Bus network optimization method, system, electronic device, and storage medium based on reinforcement learning technology

Abstract

The present application provides a bus network optimization method, system, electronic device, and storage medium based on reinforcement learning technology. The method mentions that a local optimal bus route is constructed based on passenger flow intensity, a first passenger flow intensity is obtained, a Monte Carlo tree search model is constructed, an optimal bus network is constructed for a set of bus routes, and its average passenger flow intensity is calculated to obtain a second passenger flow intensity. The first passenger flow intensity and the second passenger flow intensity are compared and judged to obtain the final bus network. Bus routes are planned based on passenger flow intensity, which can accurately match passenger flow. According to the passenger flow intensity in different regions and time periods, resources can be reasonably allocated to hot spots, and passenger flow changes can be flexibly responded to. The Monte Carlo tree search model can be used to quickly generate solutions, and the search weight is calculated using multiple attributes of the route. The route is comprehensively analyzed to make route selection and network construction more scientific, and decisions are made based on data and algorithms to reduce human experience bias.

Description

Technical Field

[0001] The present application relates to the field of public transportation technology, and in particular to a bus network optimization method, system, electronic device, and storage medium based on reinforcement learning technology. Background Art

[0002] In recent years, with the development of the economy and society and the acceleration of urbanization, urban traffic congestion and travel difficulties have become increasingly prominent. Prioritizing the development of public transportation is an inevitable requirement for alleviating traffic congestion and transforming the development model of urban transportation. The bus network is a key component of urban public transportation systems. Scientifically analyzing the conventional bus network and proposing optimization recommendations is crucial for improving the efficiency and appeal of conventional public transportation. Currently, urban public transportation passenger flow has declined significantly, with a large number of buses running idle on the road. This has put increasing pressure on bus companies and made it difficult to sustain public transportation services.

[0003] Currently, the solutions used for bus network optimization all combine operations optimization with heuristic algorithms. These methods are suitable for smaller networks. As the number of bus routes and stations increases, the number of potential bus route plans will increase exponentially. Existing methods consider balancing computational complexity and often constrain the number of possible bus route plans, so they generally cannot obtain the optimal solution. Summary of the Invention

[0004] The technical problem to be solved by this application is that the current solutions for bus network optimization are all methods that combine operations optimization with heuristic algorithms. The above methods are suitable for smaller network sizes. As the number of bus routes and stations increases, the potential bus route plans will grow exponentially. Existing methods consider balancing computational complexity and often constrain the number of possible bus route plans, so generally cannot obtain the optimal solution.

[0005] In order to solve the above problems, in order to solve the above technical problems or at least partially solve the above technical problems, the present application provides a bus network optimization method, system, electronic device, and storage medium based on reinforcement learning technology.

[0006] In a first aspect, the present invention discloses a bus network optimization method based on reinforcement learning technology, which specifically includes:

[0007] Obtain bus stops, draw bus routes between the bus stops, construct a set of candidate bus routes, and calculate the passenger flow intensity of each bus route in the set of candidate bus routes. The set of candidate bus routes must meet a preset non-linear coefficient threshold and a preset passenger flow intensity threshold.

[0008] For the set of alternative bus routes, the maximum number of bus routes included in the local optimal bus network and the minimum service bus demand ratio threshold that the local optimal bus route set needs to meet are preset;

[0009] Preprocess the candidate bus routes, sort them according to the passenger flow intensity of each route, initialize the selected route set, and initialize the local optimal bus network set;

[0010] The route with the largest passenger flow intensity in the set of alternative bus routes is selected as the first route in the local optimal bus network set, and is added to the set of selected routes. In addition, the route is removed from the set of alternative bus routes.

