A train fast and slow train operation scheme intelligent recommendation method considering capacity constraints

Abstract

The application provides a train express and slow train operation scheme intelligent recommendation method considering capacity constraints. The method firstly initializes train information of a line operation direction, expected passenger flow demand data OD, then generates an initial scheme, and sets train continuous cross-station constraint checking and station continuous jump-stop train number constraint checking, and finally solves the model through a genetic algorithm optimization. The application can intelligently recommend a scheme with balanced train full load rate under the premise of minimizing passenger waiting time, so as to save passenger travel time and improve travel quality.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent control of rail transit, and in particular to a train fast / slow train scheduling scheme intelligent recommendation method considering capacity constraints. BACKGROUND

[0002] The fast / slow train operation organization of urban rail transit is to simultaneously run the station-station stopping slow train and the cross-station fast train according to the long / short distance passenger flow characteristics and the through capacity utilization status of the line, so as to adapt the transport organization to the passenger flow characteristics. The fast train is a train stopping at part of the stations, and the slow train is a train stopping at every station in the train operation range. In view of the difference of passenger flow demand, the fast / slow train operation organization can meet the passengers with different travel demands: the fast train does not stop at part of the stations, thereby improving the travel speed and meeting the rapid demand of long-distance passengers; the slow train stops at every station, thereby meeting the direct demand of the passenger flow along the line.

[0003] The existing technologies mostly take minimizing the total travel time of all passengers or (and) the operation cost of the operation enterprise as the target. However, in the model establishment process, the scenarios considered by various models and algorithms are different. For example: 1) some technologies only consider two running modes of the fast train and the slow train, i.e. one fast train and one slow train. This limits the diversification of the running mode; 2) some technologies consider the situation of once left passengers due to the cross-station of the fast train, and consider the additional waiting time of the passengers in the optimization target, but do not consider the situation of the passengers left due to the train capacity limitation, so it cannot accurately depict the problem of the increase of the travel time caused by the left passengers of the large passenger flow line.

[0004] In the network stage of urban rail transit, the train load rate is large during the peak period, and some passengers cannot get on the train due to the train capacity limitation at some stations, so some passengers need to wait for the subsequent train or trains to get on the train. Therefore, the existing algorithm cannot solve the problem of the left passengers caused by the capacity limitation, and obviously cannot meet the demand of the scheduling scheme compilation of the large passenger flow line during the peak period. Under the condition of the network large passenger flow demand, the scheduling scheme optimization considering the capacity limitation and minimizing the passenger travel time has outstanding economic and social value.

[0005] The fast / slow train scheduling of urban rail transit through the cross-station operation mode is a train scheduling scheme adopted to match the imbalance of the passenger flow space-time distribution, under which the train skips part of the stations through the cross-station mode and reserves the train carrying capacity, so as to achieve the purpose of quickly relieving the passenger flow at the large passenger flow station, reducing the waiting time of the passengers and reducing the operation cost. The existing methods mostly optimize the modeling by minimizing the waiting time of the passengers and reducing the operation cost of the enterprise, but ignore the imbalance of the train load rate and the left passengers at the station caused by the train capacity problem. SUMMARY

[0006] The application provides a train fast / slow train scheduling scheme intelligent recommendation method considering capacity constraints, which can intelligently recommend a scheme with balanced train load rate under the premise of minimizing passenger waiting time, so as to save passenger travel time and improve travel quality.

[0007] A train fast / slow train scheduling scheme intelligent recommendation method considering capacity constraints, which comprises the following steps:

[0008] Step 1, initializing train information of a line scheduling direction and expected passenger flow demand data OD, and the specific content comprises:

[0009] Step 1.1, initializing infrastructure information

[0010] Train route information: V={1, 2, …, N} is an ordered number set of stations in a certain train direction of urban rail transit, and N is the number of stations on the route;

[0011] Train route information: k={1, 2, …, K} is an ordered departure number of trains in a research period T, and K is the total number of departures;

[0012] (1.2) Determine passenger flow demand information

[0013] The passenger flow from station i to station j is q ij , the total passenger demand at station i is and j>i; the total arrival amount at station j is and j>i;

[0014] Calculate the passenger arrival rate, and the passenger arrival rate ρ ij in the research period T is calculated by the following formula:

[0015]

