Charging scheduling method for new energy buses based on hybrid multi-objective particle swarm algorithm

Abstract

The present invention discloses a charging scheduling method for new energy buses based on a hybrid multi-objective particle swarm algorithm. First, the data set is screened and preprocessed, the outliers in the data set are deleted and filtered, and then the charging time of each bus at each charging gun is calculated, and two objective functions of charging time and charging pile usage deviation are constructed, and two fitness functions are minimized. The two objective functions are optimized using a hybrid multi-objective particle swarm algorithm, system parameters are set, the objective function is optimized, the algorithm model is deployed in the server, and the algorithm is called remotely with gRPC to obtain a nighttime charging scheduling schedule for the bus. The present invention uses a modified multi-objective particle swarm algorithm to optimize the minimum charging time and the minimum charging pile utilization deviation of the constructed new energy bus, so as to achieve the minimum charging time while minimizing the charging pile utilization deviation, generate a nighttime charging scheduling schedule, and facilitate the scheduling of charging dispatchers at bus stations.

Description

Technical Field

[0001] The present invention relates to the field of new energy transportation and environmental protection, and in particular to a new energy bus dispatching method based on a hybrid multi-objective particle swarm algorithm. Background Art

[0002] As all sectors of society pay more and more attention to energy conservation and environmental protection, new energy represented by wind and light has gradually become popular in recent years, and its share in the total annual power generation has increased year by year, and at the same time, new energy electric vehicles have been widely popularized. The proportion of new energy buses continues to increase, and the trend of adopting pure electric buses is getting stronger and stronger. However, due to the limited range of pure electric buses, more vehicles need to be charged at night. In the past, the night charging method was to charge as soon as they arrived, which would result in the failure to make good use of the advantage of lower electricity prices during the nighttime electricity price trough period, and would also result in the use of charging piles without reasonable maximum benefits. Insufficient charging at night makes it difficult to complete a day's operating tasks without charging when using pure electric buses in daily operations, and often fails to give full play to the transportation capacity of pure electric buses.

[0003] The publication number is CN 112233452 A. The scheduling process for flexible station demand includes: obtaining the time and area of ​​passenger demand and judging the weight of the objective function accordingly. The objective function F1 is the interest of the bus company measured in money, and F2 is the interest of the passenger measured in time. The intelligent algorithm is used to solve the optimal solution of the model and then plan the best bus route, and the updated route is fed back to the driver. It can be seen that this scheduling method is not a traditional fixed-time fixed-station stop, but is divided into two objective functions: the interest of the bus company measured in money and the interest of the passenger measured in time. The passenger's interest is maximized without the bus company losing money, and the passenger demand is responded to in a variable route based on this goal. This method is not based on the multi-objective particle swarm algorithm to optimize the minimum charging time and the minimum utilization deviation of the charging pile of the constructed new energy bus. This solution does not solve the problem of the night schedule of new energy buses.

[0004] Based on this, the problem arises of how to ensure that the new energy buses to be charged can be fully charged at night with the minimum time required and the minimum deviation in charging pile utilization, while meeting the new energy bus operation needs the next day. Summary of the invention

[0005] In view of the shortcomings of the above-mentioned background technology, the present invention proposes a new energy bus charging scheduling algorithm based on a hybrid multi-objective particle swarm, which solves the problems of disordered charging, low efficiency, and unbalanced resource allocation in existing new energy bus charging stations.

[0006] The solution of the present invention to solve the technical problem is: adopt a new energy bus charging scheduling method based on a hybrid multi-objective particle swarm algorithm, first screen and pre-process the basic data set of bus charging stations (including: charging gun output power, bus station capacity, bus remaining SOC value, mileage of the day, power consumption, etc.), delete and filter the abnormal values ​​in the data set, and then calculate the charging time of each bus at each charging gun, construct two objective functions of charging time and charging pile usage deviation, minimize the two objective functions, use a hybrid multi-objective particle swarm algorithm to optimize the two objective functions, set system parameters, optimize the objective function, use gRPC (google Remote Procedure Calls, gRPC) to remotely call the algorithm model to build an algorithm service, and deploy it in the server to obtain the bus night charging scheduling schedule.