[0011] Selecting the second route in the local optimal bus network set, adding each bus route in the candidate bus route set to the selected route set, calculating the average passenger flow intensity of the selected route set after the different routes are added, obtaining the route that maximizes the average passenger flow intensity of the selected route set, and adding it as the second route to the local optimal bus network set. Calculating the proportion of total bus demand served by the current local optimal bus network; the total bus demand proportion includes the passenger flow intensity of direct bus demand and the passenger flow intensity of bus demand with a single transfer on the same platform;

[0012] Updating the bus demand OD matrix of the bus terminal pair set, removing the bus demand already served by the current local optimal bus network from the bus demand OD matrix, and updating the candidate bus route set and the selected route set; updating the candidate bus route set includes deleting the newly selected route from the candidate bus route set, and deleting the routes in the candidate bus route set whose passenger flow intensity is lower than a preset passenger flow intensity threshold, to obtain an updated candidate bus route set; updating the selected route set includes keeping the selected route set consistent with the current local optimal bus network;

[0013] Repeat the above steps to obtain the next route of the local optimal bus route set, update the bus demand OD matrix, the candidate bus route set and the selected route set, and calculate the proportion of the total bus demand served by the current local optimal bus network. This process continues until the proportion of the total bus demand served by the current local optimal bus network exceeds a preset minimum service bus demand proportion threshold, or the number of bus routes in the current local optimal bus route set is equal to a preset first number of bus routes. Then, the selected route set is output as the local optimal bus route set, and the passenger flow intensity of the local optimal bus network is calculated. The passenger flow intensity of the local optimal bus network is used as the first passenger flow intensity.

[0014] A Monte Carlo tree search model is constructed. Each bus route in the candidate bus route set is assigned the same initial probability selection weight. A random search is performed to change the selected bus route suggestions consisting of a preset first number of different bus routes into the selected bus route set, which is distinguished from the candidate bus route set. The search is then modified, and the average passenger flow intensity is calculated to obtain a second passenger flow intensity.

[0015] The second passenger flow intensity is compared with the first passenger flow intensity. If it is greater than the first passenger flow intensity, the probability selection weights of the selected preset first number of bus routes are increased to obtain updated probability selection weights. Based on the updated probability selection weights, the Monte Carlo tree search is rerun to screen the selected bus routes. After repeatedly iterating the updated probability selection weights and the selected bus routes, a route combination plan with the optimal passenger flow intensity is finally generated to obtain the global optimal bus network.

[0016] Preferably, the method of obtaining bus stops, drawing bus routes between bus stops, constructing a set of candidate bus routes, and calculating the passenger flow intensity of each bus route in the set of candidate bus routes, wherein the set of candidate bus routes needs to meet a preset non-linear coefficient threshold and a preset passenger flow intensity threshold, specifically includes the following steps:

[0017] Obtain a road traffic network diagram, draw bus routes based on the stations and road sections in the diagram, establish a set of bus terminal pairs based on the stations in the road traffic network diagram, establish a bus demand OD matrix indexed by road sections, and preset non-linear coefficient thresholds and passenger flow intensity thresholds;

[0018] Expand the nodes of the first station of the bus terminal pair, expand the nodes of any first station, obtain the adjacent station list, traverse the adjacent station list, and combine the adjacent stations in the adjacent station list with the corresponding first station to form a node list, and obtain the node list of all the first stations;

[0019] Calculating the non-linear coefficient and passenger flow intensity of the node lists of all first stations to determine whether they meet the construction requirements, and constructing an initial route list from the node lists of all first stations that meet the construction requirements to obtain the initial route list. The construction requirements include that the non-linear coefficient of the node list is less than a preset non-linear coefficient threshold and the passenger flow intensity is greater than a preset passenger flow intensity threshold. The node list of the first station is the initial route.

[0020] The adjacent sites in the initial route list are used as the first sites to expand the nodes again, obtain the corresponding adjacent site list, and generate a derived route list that meets the construction requirements;

[0021] When the adjacent station in the derived route list is the last station of the bus terminal, stop the node expansion, integrate the initial route list and the derived route list into a bus route, and obtain the set of alternative bus routes.

[0022] Preferably, the construction of the Monte Carlo tree search model, assigning the same initial probability selection weight to each bus route in the candidate bus route set, randomly searching for selected bus routes consisting of a preset first number of different bus routes, calculating the average passenger flow intensity, and obtaining the second passenger flow intensity, specifically includes the following steps:

[0023] Preset the total number of Monte Carlo tree search runs and the total number of routes in the global optimal bus network solution; initialize the bus demand OD matrix, initialize the route probability selection weight set, and initialize the selected route set;

[0024] According to the route probability selection weight set, a bus route is randomly selected from the candidate route set, added to the selected route set, the candidate route set is updated, and the selected bus route is removed; the route probability selection weight set is updated, and the selected bus route is removed;

[0025] Determine whether the total number of routes in the selected route set is greater than the preset total number of routes in the global optimal bus network. If it is less than the preset value, select a bus route from the updated alternative bus route set again, and then repeat the update process until the total number of routes in the selected bus route set is equal to the preset value; calculate the average passenger flow intensity of the selected route set as the second passenger flow intensity.

