A customized bus road network optimization method and system considering service stability

Abstract

The present invention relates to the technical field of intelligent transportation, and discloses a customized bus network optimization method and system considering service stability, including the following steps: S1. Collect existing traffic flow data, estimate traffic parameters of the customized bus road network and partition the customized bus road network, and extract the vehicle speed law of the sub-areas of the customized bus road network; S2. According to the vehicle speed law, establish an optimization model of the customized bus network aiming at maximizing service stability and operating profit under the random traffic environment of the road network; S3. Use random simulation and improved NEWMAN algorithm to generate an initial customized bus network; S4. Design an alternative travel plan set for the unresponded passengers in the initial customized bus line network, and guide the unresponded passengers in the initial plan to select the final travel plan from the alternative plan set based on the passenger selection willingness; S5. Use random simulation and improved NEWMAN algorithm to conduct overall re-optimization on the initial customized bus network to obtain the optimal customized bus network. The present invention solves the problem that the prior art cannot fully consider service stability and passenger demand guidance in the actual random traffic road network environment.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent transportation, and more specifically, to a customized bus network optimization method and system considering service stability. Background Art

[0002] Today, the rapidly growing personalized and high-quality travel demands have made the contradiction between urban transportation supply and passenger travel demands increasingly prominent. Traditional fixed-route buses have been much criticized by passengers due to their disadvantages such as crowded interior space, lack of flexibility, long and uncontrollable travel time, etc. With the rapid development of mobile Internet technology, new shared bus transportation modes such as customized buses have been widely applied worldwide.

[0003] In the operation of customized buses, bus operation enterprises plan customized bus lines that meet all or part of the passenger travel demands based on specific demands, on the basis of passenger travel demands, and under the constraints of various actual conditions. If the travel demand of a passenger is covered by the customized bus line, this demand can be completed by taking the customized bus, otherwise, this demand is rejected. There are two key issues worthy of attention in the actual operation of customized buses: One is that the random traffic environment of the road network makes it difficult to ensure its service stability. "Punctuality" is a core advantage of customized bus services and an important factor affecting the attractiveness of customized buses. Service stability can be defined as the proportion of the number of passengers served on time by customized buses to the total number of passengers served. In the random traffic environment of the road network, the speed of customized bus vehicles is random, and it is very likely that customized bus vehicles cannot arrive at stations on time as expected, thus making it difficult to ensure service stability. Therefore, in the planning of customized bus networks, it is necessary to reasonably estimate the arrival time based on the randomness of the speed of customized bus vehicles and use this as an important basis to plan the routes of customized bus vehicles, so as to improve the service stability of customized buses. The other is that one-sided emphasis on passively responding to passenger travel demands in network planning leads to losses of interests for both the operation enterprises and passengers. Customized bus is a demand-responsive bus, that is, the customized bus network is planned based on the principle of meeting passenger travel demands, such as the boarding / alighting stations, boarding / alighting times, etc., which is an important way to improve the service quality and attractiveness of customized buses. However, one-sided passive response to passenger travel demands also has negative effects: On the one hand, it may lead to a substantial increase in the operation cost of customized buses, affecting the sustainability of customized bus services; on the other hand, due to the incompleteness of traffic information, the travel demands put forward by passengers may not meet the requirements of actual traffic conditions, resulting in the infeasibility or high cost of serving them, so that they can only be refused service, reducing the coverage of customized bus services.

[0004] Scholars at home and abroad have conducted relevant research on the optimization of customized bus lines: including OD area division based on hierarchical clustering, and network planning with the goal of maximizing social benefits and minimizing operating costs; also including the construction of a multi-objective optimization model for customized bus routes in response to dynamic travel requests; also including a two-stage customized bus network design framework with the goal of maximizing the operating income of the network; also including the construction of an optimization model for the customized bus network with the goal of minimizing the additional travel time of passengers and operating costs and with constraints such as travel time and vehicle capacity.

[0005] Regarding this problem, there is also a method for optimizing customized bus ridesharing considering two-sided matching in the prior art, including: determining the elements of two-sided matching for customized bus ridesharing, and the elements include passengers, vehicles, road conditions information, and stations; determining the objective functions, including the system-optimal objective function, the passenger-optimal objective function, and the operator-optimal objective function; determining the types of two-sided matching ridesharing; the optimization model for two-sided matching customized bus ridesharing; generating a preference list according to the optimization model; and the two-sided stable matching algorithm.

[0006] Through the induction and summary of the existing methods for optimizing the customized bus network, the following deficiencies can be found: First, the current network planning usually assumes that the speed or travel time of customized buses is a deterministic parameter. Although this method greatly simplifies the complexity of the travel time matching between customized buses and passengers, ignoring the randomness of the driving speed or travel time of customized buses is very likely to cause a large difference between the estimated arrival time and the actual arrival time of the vehicle, thus affecting the stability of the customized bus service. Second, the current usually emphasizes optimizing the customized bus network based on the principle of meeting passengers' travel needs, ignoring the possible unreasonable factors or infeasibility in passengers' travel needs themselves, thus having an adverse impact on operating enterprises and passengers.

[0007] However, the prior art has the problem that it cannot fully consider service stability and passenger demand guidance in the actual random traffic road network environment. Therefore, how to invent an optimization method for the customized bus network that can fully consider service stability and passenger demand guidance is an urgent problem to be solved in this technical field. Summary of the Invention

[0008] In order to solve the problem that the prior art cannot fully consider service stability and passenger demand guidance, the present invention provides an optimization method and system for a customized bus network considering service stability in a random traffic road network environment, which also has the characteristic of considering maximizing operating profit.

[0009] To achieve the above object of the present invention, the following technical solutions are adopted:

[0010] An optimization method for a customized bus network considering service stability, including the following steps:

[0011] S1. Collect existing traffic flow data, estimate traffic parameters of the customized bus network and partition the customized bus network, and extract vehicle speed patterns in the customized bus network sub-areas;

[0012] S2. According to the vehicle speed law, a customized bus network optimization model is established under the random traffic environment of the road network with the goal of maximizing service stability and operating profit;

[0013] S3. According to the optimization model, based on the speed law and passenger travel demand of the customized bus network sub-area, the initial customized bus network is generated by using random simulation and improved NEWMAN algorithm;

[0014] S4. Design a set of alternative travel plans for passengers who have not responded to the initial customized bus network, and guide the passengers who have not responded to the initial plan to choose the final travel plan from the alternative travel plans based on the passengers' choice intention;

[0015] S5. Based on the initial customized bus network, the travel plans of the unresponsive passengers in the initial plan, and the vehicle speed patterns in the customized bus network sub-areas, with the goal of maximizing service stability and operating profit, the initial customized bus network is re-optimized as a whole using random simulation and the improved NEWMAN algorithm to obtain the optimal customized bus network.

[0016] Preferably, in step S1, the traffic parameter estimation of the customized bus network is performed, specifically:

[0017]

[0018]

[0019]

[0020]

[0021]

[0022] Among them, v (b) (j,m) is the average speed of buses on section j in time period m; b represents the bus; t (b) (i,m) and d (b) (i,m) are the travel time and distance of bus trip i in time period m; T (b) (j,m) and D (b) (j,m) are the total travel time and distance of the bus in period m on section j; k (b) (j,m) is the average bus density of section j in time period m; l(j) is the length of section j; τ is the length of the time segment; k (b)(m) is the average bus density weighted by road section length within the area; J is the number of bus road sections within the area; L is the total lane length; q (b) (m) is the average bus flow within the area; s(j) is the number of lanes of road section j; T (b) (m) and D (b) (m) are the total driving time and total driving distance of buses during period m respectively; v (b) (m) is the average speed of buses in the area during period m.

[0023] Furthermore, in the step S1, for the zoning of the customized bus road network, the specific steps are as follows:

[0024] S101. Divide the road into several road sections, and calculate the road section speed of the customized bus vehicles according to the existing traffic flow data;

[0025] S102. Cluster the road sections based on the customized bus road section speed, the distance between road sections, and the road section connectivity to form several initial communities;

[0026] S103. Take the minimization of the total variance of the bus road section speed of the customized bus road network as the goal, consider the adjacency relationship between the initial communities, and merge the initial sub - communities to form several sub - communities; take the reduction of the total variance of the bus road section speed of the road network as the goal, and adjust the belonging sub - communities of the road sections located on the boundary to obtain the zoning of the customized bus road network.