[0016] Wherein, q ij is the total passenger flow between station i and station j in the research period T;

[0017] Calculate the passenger arrival amount, the departure interval is h, and the passenger amount arriving at station i before train k arrives after train k-1 leaves is:

[0018] q ijk =ρ ij h

[0019] Step 2, generate an initial scheme

[0020] For a line with N stations, the stop scheme of a single train is denoted as Ω *, if the first station departs at a fixed interval h in the research period T, the number of trains in the research period T

[0021] Step 2.1 Station stopping scheme initialization

[0022] y ik represents whether the train k stops at the station i, y ik =1 when the train k stops at the station i, otherwise y ik =0; if the station i must stop, y ik =1 is initialized, and the feasible solution is coded.

[0023] Step 2.2, initialization of the first station departure time of each train

[0024] According to the principle of equal interval departure of the first station, the first station departure time of each train in each scheme is initialized.

[0025] Step 2.3, calculation of the arrival time of each train

[0026] According to the departure time of the first station and the train running time parameters, the arrival time, departure time and travel time of each train at each station are calculated.

[0027] Step 3, constraint condition check

[0028] Step 3.1 train k continuous cross-station constraint check

[0029] Ensure that the maximum number of consecutive skipped stations of train k is t max .

[0030]

[0031] Step 3.2, constraint check of the number of consecutive skipped trains at station i

[0032] The maximum number of consecutive skipped trains at station i is s max .

[0033]

[0034] Step 3.3, filtering Ω in step 3.1 and 3.2 according to constraint conditions 3.1 and 3.2 * The schemes that do not meet the cross-station and skipping conditions in Ω are filtered to obtain the initial scheme Ω.

[0035] Preferably, the method further comprises:

[0036] Step 4, optimization solution by genetic algorithm

[0037] Step 4.1, set the evolution generation counter c=0, and set the maximum evolution generation as I max .

[0038] Step 4.2 randomly select m individuals (i.e. feasible solutions) from the initial population P(c) to form a new population P(c+1) according to the initial scheme Ω * Step 4.3 selection operation: check each individual in the population P(c) according to the "constraint condition" to see if it meets the constraint; if it meets the constraint, proceed to step 4.4, otherwise, repeat step 4.2;

[0039] Step 4.3 selection operation: check each individual in the population P(c) according to the "constraint condition" to see if it meets the constraint; if it meets the constraint, proceed to step 4.4, otherwise, repeat step 4.2;

[0040] Step 4.4 crossover operation:

[0041] Step 4.4.1 randomly select x individuals from P(c) and perform the same encoding position exchange encoding operation with a certain probability to obtain x new individuals;

[0042] Step 4.4.2 check the crossover scheme obtained in step 4.4.1 according to the "constraint condition" to see if it meets the constraint. If it meets the constraint condition, proceed to step 4.5, otherwise, repeat step 4.4.1;

[0043] Step 4.5 mutation operation:

[0044] Step 4.5.1 randomly perform partial encoding mutation operation on y (y = m - x) individuals with a certain probability according to certain rules. That is, change 1 to 0 and change 0 to 1;

[0045] Step 4.5.2 check the mutation scheme obtained in step 4.5.1 according to the "constraint condition" to see if it meets the constraint. If it meets the constraint condition, proceed to step 4.6, otherwise, repeat step 4.5.1;

[0046] Step 4.6 calculate the fitness of the population P(c), the x individuals formed after crossover, and the y individuals formed after mutation, i.e. the total travel time of passengers under each scheme, which is calculated as follows:

[0047] Step 4.6.1 calculate the passenger flow index of each individual scheme

[0048] Step 4.6.1 calculate the passenger flow index of each individual scheme

[0049] Step 4.6.2 calculate the total travel time of the scheme;

[0050] Step 4.6.3 calculate the fitness of the population P(c) and the x individuals formed after crossover and the y individuals formed after mutation. Select the m individuals with the best fitness as the new population P(c); at this time, if c < I max , then let c = c + 1 and repeat step 4.4; otherwise, end the algorithm, and the individual with the highest fitness in P(c) is the optimal scheme.