[0026] Preferably, the second passenger flow intensity is compared with the first passenger flow intensity. If the second passenger flow intensity is greater than the first passenger flow intensity, the probability selection weights of the selected preset first number of bus routes are increased to obtain updated probability selection weights. Based on the updated probability selection weights, the Monte Carlo tree search is re-run to screen the selected bus routes. After repeatedly iterating the updated probability selection weights and the selected bus routes, a route combination plan with the optimal passenger flow intensity is finally generated to obtain the global optimal bus network. Specifically, the steps include:

[0027] When the second passenger flow intensity is greater than the first passenger flow intensity, the ratio of the two is used as the factor increasing the probability selection weight of this route. For the route probability selection weight set, the probability selection weight of each route in the currently selected route set is multiplied by the weight increase factor, and the initial route probability weight set is updated.

[0028] Compare the number of Monte Carlo tree searches with the preset total number of runs. If the number of Monte Carlo tree searches is less than the preset total number of runs, save the selected route set to the global optimal bus network solution set and rerun the Monte Carlo tree search.

[0029] If the number of Monte Carlo tree searches is greater than the preset total number of runs, the set of selected routes with the second largest passenger flow intensity is output as the global optimal bus network.

[0030] Preferably, calculating the passenger flow intensity of direct bus demand specifically includes the following steps:

[0031] Split the selected bus route into multiple continuous sections, select any two sections to construct a section combination, and obtain a section combination set;

[0032] Select any section combination from the section combination set and determine whether the section combination is included in the bus demand matrix. If so, allocate the passenger demand corresponding to the bus demand matrix to the selected bus route to obtain the passenger flow of the bus route. If not, select a new section combination and mark the loaded section combination.

[0033] Obtain the number of passengers on the bus route with the largest passenger flow section, sum it with the passenger flow of the bus route, and determine whether the sum is greater than the maximum cross-section passenger capacity of the bus route;

[0034] If the sum is greater than the maximum cross-section passenger capacity of the bus line, the passenger flow value of the bus line is the passenger flow intensity of the direct bus demand of the selected bus line;

[0035] If the sum is equal to or less than the maximum cross-sectional passenger capacity of the bus line, the passenger flow intensity of the direct bus demand of the selected bus line is the maximum cross-sectional passenger capacity of the selected bus line minus the passenger volume of the largest passenger flow section on the selected bus line.

[0036] Preferably, calculating the passenger flow intensity of a bus transfer demand at the same platform specifically includes the following steps:

[0037] Determine the bus routes and transfer sections that can be transferred on the same platform for the selected bus routes, and construct a transfer route set;

[0038] Select transfer routes that pass through the same road section as the selected bus route from the transfer route set to construct a transfer road section set;

[0039] Selecting a first transfer segment from the transfer segment set and combining it with the selected bus route to construct a first preceding segment set and a first succeeding segment set, wherein the first preceding segment set includes all segments between the starting segment of the bus route and the first transfer segment, and the first succeeding segment set includes all segments between the first transfer segment and the ending segment of the bus route;

[0040] Based on the first transfer segment, the second preceding segment set and the second subsequent segment set are combined with the transfer segments in the transfer segment set to construct the second preceding segment set, wherein the second preceding segment set includes all segments between the starting segment of the transfer segment and the first transfer segment, and the second subsequent segment set includes all segments between the second transfer segment and the ending segment of the bus route;

[0041] Selecting a first selected preceding segment and a second selected subsequent segment from the first preceding segment set and the second subsequent segment set respectively to form a selected segment combination, determining whether the selected segment combination is in the bus demand matrix, and obtaining a first passenger demand;

[0042] Allocate the passenger demand of the corresponding selected segment combination to the corresponding first transfer segment and transfer line, add the passenger volume of all segments in the first selected preceding segment and the passenger volume of all segments in the second selected subsequent segment by the first passenger demand, and obtain the passenger volume set of the first selected preceding segment and the passenger volume set of the second selected subsequent segment;

