging demand. Through in-depth analysis of user travel behavior, it predicts users' travel and charging behaviors on weekdays. Based on the prediction results, it dynamically generates a schedulable time window for each user, and charges within the schedulable time window. This innovation provides an important foundation for subsequent charging scheduling optimization research, and improves scheduling flexibility and charging resource utilization efficiency.

[0014] 2. Existing charging scheduling strategies usually rely solely on electricity price signals, which can easily lead to nighttime charging peaks or increased carbon emissions, and lack comprehensive consideration of grid load fluctuations and environmental impacts. The present invention comprehensively considers the three factors of minimizing user charging costs, minimizing grid load mean square deviation, and minimizing carbon emissions, and constructs a more intelligent and comprehensive intelligent charging scheduling optimization model.

[0015] 3. Existing charging station site selection methods focus on spatial factors, while ignoring the dynamic adjustment of the temporal and spatial changes in charging demand. Due to the limited load bearing capacity of the power grid, the traditional "charge as you arrive" model is gradually unable to meet the increasing charging demand, and relying solely on spatial optimization can no longer meet future challenges. The present invention incorporates the post-scheduled charging strategy into the intelligent renovation and expansion optimization model of the charging station, comprehensively considering the user's time requirements, space requirements, and the penalty costs of uncovered users, thereby proposing a new charging station renovation and expansion optimization framework.

[0016] 4. Solving through the PSO-NSGA-III algorithm can optimize multiple objectives at the same time, generate frontier solution sets, and provide diverse decision-making solutions. In addition, this combined method guides the rapid convergence of solutions through the global search capability of PSO, overcoming the local search limitations of NSGA-III in the initial stage, while the diversity preservation mechanism of NSGA-III ensures the uniform distribution of solution sets, which can effectively improve the convergence speed and solution set quality in multi-objective optimization problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 The present invention is a flowchart of an embodiment of a method for renovating and expanding a charging station driven by user travel behavior and intelligent charging strategy.

[0018] Figure 2 Schematic diagram of the user's travel plan.

[0019] Figure 3 It is a schematic diagram of the prediction process of the user's schedulable time window in the present invention. DETAILED DESCRIPTION

[0020] The present invention is further explained below in conjunction with embodiments and drawings, but the protection scope of the present application is not limited thereto.

[0021] The method for renovating and expanding a charging station driven by user travel behavior and intelligent charging strategy of the present invention is as follows: Figure 1 As shown in the figure, it includes three parts: travel mode identification and charging demand prediction, multi-factor driven intelligent charging scheduling optimization, and intelligent renovation and expansion optimization of charging stations based on multi-dimensional needs. First, by collecting the user's travel behavior data, based on the fitting probability density function, the user's travel plan is identified, and the charging demand is predicted, and the dispatchable time window of each user is determined to provide basic data support for subsequent optimization. Secondly, an intelligent charging scheduling optimization model is constructed, and the minimum user charging cost, the minimum grid load mean square error and the minimum carbon emission are taken as the optimization goals. In combination with the constraints, the optimal solution set of the user charging strategy is solved by multi-objective optimization to provide support for intelligent charging scheduling. Then, in view of the supply and demand imbalance problem in the layout of charging stations, an intelligent renovation and expansion optimization model of charging stations is established that comprehensively considers the minimum total construction cost, the minimum total distance cost of users, and the minimum penalty cost of uncovered users. The dispatched charging strategy is incorporated into the intelligent renovation and expansion optimization model of charging stations for solution, and the renovation and expansion scheme of charging stations under different charging strategies is obtained.

[0022] Embodiment 1: The method for renovating and expanding a charging station driven by user travel behavior and intelligent charging strategy in this embodiment has the following steps: Step 1: Divide the functional areas reached by users into residential areas, work areas, and other areas except residential areas and work areas, and collect OD data of the functional areas reached by users. Obtain the first departure time K, the stay time S in each area, the distance J between areas, and the power state SOC0 at the first departure of users every day in the past year. Based on the above data, fit the probability density functions of these four parameters, and assign their respective values ​​to each user through probability extraction.

[0023] The departure time U, arrival time L and state of charge SOC at arrival of each area are calculated according to formulas (1)-(3), simulating the travel behavior pattern of users in a day.

