An Electric Bus Coordination Optimization Scheduling Method

Abstract

The present invention discloses an electric bus coordinated optimization scheduling method, belonging to the technical field of intelligent buses. The present invention comprehensively optimizes the electric bus scheduling strategy from two scales of time and space, establishes a two-layer bus scheduling planning model considering vehicle capacity, transfer problems, and the characteristics of electric buses, and solves the model according to the genetic algorithm. The present invention can generate an electric bus scheduling strategy covering both time and space, making the scheduling strategy more in line with the actual passenger flow situation and more practical benefits.

Description

Technical Field

[0001] The present invention relates to the technical field of smart public transportation, and in particular to a coordinated optimization and dispatching method for electric public transportation. Background Art

[0002] Public transport passenger flow has obvious time and space peak characteristics. Fixed and single departure plans are difficult to meet the passenger flow needs of the line network. It is necessary to re-compile driving plans for different time periods and station intervals, adjust the departure arrangements between different lines according to the transfer passenger flow, and coordinate the dispatch of the road network fleet. Specifically, it is necessary to consider the number of station passenger flow, the total service cost of public transport, etc., and implement targeted dispatching strategies such as skipping stations and adding vehicles.

[0003] However, most of the existing research on bus dispatching focuses on conventional buses and cannot adapt to the characteristics of electric buses. Due to the promotion of new energy vehicles in my country, the use of pure electric buses in most urban bus operations has gradually become the mainstream. Therefore, in dispatching, while constraining bus operation specifications, it is necessary to consider the operating characteristics of electric buses, such as battery replacement costs, battery consumption due to time-varying passenger numbers, and the determination of battery replacement needs. Based on real-time passenger flow, a dispatching optimization model is established to solve the optimal operation plan and improve the benefits of operators and passengers.

[0004] Domestic and foreign research on electric bus operation and scheduling mainly focuses on pure electric buses in real-time charging mode. However, a large number of cities also use battery-swap electric buses, which have the characteristics of stable electricity prices, fast operation without queuing, and low power waste. There are few studies on bus operation optimization based on its battery-swap characteristics and battery pack costs. For line network vehicle scheduling, existing solutions mostly study the single direction of bus departure timetables or driving mode problems, or combine the selection of service stations with departure frequency for static scheduling, so that the time dimension and space dimension of line network departure arrangements cannot be well combined for simultaneous optimization and seeking the global optimum. The existing bus multi-mode collaborative optimization cannot output a complete operation schedule, and it is difficult to dynamically fit the real-time changes in passenger flow. How to perform highly integrated dynamic collaborative scheduling in the multi-mode combination departure problem under the background of the line network still needs to be explored. Summary of the invention

[0005] In view of the shortcomings and deficiencies of the prior art, the object of the present invention is to provide a coordinated and optimized dispatching method for electric buses.

[0006] The electric bus coordinated optimization dispatching method disclosed in the present invention comprises the following steps:

[0007] Data processing steps: cleaning and preprocessing the bus operation data, route data, and passenger flow data of electric buses, and counting the passenger flow traffic start and end point data;

[0008] Steps for solving the optimized scheduling plan: Based on the data obtained in the steps of data processing, a bi-level programming model is constructed, and the genetic algorithm is used to solve it to obtain the optimal bus departure interval and the optimal bus stop plan. The bi-level programming model consists of an upper-level model and a lower-level model. The upper-level model optimizes the bus stop plan to make the service time of all vehicles as short as possible and the operating energy consumption cost reach the minimum. Its objective function is as follows:

[0009]

[0010] Among them, 、 are two normalized weighting coefficients, and ; is the total path travel time of all vehicles; is the fixed service cost of a single vehicle for one trip on line l ; The value of represents whether the l th vehicle on line k skips the stop at station i . Here, 1 means not skipping the stop, and 0 means skipping the stop; is the stopping time required for the l th vehicle on line k at station i ; is the number of boarding passengers of the l th vehicle on line k at station i ; represents whether the l th vehicle on line k needs to replace the battery after executing this shift. If the battery needs to be replaced = 1, otherwise = 0; represents the single battery replacement cost of pure electric buses; L is the total number of lines in the bus road network, I is the total number of stations on the corresponding bus line, K is the total number of trips of electric buses in the corresponding bus line;

