A bus line network optimization method based on multi-mode travel data

Abstract

The application discloses a kind of public transport line network optimization methods based on multi-mode travel data, including steps S1: collecting multi-mode traffic data information;Step S2: extracting single trip start location in traffic data information, trip end location, trip start and end time information;Step S3: grid city space, city space is divided into multiple regions according to uniform rule;Step S4: the extracted traffic data information is mapped to the grid of divided city space, and the data is classified according to demand;Step S5: determine the public transport line optimization scheme according to the demand classification result, and execute optimization according to scheme.The city traffic informationization foundation is used, and the passenger flow of bus, taxi, online car, private car, shared bicycle and other multiple traffic modes is included in passenger flow demand category, and effective potential public transport passenger flow demand is identified, and on this basis, multi-objective demand is carried out Line network optimization decision, provide a more accurate optimization method.

Description

Technical Field

[0001] This invention relates to the field of vehicle scheduling technology, specifically a method for optimizing public transport networks based on multi-modal travel data. Background Technology

[0002] Most existing public transport network optimization technologies rely on historical public transport passenger flow for optimization, rarely considering passenger flow from other modes of transportation as potential public transport users. Therefore, their optimization methods are too simplistic, insufficiently thorough, and imprecise. While some new optimization methods have been disclosed in existing technologies, some lack sufficient accuracy, while others are complex and difficult to implement.

[0003] For example, the public announcement (CN109543895A) discloses a method for optimizing public transport networks based on taxi passenger flow conversion. This method proposes to convert taxi passenger flow into public transport passenger flow. However, this method uses taxi GPS data to extract the origin-destination (OD) of taxi passenger flow, ignoring the fact that taxis are in motion whether they are picking up or looking for passengers. Due to the uncertainty of road conditions, it is difficult to distinguish between picking up and looking for passengers from GPS alone. Therefore, it is impossible to accurately extract the OD of taxi passenger flow, which will lead to inaccurate data on the conversion into potential public transport passenger flow, and the network optimization will be difficult to achieve its intended effect.

[0004] For example, the method for optimizing and adjusting conventional public transport networks based on mobile phone signaling data, disclosed in Publication No. CN106503843A, proposes to obtain the origin-destination (OD) of motorized travel based on signaling data, and use this OD as the passenger flow demand for public transport network optimization. Although this method treats all motorized travel demand as public transport travel demand, the large volume of mobile phone signaling location data and the frequent occurrence of mobile phone users holding the same or different operator SIM cards lead to high external requirements and costs for data processing, as well as significant uncertainty in expanding the mobile phone user sample. Furthermore, mobile phone signaling data is continuous data. The proposed method for obtaining the OD of motorized travel by judging it through mobile phone dwell time and travel distance, with the dwell time threshold determined comprehensively based on residents' travel habits and the time range of mobile phone data in the study area, requires conducting corresponding surveys of residents' travel habits. This not only incurs manpower and other costs, but also makes it difficult to guarantee the validity and applicability of the obtained data. Therefore, this method is costly, lacks data accuracy, fails to accurately grasp the passenger flow demand for network optimization, is unsuitable for continuous network optimization, and its practical effectiveness is difficult to guarantee. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention provides a method for optimizing public transport networks based on multi-modal travel data.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A public transport network optimization method based on multimodal travel data, including

[0008] Step S1: Collect multi-modal transportation data;

[0009] Step S2: Extract the origin location, destination location, and start and end time information of a single trip from the traffic data;

[0010] Step S3: Grid-based urban space, dividing the urban space into multiple areas according to unified rules;

[0011] Step S4: Map the extracted traffic data information onto the divided urban spatial grid and classify the data according to demand;

[0012] Step S5: Determine the bus route optimization plan based on the demand classification results, and implement the optimization according to the plan.