[0027] Even further, in the step S1, for extracting the vehicle speed pattern of the sub - areas of the customized bus road network, the specific steps are as follows:

[0028] ES1. According to the average bus flow, average density value, and average speed of each sub - area in the zoning of the customized bus road network, identify the relationship between the average bus density and average speed of the sub - area, and obtain the operation pattern of buses in the sub - area;

[0029] ES2. According to the time period of the customized bus line network optimization, find out the distribution section of the average density of the customized buses in each sub - area road network during this time period and the possible values of the speed of the customized buses corresponding to this density section, conduct a fitting of the average speed distribution of the customized buses, and extract the speed distribution pattern of the bus vehicles during the time period to be optimized of the bus line network.

[0030] Even further, in the step S2, establish an optimization model for the customized bus line network under the random traffic environment of the road network, with the goal of maximizing service stability and maximizing operation profit, specifically as follows:

[0031] a. Based on the passenger travel demand and the vehicle speed pattern of the sub - areas of the customized bus road network, seek an optimal line network optimization plan with the optimal proportion of the number of passengers seeking on - time service, and obtain the objective function for maximizing service stability:

[0032]

[0033] Among them, \(K\) is the set of vehicles in the customized bus fleet of the operating enterprise, and \(V\) b and \(V\) a are respectively the set of boarding stops and the set of alighting stops in the passenger travel demand. is the number of boarding passengers of passenger travel demand \(r\) at stop \(i\). is a 0-1 decision variable, which takes the value of 1 when vehicle \(k\) travels from stop \(i\) to stop \(j\), and 0 otherwise; is an indicator variable, which takes the value of 1 when vehicle \(k\) arrives at stops \(v\) i and \(v\) j on time, and 0 otherwise;

[0034] b. According to the passenger travel plan and the existing bus data, calculate the income of the operating enterprise and the cost paid by the operating enterprise, maximize the profit of the operating enterprise, and obtain the objective function for maximizing the operating profit:

[0035] max \(F2 = G - C\)

[0036] Among them, \(G\) is the income of the operating enterprise, and \(C\) is the cost paid by the operating enterprise.

[0037] c. According to the objective function for maximizing service stability and the objective function for maximizing operating profit, and based on the vehicle speed law, establish an optimized model for the customized bus network with the goal of maximizing service stability and operating profit under the random traffic environment of the road network;

[0038] Furthermore, in step b above, according to the passenger travel plan and the existing bus data, calculate the income of the operating enterprise and the cost paid by the operating enterprise, and maximize the profit of the operating enterprise to obtain the objective function for maximizing the operating profit. The specific steps are as follows:

[0039] b01. According to the passenger travel plan, calculate the service income \(G\) of the operating enterprise:

[0040]

[0041] Among them, \(R\) is the set of passenger travel demands; is the set of boarding and alighting stops when passenger travel demand \(r\) adopts travel plan \(s\) n ; is the assignment decision variable of passenger travel demand. When passenger travel demand \(r\) is assigned to vehicle \(k\) and adopts travel plan \(s\) n it takes the value of 1, and 0 otherwise;

[0042] b02. According to the bus network to be planned, considering the driving cost of vehicles on the line, the waiting cost of vehicles arriving at the station earlier than the passenger-specified time window, and the service cost of passengers getting on and off the vehicle, the operation cost C of the enterprise is obtained:

[0043]

[0044] Among them,

[0045]

[0046] V is the set of stations of the customized bus network to be planned; S is the set of travel plans; μ k is the fixed cost of dispatching vehicle k; η k is a 0-1 decision variable, which takes the value of 1 when vehicle k participates in operation, and 0 otherwise; λ k is the converted cost per unit time of vehicle k; t ij is the time for the vehicle to travel from station i to station j; τ i is the service time of the vehicle at station i; v0 is the vehicle yard, and it is assumed that the cost of the vehicle driving out of and returning to the vehicle yard is negligible, that is, the time for the vehicle to drive into the first station of the line from the vehicle yard and drive back to the vehicle yard from the last station of the line is 0; is the waiting time of vehicle k at station i; is the arrival time of vehicle k at station i; T r,b is the boarding time specified by passenger travel demand r;

[0047] Sb03. Obtain the objective function of maximizing the operation profit.

[0048] Furthermore, the constraint conditions of the customized bus network optimization model are:

[0049] A. Vehicle and service plan assignment constraints:

[0050] A1. It is set that each passenger travel demand can be served by at most one customized bus vehicle using one travel plan;

[0051] A2. It is set that a passenger travel demand can only be assigned to a customized bus vehicle participating in operation, and it is only allowed to drive on the line when the vehicle participates in operation;

[0052] A3. It is set that all passengers in the same demand can only be transported by the same vehicle;

[0053] B. Time constraints:

[0054] B1. For passengers who are confirmed to be provided with customized bus travel services, for each travel plan providing services for them, set the vehicle to arrive at the boarding and alighting stations within the passenger-specified time window;

[0055] B2. Set the access time between the front and rear stations;

[0056] C. Access order restriction:

[0057] C1. For the passenger travel demands to be responded to, set that the service vehicle first accesses the boarding station and then the alighting station of the passenger;

[0058] D. Constraints on the departure and arrival of customized bus vehicles:

[0059] D1. Set that each bus participating in the operation departs from the depot, and returns to the depot after completing the task;

[0060] E. Flow balance constraint:

[0061] E1. Set that if a station has pick-up and drop-off requirements at the same time or there are travel demands of passengers at different times, it is split into multiple stations with the same geographical location, and the travel time between these stations is zero;

[0062] F. Vehicle capacity constraint:

[0063] F1 Set that the number of passengers on the bus is less than the maximum passenger capacity, and set the minimum passenger capacity requirement for each line;

[0064] G. Constraint on the number of customized bus vehicles:

[0065] G1 Set that the number of customized bus vehicles put into operation cannot exceed the total number of vehicles in the fleet;

[0066] H. Constraint on the value of decision variables:

[0067] H1. Set the range of values of the decision variables of the service stability maximization model and the operation profit maximization model.

[0068] Furthermore, in step S3, according to the optimization model, based on the speed law of the customized bus road network sub-region and the passenger travel demands, a random simulation and an improved NEWMAN algorithm are used to generate an initial customized bus road network, specifically:

[0069] S301. Generate an initial route:

[0070] ES101. For each station in the customized bus road network, solve the shortest path between stations through a matrix algorithm;

[0071] ES102. Targeting travel demands, a separate route is arranged for each travel demand, which is called the initial route; each initial route is: depot - boarding stop - alighting stop - depot; it is set that the cost for the operating vehicle to drive into the first stop of the route and drive back to the depot from the last stop after completing the transportation task is 0; the initial route meets all other constraints except the minimum passenger capacity constraint of the route.

[0072] ES103. If the customized bus vehicle cannot meet the travel time requirements of passengers when driving at the maximum speed, it is determined that the customized bus cannot provide services for such passenger travel demands, and such passenger travel demands are excluded to obtain the original customized bus network.

[0073] S302. Construct a combination library:

[0074] ES201. Randomly select two initial routes. According to the shortest distance between stops on the route and the chronological order of travel times at stops, on the premise of meeting the requirements of the stop access order, splice the routes to form a new route.

[0075] ES202. If the new route violates the maximum passenger capacity constraint, terminate the route merger; otherwise, randomly select two initial routes again for merger until the new route meets the maximum passenger capacity constraint; the new route and other initial routes together form an initial solution.

[0076] ES203. Repeat the above steps M times until M route libraries are generated. The set of route libraries is called the combination library.

[0077] S303. Calculate the fitness of each route by means of random simulation:

[0078] ES301. For each route, find the sub - areas where each stop is located and the sub - area distribution of the shortest paths between each pair of consecutive stops, and calculate the driving distances of the paths in each sub - area; conduct a number of random simulations. Each time, randomly select a speed value based on the speed distribution law of the passing sub - areas, and calculate the travel times of the sub - paths between each pair of consecutive stops.

[0079] ES302. Cumulatively add the travel times between consecutive stops along the route step by step to calculate the actual arrival times of the customized bus vehicle at each stop on the route.

[0080] ES303. If the vehicle on the route can arrive at the relevant stops within the specified pick - up and drop - off time windows of the passengers, it is determined that the route can pick up and drop off the passengers on time; count the number of passengers served on time and the total number of passengers for each route in each random simulation, and calculate the service stability of each route.

[0081] ES304. Statistically calculate the profit of each route in each random simulation.