[0011] The constraint conditions of the upper-level model are:

[0012]

[0013]

[0014]

[0015]

[0016] Among them, is the departure interval between the l th vehicle on the line and the previous vehicle; k The value of represents whether the th - 1 vehicle on the line skips the station l at station k ; i represents the coefficient of energy consumption utility per unit of electricity during the operation of the th vehicle on the line; l represents the battery power required for the vehicle to operate continuously with full load for a single shift, k represents the average travel distance of all passengers; e is the battery swapping time for pure - electric buses; d

[0017] The lower - layer model optimizes the bus departure interval and regulates the departure quantity during peak hours to reduce the total waiting time of passengers. Its objective function is as follows:

[0018]

[0019] Among them, is the number of passengers who want to board the l th vehicle at station k and get off at station i on the line; j is the travel time of passengers whose trip start and end points are and who board the l th vehicle on the line; k is the waiting time of passengers whose trip start and end points are ij and who board the th vehicle on the line; l is the number of stranded passengers who cannot board the k th vehicle at station ij on the line due to limited remaining capacity; is the headway between the l th + 1 vehicle and the previous vehicle at station i on the line; k The value of represents whether the th + 1 vehicle on the line skips the station l at station k ; i The value of represents whether the th + 1 vehicle on the line skips the station l at station k ; i l k j ​ For the line l the k time headway between the i +(2)nd vehicle and the head of the previous vehicle at the station;

[0020] The constraint conditions of the lower-layer model are:

[0021]

[0022]

[0023]

[0024]

[0025] Among them, , are the minimum and maximum departure intervals respectively; is the transfer time required for the passengers on the k rd vehicle who need to transfer from line l to line m ; is the remaining capacity of the k st vehicle at the station l on line i , C is the rated passenger capacity of the vehicle; M is the network dispatching duration.

[0026] Furthermore, the total path travel time of all vehicles is calculated according to the following formula :

[0027]

[0028] Among them, is the vehicle travel time from the i -(1)st station to the i th station; θ is the acceleration or deceleration time of the vehicle at the stopping station.

[0029] Furthermore, the number of passengers who want to board the l rd vehicle at the k th station and get off at the i th station on line j is calculated according to the following formula :

[0030]

[0031]

[0032]

[0033]

[0034]

[0035]

[0036] Among them, is the arrival rate of passengers who want to board at station l and get off at station i on the line; j the headway between the th vehicle and the previous vehicle at station l on the line; k The value of i indicates whether the th vehicle on the line l skips station k -1 at station i ; The value of l indicates whether the k th vehicle on the line j skips station -1 at station l ; k is the number of passengers who want to board the i th vehicle at station j and get off at station on the line; l The time when the k th vehicle arrives at station i on the line; The time when the l th vehicle k -1 arrives at station i on the line; The time when the l th vehicle k leaves station i -1 on the line; The value of l indicates whether the k th vehicle on the line i -1 skips station ; l The time when the k th vehicle leaves station i on the line.

[0037] Furthermore, calculate the number of boarding passengers l of the k th vehicle at station i on the line according to the following formula :

[0038]

[0039]

[0040]

[0041]

[0042] Among them, is the remaining capacity of the k-1 th vehicle at the stop l on the line i ; is the number of passengers getting off at the stop l on the line k by the i th vehicle.

[0043] Furthermore, calculate the waiting time l of the passengers boarding the k th vehicle with the origin and destination of ij on the line according to the following formula:

[0044]

[0045]

[0046] Among them, is the probability of passengers on the line l transferring to the line m ; is the time when the m th vehicle arrives at the stop p on the line i .