[0007] The system parameters include the battery capacity of the electric bus, the charging efficiency of the charging pile, the total power required for the operation of the electric bus, the charging time of the charging pile, and the percentage of remaining battery power recorded for the first time by the electric bus.

[0008] The method comprises the following steps.

[0009] S1. The onboard device records the remaining SOC, mileage of the day, and power consumption of each bus. The buses entering the charging station are evaluated one by one, and whether charging is needed is determined based on the current remaining power, the next day's shift schedule, and the mileage.

[0010] S2. Filter out the buses that need to be charged, calculate the time required for each bus to reach the expected SOC based on the output power of each charging gun per hour, and build a charging matrix.

[0011] S3. Parameter initialization: Randomly generate the initial population according to the control variables and constraints, set the population size and the maximum number of iterations. Construct the objective function (1) to minimize the maximum charging time and the objective function (2) to minimize the utilization deviation of the charging piles, and iteratively find the optimal values ​​of nighttime charging to obtain the Pareto frontier of the two objectives.

[0012] Objective function:

[0013]

[0014]

[0015] The output parameters are:

[0016] CAP i The battery capacity of the electric bus numbered i;

[0017] ε jCharging efficiency of charging pile;

[0018] The total amount of electricity required for the operation of the electric bus numbered i;

[0019] c j The time the bus is charging at the charging station numbered j;

[0020] The first recorded remaining battery power percentage of the electric bus numbered j.

[0021] The constraints are:

[0022] s i,j The rechargeable status of the electric bus numbered j during period t: 0 means non-rechargeable, 1 means rechargeable;

[0023] minP j ≤P≤maxP max Ensure that the charging power of each bus is within the allowed range of the charging pile.

[0024] S4. Update the velocity and position of the particle, initialize the external archive set, and define it as the global optimal solution set.

[0025] S5, calculate objective function 1 and objective function 2 and update the external archive, determine whether there is a dominance relationship between the newly generated particle i and the particles in the external archive, if particle i is non-dominant with other particles, add particle i to the external archive, otherwise continue to iterate.

[0026] S6. Using the roulette strategy, randomly select N particles to mutate into a new population.

[0027] S7. Select excellent individuals from external archives, choose a new optimal solution and continue iterating to generate a new population.

[0028] S8. Determine whether the program meets the termination requirement. When the maximum number of iterations is reached, the loop ends and the optimal solution set is output. Otherwise, return to step S4.

[0029] S9. Output the Pareto solution set, and output the night shift schedule for new energy buses according to the particles corresponding to the solutions.

[0030] S10. The server sets a time to collect the remaining SOC of buses at new energy bus stations every day and the operation schedule for the next day. The four pieces of information, namely the remaining SOC of vehicles in the environment, battery capacity, output power of charging piles, and the schedule for the next day, are used as input. The output is the night charging schedule. The hybrid multi-objective particle swarm algorithm is used to optimize the minimum charging time and the minimum utilization deviation of charging piles for the constructed new energy buses, so as to achieve the minimum utilization deviation of charging piles while ensuring the minimum charging time.

[0031] S11. The charging scheduling is almost completed, and the night charging schedule is returned to the server.

[0032] The beneficial effects of the present invention are as follows: the present invention is applicable to the night charging scheduling plan of the current urban new energy bus charging stations. Based on the remaining power, mileage, output power during charging and the next day's operation scheduling table collected by the new energy bus and the on-board machine, the multi-objective particle swarm algorithm is used to optimize the minimum charging time and the minimum utilization deviation of the charging pile constructed for the new energy bus, so as to achieve the minimum charging time and the minimum utilization deviation of the charging pile, generate a night charging scheduling table, and facilitate the scheduling of charging dispatchers in bus stations. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a block diagram of the charging dispatching system for new energy buses of the present invention.