[0051] Preferably, the specific way of calculating the passenger flow index of each individual scheme in step 4.6.1 is as follows:

[0052] a. Calculate the number of passengers getting off

[0053] Assuming that the passengers in train k getting off at station i are in proportion to the total amount of passenger flow attracted by subsequent station j, then:

[0054]

[0055] where the total amount of passengers getting off at station j of train k is

[0056] b. Calculate the number of passengers getting on

[0057] The number of passengers getting on is considered in three situations according to the train stopping condition and the passenger carrying condition, i.e.:

[0058]

[0059] c. Calculate the number of passengers staying on

[0060] Only when v ik -a ij,k +w ij,k > v * , there are passengers staying on, and the number of passengers staying on is:

[0061] l ijk = w ijk -b ijk = w ijk -(v * -(v ik -a ijk ))

[0062] d. Calculate the number of passengers getting on

[0063] The number of passengers getting on is considered in three situations according to the train stopping condition and the passenger carrying condition, i.e.:

[0064] v ik ′= v i,k+1 = v ik +b ij,k -a ij,k

[0065] When y ik = 0, v ik = v ik ′= v i,k+1 ;

[0066] e. Calculate the number of passengers waiting

[0067] The number of passengers waiting at station i when train k arrives is composed of two parts: the number of passengers staying on l ij,k-1passengers arriving at or transferring at station i before train k arrives;

[0068] Let ρ ik denote the passenger arrival rate before train k arrives, then the arrival or transfer passenger flow during the headway h between two trains is ρ ik h, then the number of waiting passengers at station i when train k arrives is:

[0069] w ijk = l ij,k-1 + ρ ik .

[0070] Preferably, the specific way of calculating the total travel time in step 4.6.2 is as follows:

[0071] a. Calculate the waiting time

[0072] The waiting time includes two parts: the waiting time of passengers arriving at or transferring at station i, and the waiting time of passengers remaining at station i;

[0073] (i) Assuming that the time of passengers arriving at station i obeys uniform distribution, then the waiting time of passengers arriving at or transferring at station i from the time train k-1 departs to the time train k arrives is

[0074] (ii) The passengers remaining at station i when train k-1 departs need to wait for a complete headway h;

[0075] b. Calculate the travel time

[0076] The travel time of passengers includes the running time in the section and the stopping time at station;

[0077] The travel time of train k from station i to station j is:

[0078]

[0079] c. Calculate the total travel time of passengers

[0080] The optimization objective is to minimize the total travel cost of passengers, i.e. the sum of the travel time and the waiting time

[0081]

[0082] Preferably, the step 4.6.3 contains the following constraints:

[0083] Condition (i) special station stopping constraint

[0084] The starting station must stop: y 1k = 1 k = 1, 2, …, K

[0085] The terminal station must stop: y Nk= 1 k = 1,2,..., K

[0086] If all the trains in P(c) satisfy the above constraints, go to step 4.3.3; otherwise, repeat step 4.3.1.

[0087] Condition (ii) is the headway constraint between trains

[0088] The headway between train k-1 and train k should satisfy:

[0089] a ik -d i,k-1 ≥y ik y i,k-1 h i1 +(1-y i,k-1 )y ik h i2 +y i,k-1 (1-y ik )h i3 +(1-y i,k-1 )(1-y ik )h i4

[0090]

[0091]

[0092] where h i1 , h i2 , and h i3 are the minimum headways between trains k-1 and k when neither train crosses the station, the first train crosses the station, and the second train crosses the station, respectively.

[0093] If all the trains in P(c) satisfy the above constraints, go to step 4.3; otherwise, repeat step 4.3.1. BRIEF DESCRIPTION OF DRAWINGS

[0094] Figure 1 is a coding diagram of a feasible solution. DETAILED DESCRIPTION

[0095] To facilitate the depiction of the interaction between passenger flow and train flow, the following definitions are made:

[0096] b ijk is the passenger flow from station i to station j on train k;

[0097] a ijk is the passenger flow from station i to station j on train k;

[0098] l ijk is the passenger flow from station i to station j but not on train k;

[0099] v ik — the passenger load of train k when it arrives at station i;

[0100] v ik — the passenger load of train k when it departs from station i;

[0101] w ijk — the waiting passenger flow from station i to station j;

[0102] v * — the upper limit of train passenger load;

[0103] — the travel time of train k in section i (i.e. station) is

[0104] τ ik — the additional time of train k at station i;

[0105] — the stop time of train k at station i;

[0106] τ ik — the additional time of train k at station i.