[0082] ES305. Calculate the fitness of each line library in the combined library using the average service stability of each line and the profit of each line as indicators, and generate a fitness matrix;

[0083] S304. Update the line library:

[0084] ES401. Perform TOPSIS scoring on each line in the line library according to the fitness matrix;

[0085] ES402. Arrange the lines in descending order of the scores, select the top |K| lines to form the effective lines, and the remaining lines are used as the lines to be responded to;

[0086] ES403. Perform evolutionary operations on the line library through three branches:

[0087] XS1. Merge the effective lines and the lines to be responded to, that is, randomly select an effective line and a line to be responded to each time, merge them according to the rules, and check whether the maximum passenger capacity constraint is satisfied until a new line is generated, and generate a new line library;

[0088] XS2. Randomly select two lines from the effective lines for merging until a new line that satisfies the maximum passenger capacity constraint is generated, and generate a new line library;

[0089] XS3. Randomly select two lines from the lines to be responded to for merging until a new line that satisfies the maximum passenger capacity constraint is generated, and generate a new line library;

[0090] S305. Calculate the TOPSIS scores of the lines in the updated line library, and update the combined library according to the TOPSIS scores of the lines in the updated line library. Finally, retain the top M line libraries in the ranking to form the updated combined library;

[0091] S306. Repeat steps S304 - S305 to iterate the combined library until there are no lines that can be further merged;

[0092] S307. Select the optimal line library from the iterated combined library:

[0093] ES701. Calculate the fitness of the lines in each line library in the iterated combined library;

[0094] ES702. Use the TOPSIS method to score the lines in each line library and arrange the lines in descending order. Select the top |K| lines, and eliminate the lines that do not meet the minimum passenger capacity constraint to form the effective lines;

[0095] The total average service stability and total average profit of the effective lines are used as the evaluation criteria for the quality of each line library. The optimal line library is selected and used as the initial customized bus network.

[0096] Furthermore, in step S4, an alternative travel plan set is designed for the unresponded passengers in the initial customized bus line network. Based on the passengers' willingness to choose, the unresponded passengers in the initial plan are guided to select the final travel plan from the alternative travel plan set. The specific steps are as follows:

[0097] S401. Obtain the guiding requirements of the unresponded passengers in the initial customized bus line network, and generate the corresponding list of guiding requirement station pairs and the station table of the planned lines.

[0098] S402. According to the list of guiding requirement station pairs, the station table of the planned lines, and the guiding requirements, generate optional alternative travel plans by adjusting the boarding / alighting stations or boarding / alighting times, so that the unresponded passengers in the initial customized bus line network become passengers who can be responded to by the customized bus.

[0099] S403. Comprehensively consider travel time, payment cost, passengers' sensitivity to time and cost, perception error, and external environment, and calculate the perceived cost of each plan; according to the perceived cost of each plan, and based on the principle of maximizing utility, obtain the plan with the highest probability of passengers' willingness to choose, and use the plan with the highest probability of willingness to choose as the passengers' travel plan. Otherwise, the guiding requirement cannot be served.

[0100] A customized bus network optimization system considering service stability includes a data collection module, a vehicle speed rule module, a model construction module, an initial customized bus network module, a demand guidance module, and a model re-optimization module.

[0101] The data collection module is used to collect existing traffic demand and traffic flow data.

[0102] The vehicle speed rule module is used to estimate the parameters of the customized bus line network traffic and partition the customized bus road network, and extract the vehicle speed rules of the sub-regions of the customized bus road network.

[0103] The model construction module is used to establish an optimization model of the customized bus line network with the goal of maximizing service stability and operation profit under the random traffic environment of the road network according to the vehicle speed rules.

[0104] The initial customized bus network module is used to generate an initial customized bus network based on the optimization model, the speed rules of the sub-regions of the customized bus road network, and the passengers' travel demands, using random simulation and improved NEWMAN algorithm.

[0105] The demand guidance module is used to design a set of alternative travel plans for passengers who have not been responded to in the initial customized bus line network, and determine the travel plans of passengers who have not been responded to in the initial plan based on the passengers' choice intention;

[0106] The model re-optimization module is used to optimize the initial customized bus network as a whole by using random simulation and improved NEWMAN algorithm based on the initial customized bus network, the travel plans of passengers who were not responded to in the initial plan, and the vehicle speed rules of the customized bus network sub-areas, with the goal of maximizing service stability and maximizing operating profits, so as to obtain the optimal customized bus network.

[0107] The beneficial effects of the present invention are as follows:

[0108] The present invention collects existing traffic flow data, estimates the traffic parameters of the customized bus network and divides the customized bus network into zones, extracts the vehicle speed rules of the customized bus network sub-zones, and fully considers the randomness of the vehicle speeds in the customized bus network sub-zones, thereby avoiding the deviation caused by estimating the vehicle arrival time based on the average speed, and also avoiding the complexity of characterizing the vehicle speed rules at the road level and the large amount of computational workload generated thereby; the present invention also establishes a customized bus line network optimization model under a random traffic environment of the road network based on the vehicle speed rules, with the goal of maximizing service stability and maximizing operating profits; according to the optimization model, based on the speed rules of the customized bus network sub-zones and the travel needs of passengers , using random simulation and improved NEWMAN algorithm to generate the initial customized bus network, taking into account the maximization of service stability and operating profit; the present invention also designs a set of alternative travel plans for passengers who are not responded to in the initial customized bus network, and guides the passengers who are not responded to in the initial plan to select the final travel plan from the set of alternative travel plans based on the passengers' choice willingness, fully considering the guidance of passengers' travel needs, thereby expanding the range of serviceable people; finally, the present invention takes the maximization of service stability and operating profit as the goal, and uses random simulation and improved NEWMAN algorithm to re-optimize the initial customized bus network as a whole, thereby improving the optimization efficiency and accuracy of customized bus network optimization. Therefore, the present invention solves the problem that the prior art cannot fully consider service stability and passenger demand guidance in the actual road network random traffic environment, and has the characteristic of considering the maximization of operating profit. BRIEF DESCRIPTION OF THE DRAWINGS

[0109] Figure 1 It is a flow chart of a customized bus line network optimization method taking service stability into consideration according to the present invention.

[0110] Figure 2 It is a schematic diagram of fitting the customized bus speed distribution of the present invention.

[0111] Figure 3It is a schematic diagram for guiding travel demands of the present invention.

[0112] Figure 4 It is a complete algorithm flowchart of a customized bus network optimization method considering service stability of the present invention.

[0113] Figure 5 It is a road network map of the central urban area of Guangzhou.

[0114] Figure 6 It is a result map of the road network zoning of the central urban area of Guangzhou by the present invention.

[0115] Figure 7 It is a schematic diagram of travel demands for customized buses.

[0116] Figure 8 It is a trend chart of the change in the objective function value. Specific implementation manners

[0117] The present invention will be described in detail below in conjunction with the accompanying drawings and specific implementation manners.

[0118] Example 1

[0119] As Figure 1 shown, a customized bus network optimization method considering service stability includes the following steps:

[0120] S1. Collect existing traffic flow data, estimate traffic parameters of the customized bus road network and zone the customized bus road network, and extract the vehicle speed rules of the sub - zones of the customized bus road network;

[0121] S2. According to the vehicle speed rules, establish an optimization model for the customized bus network aiming at maximizing service stability and operation profit under the random traffic environment of the road network;

[0122] S3. According to the optimization model, based on the speed rules of the sub - zones of the customized bus road network and passengers' travel demands, use random simulation and improved NEWMAN algorithm to generate an initial customized bus network;

[0123] S4. Design an alternative travel plan set for the passengers not responded to in the initial customized bus network, and guide the passengers not responded to in the initial plan to select the final travel plan from the alternative plan set based on the passengers' choice willingness;

[0124] S5. According to the initial customized bus network, the travel plans of the passengers not responded to in the initial plan and the vehicle speed rules of the sub - zones of the customized bus road network, aiming at maximizing service stability and operation profit, use random simulation and improved NEWMAN algorithm to overall re - optimize the initial customized bus network to obtain the optimal customized bus network.

[0125] Example 2

[0126] More specifically, in a specific embodiment, in step S1, estimating the traffic parameters of the customized bus road network specifically includes:

[0127]

[0128]

[0129]

[0130]

[0131]

[0132] where v (b) (j,m) is the average speed of the bus on section j during time period m; b represents the bus; t (b) (i,m) and d (b) (i,m) are respectively the travel time and travel distance of bus trip i during time period m; T (b) (j,m) and D (b) (j,m) are respectively the total travel time and travel distance of the bus on section j during time period m; k (b) (j,m) is the average density of the bus on section j during time period m; l(j) is the length of section j; τ is the length of the aggregation time period; k (b) (m) is the average density of the bus weighted by the section length in the area; J is the number of bus sections in the area; L is the total lane length; q (b) (m) is the average flow of the bus in the area; s(j) is the number of lanes of section j; T (b) (m) and D (b) (m) are respectively the total travel time and total travel distance of the bus during time period m; v (b) (m) is the average speed of the bus in the area during time period m.