[0024] (1) (2) (3) Among them U a is the departure time from area a, L a is the arrival time at area a, S a is the residence time in area a, L b is the arrival time from area a to area b, J ab is the distance between area a and area b, speed is the vehicle speed, SOC b is the state of charge when reaching area b from area a, SOC a is the power state when reaching area a, Q 100 It is the power consumption of electric vehicles per 100 kilometers under standard working conditions, and Q0 is the rated power of electric vehicles.

[0025] Step 2: The present invention divides the user's travel plans into three types, such as Figure 2 As shown in the figure, the three types are: the first type is from the residential area to the work area and then back to the residential area; the second type is from the residential area to the work area and then to other areas and finally back to the residential area; the third type is from the residential area to the work area and then to other areas, from other areas back to the work area and finally back to the residential area. Through cyclic calculation, the dispatchable time window of each user who needs to charge is predicted.

[0026] The schedulable time window prediction process of the user of the present invention is as follows Figure 3As shown in the figure, n represents the user. Each user will randomly generate g, where g is 1, 2 or 3, representing three travel plans. The users of the three travel plans initially travel from the residential area H to the work area W. Through the probability density function of the user's first departure time K of each day, the probability density function of the distance J between the areas, and the probability density function of the power state SOC0 at the first departure fitted in step 1, the user's first departure time K1, the distance J between the residential area and the work area are extracted. hw , the state of charge SOC0 at the first departure, and substitute it into formulas (1)-(3) to calculate the arrival time L1 from the residential area to the work area, the state of charge SOC1 when arriving from the residential area to the work area, and the departure time U1 from the work area; and extract the first stay time S in the work area from the probability density function of the stay time S in each area w1 . Determine whether the SOC1 at this time is lower than 30%. If so, the schedulable time window of user n is L1-U1 (arrival time to the work area ~ departure time from the work area). During this time, user n can participate in charging scheduling, and then let n+1, which represents the start of the prediction for the next user; if not, determine whether g is equal to 1, which is to determine whether user n adopts the first travel plan.

[0027] If g is equal to 1, calculate the arrival time L2 from the work area to the residential area H and the power state SOC2 at the time of arrival, and determine whether the SOC2 at this time is lower than 30%. If so, the schedulable time window of user n is L2-K1 (arrival time at the residential area ~ first departure time of the next day). During this time, the user can participate in the charging scheduling, and then let n+1 make a prediction for the next user; if not, it means that the user does not need to charge today, and directly let n+1 make a prediction for the next user.

[0028] If g is not equal to 1, it means that the user does not adopt the first travel plan. Regardless of whether the second or third travel plan is adopted, it is necessary to calculate the arrival time L3 from the working area to other areas O and the power state SOC3 at the time of arrival, and the departure time U2 from other areas. The residence time S0 to other areas is extracted from the probability density function of the residence time S in each area, and then it is determined whether SOC3 is lower than 30%. If so, the schedulable time window of user n is L3-U2 (arrival time to other areas ~ departure time from other areas). During this time, the user can participate in charging scheduling, and then let n+1, which represents the start of the prediction for the next user; if not, it is determined whether g is equal to 2, which is to determine whether user n adopts the second travel method.

[0029] If g is equal to 2, calculate the arrival time L4 from other areas to the residential area and the power state SOC4 at the time of arrival, and determine whether the SOC4 at this time is lower than 30%. If so, the schedulable time window of user n is L4-K1 (arrival time at the residential area ~ first departure time of the next day). After the user arrives at the residential area again, he will no longer set out on the same day. During this period, he can participate in the charging scheduling, and then let n+1 make a prediction for the next user; if not, it means that user n does not need to charge today, and directly let n+1 make a prediction for the next user.

[0030] If g is not equal to 2, it means that user n adopts the third travel plan. The arrival time L5 and the power state SOC5 of the second arrival at the work area from other areas are calculated, as well as the departure time U3 from the work area again. At this time, the number of ferries is three. The second stay time S of the work area is extracted from the probability density function of the stay time S in each area. w2 , determine whether the SOC5 at this time is lower than 30%. If so, the schedulable time window of user n is L5-U3 (arrival time to the work area ~ departure time from the work area). During this time, he can participate in charging scheduling, and then let n+1 predict the next user; if SOC5 is not lower than 30%, then calculate the arrival time L6 from the work area to the residential area H and the power state SOC6 at the time of arrival. At this time, the number of ferries is four. Determine whether the SOC6 at this time is lower than 30%. If so, the schedulable time window of user n is L6-K1 (arrival time to the residential area ~ first departure time of the next day). During this time, he can participate in charging scheduling; if SOC6 is not lower than 30%, it means that user n does not need to charge today. Regardless of charging or not, n+1 needs to be asked to predict the next user. Judgment n Is it greater than N* If not, regenerate the next user's g ; If yes, exit the program, all users are predicted, and the total number of users N who need to charge in the planning area is recorded. N* is the number of all users that need to be predicted.