[0107] A specific intelligent recommendation method for train express and slow train scheduling considering capacity constraints is as follows:

[0108] Step 1, initialize the train information of the line running direction and the expected passenger flow demand data OD, which includes the following specific contents:

[0109] (1) initialize infrastructure information

[0110] 1) Train route information: V = {1, 2, …, N} is an ordered set of station numbers of a certain running direction of urban rail transit, and N is the number of stations on the route;

[0111] 2) Train route information: k = {1, 2, …, K} is an ordered set of train numbers in the study period T, and K is the total number of trains.

[0112] (2) Passenger flow demand information

[0113] q ij is the travel demand between stations, i.e. the passenger flow from station i to station j is q ij . The total passenger demand at station i is and j > i; the total arrival at station j is and j > i

[0114] 1) Passenger arrival rate

[0115] Assuming that the passenger arrival rate is independent of the train operation scheme, and the passengers arrive evenly between each OD pair, the passenger arrival rate ρ ij (the arrival amount per unit time) is calculated by the following formula:

[0116]

[0117] where q ij is the total passenger flow between station i and station j in the study period T.

[0118] 2) Passenger arrival amount

[0119] Let q ij be the passenger flow between station i and station j; the passenger amount that arrives at station i or changes into the train k before the train k-1 departs is:

[0120] q ijk = ρ ij h

[0121] where h is the departure interval.

[0122] Step 2, generate an initial scheme

[0123] For a line with N stations, there are theoretically 2 N feasible stop station schemes for a single train, denoted as Ω * . If the first station departs at a fixed departure interval h in the study period T, the number of train operations in the study period is (T divided by the departure interval h, rounded down). At this time, the complete scheme Ω * contains (2 N ) K combinations of train operation schemes.

[0124] Step 2.1 Stop station scheme initialization

[0125] Let y ik represent whether the train k stops at station i; when the train k stops at station i, y ik = 1, otherwise y ik = 0. If a station i (such as the first and last stations, transfer stations, and other large passenger flow stations) must stop, initialize y ik = 1. Encode the feasible solution, and the encoding of any feasible solution is shown as follows. Figure 1

[0126] Step 2.2 Initialize the first station departure time of each train

[0127] According to the principle of equal interval departure at the first station, initialize the first station departure time of each train under each scheme.

[0128] ​Step 2.3 Calculate the arrival time of each train at each station

[0129] According to the departure time of the first station and the train running time parameters, the arrival time, departure time and travel time of each train at each station are calculated.

[0130] Step 3, constraint check

[0131] Due to the characteristics of urban rail transit operation service, the number of trains continuously crossing stations and the number of stations skipping trains (trains passing through stations without stopping) can be deleted to preliminarily narrow the range of feasible solutions and save the calculation capacity of subsequent steps.

[0132] Step 3.1 Train k continuous crossing station constraint check

[0133] Ensure that the maximum number of consecutive skipping stations for train k is t max .

[0134]

[0135] Step 3.2 Station i continuous skipping train number constraint check

[0136] The maximum number of consecutive trains allowed to skip at station i is s max .

[0137]

[0138] Step 3.3 Filter Ω in Step 3.1 and 3.2 according to constraint conditions * that do not meet the crossing station and skipping conditions to obtain the initial scheme Ω.

[0139] Step 4, genetic algorithm optimization solution

[0140] Step 4.1 Set the evolution generation counter c = 0 and set the maximum evolution generation as I max ;

[0141] Step 4.2 Randomly select m individuals (i.e. feasible solutions) from the initial scheme Ω * to form the initial population P(c).

[0142] Step 4.3 Selection operation: check whether each individual in the population P(c) meets the constraints. If the constraints are met, proceed to Step 4.4, otherwise, re-perform Step 4.2.

[0143] Step 4.4 crossover operation:

[0144] Step 4.4.1 Randomly select x individuals in P(c) and perform the same encoding position exchange encoding operation with a certain probability to obtain x new individuals.

[0145] Step 4.4.2 Check whether the crossover scheme obtained in step 4.4.1 satisfies the constraints. If the constraints are satisfied, go to step 4.5, otherwise, go back to step 4.4.1.