[0034] Figure 2 It is a structural diagram of a new energy bus station of the present invention.

[0035] Figure 3 It is a flow chart of charging scheduling of new energy buses using the hybrid multi-objective particle swarm algorithm of the present invention.

[0036] Figure 4 This is an example diagram of a nighttime charging scheduling plan obtained by the method of the present invention. DETAILED DESCRIPTION

[0037] The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0038] Example 1: Figure 1 As shown, a new energy bus charging scheduling system based on a hybrid multi-objective particle swarm algorithm is provided, which integrates a new energy bus charging scheduling platform, including a new energy bus charging scheduling platform, an on-board data acquisition module, a charging pile output power module, a scheduling plan module, a learning optimization module and an alarm module.

[0039] The multi-objective particle swarm optimization algorithm is a kind of heuristic algorithm, and the particle swarm optimization algorithm has its own advantages in solving multi-objective problems. First, the particle swarm optimization algorithm can use efficient clusters to search for non-inferior solutions in parallel, and in each iteration, the particle swarm optimization algorithm can generate multiple non-inferior solutions; secondly, the memory function of the particle swarm optimization algorithm makes the particle swarm optimization algorithm have high computational efficiency and execution speed. More importantly, the particle swarm optimization algorithm does not rely on the function characteristics of the problem and the form of the solution, so its application becomes very extensive and can be applied to most multi-objective problems.

[0040] However, if the initial particle swarm optimization algorithm is directly used to solve multi-objective problems, due to the characteristics of the particles themselves, the particles will track a solution that is not on the Pareto frontier, causing the algorithm to converge to the local range of non-inferior solutions without reaching the Pareto frontier. Therefore, when using the particle swarm optimization algorithm to solve multi-objective problems, we must solve the problem of choosing the global optimal position and the optimal position of the algorithm. For the selection of the global optimal position, we require the algorithm to have a good convergence speed and be able to converge to the Pareto frontier; and for the selection of the optimal position of the algorithm, we need to simplify the complexity of the optimal position of the algorithm as much as possible while optimizing the algorithm.

[0041] like Figure 2 As shown, the new energy bus charging scheduling based on the hybrid multi-objective particle swarm algorithm is characterized by the following steps.

[0042] S1. The onboard device records the remaining SOC, mileage of the day, and power consumption of each bus. The buses entering the charging station are evaluated one by one, and whether charging is needed is determined based on the current remaining power, the next day's shift schedule, and the mileage.

[0043] S2. Filter out the buses that need to be charged, calculate the time required for each bus to reach the expected SOC based on the output power of each charging gun per hour, and build a charging matrix;

[0044] S3. Parameter initialization: Randomly generate the initial population according to the control variables and constraints, set the population size and the maximum number of iterations. Construct the objective function (1) to minimize the maximum charging time and the objective function (2) to minimize the utilization deviation of the charging piles, and iteratively find the optimal values ​​of nighttime charging to obtain the Pareto frontier of the two objectives.

[0045] Objective function:

[0046]

[0047]

[0048] The output parameters are:

[0049] CAP i The battery capacity of the electric bus numbered i;

[0050] ε j Charging efficiency of charging pile;

[0051] The total amount of electricity required for the operation of the electric bus numbered i;

[0052] c j The time the bus is charging at the charging station numbered j;

[0053] The first recorded remaining battery power percentage of the electric bus numbered j;

[0054] The constraints are:

[0055] s i,j The rechargeable status of the electric bus numbered j during period t: 0 means non-rechargeable, 1 means rechargeable;

[0056] minP j ≤P≤maxP max Ensure that the charging power of each bus is within the allowed range of the charging pile.