[0031] Step 3: Divide the 24 hours of a day into 72 equal-length time periods (T=72), each of which is 20 minutes long, and obtain the electricity price W for different time periods t 2. Basic load of power grid in different time periods , Charging power of user n in time period t , the power generation of each power plant in different time periods and charging efficiency Based on the above parameters, the charging amount Q of the nth user is calculated using formula (4): n, use formula (5) to calculate the average load P of the power grid avr , use formula (6)-(11) to calculate the carbon emission factor CEF for time period t t .

[0032]

[0033] Among them, the variable is a binary variable, where =1 means user n charges in time period t, =0 means not charged; z n is the charging time of user n, G t for The total power generation of the power plant during the time period t, G t-1 for The total power generation of the power plant during the time period t-1, E t for The total emissions generated by a power plant during time period t due to power generation, E t-1 for The total emissions generated by the power plant during time period t-1 due to power generation, Indicates The emissions generated by a power plant during time period t due to power generation, For the Average emission factor for each power plant, ΔE t and ΔG t are the changes in carbon emissions and power generation between two adjacent time periods, respectively. Formula (9) represents a carbon emission factor calculation model with a slope of α and an intercept of δ. The carbon emission factor calculation model describes the change in hourly carbon emissions with power generation. The slope α of the carbon emission factor calculation model is CEF t .

[0034] Step 4: Minimize user charging cost, grid load mean square error and carbon emission as optimization objectives, and use each user’s charging time z as the optimization target. n The user's power state before and after charging must be within the schedulable time window and between the upper and lower limits of the power state. A multi-factor driven intelligent charging scheduling optimization model is constructed to solve the optimal solution set of the user's charging strategy. The optimal solution set includes the charging time and charging power after each user is scheduled. The schedulable time window of the user in step 2 is considered in the construction of this intelligent charging scheduling optimization model. The specific optimization objective functions f1-f3 are shown in formulas (12)-(14), and the constraints are shown in formulas (15)-(16).

[0035] (12) (13) (14) (15) (16) Among them, the objective function f1 represents the minimization of user charging cost, f2 represents the minimization of grid load mean square error, and f3 represents the minimization of carbon emissions. Formula (15) constrains the charging time z of the nth user n The schedulable time window must be satisfied, where A n is the left time window of the schedulable time window, B n is the right time window of the schedulable time window; Formula (16) constrains the state of charge SOC of user n before and after charging n Both need to be between the upper and lower limits of the state of charge, where SOC min is the minimum value of the state of charge, SOC max It is the maximum value of the power state.

[0036] Step 5: Obtain the latitude and longitude coordinates of the candidate site and its unit land price , the latitude and longitude coordinate data of each charging station and the number of charging piles in the charging station m , the construction cost of charging stations , the price of a single charging pile C cp , the maintenance cost of a single charging pile is C m . Based on the above data and the optimal solution set of the user charging strategy obtained in step 4, the optimization objectives are to minimize the total construction cost, minimize the total distance cost of users, and minimize the penalty cost of uncovered users. The distance to the charging station cannot exceed the service radius of the charging station, and the number of charging piles in the charging station cannot exceed the maximum capacity specified in the charging station design. An optimization model for intelligent renovation and expansion of charging stations based on multi-dimensional needs is established. Through multi-objective optimization solution, the renovation and expansion plans of charging stations under different charging strategies are obtained. Each renovation and expansion plan includes the optimal location and optimal capacity of the charging station. The capacity of a charging station refers to the number of charging piles in the charging station. Compare the optimal capacity of the charging station obtained with the original number of charging piles in the charging station (O). m The size of the charging station determines the supply and demand relationship, and expansion is carried out when supply is less than demand. When supply exceeds demand, the use probability and maintenance cost of the charging station are taken as input variables, the Mamdani fuzzy reasoner is introduced and the corresponding fuzzy rules are set to deduce the possibility of dismantling the charging station.

[0037] The optimization objective functions f4-f6 of the optimization model for intelligent renovation and expansion of charging stations are shown in formulas (17)-(19). The constraints are shown in formulas (22)-(23).