[0146] Step 4.5 Mutation operation:

[0147] Step 4.5.1 Perform partial coding mutation operation on y (y = m - x) individuals randomly with a certain probability according to a certain rule. That is, change 1 to 0 and change 0 to 1.

[0148] Step 4.5.2 Check whether the mutation scheme obtained in step 4.5.1 satisfies the constraints. If the constraints are satisfied, go to step 4.6, otherwise, go back to step 4.5.1.

[0149] Step 4.6 Calculate the fitness of the x individuals formed after crossover and the y individuals formed after mutation in the population P(c), that is, the total travel time of passengers under each scheme. The calculation method is as follows:

[0150] Step 4.6.1 Calculate the passenger flow index of each individual scheme

[0151] According to the following method, the number of passengers getting off, getting on, staying, leaving the station, and waiting at each station for each train is calculated step by step.

[0152] a. Getting off passengers

[0153] Assuming that the passengers in train k getting off at station i are in proportion to the total amount of passenger flow attracted by subsequent station j, then:

[0154]

[0155] where v 0k = 0. The total amount of passengers getting off at station j for train k is

[0156] b. Getting on passengers

[0157] The getting on passengers are considered in three cases according to the train stop condition and the passenger carrying condition, that is:

[0158]

[0159] c. Staying passengers

[0160] Only when v ik -a ij,k +w ij,k > v * , passengers stay, and the staying amount is:

[0161] l ijk = wijk -b ijk = w ijk - (v * - (v ik - a ijk )

[0162] and the first station has no passenger flow l 1jk = 0.

[0163] d. Off-board passenger load

[0164] The off-board passenger load of train k is:

[0165] v ik = v i,k+1 = v ik + b ij,k - a ij,k

[0166] When y ik = 0, v ik = v ik = v i,k+1 .

[0167] e. Waiting passengers

[0168] The waiting passengers at station i when train k arrives consists of two parts: the passengers who stay l ij,k-1 , and the passengers who arrive or transfer before train k arrives. If we denote the arrival rate of passengers before train k arrives by p ik , then the arrival or transfer passengers between two trains is p ik h. Then the waiting passengers at station i when train k arrives is:

[0169] w ijk = l ij,k-1 + p ik

[0170] Step 4.6.2 Calculate the fitness value of the solution, i.e. the total travel time. The steps include:

[0171] a. Waiting time

[0172] The waiting time includes two parts: the waiting time of passengers who arrive or transfer, and the waiting time of passengers who stay.

[0173] (1) Assuming that the arrival time of passengers at the station obeys a uniform distribution, then the waiting time of passengers who arrive or transfer between the departure of train k-1 and the arrival of train k is

[0174] (2) The passengers who stay at the time of the departure of train k-1 need to wait for a complete departure interval, i.e. h.

[0175] b. Travel time

[0176] Passenger travel time consists of section travel time and dwell time (including its dwell allowance).

[0177] The travel time of train k from station i to station j is:

[0178]

[0179] c. Total passenger travel time

[0180] The optimization objective is to minimize the total passenger travel cost, which is the sum of train travel time and waiting time.

[0181]

[0182] Step 4.6.3 Calculate the fitness of the x individuals formed after crossover and the y individuals formed after mutation in population P(c). Select the m individuals with the best fitness as the new population P(c). At this time, if c max , let c = c + 1, repeat step 4.4; otherwise, end the algorithm, and the individual with the highest fitness in P(c) is the optimal solution.

[0183] Constraints:

[0184] Condition 1 Special station dwell constraint

[0185] To facilitate passenger flow calculation and reduce problem solving scale, the model stipulates that all trains need to stop at the origin and destination stations and other stations.

[0186] Starting station must stop: y 1k = 1 k = 1, 2, …, K

[0187] Terminal station must stop: y Nk = 1 k = 1, 2, …, K

[0188] If all trains in P(c) satisfy the above constraints, continue to step 4.3.3; otherwise, repeat step 4.3.1

[0189] Condition 2 Tracking interval constraint

[0190] Tracking train interval refers to the minimum interval time between tracking trains running in the same direction without interfering with each other. To ensure safe operation, the departure interval between train k-1 and train k should satisfy:

[0191] a ik -d i,k-1 ≥ y ik y i,k-1 h i1 + (1-yi,k-1 y ik h i2 +y i,k-1 (1-y ik )h i3 +(1-y i,k-1 )(1-y ik )h i4

[0192]

[0193]

[0194] where h i1 , h i2 , h i3 are the minimum headway of the trains k-1 and k at the station without crossing, the front train crossing, the rear train crossing, respectively.