[0133] In a specific embodiment, in step S1, the partitioning of the customized bus road network specifically includes the following steps:

[0134] S101. Divide the road into several sections, and calculate the speed of the customized bus section according to the existing traffic flow data;

[0135] S102. Cluster the sections based on the speed of the customized bus section, the distance between sections, and the section connectivity to form several initial cells;

[0136] S103. With the goal of minimizing the total variance of the bus section speeds in the customized bus road network, considering the adjacency relationship of the initial sub - areas, the initial sub - areas are merged to form several sub - areas; with the goal of reducing the total variance of the bus section speeds in the road network, the belonging sub - areas of the sections located on the boundary are adjusted to obtain the customized bus road network partition.

[0137] In a specific embodiment, in step S1, to extract the vehicle speed pattern of the customized bus road network sub - areas, the specific steps are as follows:

[0138] ES1. According to the average bus flow, average density value, and average speed of each sub - area in the customized bus road network partition, identify the relationship between the average density and average speed of buses in the sub - area, and obtain the operation pattern of buses in the sub - area.

[0139] ES2. According to the time period of the customized bus line network optimization, find out the distribution section of the average density of customized buses in each sub - area road network within this time period and the possible values of the speeds of customized buses corresponding to this density section, perform the fitting of the average speed distribution of customized buses, and extract the speed distribution pattern of bus vehicles within the time period of the bus line network optimization. The graph after the speed distribution fitting is as Figure 2 shown.

[0140] In a specific embodiment, in step S2, according to the vehicle speed pattern, establish an optimization model for the customized bus line network with the goals of maximizing service stability and maximizing operation profit in a random traffic environment of the road network, specifically as follows:

[0141] a. Based on the passenger travel demand and the vehicle speed pattern of the customized bus road network sub - areas, seek an optimal line network optimization plan with the highest proportion of passengers who receive on - time service, and obtain the objective function for maximizing service stability:

[0142]

[0143] where K is the set of vehicles in the customized bus fleet of the operating enterprise, V b and V a are respectively the sets of boarding and alighting stops in the passenger travel demand, is the number of boarding passengers of passenger travel demand r at stop i, is a 0 - 1 decision variable, which takes the value of 1 when vehicle k travels from stop i to stop j, and 0 otherwise; is an indicator variable, which takes the value of 1 when vehicle k arrives at stops v i and v j on time, and 0 otherwise;

[0144] b. According to the travel plans of passengers and the existing bus data, calculate the income of the operating enterprise and the costs paid by the operating enterprise, and maximize the profit of the operating enterprise to obtain the objective function for maximizing operating profit:

[0145] max F2 = G - C

[0146] Wherein, G is the income of the operating enterprise, and C is the cost paid by the operating enterprise.

[0147] c. According to the objective functions of maximizing service stability and operating profit, and based on the vehicle speed law, establish an optimization model for the customized bus network with the objectives of maximizing service stability and operating profit in a random traffic environment of the road network;

[0148] In a specific embodiment, in step b, according to the travel plans of passengers and the existing bus data, calculate the income of the operating enterprise and the cost paid by the operating enterprise, maximize the profit of the operating enterprise, and obtain the objective function of maximizing operating profit. The specific steps are as follows:

[0149] b01. According to the travel plans of passengers, calculate the service income G of the operating enterprise:

[0150]

[0151] Wherein, R is the set of passenger travel demands; is the set of boarding and alighting stops when the passenger travel demand r adopts the travel plan s n ; is the assignment decision variable of the passenger travel demand. When the passenger travel demand r is assigned to vehicle k and adopts the travel plan s n , it takes the value of 1, otherwise it is 0;

[0152] b02. According to the bus network to be planned, considering the driving cost of the vehicle on the line, the waiting cost of the vehicle arriving at the station earlier than the passenger-specified time window, and the service cost of passengers boarding and alighting, obtain the operating enterprise cost C:

[0153]

[0154] Wherein,

[0155]

[0156] V is the set of stops of the customized bus network to be planned; S is the set of travel plans; μ k is the fixed cost of dispatching vehicle k; η k is a 0-1 decision variable. When vehicle k participates in operation, it takes the value of 1, otherwise it is 0; λ k is the converted cost per unit time of vehicle k; t ij is the time for the vehicle to travel from stop i to stop j; τ iThe service time of the vehicle at station i; v0 is the depot, and it is assumed that the cost of the vehicle driving out of and returning to the depot is negligible, that is, the time for the vehicle to drive into the first station of the line and drive back to the depot from the last station of the line is 0; is the waiting time of vehicle k at station i; is the arrival time of vehicle k at station i; T r,b is the boarding time specified for passenger travel demand r;

[0157] Sb03. Obtain the objective function for maximizing the operating profit.

[0158] In a specific embodiment, the constraint conditions of the customized bus network optimization model are as follows:

[0159] A. Vehicle and service plan assignment constraints:

[0160] A1. It is set that each passenger travel demand can be served by at most one customized bus vehicle using one travel plan, and each demand can only be completed by the original travel plan or an alternative plan:

[0161]

[0162] A2. It is set that a passenger travel demand can only be assigned to the customized bus vehicles participating in the operation, and only when the vehicle participates in the operation is it allowed to drive on the line:

[0163]

[0164]

[0165] A3. Under normal circumstances, it is not allowed for the customized bus to complete the passenger travel service by means of transfer, that is, the passengers in each demand can only be transported by the same vehicle. It is set that the passengers in each demand can only be transported by the same vehicle:

[0166]

[0167]

[0168] B. Time constraints:

[0169] B1. For the passengers who are confirmed to be provided with customized bus travel services, for each travel plan providing services for them, set that the vehicle arrives at the boarding and alighting stations within the time window specified by the passengers:

[0170]

[0171]

[0172]

[0173]

[0174] B2. Set the access time between the front and rear stations; when vehicle k successively visits stations i and j, and station j is the immediate successor station of station i, the time constraint must be satisfied, where W is a sufficiently large positive number. This constraint means that the arrival time of vehicle k at station j is not earlier than the sum of the arrival time of vehicle k at station i, the waiting time of the vehicle, the service time, and the travel time from station i to station j:

[0175]

[0176] C. Access order restriction:

[0177] C1. For the passenger travel demands to be responded to, set that the service vehicle first visits the boarding station and then visits the alighting station:

[0178]

[0179] In the formula, is the total operating time of vehicle k between stations i and j;

[0180] D. Constraints on the departure and arrival of customized bus vehicles:

[0181] D1. Set that each bus participating in the operation departs from the depot and returns to the depot after completing the task:

[0182]

[0183]

[0184] E. Flow balance constraint:

[0185] E1. In the present invention, if a station has both pick-up and drop-off requirements or travel demands at different times, then the station is split into multiple stations with the same geographical location, and the travel time between these stations is zero. In this way, each station has only one boarding requirement or one alighting requirement. In this case, each station needs to be visited by the customized bus vehicle at most once. In addition, it is set that the customized bus vehicle participating in the operation must leave after visiting the stations other than the depot on the line:

[0186]

[0187]

[0188] F. Vehicle capacity constraint:

[0189] Set the number of passengers on the bus less than the maximum passenger capacity, and set the minimum passenger capacity requirement for each route:

[0190]

[0191]

[0192] In the formula, P i k is the number of people in vehicle k at stop i; is the number of boarding passengers of travel demand r at stop j; is the boarding stop of travel demand r; P kmax is the maximum passenger capacity of vehicle k; P kmin is the minimum passenger capacity requirement of vehicle k;

[0193] G. Constraints on the number of customized bus vehicles:

[0194] G1 Set that the number of customized bus vehicles put into operation cannot exceed the total number of vehicles in the fleet:

[0195]

[0196] In the formula, |K| represents the number of elements in set K, that is, the total number of vehicles in the fleet;

[0197] H. Constraints on the value of decision variables:

[0198] H1. Set the range of values of the decision variables of the service stability maximization model and the operation profit maximization model.

[0199]

[0200]

[0201]

[0202] In a specific embodiment, in step S3, according to the optimization model, based on the speed law of the customized bus road network sub-region and the travel demand of passengers, a random simulation and an improved NEWMAN algorithm are used to generate an initial customized bus line network, specifically:

[0203] S301. Generate an initial route:

[0204] ES101. For each stop in the customized bus road network, solve the shortest path between each stop through a matrix algorithm;

[0205] ES102. Targeting travel demands, a separate route is arranged for each travel demand, which is called the initial route; each initial route is: depot - boarding stop - alighting stop - depot; it is set that the cost for the operating vehicle to drive into the first stop of the route and drive back to the depot from the last stop after completing the transportation task is 0; the initial route meets all other constraints except the minimum passenger capacity constraint of the route.