[0043] Obtain the maximum cross-sectional passenger capacity of the line corresponding to the first selected preceding section and the second selected subsequent section, obtain the maximum passenger flow section and maximum passenger flow of the first selected preceding section and the maximum passenger flow section and maximum passenger flow of the second selected subsequent section from the passenger volume set of the first selected preceding section and the passenger volume set of the second selected subsequent section, compare the maximum passenger flow of the first selected preceding section with the maximum cross-sectional passenger capacity of the line of the first selected preceding section, and compare the maximum passenger flow of the second selected subsequent section with the maximum cross-sectional passenger capacity of the line of the second selected subsequent section, when the maximum passenger flow of the first selected preceding section is the maximum passenger flow of the second selected subsequent section, When the maximum passenger flow is less than the maximum cross-sectional passenger capacity of the line, the passenger flow intensity of the bus transfer demand between the first selected preceding section and the second selected subsequent section is the first passenger demand; when the maximum passenger flow is greater than the maximum cross-sectional passenger capacity of the line, the passenger flow intensity of the bus transfer demand between the first selected preceding section and the second selected subsequent section is the difference between the maximum cross-sectional passenger capacity of the line and the passenger volume of all sections within the first selected preceding section, and the difference between the maximum cross-sectional passenger capacity of the line and the passenger volume of all sections within the second selected subsequent section;

[0044] Repeat the above process for the first set of subsequent road segments and the second set of subsequent road segments to calculate the passenger flow intensity of the corresponding one-time same-platform transfer bus demand.

[0045] Preferably, the method further comprises the following steps:

[0046] For the final bus network, obtain the passenger flow of each station in real time;

[0047] Develop differentiated bus departure time adjustment strategies based on the travel characteristics and needs of different groups of people;

[0048] For routes with high passenger flow determined through data analysis and field investigations, the frequency of routes will be increased during peak hours. At the same time, shortened routes will be designed and added.

[0049] In a second aspect, the present invention discloses a bus network optimization system based on reinforcement learning technology, which includes a bus network optimization method based on reinforcement learning technology.

[0050] In a third aspect, the present invention discloses an electronic device comprising a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus;

[0051] Memory for storing computer programs;

[0052] The processor is used to implement the steps of the bus network optimization method based on reinforcement learning technology when executing the program stored in the memory.

[0053] In a fourth aspect, the present invention discloses a storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of a bus network optimization method based on reinforcement learning technology.

[0054] The above technical solution provided by this application has the following advantages compared with the existing technology:

[0055] The present application provides a bus network optimization method, system, electronic device, and storage medium based on reinforcement learning technology. The method mentions that a local optimal bus network is constructed based on passenger flow intensity, a first passenger flow intensity is obtained, a Monte Carlo tree search model is constructed, a global optimal bus network is constructed for a set of bus routes, and its average passenger flow intensity is calculated to obtain a second passenger flow intensity. The first passenger flow intensity and the second passenger flow intensity are compared and judged to obtain a bus network with a global optimal passenger flow intensity. Bus routes are planned based on passenger flow intensity, which can accurately match passenger flow. According to the passenger flow intensity in different regions and time periods, resources can be reasonably allocated to hot spots, and passenger flow changes can be flexibly responded to. The Monte Carlo tree search model can be used to quickly generate solutions, and the search weight can be calculated using multiple attributes of the route. The route is comprehensively analyzed to make route selection and network construction more scientific, and decisions are made based on data and algorithms to reduce human experience bias.

[0056] Furthermore, the method can optimize the bus network structure and greatly reduce the phenomenon of empty running of buses. On the basis of meeting the original bus passenger service needs, the overall mileage of bus routes will be reduced, and the passenger flow intensity of the bus network can be increased to twice the original level, which will greatly improve the operating efficiency of bus companies and achieve cost reduction and efficiency improvement. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0058] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0059] Figure 1 A flowchart of a bus network optimization method based on reinforcement learning technology provided in this application;

[0060] Figure 2 This is a schematic diagram of the specific process of step S1 of a bus network optimization method based on reinforcement learning technology provided in this application;

[0061] Figure 3 This is a schematic diagram of the specific process of step S2 of a bus network optimization method based on reinforcement learning technology provided in this application;

[0062] Figure 4 This is a schematic diagram of the specific process of step S3 of a bus network optimization method based on reinforcement learning technology provided in this application;

[0063] Figure 5 This is a specific flow chart of step S4 of a bus network optimization method based on reinforcement learning technology provided in this application. DETAILED DESCRIPTION

[0064] To make the objectives, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments of this application, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of this application.