[0013] Further, step S1 includes step S11: obtaining public transport passenger flow data by using the transportation business system or by conducting routine passenger flow OD derivation; step S12: obtaining taxi and ride-hailing passenger flow data by using taxi passenger pick-up and drop-off pressure meter data and ride-hailing order data; step S13: obtaining private car flow data by using private car entry and exit parking lot data; and step S14: obtaining shared bicycle passenger flow data by using shared bicycle order data and the number of vehicles in shared bicycle parking areas.

[0014] Furthermore, in step S2, the origin, destination and corresponding time of a complete single bus trip are extracted. A passenger's multiple consecutive internal bus transfers in a single trip are considered as a single bus trip. If a passenger can reach point D from point O through one or more bus routes, then a bus link is considered to exist between grid point O and point D.

[0015] Furthermore, in step S3, the grid size of the urban area is smaller than the grid size of the suburban area.

[0016] Further, step S4 includes step S41: mapping the origin and destination of a single trip of multiple types of transportation to the urban spatial grid; step S42: based on the arrival time at the destination, dividing and aggregating the travel behavior data of the same OD pair every ten minutes to obtain the travel demand matrix of multiple modes of transportation.

[0017] Furthermore, the demand matrix is ​​as follows:

[0018]

[0019]

[0020] Among them, the maximum value of o and d is equal to the number of city grids, and the maximum value of t is 144.

[0021] Furthermore, it also includes step S43: extracting long-distance travel data of shared bicycles to obtain the long-distance shared bicycle travel demand matrix E1; step S44: extracting shared bicycle supply shortage travel data to obtain the shared bicycle travel demand matrix E2 for different time periods throughout the day; step S45: removing duplicate data in E2 and E1 to obtain the shared bicycle travel demand matrix E3.

[0022] Furthermore, step S5 includes step S51: mining potential passenger flow demand for public transportation from multi-modal travel data; step S52: optimizing the public transportation network based on potential demand; step S53: optimizing the public transportation network based on remaining capacity; and step S54: determining public transportation route optimization measures.

[0023] Furthermore, step S52 includes two optimization methods, one of which is capacity expansion, when... That is, the bus link between point O and point D is connected, and simultaneously satisfies:

[0024]

[0025]

[0026]

[0027] Then 50% As a potential demand, the link between O and D needs to be expanded; the second option is to build a new one. If there is no bus link between point O and point D, then 50% A new channel will be built to connect points o and d as a potential demand.

[0028] Furthermore, step S53 includes two optimization methods: first, for links with remaining capacity that are guaranteed lines, the original channel capacity is reduced using the larger of the actual demand and the guaranteed demand for each time period as the capacity reference value; second, for links with remaining capacity that are not guaranteed lines, when... Then cancel the bus link between point O and point D; when At that time, based on actual needs The original channel capacity was reduced as a capacity reference value.

[0029] The beneficial effects of this invention are:

[0030] This invention provides a public transport network optimization method based on multi-modal travel data, breaking through the limitations of previous public transport passenger flow demand analysis. Utilizing the urban transportation information infrastructure, it incorporates passenger flow from various modes of transportation, including buses, taxis, ride-hailing services, private cars, and shared bicycles, into the passenger flow demand scope, identifying effective potential public transport passenger flow demand. Based on this, it conducts network optimization decisions based on multi-objective demand, providing a more precise optimization method. This method leverages existing transportation business systems to collect data, taking advantage of high data quality and low processing costs, and progressively carries out network optimization work, providing a new approach and method for public transport network optimization. Attached Figure Description

[0031] Figure 1 This is a flowchart of the public transport network optimization method based on multi-modal travel data according to the present invention. Detailed Implementation

[0032] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0033] Please refer to Figure 1 A public transport network optimization method based on multi-modal travel data, including

[0034] Step S1: Collect multi-modal transportation data;

[0035] Step S2: Extract the origin location, destination location, and start and end time information of a single trip from the traffic data;

[0036] Step S3: Grid-based urban space, dividing the urban space into multiple areas according to unified rules;

[0037] Step S4: Map the extracted traffic data information onto the divided urban spatial grid and classify the data according to demand;

[0038] Step S5: Determine the bus route optimization plan based on the demand classification results, and implement the optimization according to the plan.