[0206] ES103. If the customized bus vehicle cannot meet the travel time requirements of passengers when driving at the maximum speed, it is determined that the customized bus cannot provide services for such passenger travel demands, and such passenger travel demands are excluded to obtain the original customized bus network.

[0207] S302. Construct a combination library:

[0208] ES201. Randomly select two initial routes, and according to the shortest distance between stops on the route and the chronological order of travel times at stops, on the premise of meeting the requirements of the stop access order, splice the routes to form a new route.

[0209] ES202. If the new route violates the maximum passenger capacity constraint, terminate the route merger; otherwise, randomly select two initial routes again for merger until the new route meets the maximum passenger capacity constraint; the new route and other initial routes together form an initial solution.

[0210] ES203. Repeat the above steps M times until M route libraries are generated, and the set of route libraries is called the combination library.

[0211] S303. Calculate the fitness of each route by means of random simulation:

[0212] ES301. For each route, find the sub - areas where each stop is located and the sub - area distribution of the shortest paths between each pair of consecutive stops, and calculate the driving distances of the paths in each sub - area; conduct a number of random simulations, and each time randomly select a speed value based on the speed distribution law of the passing sub - areas, and calculate the travel times of the sub - paths between each pair of consecutive stops.

[0213] ES302. Gradually accumulate the travel times between consecutive stops along the route to calculate the actual arrival times of the customized bus vehicle at each stop on the route.

[0214] ES303. If the vehicle on the route can arrive at the relevant stops within the specified pick - up and drop - off time windows of passengers, it is determined that the route can pick up and drop off the passengers on time; count the number of passengers served on time and the total number of passengers for each route in each random simulation, and calculate the service stability of each route.

[0215] ES304. Statistically calculate the profit of each route in each random simulation:

[0216]

[0217] Among them, J is the number of simulations, I is the total number of line requirements for this simulation, and r ij is the profit of the i-th requirement in the j-th simulation, and u ij indicates whether the requirement i is met on time in the j-th simulation, which is a 0-1 variable.

[0218] ES305. Calculate the fitness of each line library in the combined library with the average service stability of each line and the average profit of each line as indicators, and generate a fitness matrix;

[0219] S304. Update the line library:

[0220] ES401. Perform TOPSIS scoring on each line in the line library according to the fitness matrix:

[0221] ① Calculate the distances of the positive and negative ideal solutions of each line library:

[0222]

[0223]

[0224] In the formula, is the distance from line l to the positive ideal solution; is the distance from line l to the negative ideal solution; υ j is the weight of the j-th index, F lj is the index value of line l on the j-th index; are the positive and negative ideal solutions of the j-th index.

[0225] ② Calculate the comprehensive score C l , C l →1 indicates that line l is better:

[0226]

[0227] ES402. Arrange the lines in descending order of score, select the first |K| lines to form valid lines, and the remaining lines as lines to be responded to;

[0228] ES403. Perform evolutionary operations on the line library through three branches:

[0229] XS1. Merge the valid lines and the lines to be responded to, that is, randomly select a valid line and a line to be responded to each time, merge them according to the rules, and check whether the maximum passenger capacity constraint is met until a new line is generated, and generate a new line library;

[0230] XS2. Randomly select two lines from the valid lines for merging until new lines that meet the maximum passenger capacity constraint are generated, and a new line library is produced;

[0231] XS3. Randomly select two lines from the lines to be responded for merging until new lines that meet the maximum passenger capacity constraint are generated, and a new line library is produced;

[0232] In this embodiment, each time the line library is updated, the total number of lines is reduced by one.

[0233] In this embodiment, when checking whether the new line meets the vehicle's maximum passenger capacity constraint, it is necessary to consider that the maximum capacities of different vehicle models are different. Assume that the vehicle's maximum passenger capacity is P1 max and and When the number of passengers on the vehicle is still less than P1 after merging two certain lines max , the lines can be merged; when the number of passengers on the vehicle exceeds after line merging, the merging of these two lines should be prohibited; when the number of passengers on the vehicle after line merging, it is necessary to compare the advantages and disadvantages of using small-capacity and large-capacity vehicle models. Therefore, the line merging branches with maximum passenger capacities of P1 max and are retained simultaneously: one is with P1 max as the maximum passenger capacity constraint. Since the number of passengers on this line will exceed P1 max after merging, the merging of this line must be prohibited, and other lines are selected for merging to complete the update of the line library. To expand the search scope during the search process, the line library obtained in this way is allowed not to be eliminated within the specified number of generations. The other is with as the maximum passenger capacity constraint, allowing the merging of this line to complete the update of the line library.

[0234] S305. Calculate the TOPSIS scores of the lines in the updated line library, and update the combination library according to the TOPSIS scores of the lines in the updated line library. Finally, retain the top M line libraries in the ranking to form the updated combination library:

[0235] (1) Perform the following operations on the line library updated with as the maximum passenger capacity constraint:

[0236] ① Calculate the fitness of the lines in each line library;

[0237] ② Use the TOPSIS method to score the lines in each line library and sort the lines in descending order. Select the top |K| lines to form the valid lines of this line library;

[0238] ③Use the total average service stability and total average profit of the effective lines as the evaluation criteria for the quality of each line library. The TOPSIS method is used to score each line library.

[0239] ④Retain the top M line libraries in terms of ranking.

[0240] (2) Perform the following operations on the line library updated with P1 max as the maximum passenger capacity constraint:

[0241] Allow this line library not to eliminate the good and keep the bad according to the TOPSIS score within the specified number of iterations; when the specified number of iterations is exceeded, merge it into the line library updated with as the maximum passenger capacity constraint. According to the TOPSIS scores of each line library, finally retain the top M line libraries in terms of ranking.

[0242] S306. Repeat steps S304 - S305 to iterate the combined library until there are no lines that can be further merged;

[0243] S307. Select the optimal line library from the iterated combined library:

[0244] ES701. Calculate the fitness of the lines in each line library in the iterated combined library;

[0245] ES702. Use the TOPSIS method to score the lines in each line library and sort the lines in descending order. Select the top |K| lines, and after eliminating the lines that do not meet the minimum passenger capacity constraint, form the effective lines;

[0246] ES703. Use the total average service stability and total average profit of the effective lines as the evaluation criteria for the quality of each line library, select the optimal line library, and use the optimal line library as the initial customized bus network:

[0247]

[0248] where m * is the optimal line library, and C m is the TOPSIS score of line library m.

[0249] In a specific embodiment, in step S4, for the unresponded passengers in the initial customized bus network, design an alternative travel plan set, and based on the passenger selection intention, guide the unresponded passengers in the initial plan to select the final travel plan from the alternative travel plan set. The specific steps are as follows:

[0250] S401. Obtain the guiding requirements of the unresponded passengers in the initial customized bus network, generate the corresponding list of guiding requirement station pairs and the station list of the planned lines;

[0251] S402. According to the list of demand sites to be guided, the site table of the planned route, and the demand to be guided, generate optional alternative travel plans by adjusting the boarding / alighting sites or boarding / alighting times, so that the passengers not responded to in the initial customized bus network become passengers who can be responded to by the customized bus;

[0252] S403. Comprehensively consider travel time, payment cost, passengers' sensitivity to time and cost, perception error, and external environment, and calculate the perceived cost of each plan; according to the perceived cost of each plan, and based on the principle of maximizing utility, obtain the plan with the highest probability of passengers' choice willingness, and take the plan with the highest probability of choice willingness as the travel plan of the passengers, otherwise the demand to be guided cannot be served.

[0253] In this embodiment, as Figure 3 shown, s1 is the original travel plan of travel demand r, and the specified boarding / alighting sites and boarding / alighting time windows are {i, j} and [T r,b ±Δ, T r,a ±Δ], where Δ is the time deviation allowed by the passenger, is the time when vehicle k arrives at site i. When the customized bus system cannot provide service for it with the original travel plan s1, it will try to adjust the boarding / alighting sites or boarding / alighting times to generate alternative travel plans s2, s3 or s4. In plan s2, the boarding / alighting sites are {i', j}, and the boarding time window is is the time length for the passenger to travel from i to i' by other means, and the alighting time window remains unchanged; in plan s3, the boarding / alighting sites are {i, j'}, the passenger's boarding time window remains unchanged, and the alighting time window is is the time length for the passenger to travel from j to j' by other means. In plan s4, the boarding / alighting sites are {i', j'}, and the passenger's boarding / alighting time windows are and

[0254] In this embodiment, since alternative plans s2, s3 and s4 all change the original travel requirements of the passengers and some distances need to be completed by other means of transportation (such as walking or shared bicycles, etc.), which reduces the comfort of the passengers' travel. Therefore, when the customized bus system cannot provide service for the passengers according to their original travel requirements, compensation can be provided through measures such as generalized fare discounts to guide the passengers to continue to choose alternative plans s2, s3 or s4 for travel.