[0065] First, see Figure 1-5 The present invention discloses a bus network optimization method based on reinforcement learning technology, which specifically includes:

[0066] Step S1: Obtain bus stops, draw bus routes between the bus stops, construct a set of candidate bus routes, and calculate the passenger flow intensity of each bus route in the set of candidate bus routes. The set of candidate bus routes must meet a preset non-linear coefficient threshold and a preset passenger flow intensity threshold.

[0067] Step S2: Filter bus routes one by one from the candidate bus route set, add them to the selected bus route set, and calculate the average passenger flow intensity of the selected bus route set, obtaining the selected bus route set with the largest average passenger flow intensity under the corresponding number of bus routes. When the total number of bus routes in the selected bus route set reaches a preset first number of bus routes, or the total bus demand ratio of the selected bus route set is greater than the preset total bus demand ratio, stop selecting bus routes, obtain a local optimal bus network, calculate the passenger flow intensity of the local optimal bus network, and obtain a first passenger flow intensity.

[0068] Step S3: Constructing a Monte Carlo tree search model, assigning the same initial probability selection weight to each bus route in the candidate bus route set, randomly searching for selected bus routes consisting of a preset first number of different bus routes, calculating the average passenger flow intensity, and obtaining a second passenger flow intensity;

[0069] Step S4: Compare the second passenger flow intensity with the first passenger flow intensity. If it is greater than the first passenger flow intensity, increase the probability selection weights of the selected preset first number of bus routes to obtain updated probability selection weights. Based on the updated probability selection weights, re-run the Monte Carlo tree search to screen the selected bus routes. After repeatedly iterating the updated probability selection weights and the selected bus routes, finally generate a route combination plan with the optimal passenger flow intensity to obtain the global optimal bus network.

[0070] Specifically, in step S1, bus stop information is obtained and a large set of candidate bus routes is constructed to facilitate the subsequent reinforcement learning process in searching for route combinations. When constructing this set of candidate bus routes, the geometric constraints and passenger flow intensity constraints of the routes are comprehensively considered. A node expansion approach is adopted, and at each step, the currently expanded route is evaluated to see whether it satisfies the nonlinear coefficient constraints of the route and the passenger flow intensity constraints of the bus route service. Route passenger flow intensity can reflect the distribution of bus passenger flow and help understand passenger flow demand in different regions.

[0071] Specifically, in step S2, the passenger flow intensity of a bus route refers to the number of passengers carried per unit length of each public transportation route. This, to a certain extent, reflects the operational efficiency of the bus network. Using a heuristic approach, a set of locally optimal bus routes is initially constructed, and their passenger flow intensities are obtained, resulting in a first passenger flow intensity. The core idea is to ensure that each step in bus route selection is a locally optimal solution, ensuring that newly added routes maximize the passenger flow intensity of the existing network.

[0072] Specifically, in step S3, a Monte Carlo tree search model is constructed. The bus route set is input into the model's search space. The search weights of the routes within the search space are calculated to form a model route set and a spatial route search weight set. The bus demand matrix is initialized, and routes are selected from the model route set to be added to the public transportation network. The relevant set and matrix are updated with each added route until the route counter exceeds the preset total number of bus network routes. The routes added to the public transportation network are then integrated to obtain the initial optimal bus network set, and their average passenger flow intensity, i.e., the second passenger flow intensity, is calculated. The Monte Carlo tree search model comprehensively considers various route combinations and passenger flow factors, optimizing the bus network from a global perspective and improving the overall efficiency of the network.