[0038] (17) (18) (19) (20) (twenty one) (twenty two) (twenty three) Among them, the objective function f4 represents the minimization of the total travel cost of users, f5 represents the minimization of the total construction cost, and f6 represents the minimization of the penalty cost of uncovered users. represents the distance from the starting point O to the charging station m for the nth user, represents the distance from the charging station m to the end point D of the nth user, is a binary variable, =1 means the nth user is charging at the mth charging station, =0 means no charging. r is the discount rate, year is the service life of a charging station, N is the total number of users who need to charge in the planning area, C u is the unit penalty cost for failing to cover the user. M is the number of charging stations in the planning area, which is calculated and determined by formula (21); Represents the planning area, R ser Represents the service radius of the charging station. C c,m represents the construction cost of the mth charging station, which is calculated as shown in formula (20): m Represents the site construction area, y m is a binary variable, y m =0 means that charging station m originally has a charging station, y m =1 means there is no charging station, AM m represents the demand of charging station m (charging time and charging power of different users in different time periods), which needs to meet the charging demand in the peak time period. The demand in the peak time period will be affected by the charging strategies of different users given by the intelligent charging scheduling optimization model and will show dynamic changes. Formula (22) constrains that the distance to the charging station cannot exceed the service radius of the charging station, and formula (23) constrains that the number of charging piles in charging station m must not exceed the maximum capacity H specified in the design of the charging station. max .

[0039] Embodiment 2: In this embodiment, the intelligent charging scheduling optimization model and the charging station intelligent renovation and expansion optimization model are optimized and solved by combining the particle swarm optimization algorithm with the third-generation non-dominated sorting genetic algorithm (PSO-NSGA-Ⅲ algorithm). The specific process is: Step 6.1: For the intelligent charging scheduling optimization model, the user charging time is encoded as an independent variable in the form of double-layer encoding, the first layer is the start charging time, and the second layer is the charging duration. Specifically, the chromosome length is N, where N represents the total number of users who need to charge in the planning area. The first layer encoding form is such as {2, 5, 15, ..., 10}, and the specific encoding value is obtained by randomly generating an integer from 1 to 72. Because the present invention divides 24 hours a day into 72 segments, this code should be randomly determined from integers from 1 to 72, and the corresponding user's start charging time is 0:40, 1:40, 5:00, ..., 3:20. The second layer encoding form is such as {20, 45, 100, ..., 30}, which respectively represents that the user's charging time is 20 minutes, 45 minutes, 100 minutes, ..., 30 minutes.

[0040] Step 6.2: For the charging station intelligent renovation and expansion optimization model, the location of the charging station is encoded as an independent variable. Specifically, the chromosome length is M, where M represents the number of charging stations, and the location number of the charging station is used as the encoding. The encoding format is {2, 5, 8, ..., 20}, which corresponds to the location number of the charging station, that is, the locations numbered 2, 5, 8, ..., 20 in the candidate sites.

[0041] Step 6.3: Use PSO-NSGA-Ⅲ algorithm to solve, specifically: (1) Parameter initialization: Set the population size, maximum number of iterations, crossover probability, mutation probability, inertia weight, and learning factor to define a uniformly distributed set of reference points for NSGA-III. This set covers the target space and ensures that the Pareto front solutions of all objective functions can be effectively mapped to these reference points.

[0042] (2) Population initialization: Double-layer coding form: the charging start time is used as the first layer of coding, and the charging duration is used as the second layer of coding. The charging start time and the generated charging duration are input, and the initial population is randomly generated as the parent generation in the iteration process, and the fitness value of each chromosome of the parent generation is calculated; Single-layer coding form: the charging station location number is randomly generated as the parent generation in the inheritance process, and the fitness value of each chromosome of the parent generation is calculated.

[0043] (3) Genetic algorithm operation (applicable to the second layer of double-layer coding): The roulette wheel selection method is used to select parent individuals, and two-point crossover is used to generate offspring according to the crossover probability. The genes in the chromosomes are randomly perturbed according to the mutation probability to generate mutated offspring.