[0047] Furthermore, calculate the number of stranded passengers l who cannot board the i th vehicle due to limited remaining capacity at the stop k on the line according to the following formula:

[0048]

[0049] Furthermore, the method for obtaining the optimal bus departure interval and the optimal bus stop plan by solving through a genetic algorithm includes the following steps:

[0050] Initialization parameter step: Set the maximum number pop of the population size, randomly generate individuals of the bus departure interval and the bus stop plan, then set the maximum number of generations max, and set the generation counter to 1;

[0051] Coding and initial solution steps: Code the variable headway and stop plan, and use random initial values to form the genes of the chromosome; if the variable meets the constraint conditions, go to the calculation and selection steps; if not, the initial solution should be regenerated;

[0052] Calculation and selection steps: Calculate the fitness values of all chromosomes, and select chromosomes by the roulette wheel method. If the fitness of the chromosomes in this generation is higher than that of the previous generation, the chromosomes in this generation should be retained as the current best solution; if it is lower than the previous generation, then discard the selection of the chromosomes in this generation;

[0053] Reproduction steps: The current chromosomes generate the next generation of individuals through crossover and mutation behaviors; if each individual meets the constraint conditions, go to the stop steps; if not, reproduce the individuals again;

[0054] Stop steps: If the current number of evolutionary generations is equal to max, stop the loop and obtain the optimal solution; if not, return to the calculation and selection steps.

[0055] The beneficial effects of the present invention are as follows: The present invention comprehensively optimizes the electric bus dispatching strategy from two scales of time and space, and can obtain a complete, directly feasible operation time and space scheduling plan for road network vehicles, including two spatio-temporal dimensions, which is more convenient and efficient. Moreover, a highly coupled and tight whole is formed among different levels of the model, and it has better global optimality than an isolated optimization model. The present invention takes into account the timetable optimization problem in electric bus dispatching, has the characteristics of real-time response, can realize the dynamic control of bus dispatching, has stronger robustness in dealing with the actual vehicle capacity update, has better fitting for the passenger flow changes of the line network, and has better matching and adaptability to irregular passenger flows compared with the traditional timetable with equal headway, making the capacity delivery more accurate, thus reducing the waste of resources. The present invention establishes an electric bus dynamic collaborative dispatching model with vehicle capacity constraints added, recalculating the waiting time of transfer passengers, and considering the energy consumption cost of electric buses, making the model more accurate and practical, optimizing the line departure plan, greatly reducing the costs of the operator and the travel time of passengers, making the connection of driving arrangements between multiple lines stronger, and the convenience of passenger transfer higher. The present invention solves the problem of mismatch between dynamic passenger flow and bus capacity, makes the dispatching strategy more in line with the actual passenger flow situation and more practical, and has a wide application prospect in urban bus line networks Description of the Drawings

[0056] Figure 1 It is a schematic structural diagram of the bilevel programming model used in the present invention.

[0057] Figure 2 It is a schematic flow diagram of the genetic optimization algorithm used in the present invention. Detailed Embodiments

[0058] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0059] In one embodiment of the present invention, a two - layer programming model as shown in Figure 1 is mainly used to solve the coordinated optimization scheduling of electric buses, specifically including the following steps:

[0060] S1. Clean and pre - process the bus operation data, line data, and passenger flow data required for the model algorithm, and further statistically calculate the passenger flow OD (origin - destination) data based on the IC card data;

[0061] S2. Design the expression forms of various variables required for the model (such as the total path travel time of all vehicles , the number of people who fail to board due to skipping stops , the waiting time of passengers , the number of stranded passengers , the headway , the number of boarding passengers , the number of alighting passengers , whether to change the battery , etc.);

[0062] S3. Based on the data in S1 and the variable expression forms designed in S2, construct the upper - layer model in the two - layer programming model. In the upper - layer model, by optimizing the bus stop plan, the service time of all vehicles is made as short as possible and the operating energy consumption cost is minimized;

[0063] S4. Based on the data in S1 and the variable expression forms designed in S2, construct the lower - layer model in the two - layer programming model. In the lower - layer model, by optimizing the bus departure interval and regulating the departure quantity during peak hours, the total waiting time of passengers is expected to be reduced;

[0064] S5. Initialize the algorithm parameters, design a genetic algorithm for solving based on the two - layer programming model established in S3 and S4, and finally obtain the optimal bus scheduling plan.

[0065] Furthermore, the specific content of S2 includes:

[0066] S201. Calculate the total path travel time of all vehicles :

[0067]

[0068] Among them, is the vehicle travel time from i -1 stop to i stop; θ is the acceleration or deceleration time of the vehicle when stopping at the station; The value of represents the line l at the kThe vehicle at the station i Whether it skips a station, where 1 means it does not skip and 0 means it skips.