[0073] Specifically, in step S4, when the second passenger flow intensity is greater than the first passenger flow intensity, the ratio of the two is used as the factor increasing the probability selection weight of this route. For the route probability selection weight set, the probability selection weight of each route in the currently selected route set is multiplied by the weight increase factor, and the initial route probability selection weight set is updated. It is determined whether the number of Monte Carlo tree searches is less than the preset total number of runs. If so, the selected route set is saved in the global optimal bus network solution set, and the Monte Carlo tree search is rerun. If so, the selected route set with the largest second passenger flow intensity is output as the global optimal bus network solution. By comparing the passenger flow intensities of the local and global optimized networks, it is ensured that the final output bus network achieves global optimization while taking into account local needs, improves the overall utilization efficiency of bus resources, and ensures that the bus network achieves the best operating effect while meeting the passenger flow needs of different regions. Furthermore, if the ratio of the first passenger flow intensity to the second passenger flow intensity is less than the preset weight increase factor value, the global optimized bus network is abandoned, the global optimized bus network is reconstructed, and the optimization is performed again to obtain a globally optimized bus network that meets the requirements, avoid adopting a global optimization scheme that may not be effective, prevent the destruction of the originally good local bus network due to unreasonable global optimization, ensure the overall stability and effectiveness of the bus network, and ensure that the bus network optimization is always guided by actual passenger flow demand and operational results, and do not blindly pursue global optimization while ignoring local needs.

[0074] It can be understood that based on the passenger flow intensity, a local optimal bus network is constructed, the first passenger flow intensity is obtained, a Monte Carlo tree search model is constructed, the global optimal bus network is constructed for the bus route set, the average passenger flow intensity is calculated, and the second passenger flow intensity is obtained. The first passenger flow intensity and the second passenger flow intensity are compared and judged to obtain the global optimal bus network for passenger flow intensity. Bus routes are planned based on passenger flow intensity, which can accurately match passenger flow. According to the passenger flow intensity in different regions and time periods, resources can be reasonably allocated to hot spots, and passenger flow changes can be flexibly responded to. The Monte Carlo tree search model can be used to quickly generate solutions, calculate search weights using multiple attributes of routes, and comprehensively analyze routes to make route selection and network construction more scientific, make decisions based on data and algorithms, and reduce human experience bias.

[0075] Furthermore, the method can optimize the bus network structure and greatly reduce the phenomenon of empty running of buses. On the basis of meeting the original bus passenger service needs, the overall mileage of bus routes will be reduced, and the passenger flow intensity of the bus network can be increased to twice the original level, which will greatly improve the operating efficiency of bus companies and achieve cost reduction and efficiency improvement.

[0076] As an embodiment, for the final bus network, the passenger flow of each station is obtained in real time; differentiated bus departure time adjustment strategies are formulated based on the travel characteristics and needs of different groups of people; for routes with large passenger flow determined through data analysis and field investigations, the route frequency is increased during peak hours, and at the same time, shortened travel routes are designed and added.

[0077] Specifically, passenger flow at each stop is obtained through high-precision passenger flow statistics equipment installed at bus stops, including infrared sensor counters, video analysis systems, and a fusion of multiple data sources such as smart bus card systems and mobile phone app ride data. The frequency of collecting various types of data is no less than once a minute to ensure the real-time and accuracy of passenger flow data. At the same time, image recognition technology is used to analyze passengers' facial features and clothing styles, and combined with age, gender and other data from card swipe information, the identities of the passengers at the corresponding stop are determined, including students, office workers, senior citizens, tourists, etc. Among them, bus departure time adjustment strategies are formulated based on the identities of different groups. For example, for students, the number of buses will be increased by 20 to 30 minutes on school mornings, and the departure interval on routes around schools will be shortened from 15 minutes to 8 to 10 minutes during school hours. For office workers, the departure interval on routes connecting major residential areas with commercial and office areas will be shortened to 5 to 8 minutes during peak hours on weekdays, and the departure time will be dynamically adjusted according to real-time road conditions to avoid congestion. For senior citizens, the departure time will be adjusted from 9:00 to 11:00 a.m. and from 2:00 to 4:00 p.m. For routes with high passenger flow, shortened routes can be designed. For example, for routes that pass through multiple commercial and residential areas and have sections with high passenger flow, short routes can be designed that only run between these commercial areas and major stations around them, with departure intervals set at 8 to 12 minutes. This will increase the frequency of routes in these sections to meet the needs of high passenger flow.