[0044] (4) Particle swarm update (applicable to the first layer of double-layer coding and single-layer coding): Update the particle speed and position as the next generation population, calculate the updated fitness value, and update the individual optimum and global optimum of the particles. The update method is as follows:

[0045] Formula (24) is the speed update formula, where is the i-th particle in the The speed in the iteration determines the moving direction and step size of the particle in the search space; is the i-th particle in the The speed in iterations; is the inertia weight, which controls the degree of retention of the particle's current velocity; c1 and c2 are individual learning factors and group learning factors; r1 and r2 are two random numbers randomly generated in the interval [0, 1]; pbest i is the optimal position found by the ith particle in the historical iteration; gbest is the global optimal position. Formula (25) is the position update formula, where is the i-th particle in the The position in the iteration represents the new position of the particle in the solution space; is the i-th particle in the The position in the iteration.

[0046] (5) Non-dominated sorting: Perform non-dominated sorting on the individuals in the current population, divide the population into multiple non-dominated levels, and retain the best individuals for the next generation of evolution. The number of individuals depends on the size of the population.

[0047] (6) Use reference points for next generation selection: Setting reference points: Reference points are set by dividing the target space (a multidimensional space consisting of multiple objective function values) into a uniform grid. First, the minimum and maximum values ​​of each objective function are determined, and then the target space is divided into several equally spaced intervals to generate corresponding reference points. The generated reference points are evenly distributed in the target space to ensure that the values ​​of each objective function can be reasonably covered.

[0048] Reference point mapping: Normalize all objective function values ​​of the current population and assign each normalized objective function to its nearest reference point to form a reference point mapping; Reference point associated solutions: For each reference point, calculate the number of solutions associated with it, and ensure that the less occupied reference points have priority in selecting their associated solutions. If the number of associated solutions for a reference point is insufficient, it will be allocated from other solution sets first; Priority selection of solutions: During the allocation process, if a reference point is associated with multiple solutions, the solution closest to the reference point in the target space is selected first, and this process is repeated until the new population reaches the predetermined size.

[0049] The calculation method for normalizing the objective function value is as follows:

[0050] in is the normalized value of the jth objective function at the ith particle, f j，i is the original value of the jth objective function at the ith particle, f j,min is the minimum value of the jth objective function corresponding to all particles, f j,max It is the maximum value of the jth objective function of all particles.

[0051] (7) Iteration stop condition: For the intelligent charging scheduling optimization model, repeat steps (3) to (6) until the maximum number of iterations is reached; for the charging station intelligent renovation and expansion optimization model, repeat steps (4) to (6) until the maximum number of iterations is reached.

[0052] (8) Output the optimal solution: Output the Pareto front solution covered by the reference point, record the charging time and charging power of each user obtained by the double-layer coding, and the optimal location and optimal capacity of the charging station obtained by the single-layer coding.

[0053] Any matters not described in the present invention are applicable to the prior art.

Claims

1. A method for renovating and expanding a charging station driven by user travel behavior and intelligent charging strategy, characterized in that: The charging station reconstruction and expansion method comprises the following contents: The functional areas reached by users are divided into residential areas, work areas, and other areas except residential areas and work areas, and the OD data of the functional areas reached by users are collected; the first departure time K, the stay time S in each area, the distance J between areas, and the power state SOC0 at the first departure of users are obtained every day in the past year, and the probability density functions of these four parameters are fitted, and their respective values ​​are assigned to each user through probability extraction; The user travel plans are divided into three types: the first type is from the residential area to the work area and then back to the residential area; the second type is from the residential area to the work area and then to other areas and finally back to the residential area; the third type is from the residential area to the work area and then to other areas, from other areas back to the work area and finally back to the residential area; For each user, a travel plan is randomly generated from the three types of user travel plans, and the battery status of the user when arriving at the next area is repeatedly judged based on the number of ferries in different areas on the same day in the travel plan. When the battery status at the area is lower than 30%, charging is required, and the schedulable time window is: the arrival time at the area ~ the departure time from the area; if the user does not depart on the day of arriving at the area, the schedulable time window is: the arrival time at the area ~ the first departure time of the next day; Obtain the predicted schedulable time windows of all users and the total number of users N who need to charge in the planning area; The optimization objectives are to minimize the user charging cost, minimize the mean square error of the grid load and minimize the carbon emission, and to use the charging time z of each user as the n It is necessary to build an intelligent charging scheduling optimization model within the schedulable time window and the power state of the user before and after charging must be between the upper and lower limits of the power state. The optimal solution set of the user's charging strategy is solved. The optimal solution set includes the charging time and charging power of each user after scheduling. Combined with the optimal solution set of user charging strategies obtained by the intelligent charging scheduling optimization model, and taking the minimum total construction cost, the minimum total distance cost of users, and the minimum penalty cost of uncovered users as the optimization objectives, an optimization model for intelligent renovation and expansion of charging stations is constructed; Solve the optimization model of intelligent renovation and expansion of charging stations, and output renovation and expansion plans for charging stations under different charging strategies. Each renovation and expansion plan includes the optimal location and optimal capacity of the charging station. Compare the optimal capacity of the charging station obtained with the number of charging piles in the charging station m The size of the charging station determines the supply and demand relationship, and expansion is carried out when supply is less than demand. When supply exceeds demand, the use probability and maintenance cost of the charging station are used as input variables, and the Mamdani fuzzy reasoner is used to obtain the possibility of dismantling the charging station.