[0039] Specifically, step S1 requires collecting bus passenger flow data, taxi passenger flow data, private car passenger flow data, and shared bicycle passenger flow data. This includes step S11: obtaining bus passenger flow data using the transportation business system or by conducting routine passenger flow OD derivation; step S12: obtaining taxi or ride-hailing passenger flow data using taxi pick-up and drop-off meter data and ride-hailing order data; step S13: obtaining private car passenger flow data using private car entry and exit parking lot data; and step S14: obtaining shared bicycle passenger flow data using shared bicycle order data and the number of shared bicycles in shared bicycle parking areas.

[0040] Due to the different nature of parking lots, the different information collected at entrances and exits, and the difficulty in determining whether it is a single trip or multiple trips, the data on private cars entering and exiting parking lots needs to be processed by a professional system. The data on private cars entering and exiting parking lots here refers to the complete data on a single trip of a vehicle entering and exiting the parking lot after being extracted and processed by the professional system. If the city's parking lot travel data has not yet been processed by a relevant system, then this step only needs to extract data on buses, taxis, ride-hailing vehicles, and shared bicycles.

[0041] In step S2, the multi-mode data collected in step S1 is extracted. The extraction rules are as follows: extract the starting point location, ending point location, and start and end time information of each trip from each data point. Among them, the extraction of bus information is based on the complete single trip, extracting the starting point, ending point, and corresponding time. A passenger's multiple consecutive internal bus transfers in a single trip are considered as one bus trip. If a passenger can reach point D from point O through one or more bus routes, then a bus link is considered to exist between grid point O and point D.

[0042] In step S3, the purpose of gridding the urban space is to integrate the origin and destination information of the five different modes of transportation mentioned above. The entire city will be uniformly divided into grids according to relevant urban transportation planning and management standards. If the transportation management service department already has relevant division standards, they can be directly adopted. Otherwise, during the division process, the grid size in the central urban area will be smaller than that in the suburbs. Overly large grids in the central urban area will, on the one hand, prolong the walking time for passengers after disembarking, and on the other hand, cause traffic congestion due to relatively scarce road resources and concentrated bus arrivals at stations.

[0043] Step S4 analyzes travel demand, including step S41: mapping the origin and destination of a single trip for multiple types of transportation to the urban spatial grid; step S42: based on the arrival time at the destination, dividing and aggregating the travel behavior data of the same OD pair every ten minutes to obtain travel demand matrices for multiple modes of transportation.

[0044] Specifically, the origin and destination of a single trip in the fields of public transportation, taxis, ride-hailing, private cars, and shared bicycles are mapped onto the urban spatial grid. Based on the time of arrival at the destination, the travel behavior data of the same origin and destination are divided and collected in ten-minute intervals to obtain the travel demand matrices of the above five modes of transportation, which are named matrices A, B, C, D, and E respectively.

[0045] Because all modes of transportation share a unified spatial grid and a unified time division, the travel demand matrix for each mode of transportation takes the following form, taking matrix A as an example:

[0046]

[0047]

[0048] The maximum value of o and d is equal to the number of city grids, and the maximum value of t is 144.

[0049] Step S4 also includes step S43: extracting long-distance travel data of shared bicycles to obtain the long-distance shared bicycle travel demand matrix E1; step S44: extracting shared bicycle supply shortage travel data to obtain the shared bicycle travel demand matrix E2 for different time periods throughout the day; step S45: removing duplicate data in E2 and E1 to obtain the shared bicycle travel demand matrix E3.