[0078] Furthermore, the distribution of some schools along routes will be determined by considering factors such as route length, capacity, road capacity, and surrounding traffic conditions. A real-time traffic monitoring system will be established, combining bus operation data with information from traffic management departments to monitor route traffic conditions and passenger boarding. When it is determined that the excessive number of schools on a route causes traffic congestion and difficulty boarding during school hours (e.g., significant decreases in vehicle speeds, long queues at stations, or excessive vehicle occupancy), departure schedules will be adjusted. For example, bus routes under 10 kilometers in length, with narrow roads and limited capacity, will be limited to passing no more than three schools. Bus routes on major arterial roads will be limited to passing no more than five schools, taking into account school start and end times and student populations. During peak school hours, fixed departure intervals will be adjusted to dynamic intervals, adjusting departure intervals from 10 minutes to 6 to 8 minutes based on real-time passenger flow and road conditions. At the same time, the departure time of vehicles should be arranged reasonably, and some buses should be advanced or delayed by 10 to 15 minutes so that they can pass through the school area when the roads are relatively unobstructed.

[0079] Finally, through collaboration with urban planning departments and community management agencies, the distribution of schools and senior communities along the routes was analyzed, with detailed information on school type, size, start and end times, and the population size and age structure of senior communities. Furthermore, using high-definition cameras on buses and image analysis algorithms, real-time monitoring of boarding age ranges, number of stops, and other information was used to determine the demographics of passengers on the routes. Based on the distribution of schools and senior communities along the routes and real-time demographic information, fare adjustment strategies were formulated. For routes with a large number of elderly people, comfortable buses equipped with soft seats, spacious aisles, and accessible facilities (such as wheelchair ramps and handrails) were selected.

[0080] Specifically, since students and the elderly have a greater demand for public transportation, fares can be adjusted for routes with high numbers of these groups. While students receive discounts, some longer routes can result in higher fares. Furthermore, senior citizens may have discounts or even free rides with senior citizen cards, but there are still some elderly people of similar age or those with disabilities. For these groups, fares can be reduced accordingly, making bus routes more user-friendly. For example, for routes that pass by many schools, discounted fares can be offered to students during rush hour, such as half-price or discounted fares for students with student cards. For routes with a high concentration of elderly communities, discounts can be offered to senior citizens with cardholders, or monthly or seasonal pass packages can be introduced. When these elderly people travel, overall fares can be adjusted accordingly, such as by 20%-40%. Furthermore, fares can be adjusted dynamically based on real-time demographics, such as further reductions when there are more senior citizens and appropriate increases when there are high volumes of commuters. Furthermore, on routes with a large elderly population, more experienced drivers are assigned to drive. Through a multi-dimensional assessment of drivers' driving experience, safety records, and passenger reviews, these routes are matched with drivers with higher driving experience and higher ratings. Drivers are also required to undergo specialized training tailored to the travel characteristics and needs of the elderly. Speed limits are also implemented on these routes to prevent drivers from driving too fast, resulting in a poor passenger experience. Furthermore, buses are labeled for elderly routes to facilitate identification. Furthermore, vehicle types are dynamically adjusted based on real-time demographic information. For example, when a large number of passengers are carrying large luggage, vehicles that can accommodate these items are deployed.