[0255] In this embodiment, when a passenger cannot travel according to the original plan by customized bus, the travel plan will be reselected according to the perceived costs of different alternative plans. The perceived travel cost is related to the actual travel time and payment cost, and is affected by the passenger's sensitivity to time and cost as well as the perceived error. In addition, the external environment (such as weather conditions like temperature, whether it is raining, etc.) will also significantly affect the passenger's choice decision. Under weather conditions such as high temperature, low temperature or rain, passengers are more willing to travel by pure customized bus. Let: f ij be the travel cost from i to j by customized bus; be the travel cost from i to i' by other means; β r be the time sensitivity coefficient of travel demand r; γ r be the cost sensitivity coefficient of travel demand r; θ1 and θ2 are the generalized fare discount coefficients when adjusting one or two stops in the alternative plan respectively; ω1 and ω2 are the weights of travel time and travel cost respectively, and ω1 + ω2 = 1; δ ij be the perceived error of the travel cost for the passenger; α is the weather impact factor. When in weather conditions such as high temperature, low temperature or rain, α is a sufficiently large positive number, meaning that the passenger will not choose the alternative plan to travel; otherwise, α takes the value of 1, that is, the weather conditions do not affect the passenger's travel plan selection behavior. Ignoring the time loss of getting on and off the vehicle, the expected value of the perceived utility of passenger r adopting the alternative travel plan is as follows:

[0256]

[0257]

[0258]

[0259]

[0260] In the formula, and are the costs of passenger r adopting four travel plans, and are expressed in the form of generalized costs; therefore, the negative of it can be understood as the deterministic term of utility. There is an error in the passenger's perception of the cost. Therefore, assume that δ ij , δ i'j , δ ij' and δ i'j' are the error terms of the perceived cost, which follow the type-I extreme value distribution with parameter τ and are independently and identically distributed. The variance of this error term is According to this assumption, an explicit expression of the probability selection of the travel plan, that is, the logit model, can be obtained. Therefore, the probability n that passenger s chooses the travel plan can be estimated as:

[0261]

[0262] According to the principle of utility maximization, if the customized bus system cannot provide services to passengers based on the original travel plan, then on the premise that the generalized cost of the alternative plan is not higher than that of the original travel plan, passengers will choose the alternative plan with the highest probability as the final travel plan.

[0263] In this embodiment, S402. Generating optional alternative travel plans by adjusting the boarding / alighting stops or boarding / alighting times according to the list of stations with guiding requirements, the station list of the planned route, and the guiding requirements is specifically as follows:

[0264] Let the station pair in the guiding requirement be [A, B]. It is possible that there are multiple alternative travel plans or there may be no alternative travel plan. Specifically as follows:

[0265] ① If station A coincides with a certain station C on the planned route in terms of geographical location, then search for another station D on this planned route whose distance from station B does not exceed a given value, and try to guide the guiding requirement B to D. Calculating the boarding / alighting time window of the passenger at station D at this time based on the principle of not affecting the time for the passenger to reach the final destination. If there is an intersection between this boarding / alighting time window and the time window at the corresponding station on the planned route, then the alternative travel plan s2 can be generated.

[0266] ② If station B coincides with a certain station C on the planned route in terms of geographical location, then try to generate the alternative travel plan s3.

[0267] ③ If neither station A nor station B coincides with the stations on the planned route, then search for the stations on the planned route within a specified range from station A or B; then, search whether there is a station on this route whose distance from the other station of the guiding requirement meets the requirements. If so, calculate the boarding / alighting time window after adjusting the stations with guiding requirements. If there is an intersection between this time window and the time window at the corresponding station on the planned route, then the alternative travel plan s4 can be generated.

[0268] ④ If no alternative travel plan can be searched without adding routes, then this guiding requirement cannot be served.

[0269] As Figure 4As shown in the figure, the present invention collects existing traffic flow data, estimates traffic parameters of the customized bus road network and partitions the customized bus road network, extracts the vehicle speed law of the sub-areas of the customized bus road network, fully considering the randomness of the vehicle speed of the customized bus, which not only avoids the deviation caused by predicting the vehicle arrival time based on the average speed, but also circumvents the complexity of characterizing the vehicle speed law at the road level and the resulting large amount of computational work. The present invention also establishes an optimized model of the customized bus line network with the maximization of service stability and operation profit as the goals based on the actual random traffic road network environment. According to the optimized model, based on the speed law of the sub-areas of the customized bus road network and the travel demands of passengers, a random simulation and an improved NEWMAN algorithm are used to generate an initial customized bus line network, considering the maximization of service stability and operation profit. The present invention also designs an alternative travel plan set for the passengers not responded in the initial customized bus line network, and guides the passengers not responded in the initial plan to select the final travel plan from the alternative travel plans based on the passenger selection intention, fully considering the travel demand adjustment intention of the passengers not responded, thereby expanding the scope of the servable population. Finally, with the maximization of service stability and operation profit as the goals, the present invention uses a random simulation and an improved NEWMAN algorithm to overall re-optimize the initial customized bus line network, improving the optimization efficiency and accuracy of the customized bus line network. The present invention thus achieves the following: 1. The vehicle speed of the customized bus is regarded as a random variable, and the speed law is extracted based on the sub-area partition theory, which is both in line with the actual situation and reduces the complexity of characterizing the vehicle speed at the road level. 2. By means of random simulation, a customized bus line network plan with service stability and optimal profit under the condition of random speed is searched, which is conducive to the customized bus system to provide services with the planned routes and service quality, ensuring the service effect of the customized bus. 3. Breaking through the limitation of passively responding to the travel demands of passengers, an alternative travel plan set is designed for the passengers not responded in the initial line network, and the passengers of this type are guided to determine the final travel plan from the alternative travel plan set based on the passenger selection intention. On the one hand, the number of passengers served by the customized bus is increased, and on the other hand, more passengers are provided with environmentally friendly and high-quality bus services.

[0270] Embodiment 3

[0271] A customized bus line network optimization system considering service stability, comprising a data collection module, a vehicle speed law module, a model construction module, an initial customized bus line network module, a demand guidance module, and a model re-optimization module;

[0272] The data collection module is used to collect existing traffic demands and traffic flow data;

[0273] The vehicle speed law module is used to estimate traffic parameters of the customized bus road network and partition the customized bus road network, and extract the vehicle speed law of the sub-areas of the customized bus road network;

[0274] The model building module is used to establish a customized bus network optimization model under a random traffic environment of a road network according to the vehicle speed law, with the goal of maximizing service stability and maximizing operating profit;

[0275] The initial customized bus network module is used to generate the initial customized bus network according to the optimization model, based on the speed law of the customized bus network sub-area and the travel demand of passengers, using random simulation and improved NEWMAN algorithm;

[0276] The demand guidance module is used to design a set of alternative travel plans for passengers who have not responded in the initial customized bus line network, and guide the passengers who have not responded in the initial plan to select a final travel plan from the alternative travel plans based on the passengers' choice intention;

[0277] The model re-optimization module is used to optimize the initial customized bus network as a whole by using random simulation and improved NEWMAN algorithm based on the initial customized bus network, the travel plans of passengers who were not responded to in the initial plan, and the vehicle speed rules of the customized bus network sub-areas, with the goal of maximizing service stability and maximizing operating profits, so as to obtain the optimal customized bus network.

[0278] In this embodiment, Figure 5 , Figure 6 As shown, the present invention analyzes the driving rules of the bus network based on the GPS data of Guangzhou buses, and uses a partitioning algorithm to divide the road network in the central urban area of ​​Guangzhou into sub-areas. Figure 6 As shown, after partitioning, the road network is divided into 5 homogeneous sub-areas.

[0279] In this embodiment, according to the relationship between the average density and speed of buses in each sub-area, the speed distribution of each sub-area within the customized bus network planning period is fitted. For example, the speed value of sub-area 4 is fitted by using Laplace distribution, t distribution and normal distribution, as shown in FIG. Figure 8 Table 1 shows the speed distribution of this area in each model fitting performance. The results show that the normal distribution has the best fitting effect, so the normal distribution is used to fit the speed law of this sub-area. The bus speeds of each sub-area can be fitted by the normal distribution. The mean and standard deviation of the bus speeds are shown in Table 2.

[0280] Table 1 Fitting performance of each model in sub-area 4

[0281]

[0282] Note: SSE refers to the sum of squared errors; AIC refers to Akaike information criterion; BIC refers to Bayesian information.