[0050] Specifically, the first step is to identify the potential demand areas for public transportation. Following the public transportation priority development strategy, efforts should be made to provide public transportation services to meet residents' travel needs as much as possible. Subways and buses are the mainstays of urban public transportation. Subways, due to their timeliness and punctuality, are the preferred choice for most public transportation passengers. Potential public transportation demand is mainly explored from individualized travel modes such as taxis, ride-hailing services, private cars, and shared bicycles. Attracting passengers from personalized motorized travel modes like taxis and ride-hailing services to public transportation is a direct measure to implement public transportation priority. For shared bicycles, as an environmentally friendly and economical mode of transportation, the goal of exploring potential public transportation demand is not simply to attract passengers, but to uncover unmet travel needs through shared bicycle usage data.

[0051] Analyzing shared bike data reveals two types of implicit public transportation demand: The first is the demand for relatively long-distance shared bike trips. If passengers have convenient public transportation options, this demand may translate into bus passenger flow. The second is the travel demand that the existing shared bike supply cannot meet. For example, passengers want to ride a shared bike, but there are not enough shared bikes available nearby. This phenomenon is very obvious at subway stations in the suburbs and generally exhibits tidal characteristics.

[0052] Before proceeding with step S43, it is necessary to distinguish between actual demand and potential demand. Actual public transport travel behavior that has already occurred is considered actual public transport demand, while non-occurring public transport travel behavior is considered potential demand.

[0053] In step S43, long-distance travel data for shared bicycles is extracted. According to the "2022 Report on Shared Bicycle / Electric Bicycle Riding in Major Chinese Cities," the average distance of a single shared bicycle ride is 1.5 kilometers. Considering that this distance is gradually increasing, the OD (Original Distance) pairs in the shared bicycle travel demand matrix are filtered out for distances greater than or equal to 2 kilometers, and shared electric bicycles are filtered out for distances greater than or equal to 3 kilometers, resulting in the long-distance shared bicycle travel demand matrix E1.

[0054] In step S44, data on shared bicycle demand imbalance is extracted. The outflow and stock data of shared bicycles within the same spatial grid are observed; that is, the number of shared bicycle orders starting from that grid and the available shared bicycle data within that grid are analyzed on the same time axis. If, within a short period of time, such as ten minutes, the outflow data of shared bicycles continuously accumulates, causing the stock of shared bicycles in that grid to decrease to zero, it reflects that the demand for shared bicycles in that grid during that time period is insufficient, and there is a need to operate bus routes. This determines the starting point O of potential bus demand. The ending point D can be analyzed based on the ending point of shared bicycle orders during that time period. The order ending points during that time period are aggregated according to the grid to obtain the shared bicycle travel OD matrix for that time period. The same process yields the shared bicycle travel demand matrix E2 for different time periods throughout the day. Since E2 contains OD pairs with distances greater than or equal to 2 kilometers, to avoid double counting and affecting data accuracy, data satisfying the E1 condition in E2 are removed, resulting in the shared bicycle travel demand matrix E3.

[0055] Step S5 includes step S51: mining potential passenger flow demand for public transportation from multi-modal travel data; step S52: optimizing the public transportation network based on potential demand; step S53: optimizing the public transportation network based on remaining capacity; and step S54: determining public transportation route optimization measures.

[0056] In step S51, passenger data with the same origin-destination (OD) pair from the relevant travel data of taxis, ride-hailing services, private cars, and shared bicycles are merged and calculated as F = B∪C∪D∪E1∪E3 to obtain the combined travel demand matrix F. In reality, urban passenger transport modalities are a complex game between various supply schemes and numerous service demands, influenced by factors such as travel preferences, road network structure, road conditions, and weather. A large number of passengers in each OD pair in the combined travel demand matrix F does not necessarily indicate potential public transport passenger demand.

[0057] This application considers a scenario where the actual demand for public transportation at a certain origin-destination (OD) exceeds the current public transportation capacity. In this case, other modes of transportation for that OD may be converted into potential public transportation demand. For the same OD, if public transportation capacity is sufficient, then passenger flow from other modes of transportation is not within the scope of this conversion. Therefore, if... but There is potential passenger flow for public transportation in the area. This refers to the passenger flow capacity from origin o to destination d within time period t in the current public transportation service supply. This capacity can be calculated from the number of bus trips and the capacity of the buses within the time period. This represents the actual passenger flow from the starting point o to the ending point d within time period t.