2. The method for rebuilding and expanding a charging station according to claim 1, characterized in that: The intelligent charging scheduling optimization model and the charging station intelligent renovation and expansion optimization model are solved by particle swarm optimization algorithm and the third-generation non-dominated sorting genetic algorithm; For the intelligent charging scheduling optimization model, the user charging time is encoded as an independent variable, using a two-layer encoding form. The first layer is the start time of charging, and the second layer is the charging duration. The chromosome length is the total number of users who need to charge in the planning area. For the optimization model of intelligent renovation and expansion of charging stations, the location of the charging stations is encoded as an independent variable, and the chromosome length is the number of charging stations.

3. The method for rebuilding and expanding a charging station according to claim 1, characterized in that: The intelligent renovation and expansion optimization model of the charging station is subject to the constraints that the distance to the charging station cannot exceed the service radius of the charging station and the number of charging piles in the charging station cannot exceed the maximum capacity specified in the design of the charging station.

4. The method for rebuilding and expanding a charging station according to claim 1, characterized in that: The optimization objective functions f1-f3 of the intelligent charging scheduling optimization model are formulas (12)-(14), and the constraints are formulas (15)-(16): （12） （13） （14） （15） （16） Among them, the objective function f1 represents the minimization of user charging cost, f2 represents the minimization of grid load mean square error, and f3 represents the minimization of carbon emissions. Formula (15) constrains the charging time z of the nth user n Must meet its schedulable time window range, A n is the left time window of the schedulable time window, B n is the right time window of the schedulable time window; Formula (16) constrains the state of charge SOC of user n before and after charging n Both are between the upper and lower limits of the state of charge, SOC min is the minimum value of the state of charge, SOC max is the maximum value of the state of charge; variable is a binary variable, where =1 means user n charges in time period t, =0 means no charging; N is the total number of users who need to charge in the planning area; W t The electricity price in different time periods; It is the basic load of the power grid in different time periods; is the charging power of user n in time period t; P avr is the average load of the power grid; T is the total number of time periods; Q n Charge amount for the nth user; CEF t is the carbon emission factor in time period t.

5. The method for rebuilding and expanding a charging station according to claim 4, characterized in that: The optimization objective functions f4-f6 of the charging station intelligent renovation and expansion optimization model are as follows: Formulas (17)-(19), and the constraints are as follows: Formulas (22)-(23): （17） （18） （19） （20） （21） （22） （23） Among them, the objective function f4 represents the minimization of the total travel cost of users, f5 represents the minimization of the total construction cost, and f6 represents the minimization of the penalty cost of uncovered users; represents the distance from the starting point O to the charging station m for the nth user, represents the distance from the charging station m to the end point D of the nth user, is a binary variable, =1 means the nth user is charging at the mth charging station, =0 means no charging; r is the discount rate, year is the service years of the charging pile; N is the total number of users who need charging in the planning area; C u is the unit penalty cost for failing to cover users; M is the number of charging stations in the planned area; Represents the planning area, R ser Represents the service radius of the charging station; C c,m represents the construction cost of the mth charging station; ARER m Represents the site construction area; y m is a binary variable, y m =0 means that charging station m originally has a charging station, y m =1 means there is no charging station; AM m represents the demand for charging station m, which changes dynamically due to the influence of different user charging strategies given by the intelligent charging scheduling optimization model; is the land price per unit area, O m is the number of original charging piles at the charging station, is the construction cost of the charging station, C cp is the price of a single charging pile, C m is the maintenance cost of a single charging pile; Formula (22) constrains that the distance to the charging station cannot exceed the service radius of the charging station; Formula (23) constrains that the number of charging piles in the charging station m cannot exceed the maximum capacity H specified in the design of the charging station max .

Images