[0283] Table 2 Distribution of vehicle speeds in different zones

[0284]

[0285] In this embodiment, the present invention obtains the shortest distances between all road network connection nodes in the topological road network through a matrix algorithm, and constructs a shortest distance matrix. By matching the passenger demand points to the road network nodes and using them as indexes, the shortest distance information between road network nodes can be searched. Then, based on the sub-areas corresponding to the inter-station paths and the driving distances within these sub-areas, a speed simulation experiment based on the partition speed distribution is carried out to obtain the operating speed of the customized bus vehicles, and the travel time of the links passed by the line is calculated accordingly.

[0286] In this embodiment, based on the randomness of the customized bus speed, aiming at the maximum service stability and the maximum profit of the operating enterprise, considering several actual constraint conditions, a multi-model and capacity-limited customized bus network optimization scheme can be obtained by using random simulation and the improved NEWMAN algorithm.

[0287] In this embodiment, the method proposed by the present invention is used to conduct a simulation analysis on the optimization of the customized bus network in Guangzhou. The number of customized bus travel demands during a certain period on a certain day in a certain month and year in Guangzhou is 455, and some detailed demand information is as Figure 7 shown. The fleet size is 85 vehicles, among which 16 are 22-seat vehicles and 69 are 45-seat vehicles. The fixed cost of the 22-seat vehicle is 130 yuan per trip, and the variable cost of the vehicle is 2.3 yuan / km. The fixed cost of the 45-seat vehicle is 150 yuan per trip, and the variable cost of the vehicle is 3.2 yuan / km. The time consumption for passengers to get on and off the vehicle is 2 seconds per time. It is assumed that the sensitivity degree (β) of passengers to travel time is 0.4, and the sensitivity degree (γ) to the customized bus fare is 0.6. The weights ω1 and ω2 of travel time and bus fare are 0.4 and 0.6 respectively. The generalized fare discount coefficients θ1 and θ2 when adjusting one or two stops for alternative travel plans are 0.8 and 0.6 respectively. The weather influence factor α is taken as 1, and the customized bus fare rule is 4 yuan within 3 km, and 2 yuan / km for the part exceeding 3 km.

[0288] In this embodiment, by using the method proposed by the present invention, an optimized customized bus network is obtained. The service stability of the network is 87.58%, and the income of the operating enterprise is 3,131.540 yuan. The change trend of the objective function value is as Figure 8 shown.

[0289] In this embodiment, the network optimization effect obtained by the method proposed by the present invention is compared with the network optimization effects of a single-model fleet, using the average speed of the customized bus partition road network, and only based on the principle of passive demand response. As shown in Table 3. It can be seen from the table that by using the method proposed by the present invention, a customized bus network scheme with better stability and profit can be obtained, and the effect obtained by providing services with a multi-model fleet in the operation of customized buses is better.

[0290] Table 3 Comparison Table of Maximum Benefits and Stabilities of Different Vehicle Models and Speeds

[0291]

[0292] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, rather than limitations on the implementation manners 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. A customized bus network optimization method considering service stability, characterized in that: The following steps are involved: S1. Collect existing traffic flow data, estimate traffic parameters of the customized bus network and partition the customized bus network, and extract vehicle speed patterns in the customized bus network sub-areas; S2. According to the vehicle speed law, a customized bus network optimization model is established under the random traffic environment of the road network with the goal of maximizing service stability and operating profit; Specifically:

201. Based on passenger travel demand and customized vehicle speed patterns in bus network sub-areas, we seek a network optimization solution that optimizes the proportion of passengers served on time and obtain the objective function for maximizing service stability: Among them, \(K\) is the set of vehicles in the customized bus fleet of the operating enterprise, \(V\) b and \(V\) a are respectively the set of boarding stops and the set of alighting stops of the traveling passengers. is the number of boarding passengers of passenger travel demand \(r\) at stop \(i\). is a 0-1 decision variable, which takes the value of 1 when vehicle \(k\) travels from stop \(i\) to stop \(j\), and 0 otherwise. is an indicator variable, which takes the value of 1 when vehicle \(k\) arrives at stops \(v\) i and \(v\) j on time, and 0 otherwise.

202. According to the passengers' travel plans and the existing bus data, the operating company's revenue and the cost paid by the operating company are calculated to maximize the operating company's profit, and the operating profit maximization objective function is obtained: max F2=GC Among them, G is the income of the operating enterprise, and C is the cost paid by the operating enterprise; 203. Based on the objective function of maximizing service stability and operating profit, and according to the law of vehicle speed, a customized bus network optimization model with the goal of maximizing service stability and operating profit is established under the random traffic environment of the road network; S3. According to the customized bus network optimization model, based on the speed law and passenger travel demand of the customized bus network sub-area, random simulation and improved NEWMAN algorithm are used to generate the initial customized bus network; S4. Design a set of alternative travel plans for passengers who have not responded to the initial customized bus network, and guide the passengers who have not responded to the initial plan to choose the final travel plan from the set of alternative travel plans based on the passengers' choice intention; S5. Based on the initial customized bus network, the travel plans of the unresponsive passengers in the initial plan, and the vehicle speed patterns in the customized bus network sub-areas, with the goal of maximizing service stability and operating profit, the initial customized bus network is re-optimized as a whole using random simulation and the improved NEWMAN algorithm to obtain the optimal customized bus network.

2. The customized bus network optimization method considering service stability according to claim 1, characterized in that: In the step S1, the traffic parameters of the customized bus network are estimated, specifically: Among them, v (b) (j1, m) is the average speed of the bus on section j1 during period m; b represents the bus; t (b) (i1, m) and d (b) (i1, m) is the driving time and driving distance of bus trip i1 during period m, respectively; T (b) (j1, m) and D (b) (j1, m) are the total driving time and driving distance of buses on section j1 during period m; k (b) (j1, m) is the average density of buses on section j1 during period m; l(j1) is the length of section j1; τ is the length of the aggregation period; k (b) (m) is the average density of buses weighted by section length in the area; J is the number of bus sections in the area; L is the total lane length; q (b) (m) is the average bus flow in the area; s(j1) is the number of lanes of section j1; T (b) (m) and D (b) (m) are the total driving time and total driving distance of buses during period m respectively; v (b) (m) is the average speed of buses in the area during period m.

3. The customized bus network optimization method considering service stability according to claim 1, characterized in that: In step S1, the partitions of the bus network are customized, and the specific steps are as follows: S101. Divide the road into several sections and calculate the section speed of the customized bus according to the existing traffic flow data; S102. Clustering the road sections based on the customized bus section speed, the distance between the road sections and the road section connectivity to form several initial communities; S103. Taking the minimization of the total variance of the bus route speed of the customized bus route network as the goal, considering the adjacency relationship between the initial sub-areas, the initial sub-areas are merged to form several sub-areas; With the goal of reducing the total speed variance of bus route sections in the road network, the sub-areas to which the route sections on the boundary belong are adjusted to obtain customized bus route network partitions.

4. The customized bus network optimization method considering service stability according to claim 1, characterized in that: In the step S1, the vehicle speed pattern of the customized bus network sub-area is extracted, and the specific steps are as follows: ES1. According to the average bus flow, average density and average speed of each sub-area of ​​the customized bus network partition, the relationship between the average density and average speed of the sub-area buses is identified to obtain the operation rules of the bus sub-area; ES2. According to the time period optimized for the customized bus network, find out the distribution section of the average density of customized bus vehicles in the road network of each sub-region within this time period and the possible values of the speed of customized buses corresponding to this density section, conduct a fitting of the average speed distribution of customized buses, and extract the speed distribution law of bus vehicles within the time period to be optimized for the bus network.

5. The customized bus network optimization method considering service stability according to claim 1, characterized in that: In step 202, according to the travel plans of passengers and the existing bus data, calculate the revenue of the operating enterprise and the costs paid by the operating enterprise, maximize the profit of the operating enterprise, and obtain the objective function for maximizing the profit of the operating enterprise. The specific steps are as follows: b01. According to the travel plans of passengers, calculate the service revenue G of the operating enterprise: where $R$ is the set of passenger travel demands; is the set of boarding and alighting stops when travel plan $s$ is adopted for passenger travel demand $r$ n ; is the assignment decision variable for passenger travel demand. It takes the value of 1 when passenger travel demand $r$ is assigned to vehicle $k$ and travel plan $s$ n is adopted, and 0 otherwise; b02. According to the bus network to be planned, considering the driving costs of vehicles on the line, the waiting costs of vehicles arriving at the stations earlier than the specified time window of passengers, and the service costs of passengers getting on and off, obtain the cost C of the operating enterprise: Among them, V is the set of stops for the proposed customized bus network; S is the set of travel plans; μ k is the fixed cost of scheduling vehicle k; η k is a 0-1 decision variable, which takes the value of 1 when vehicle k participates in operation, otherwise 0; λ k is the converted cost per unit time of vehicle k; t ij is the time for the vehicle to travel from stop i to stop j; τ i The service time of the vehicle at stop i; v0 is the depot, and it is assumed that the cost of the vehicle driving out of and returning to the depot is negligible, that is, the time for the vehicle to drive into the first stop of the line from the depot and the time to drive back to the depot from the last stop of the line are 0; is the waiting time of vehicle k at stop i; is the arrival time of vehicle k at stop i; T r,b is the boarding time specified by the passenger travel demand r; Sb03. Obtain the objective function for maximizing the operating profit.