[0058] In step S52, the bus network not only achieves geographical connectivity but also needs to meet travel demand during specific time periods in terms of corridor capacity. Considering residents' habits, the bus network generally won't undergo major adjustments in a short period. To meet potential demand, it's necessary to optimize the existing bus network X... 0 Adding capacity to existing routes can be divided into two categories. The first is to expand the capacity of existing routes, which is common in the optimization of bus networks. The second is to build new routes, which is mainly applicable to new urban areas or suburban areas.

[0059] Regarding expansion, when That is, the bus link between point O and point D is connected, and simultaneously satisfies:

[0060]

[0061]

[0062]

[0063] Then 50% As a potential requirement, the link between O and D needs to be expanded.

[0064] In the above method This refers to the actual bus passenger flow from origin o to destination d within time period t. This refers to the passenger flow capacity of public transport from origin o to destination d within time period t, within the current public transport service supply. This refers to the passenger flow demand from origin o to destination d within time period t in the combined travel demand matrix F. This refers to the initial potential demand for public transportation that is waiting to be met. For new construction, when... If there is no bus link between point O and point D, then 50% A new channel will be built to connect points o and d as a potential demand.

[0065] In step S53, to match public transport demand, links with remaining capacity can be connected to the existing public transport network X. 0 The subtraction is divided into two situations. The first is that the links with remaining capacity belong to the guaranteed lines, since bus lines are the foundation of people's livelihood guarantee services, and the minimum requirement is to meet the guarantee needs. The second is that the links with remaining capacity do not belong to the guaranteed lines, and the capacity of the conditional channels is adjusted according to actual needs.

[0066] For links with remaining capacity that are considered guaranteed lines, the original channel capacity is reduced based on the larger of the actual demand and the guaranteed demand for each time period.

[0067] For links with remaining capacity that are not considered guaranteed lines, when Then cancel the bus link between point O and point D; when At that time, based on actual needs The original channel capacity was reduced as a capacity reference value.

[0068] In step S54, the bus network X 0 The target net X is obtained after processing in steps S52 and S53. 1 The network includes connectivity information in the spatial dimension and capacity information in the temporal dimension. In order to apply optimization schemes such as link expansion, reduction, construction, and cancellation to bus routes, and considering the public's usage habits of the existing bus network, major route adjustments should be avoided as much as possible. Instead, commonly used measures should be adjusted on existing bus routes.

[0069] To address the expansion and reduction of bus routes, adjustments are made to the frequency of the affected bus routes, along with the introduction of shorter routes, to adjust the capacity of the corresponding routes during specific time periods. If the expanded route is long-distance, involves high passenger flow, and exhibits clear patterns (e.g., the distance between the origin and destination is more than ten kilometers, passenger flow is small-scale within a unit of time, and there is normal passenger flow demand during weekday morning and evening peak hours), then direct bus routes can be added during the corresponding time periods.

[0070] For newly established bus routes, adjustments can be made by opening new bus lines or extending existing ones. If the new route is long, direct customized bus lines can be opened; if the new route is short, it can be achieved by extending existing routes or opening micro-circulation routes.

[0071] In response to link cancellations, unnecessary investment can be reduced by canceling existing routes or launching shorter routes, and the saved capacity can be invested in routes involved in expansion and new link construction.

[0072] In step S54, since urban passenger transport modal share is a complex interplay between various supply options and numerous service demands, the supply of public transport, travel demand, and the supply of other modes of transport can all alter the urban passenger transport modal share. Therefore, steps S1 to S5 are repeated to promptly track and understand the demand and supply of public transport, rationally allocate transport resources, and continuously optimize the public transport network on a regular basis.

[0073] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.