[0069] S202. Calculate the calculation line l The number of passengers who want to board the k vehicle at i the station and get off at j the station :

[0070]

[0071]

[0072]

[0073]

[0074]

[0075]

[0076] Among them, is the arrival rate of passengers who want to board at l the station and get off at i the station on line j ; is the headway between the l th k vehicle and the previous vehicle at i the station on line The value of l indicates whether the k th - 1 vehicle on line i skips the station; The value of l indicates whether the k th - 1 vehicle on line j skips the station; is the number of passengers who want to board the l th - 1 vehicle at k the station and get off at i the station on line j ; is the arrival time of the l th k vehicle at i the station on line is the arrival time of the l th k th - 1 vehicle at i the station on line is the time when the l th k vehicle leaves the stationi The moment of -1; The value represents the line l On the k Whether the i vehicle at stop For the line l On the k The moment when the i vehicle departs from stop For the line l On the k The headway between the For the line l On the k The i vehicle needs to stop at stop

[0077] S203. Calculate the number of boarding and alighting passengers of the l On the k vehicle at stop i : , :

[0078]

[0079]

[0080]

[0081]

[0082] Among them, Is the k remaining capacity of the l vehicle on line i at stop Is the k-1 remaining capacity of the l vehicle on line i at stop For the line l On the k The i number of alighting passengers of the vehicle at stop C Is the rated load of the vehicle.

[0083] S204. Calculate the energy consumption cost of electric buses during the period

[0084]

[0085] For pure electric bus vehicles, the energy consumption cost is specifically the battery swapping cost of the bus in this study, and the real-time passenger load is related to battery loss. Among them, Represents the single battery replacement cost of a pure electric bus (yuan / time). Indicates that after the bus completes the shift, if the battery needs to be replaced = 1, otherwise = 0;

[0086] S205. Calculate the waiting time of passengers at the OD points (origin and destination of traffic) of the l th vehicle on the route k : ij For passengers on the route :

[0087]

[0088]

[0089] Among them, Is the probability of passengers on route l transferring to route m ; Is the arrival time of the m th vehicle at station p on route i . Is the transfer time required for passengers on the k th vehicle who need to transfer from route l to route m .

[0090] S206. Calculate the number of stranded passengers at station l on route i who cannot board the k th vehicle due to limited remaining capacity :

[0091]

[0092] Furthermore, the specific steps of S3 are as follows:

[0093] S301. Based on the data of S1 and the variable expression forms designed in S2, construct the objective function of the upper-level model:

[0094]

[0095] Is the fixed service cost of a single vehicle for a round trip on route l .

[0096] For this kind of bi-objective nonlinear optimization problem, it can be calculated uniformly by taking different weight values for the two objectives, thereby achieving the effect of reducing the computational amount. 、 Are the normalized weighting coefficients of the two objectives respectively, and 。

[0097] S302. Add corresponding constraint conditions to the objective function based on S301:

[0098] To ensure the headway and avoid driving conflicts with the previous vehicle, the travel time reduced by a vehicle through skipping stops during a trip cannot exceed the departure interval of this shift. Therefore, add the following constraint to the objective function:

[0099]

[0100] To prevent the situation that a certain stop is continuously skipped, resulting in too long waiting time for some passengers or even being unable to board the vehicle, add constraints to ensure that each stop will not be continuously skipped:

[0101]

[0102] The battery energy consumption of pure - electric vehicles is related to factors such as the speed and load during vehicle operation. After each shift is completed, determine whether battery swapping is required, and perform the following constraints:

[0103]

[0104] Among them, represents the unit - electricity - consumption utility coefficient during the operation of pure - electric buses, e represents the battery power required for a vehicle to operate a single shift with continuous full load, d represents the average riding distance of all passengers.

[0105] All buses operating in the research area are pure - electric buses, and the battery - swapping mode is adopted for charging. The average battery - swapping time is 10 min / vehicle. To prevent vehicle failures, it is necessary to ensure that the departure time interval between adjacent shifts is greater than the battery - swapping time. The constraint formula is as follows:

[0106]

[0107] Among them is the battery - swapping time of pure - electric buses.