[0057] S4, the update particle uses the cross and compile strategy to initialize the external archive set and define it as the global optimal solution set.

[0058] S5, calculate objective function 1 and objective function 2 and update the external archive, determine whether there is a dominance relationship between the newly generated particle i and the particles in the external archive, if particle i is non-dominant with other particles, add particle i to the external archive, otherwise continue to iterate.

[0059] S6. Using the roulette strategy, randomly select N particles to mutate into a new population.

[0060] S7. Select excellent individuals from external archives, choose a new optimal solution and continue iterating to generate a new population.

[0061] S8. Determine whether the program meets the termination requirement. When the maximum number of iterations is reached, the loop ends and the optimal solution set is output. Otherwise, return to step S4.

[0062] S9. Output the Pareto solution set, and output the night shift schedule for new energy buses according to the particles corresponding to the solutions.

[0063] S10, the server sets a time to collect the remaining SOC of buses at new energy bus stations every day and the next day's operation schedule, and takes the remaining SOC of vehicles in the environment, battery capacity, charging pile output power, and the next day's schedule as four-way information as input, and outputs the night charging schedule, and uses a hybrid multi-objective particle swarm algorithm to optimize the minimum charging time and the minimum charging pile utilization deviation of the constructed new energy bus, so as to achieve the minimum charging time and the minimum charging pile utilization deviation;

[0064] S11. The charging scheduling plan is completed, and the night charging schedule is returned to the server.

[0065] The design process of the present invention is as follows: first, the data set is screened and preprocessed, outliers in the data are deleted and filtered, and then the charging time of each bus at each charging gun is calculated, two objective functions of charging time and charging pile usage deviation are constructed, and two fitness functions are minimized. The two objective functions are optimized using a hybrid multi-objective particle swarm algorithm. System parameters are set and the objective function is optimized. The algorithm model is deployed on the server, and the algorithm is remotely called using gRPC, so that the bus night charging scheduling schedule can be obtained.

[0066] Figure 4 As shown, when the method is executed, the SOCstart and SOCend of each vehicle are first input, and the expected charging time of each vehicle at each charging pile is calculated according to the hourly charging power of each charging gun to obtain the charging matrix.

[0067] According to the system parameters, particle swarm parameters and external file set initialization, the bus charging matrix is ​​constructed, the optimal individual is initialized, and the fitness value, i.e., the charging time and charging pile utilization deviation, is calculated according to formula (1) and formula (2), and the optimal individual solution is stored in the external archive.

[0068] The solution in the solution space is cross-strategized and cross-strategy is used to generate new solutions. The fitness value of the particle is calculated according to formulas (1) and (2), which is the charging time and charging pile utilization deviation.

[0069] After the update, the individual optimal solution, global optimal solution and external archive set are updated according to whether the particle meets the optimal position. When the particle meets the optimal position, the individual optimal solution, global optimal solution and external archive set are updated according to the particle fitness value and dominance relationship. When the number of iterations iter is greater than the maximum number of iterations MCY, that is, Iter>MCY, the system outputs the Pareto solution set and then outputs the night shift schedule for new energy buses.

[0070] When the updated particle does not meet the optimal position, starting from the second particle, determine whether the i-th particle dominates the particles in the external file. If it is dominated, the i-th particle is not added to the file. If it is dominated, the first particle is directly stored in the external file, and the i-th particle is added to the external file. If it is not dominated, the first particle is directly stored in the external file, and the i-th particle is added to the external file.

[0071] Based on the above scheme, this embodiment uses the modified multi-objective particle swarm algorithm to optimize the minimum charging time and the minimum utilization deviation of the charging pile constructed for the new energy bus based on the remaining power, mileage, output power during charging, and the next day's operation schedule collected by the new energy bus and the on-board machine, so as to achieve the minimum charging time while minimizing the utilization deviation of the charging pile, generate a night charging schedule, and facilitate the scheduling of charging dispatchers at bus stations.