6. The customized bus network optimization method considering service stability according to claim 1, characterized in that: The constraint conditions of the customized bus network optimization model are as follows: A. Vehicle and service plan assignment constraints: A1. It is set that each passenger travel demand can be served by at most one customized bus vehicle using one travel plan. A2. It is set that a passenger travel demand can only be assigned to the customized bus vehicles participating in the operation, and it is only allowed to drive on the line when the vehicle participates in the operation. A3. It is set that all passengers in the same demand can only be transported by the same vehicle. B. Time constraints: B1. For passengers who are confirmed to be provided with customized bus travel services, for each travel plan providing services to them, set the vehicle to arrive at the boarding and alighting stations within the specified time window of the passengers. B2. Set the access time between the front and rear stations. C. Access order restrictions: C1. For the passenger travel demands that are responded to, set the service vehicle to first visit its boarding station and then its alighting station. D. Customized bus vehicle departure and arrival constraints: D1. It is set that each bus participating in the operation departs from the depot and returns to the depot after completing the task. E. Flow balance constraints: E1. It is set that if a station has pick-up and drop-off requirements at the same time or travel demands of passengers at different times, it is split into multiple stations with the same geographical location, and the travel time between these stations is zero. F. Vehicle capacity constraints: F1 It is set that the number of passengers on the bus is less than the maximum passenger capacity, and the minimum passenger capacity requirement for each line is set. G. Customized bus vehicle quantity constraints: G1 It is set that the number of customized bus vehicles put into operation cannot exceed the total number of vehicles in the fleet. H. Decision variable value constraints: H1. Set the range of values of the decision variables of the service stability maximization model and the operating profit maximization model.

7. The customized bus network optimization method considering service stability according to claim 5, characterized in that: In step S3, according to the optimization model, based on the speed law of the customized bus road network sub-region and the travel demands of passengers, use the stochastic simulation and improved NEWMAN algorithm to generate an initial customized bus network. Specifically: S301. Generate initial routes: ES101. For each station in the customized bus road network, solve the shortest path between each station through the matrix algorithm. ES102. Taking travel demands as the object, a separate route is arranged for each travel demand, which is called the initial route; each initial route is: depot - boarding station - alighting station - depot; it is set that the cost for the operating vehicle to drive into the first station of the route and drive back to the depot from the last station after completing the transportation task is 0; the initial route satisfies all other constraints except the minimum passenger capacity constraint of the route. ES103. If the customized bus vehicle cannot meet the travel time requirements of passengers when driving at the maximum speed, it is judged that the customized bus cannot provide services for such passenger travel demands, and such passenger travel demands are excluded to obtain the original customized bus network. S302. Construct a combination library: ES201. Randomly select two initial routes. According to the shortest distance between stations on the routes and the chronological order of travel times at stations, on the premise of meeting the requirements of the station access order, splice the routes to form a new route. ES202. If the new route violates the maximum passenger capacity constraint, terminate the route merger; otherwise, randomly select two initial routes again for merger until the new route meets the maximum passenger capacity constraint. The new route and other initial routes together form an initial solution. ES203. Repeat the above steps M times until M route libraries are generated. The set of route libraries is called the combination library. S303. Calculate the fitness of each route by means of random simulation: ES301. For each route, find out the sub - areas where each station is located and the sub - area distribution of the shortest paths between each consecutive pair of stations, and calculate the driving distances of the paths in each sub - area; conduct a number of random simulations. Each time, randomly select a speed value based on the speed distribution law of the passing sub - areas, and calculate the travel times of the sub - paths between each consecutive pair of stations. ES302. Gradually accumulate the travel times between consecutive stations along the route to calculate the actual arrival times of the customized bus vehicles at each station on the route. ES303. If the vehicle on the route can arrive at the relevant stations within the specified pick - up and drop - off time windows of passengers, it is judged that the route can pick up and drop off the passengers on time. Count the number of passengers served on time and the total number of passengers for each route in the random simulation, and calculate the service stability of each route. ES304. Count the profits of each route in each random simulation. ES305. Calculate the fitness of each route library in the combination library with the mean service stability of each route and the profit of each route as indicators, and generate a fitness matrix. S304. Update the route library: ES401. According to the fitness matrix, conduct TOPSIS scoring for each route in the route library. ES402. Sort the lines in descending order of score, and select the first |k 1| lines to form valid lines, and the remaining lines are used as lines to be responded to; ES403. Perform evolutionary operations on the route library through three branches: XS1. Merge the effective routes and the routes to be responded to, that is, randomly select an effective route and a route to be responded to each time, merge them according to the rules, and check whether the maximum passenger capacity constraint is met until a new route is generated, generating a new route library. XS2. Randomly select two routes from the effective routes for merger until a new route that meets the maximum passenger capacity constraint is generated, generating a new route library. XS3. Randomly select two routes from the waiting response routes and merge them until a new route that meets the maximum passenger capacity constraint is generated, and a new route library is generated; S305. Calculate the TOPSIS scores of the routes in the updated route library, and update the combination library according to the TOPSIS scores of the routes in the updated route library, and finally retain the route libraries ranked in the top M to form the updated combination library; S306. Repeat steps S304-S305 to iterate the combination library until there are no more lines that can be merged; S307. Select the optimal circuit library from the iterated combination library: ES701. Calculate the fitness of each circuit in the iterated combination library; ES702. Use the TOPSIS method to score the lines in each line library, sort the lines in descending order, and select the lines ranked in the top |k 1| places. After excluding the lines that do not meet the minimum passenger capacity constraint, form the effective lines; ES703. The total average service stability and total average profit of effective routes are used as the evaluation criteria for the quality of each route library, and the optimal route library is selected and used as the initial customized bus line network.

8. The customized bus network optimization method considering service stability according to claim 1, characterized in that: In step S4, alternative travel plans are designed for passengers who have not responded in the initial customized bus line network, and the passengers who have not responded in the initial plan are guided to select the final travel plan from the alternative travel plans based on the passengers' choice intention. The specific steps are: S401. Obtain the unanswered passenger guidance needs in the initial customized bus line network, and generate a corresponding list of station pairs to be guided and a station table of the planned route; S402. According to the list of sites to be guided, the site table of the planned route, and the demand to be guided, an optional alternative travel plan is generated by adjusting the boarding / disembarking sites or boarding / disembarking times, so that the passengers who are not responded to in the initial customized bus line network become passengers who can be responded to by the customized bus; S403. Calculate the perceived cost of each plan by comprehensively considering travel time, payment fees, passenger sensitivity to time and fees, perception error, and external environment; obtain the plan with the highest probability of passenger willingness to choose based on the perceived cost of each plan and the principle of utility maximization, and use the plan with the highest probability of passenger willingness to choose as the passenger's travel plan, otherwise the demand to be guided cannot be served.

9. A customized bus network optimization system considering service stability, characterized in that: Used to implement the customized bus network optimization method as described in any one of claims 1 to 8, comprising a data acquisition module, a vehicle speed law module, a model building module, an initial customized bus network module, a demand guidance module, and a model re-optimization module; The data acquisition module is used to collect existing traffic demand and traffic flow data; The vehicle speed law module is used to estimate the traffic parameters of the customized bus network and partition the customized bus network, and extract the vehicle speed law of the customized bus network sub-area; The model building module is used to establish a customized bus network optimization model under a random traffic environment of a road network according to the vehicle speed law, with the goal of maximizing service stability and maximizing operating profit; The initial customized bus network module is used to generate the initial customized bus network according to the optimization model, based on the speed law of the customized bus network sub-area and the travel demand of passengers, using random simulation and improved NEWMAN algorithm; The demand guidance module is used to design a set of alternative travel plans for passengers who have not been responded to in the initial customized bus line network, and determine the travel plans of passengers who have not been responded to in the initial plan based on the passengers' choice intention; The network re-optimization module is used to optimize the initial customized bus network as a whole by using random simulation and improved NEWMAN algorithm based on the initial customized bus network, the travel plans of passengers who were not responded to in the initial plan, and the vehicle speed rules of the customized bus network sub-areas, with the goal of maximizing service stability and operating profit, to obtain the optimal customized bus network.

Images