[0195] If all the trains in P(c) satisfy the above constraints, continue to step 4.3; otherwise, repeat step 4.3.1.

Claims

1. A method for intelligent recommendation of train express and slow running schemes considering capacity constraints, characterized in that, The method comprises the following steps: Step 1, initializing train information of a line running direction and expected passenger flow demand data OD, and the specific content comprises: Step 1.1, initializing infrastructure information Train route information: k={1, 2, L, K} is the ordered number set of trains in the research period T, and K is the total number of trains; (1.2) determine passenger flow demand information Calculate the passenger arrival amount, the train interval is h, and the passenger amount of the train k arriving at the station i before the train k-1 leaves is: The passenger flow from station i to station j is q ij The total demand for passengers at station i is and j>i; and the total amount of passengers arriving at station j is and j>i; Calculate the passenger arrival rate, the rate of passengers arriving in the study period T, ρ ij Calculate by the following formula: where q ij is the total passenger flow between station i and station j in the study period T; Step 2, generate an initial scheme q ijk = p ij h Step 2.1, initialize the stop station scheme For a line with N stations, the stop plan of a single train is denoted as Ω * If the first station departs trains at a fixed interval of h in a study period T, the number of trains in the study period T is Step 2.2, initialize the departure time of each train at the first station Let y ik denote whether train k stops at station i or not, y ik = 1 if train k stops at station i, otherwise y ik = 0; if a station i must be stopped, then initialize y ik = 1, encode the feasible solution; According to the principle of equal interval departure at the first station, the departure time of each train at the first station in each scheme is initialized; Step 2.3, calculate the arrival time of each train at each station According to the departure time of the first station and the train running time parameter, the arrival time, departure time and travel time of each train at each station are calculated; Step 3, constraint condition check Step 3.1, continuous crossing station constraint check of train k Step 3.2, continuous skip-stop train number constraint check of station i Guarantee the maximum number of consecutive crossing stations of train k is t max ; Step 4, optimization solution by genetic algorithm, The maximum number of consecutive stop trains allowed at station i is s max ; Step 3.

3. Filter Ω according to constraints 3.1 and 3.2 * The initial solution Ω is obtained from the solutions that do not satisfy the cross-site and skip-stop conditions. The algorithm constraint condition is: Condition 1, special station stop constraint Condition 2, tracking interval constraint Originating station must park: y 1k = 1 k = 1,2, L, K Terminal must park: y Nk = 1 k = 1,2, L, K The tracking train interval refers to the minimum interval time of tracking trains running in the same direction without interfering with each other, and in order to ensure the running safety, the departure interval between the train k-1 and the train k should satisfy: The specific steps of the step 4 using the genetic algorithm for optimization solution further comprise: wherein: h i1 h i2 h i3 are the minimum headway for the trains k-1 and k, respectively, with no crossing at the station, with the first train crossing the station, and with the second train crossing the station.