[0108] Furthermore, the specific content of S4 includes:

[0109] S401. Based on the data of S1 and the variable expression form designed in S2, construct the objective function of the lower - layer model:

[0110]

[0111] Among them, is the line l on which thek The starting and ending points of the vehicle traffic are ij the boarding time of the passengers.

[0112] S402. Add corresponding constraint conditions to the objective function based on S301:

[0113]

[0114] Among them, , are the minimum and maximum headways respectively, obtained from historical data;

[0115] The transfer behavior at the transfer point is only certified within one maximum headway. Otherwise, it is regarded as two independent boarding behaviors and no separate calculation constraint is applied. Therefore, add the constraint:

[0116]

[0117] To ensure that the remaining capacity is within a reasonable range, set the constraint:

[0118]

[0119] To ensure that there are continuously vehicles in service on the line within the operation time range, add the constraint:

[0120]

[0121] Among them, M is the network scheduling duration.

[0122] Furthermore, the S5 specifically includes the following steps, as Figure 2 shown:

[0123] S501. Initialize parameters. Set the maximum number pop of the population size, and randomly generate individuals of the headway and stop plan. Then set the maximum number of evolutionary generations max, and set the evolutionary generation counter to 1.

[0124] S502. Coding and initial solution. Encode the variables of the headway and stop plan, and form the genes of the chromosome with random initial values. If the variables meet the constraint conditions, go to S503. If not, the initial solution should be regenerated.

[0125] S503. Calculation and selection. Calculate the fitness values of all chromosomes. Select chromosomes by the roulette wheel method. If the fitness of the chromosomes in this generation is higher than that of the previous generation, keep the chromosomes in this generation as the current best solution. If it is lower than the previous generation, then abandon the selection of the chromosomes in this generation.

[0126] S504, Reproduction. The current chromosome generates the next generation of individuals through crossover and mutation behaviors. If each individual meets the constraint conditions, go to S505; if not, reproduce individuals again.

[0127] S505, If the current generation number of evolution is equal to max, stop the loop and obtain the optimal solution. If not, return to step S503.

[0128] The above are only the preferred embodiments of the present invention and do not constitute a limitation on the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the claims of the present invention.