Claims

1. A method for optimizing public transport networks based on multi-modal travel data, characterized in that: include Step S1: Collect multi-modal transportation data, including public transportation, taxis, ride-hailing services, private cars, and shared bicycles; Step S2: Extract the origin location, destination location, and start and end time information of a single trip from the traffic data; Step S3: Grid-based urban space, dividing the urban space into multiple areas according to a unified rule; wherein the grid size of the urban area is smaller than that of the suburban area; Step S4: Map the extracted traffic data information onto the divided urban spatial grid, and classify the data according to demand; Step S4 includes: Step S41: Map the origin and destination of a single trip for multiple types of transportation to the urban spatial grid; Step S42: Based on the arrival time at the destination, the travel behavior data of the same OD pair is divided and collected in ten-minute intervals to obtain travel demand matrices A, B, C, D, and E for buses, taxis, ride-hailing services, private cars, and shared bicycles, respectively. Step S43: Extract long-distance travel data of shared bicycles to obtain the long-distance shared bicycle travel demand matrix E1; Step S44: Extract shared bicycle demand data to obtain shared bicycle travel demand matrix E2 for different time periods throughout the day; Step S45: Remove the duplicate data in E2 that are the same as those in E1 to obtain the shared bicycle travel demand matrix E3; Step S5: Determine the bus route optimization plan based on the demand classification results, and implement the optimization according to the plan; Step S5 includes: Step S51: Merge and calculate passenger data with the same origin-destination (OD) from the relevant travel data of taxis, ride-hailing services, private cars, and shared bicycles. We obtain the combined travel demand matrix F, and mine the potential passenger flow demand of public transportation from multi-modal travel data. Step S52: Optimize the bus network based on potential demand, including two optimization methods; One is expansion, when ， That is, the bus link between point O and point D is connected, and simultaneously satisfies: Then 50% As a potential demand, the link between O and D needs to be expanded; The second is newly built, when ， If there is no bus link between point O and point D, then 50% A new channel will be built to connect points o and d as a potential demand. In the formula It refers to t From the starting point within the time period o To the finish line d Actual bus passenger flow This refers to the current supply of public transportation services. t From the starting point within the time period o To the finish line d Public transport passenger capacity This refers to the combined travel demand matrix F. t From the starting point within the time period o To the finish line d Passenger demand, This refers to the initial potential demand for public transportation that is waiting to be met; Step S53: Optimize the bus network based on remaining capacity, including two optimization methods; One approach is to reduce the original channel capacity for links with remaining capacity that are considered guaranteed lines, using the larger of the actual demand and the guaranteed demand for each time period as the capacity reference value. Secondly, links with remaining capacity are not considered guaranteed lines. Then cancel the bus link between point O and point D; when At that time, based on actual needs The original channel capacity is reduced as a capacity reference value; Step S54: Determine bus route optimization measures.

2. The public transport network optimization method based on multi-modal travel data according to claim 1, characterized in that: Step S1 includes: Step S11: Obtain public transport passenger flow data by utilizing the transportation business system or by conducting routine passenger flow OD derivation; Step S12: Obtain taxi and ride-hailing passenger flow data using taxi pick-up and drop-off meter data and ride-hailing order data; Step S13: Obtain private vehicle flow data using private car entry and exit parking lot data; Step S14: Obtain shared bicycle passenger flow data using shared bicycle order data and the number of shared bicycles in the parking area.

3. The public transport network optimization method based on multi-modal travel data according to claim 2, characterized in that: In step S2, the origin, destination and corresponding time of a complete single bus trip are extracted. A passenger’s multiple consecutive internal bus transfers in a single trip are considered as a single bus trip. If a passenger can reach point D from point O through one or more bus routes, then a bus link is considered to exist between grid point O and point D.

4. The public transport network optimization method based on multi-modal travel data according to claim 1, characterized in that: The public transport travel demand matrix is ​​as follows: A= （1） = （2） Among them, the maximum value of o and d is equal to the number of city grids, and the maximum value of t is 144.

Images