[0072] The above specific embodiments of the present invention are only used to illustrate or explain the principles of the present invention, and do not constitute a limitation of the present invention. Therefore, any modification, equivalent substitution, improvement, etc. made without departing from the spirit and scope of the present invention should be included in the protection scope of the present invention.

Claims

1. A new energy bus charging scheduling method based on a hybrid multi-objective particle swarm algorithm, characterized in that: First, the basic data set of bus charging stations is screened and preprocessed, and outliers in the data set are deleted and filtered. Then, the charging time of each bus at each charging gun is calculated, and two objective functions of charging time and charging pile usage deviation are constructed. The two objective functions are minimized, and the two objective functions are optimized using a hybrid multi-objective particle swarm algorithm. The system parameters are set, the objective functions are optimized, and the algorithm model is constructed using gRPC remote call algorithm service, which is deployed in the server to obtain the bus night charging scheduling schedule; the system parameters include the battery capacity of the electric bus, the charging efficiency of the charging pile, the total power required for the operation of the electric bus, the charging time of the charging pile, and the percentage of the remaining battery power of the electric bus recorded for the first time; The new energy bus charging scheduling method comprises the following steps: S1. The onboard device records the remaining SOC, mileage of the day, and power consumption of each bus. The buses entering the charging station are evaluated one by one, and whether charging is needed is determined based on the current remaining power, the next day's shift schedule, and the mileage. S2. Filter out the buses that need to be charged, calculate the time required for each bus to reach the expected SOC based on the output power of each charging gun per hour, and build a charging matrix; S3, parameter initialization: randomly generate the initial population according to the control variables and constraints, set the population size and the maximum number of iterations; construct the objective function (1) to minimize the maximum charging time and the objective function (2) to minimize the utilization deviation of the charging piles, and iteratively find the optimal values ​​of nighttime charging to obtain the Pareto frontier of the two objectives; Objective function: The output parameters are: CAP i The battery capacity of the electric bus numbered i; ε j Charging efficiency of charging pile; The total amount of electricity required for the operation of the electric bus numbered i; c j The time the bus is charging at the charging station numbered j; The first recorded remaining battery power percentage of the electric bus numbered j; The constraints are: s i,j The rechargeable status of the electric bus numbered j during period t: 0 means non-rechargeable, 1 means rechargeable; minP j ≤P≤maxP max Ensure that the charging power of each bus is within the allowed range of the charging pile; S4, update the velocity and position of the particle, initialize the external archive set, and define it as the global optimal solution set; S5, calculate objective function 1 and objective function 2 and update the external archive, determine whether there is a dominance relationship between the newly generated particle i and the particles in the external archive, if particle i is non-dominant with other particles, add particle i to the external archive, otherwise continue iteration; S6, using the roulette strategy, randomly select N particles to mutate into a new population; S7, select the excellent individuals in the external archive, choose the new optimal solution and continue to iterate to generate a new population; S8, determine whether the program meets the termination requirement, when the maximum number of iterations is reached, the loop ends and the optimal solution set is output, otherwise, return to step S4; S9. Output the Pareto solution set, and output the night shift schedule for new energy buses according to the particles corresponding to the solutions.

2. The new energy bus charging scheduling method according to claim 1 is characterized in that: The server sets a time to collect the remaining SOC of buses at new energy bus stations every day and the next day's operation schedule. The four pieces of information in the environment, namely the remaining SOC of vehicles, battery capacity, output power of charging piles, and the next day's schedule, are used as input, and the output is the night charging schedule. The hybrid multi-objective particle swarm algorithm is used to optimize the minimum charging time and the minimum utilization deviation of charging piles for the constructed new energy buses, so as to achieve the minimum charging time and the minimum utilization deviation of charging piles.

3. The new energy bus charging scheduling method according to claim 1 or 2, characterized in that: The charging scheduling plan is completed and the night charging schedule is returned to the server.

Images