Claims

1. An electric bus coordinated optimization scheduling method, characterized in that, Including the following steps: Steps of data processing: cleaning, preprocessing the bus operation data, route data, and passenger flow data of electric buses, and counting the starting and ending points of passenger flow traffic; Steps of solving the optimized dispatching plan: Based on the data obtained in the steps of data processing, construct a bi-level programming model, and solve it through a genetic algorithm to obtain the optimal bus departure interval and the optimal bus stop plan. The bi-level programming model consists of an upper-level model and a lower-level model. Among them, the upper-level model optimizes the bus stop plan to make the service time of all vehicles as short as possible and the operating energy consumption cost reach the lowest. Its objective function is as follows: Among them, α and β are two normalized weighting coefficients respectively, and α + β = 1; Δ is the total route travel time of all vehicles; Ql is the fixed service cost for a single vehicle trip on route l; The value of represents whether the k-th vehicle on route l skips stop i. Here, 1 means not skipping the stop, and 0 means skipping the stop; is the dwell time required for the k-th vehicle on route l at stop i; is the number of boarding passengers of the k-th vehicle on route l at stop i; represents whether the k-th vehicle on route l needs to replace the battery after executing this shift. If the battery needs to be replaced otherwise C battery represents the single battery replacement cost of a pure electric bus vehicle; L is the total number of routes in the bus road network, I is the total number of stops of the corresponding bus routes, and K is the total number of trips of electric buses in the corresponding bus routes; The constraint conditions of the upper-level model are: Among them, H l，k is the departure interval between the k-th vehicle and the previous vehicle on line l; The value of indicates whether the (k - 1)-th vehicle on line l skips a stop at station i; represents the unit power consumption utility coefficient during the operation of the k-th vehicle on line l; e represents the battery power required for the vehicle to run continuously with full load for a single shift, d represents the average riding distance of all passengers; t b is the battery swapping time for pure electric bus vehicles; The lower-level model optimizes the bus departure interval and regulates the departure quantity during peak hours in order to reduce the total waiting time of passengers. Its objective function is as follows: Among them, is the number of passengers who want to take the k-th vehicle on line l, get on at station i and get off at station j; is the travel time of passengers on line l whose transportation starting and ending points are ij when taking the k-th vehicle; is the waiting time of passengers on line l whose transportation starting and ending points are ij when taking the k-th vehicle; is the number of stranded passengers who fail to take the k-th vehicle at station i on line l due to limited remaining capacity; is the headway between the (k + 1)-th vehicle and the previous vehicle at station i on line l; The value of indicates whether the (k + 1)-th vehicle skips station i on line l; The value of indicates whether the (k + 1)-th vehicle skips station j on line l; is the headway between the (k + 2)-th vehicle and the previous vehicle at station i on line l; The constraint conditions of the lower-level model are: where h min and h max are the minimum and maximum headways respectively; is the transfer time required for the passengers on the k-th vehicle who need to transfer from line l to line m; is the remaining capacity of the k-th vehicle at stop i on line l, C is the rated passenger capacity of the vehicle; M is the network scheduling duration; The steps of solving the optimal bus departure interval and the optimal bus stop plan through a genetic algorithm include the following steps: Initializing parameters step: Set the maximum number pop of the population size, randomly generate individuals of the bus departure interval and the bus stop plan, then set the maximum number of generations max, and set the generation counter to 1; Encoding and initial solution step: Encode the variables of the departure interval and the stop plan, and form the genes of the chromosome with random initial values; if the variables meet the constraint conditions, go to the calculation and selection step; if not, the initial solution should be regenerated; Calculation and selection step: Calculate the fitness values of all chromosomes, select chromosomes through the roulette wheel method. If the fitness of the chromosomes in this generation is higher than that of the previous generation, keep the chromosomes in this generation as the current best solution; if it is lower than the previous generation, then give up selecting the chromosomes in this generation; Reproduction step: The current chromosomes generate the next generation of individuals through crossover and mutation behaviors; if each individual meets the constraint conditions, go to the stop step; if not, reproduce the individuals again; Stop step: If the current generation number is equal to max, stop the loop and obtain the optimal solution; if not, return to the calculation and selection step.

2. The electric bus coordinated optimization scheduling method according to claim 1, characterized in that, Calculate the total path travel time Δ of all vehicles according to the following formula: Among them, δ i-1，i is the vehicle travel time from station i - 1 to station i; θ is the acceleration or deceleration time of the vehicle at the stop.

3. The electric bus coordinated optimization scheduling method according to claim 2, characterized in that, Calculate the number of passengers who want to take the k-th vehicle on line l, board at station i, and get off at station j according to the following formula H l，1 ＝0 Among them, λ l，ij is the arrival rate of passengers who want to board at station i and alight at station j on line l; is the headway between the k-th vehicle and the previous vehicle at station i on line l; The value of indicates whether the (k - 1)-th vehicle on line l skips station i; The value of indicates whether the (k - 1)-th vehicle on line l skips station j; is the number of passengers who want to board the (k - 1)-th vehicle at station i and alight at station j on line l; is the arrival time of the k-th vehicle at station i on line l; is the arrival time of the (k - 1)-th vehicle at station i on line l; is the departure time of the k-th vehicle from station i - 1 on line l; The value of indicates whether the k-th vehicle on line l skips station i - 1; is the departure time of the k-th vehicle from station i on line l.

4. The electric bus coordinated optimization scheduling method according to claim 3, characterized in that, Calculate the number of passengers boarding the k-th vehicle on line l at station i according to the following formula wherein, is the remaining capacity of the (k - 1)-th vehicle at station i on line l; is the number of passengers getting off the k-th vehicle at station i on line l.

5. The electric bus coordinated optimization scheduling method according to claim 3, characterized in that, Calculate the waiting time of passengers boarding the k-th vehicle on line l with the origin and destination being ij according to the following formula Among them, ζ l，m is the probability that passengers on line l transfer to line m; is the arrival time of the p-th vehicle on line m at station i.

6. The electric bus coordinated optimization scheduling method according to claim 4, characterized in that Calculate the number of stranded passengers at station \(i\) on line \(l\) who fail to board the \(k\)-th vehicle due to limited remaining capacity according to the following formula

Images