2. The method of claim 1, wherein, Step 4.3, selection operation: according to the "constraint condition", check whether each individual in the group P(c) satisfies the constraint; if the constraint is satisfied, step 4.4 is performed, otherwise step 4.2 is performed again; Step 4.1 Set evolutionary generation counter c = 0, set maximum evolutionary generations as I max ; Step 4.2 in the initial scheme Ω * an initial population P(c) of m individuals is randomly selected; Step 4.4, crossover operation: Step 4.4.1, randomly select x individuals in P(c), and perform the same coding position exchange coding operation according to a certain probability to obtain x new individuals; Step 4.4.2, according to the "constraint condition", check whether the crossover scheme obtained in step 4.4.1 satisfies the constraint, if the constraint condition is satisfied, step 4.5 is performed, otherwise step 4.4.1 is performed again; Step 4.5, mutation operation: Step 4.5.1, according to a certain rule, randomly perform partial coding mutation operation on y (y=m-x) individuals with a certain probability, that is, change 1 to 0 and change 0 to 1; Step 4.5.2, according to the "constraint condition", check whether the mutation scheme obtained in step 4.5.1 satisfies the constraint; if the constraint condition is satisfied, step 4.6 is performed, otherwise step 4.5.1 is performed again; Step 4.6, calculate the fitness of the x individuals formed after the crossover and the y individuals formed after the mutation of the group P(c), that is, the total travel time of passengers under each scheme, and the calculation method is as follows: Step 4.6.1, calculate the passenger flow index of each individual scheme Step 4.6.2, calculate the total travel time of the scheme; ​ ​ Step 4.6.3 Calculate the fitness of the x individuals formed after crossover and y individuals formed after mutation in the population P(c), and select the m individuals with the best fitness as the new population P(c); at this time, if c < I max then let c = c + 1, repeat step 4.4; otherwise end the algorithm, and the individual with the highest fitness in P(c) is the optimal solution. 3.The method of claim 1, wherein the step 4.6.1 is performed by the following steps: a. calculating the number of passengers getting off at each station; b. calculating the number of passengers getting on at each station; c. calculating the number of passengers staying on at each station; d. calculating the number of passengers on board at each station; and e. calculating the number of passengers waiting at each station. 4.The method of claim 1, wherein the step 4.6.2 is performed by the following steps: a. calculating the waiting time of passengers at each station; b. calculating the travel time of passengers at each station; and c. calculating the total travel time of passengers in the scheme. 5.The method of claim 1, wherein the step 4.6.3 is performed by the following steps: a. setting the following constraints: (i) special station stop constraint; and (ii) headway constraint; b. repeating the step 4.3.1 if the constraints are not met; and c. continuing the step 4.3 if the constraints are met. 6.The method of claim 1, wherein the step 4.6.4 is performed by the following steps: a. calculating the total travel cost of passengers in the scheme; and b. repeating the step 4.3.1 if the total travel cost is not minimized; and c. continuing the step 4.3 if the total travel cost is minimized. wherein the total amount of passengers getting off at the station j of the train k is 7.The method of claim 1, wherein the step 4.6.5 is performed by the following steps: a. calculating the total travel cost of passengers in the scheme; and b. repeating the step 4.3.1 if the total travel cost is not minimized; and c. continuing the step 4.3 if the total travel cost is minimized. 8.The method of claim 1, wherein the step 4.6.6 is performed by the following steps: a. calculating the total travel cost of passengers in the scheme; and b. repeating the step 4.3.1 if the total travel cost is not minimized; and c. continuing the step 4.3 if the total travel cost is minimized. 9.The method of claim 1, wherein the step 4.6.7 is performed by the following steps: a. calculating the total travel cost of passengers in the scheme; and b. repeating the step 4.3.1 if the total travel cost is not minimized; and c. continuing the step 4.3 if the total travel cost is minimized. Only when v ik - a ij,k + w ij,k > v * Passengers stay, stay for the amount: l ijk = w ijk - b ijk = w ijk - (v * - (v ik - a ijk )) 10.The method of claim 1, wherein the step 4.6.8 is performed by the following steps: a. calculating the total travel cost of passengers in the scheme; and b. repeating the step 4.3.1 if the total travel cost is not minimized; and c. continuing the step 4.3 if the total travel cost is minimized. ​ v ik ' = v i,k+1 = v ik + b ij,k - a ij,k When y ik = 0, v ik = v ik ' = v i,k+1 ; ​ The waiting passengers at station i when train k arrives consist of two parts: the staying passengers l ij,k-1 and the arriving or transfer passengers before train k-1 leaves and train k arrives. Let ρ ik denote the passenger arrival rate before train k arrives, then the arrival or transfer passenger flow of two trains with interval h is ρ ik h, then the waiting passenger amount of station i when train k arrives is: w ijk = l ij,k-1 + p ik . ​ ​ ​ ​ (i) Assuming that the time of arrival of passengers at the station obeys a uniform distribution, the waiting time of passengers arriving or transferring passengers from train k-1 to train k is ​ ​ ​ ​ ​ ​ ​ ​ ​ Originating station must park: y 1k = 1 k = 1,2, L, K Terminal must park: y Nk = 1 k = 1,2, L, K ​ ​ ​ wherein: h i1 h i2 h i3 are the minimum headway for the preceding and following trains k-1 and k not crossing the station, the preceding train crossing the station, and the following train crossing the station, respectively. ​ ​

Images