[0081] As an example, to address emergencies requiring real-time adjustments to bus routes, a comprehensive, multi-layered emergency monitoring system is being established. This system integrates real-time traffic data and accident reports from traffic management departments, weather forecasts and disaster warnings from meteorological departments, and real-time news reports from news media. Furthermore, it integrates data from city surveillance cameras and sensor data on buses (such as vehicle driving status and location information) to provide 24-hour, uninterrupted, real-time monitoring of emergencies that could impact bus operations. These emergencies include, but are not limited to, natural disasters (such as heavy rain, snow, earthquakes, and typhoons) and traffic accidents (such as vehicle collisions and road construction). A scientific, rigorous, and tiered early warning mechanism is being developed. Warning signals are categorized into different levels based on multiple factors, including the severity, scope, and development trends of the incident. These levels include blue (for general events, potentially causing minor impacts on bus operations), yellow (for major events, potentially disrupting some bus routes), orange (for major events, potentially causing multiple bus route suspensions or significant delays), and red (for particularly serious events, potentially paralyzing the entire bus network). Clearly define the corresponding release process, channels, and scope for different levels of warning signals. For example, blue warnings can be pushed through bus apps and displayed on bus stop displays. Orange warnings and above, in addition to these channels, must also be widely disseminated through radio, television, and other media to ensure that bus operators, drivers, and passengers receive timely and accurate warning information, allowing sufficient preparation time for emergency response. Develop detailed, specific, and actionable emergency response strategies for different types of emergency events. In the event of a natural disaster (such as road flooding caused by heavy rain): Immediately activate the road flooding monitoring mechanism, using water level sensors installed at key locations to provide real-time feedback on flood depth. For sections of road where flooding exceeds safety thresholds, quickly adjust the routing of affected bus routes. Utilize geographic information systems (GIS) and real-time road condition data to plan and select alternative routes to avoid flooded sections. Furthermore, adjust vehicle operating speeds and departure intervals based on flooding conditions and road capacity to ensure driving safety. For bus routes where route adjustments are not possible, promptly issue suspension or diversion notices, and deploy staff at relevant stops to provide guidance and explanations. In the event of a traffic accident: Obtain detailed information immediately, including the location, severity, and types of vehicles involved. Take appropriate measures based on the impact of the accident on road traffic. If the accident results in partial lane closures, adjust the routes of the affected routes and redirect vehicles. If the accident is more serious, resulting in complete road congestion, suspend the relevant routes and promptly release information to passengers through public transportation apps, broadcasts, and other channels, informing them of the resumption time of the suspended routes and alternative travel suggestions.At the same time, cooperate with the traffic management department to handle accidents and on-site guidance to ensure that rescue vehicles and personnel can arrive at the scene smoothly.

[0082] As an example, we can strengthen the integration of bus networks with other modes of transportation, such as urban rail transit, shared bicycles, and taxis. By establishing a comprehensive transportation information platform, we can achieve data sharing and coordinated scheduling among different modes of transportation; and by formulating a unified transportation plan, we can promote the coordinated development of different modes of transportation.

[0083] For example, bus stops and subway stations can be rationally arranged to achieve seamless transfers. Shared bicycles can be provided with dedicated parking areas and connected to bus stops to facilitate short-distance travel. Cooperating with taxi companies can ensure that taxis and buses operate in a complementary manner, improving the overall efficiency of urban transportation. By adjusting bus fares and offering discounts, passengers can be encouraged to prioritize public transportation. Shared bicycle companies and taxi companies can be encouraged to adopt green and environmentally friendly operating models, contributing to the sustainable development of the city. Furthermore, the construction and management of integrated transportation hubs can be strengthened, improving their service levels and transfer efficiency, and providing passengers with a more convenient travel experience.

[0084] Step S1 specifically includes the following steps:

[0085] Step S11: Obtain a road traffic network diagram, draw bus routes based on the stations and road sections in the diagram, establish a set of bus terminal pairs based on the stations in the road traffic network diagram, establish a bus demand OD matrix indexed by the road sections, and preset a non-linear coefficient threshold and a preset passenger flow intensity threshold;

[0086] Step S12: Perform node expansion on the first station of the bus terminal pair set, perform node expansion on any first station, obtain a list of adjacent stations, traverse the list of adjacent stations, and form a node list with the adjacent stations in the adjacent station list and the corresponding first station to obtain a node list of all the first stations;

[0087] Step S13: Calculate the non-linear coefficient and passenger flow intensity of the node lists of all first stations, determine whether they meet the construction requirements, and construct an initial route list from the node lists of all first stations that meet the construction requirements to obtain the initial route list. The construction requirements include that the non-linear coefficient of the node list is less than a preset non-linear coefficient threshold and the passenger flow intensity is greater than a preset passenger flow intensity threshold. The node list of the first station is the initial route.

[0088] Step S14: The adjacent sites in the initial route list are used as the first sites to perform node expansion again, obtain the corresponding adjacent site list, and generate a derived route list that meets the construction requirements;

[0089] Step S15: When the adjacent station in the derived route list is the last station of the bus terminal, stop node expansion, integrate the initial route list and the derived route list into bus routes, and obtain a set of candidate bus routes.

[0090] As can be understood, using a high-precision geographic information system (GIS) and real-time traffic data, we can obtain precise location information for bus stops and the surrounding road network, thereby creating an accurate road traffic network map. This map includes bus stops and bus routes. Bus routes are then mapped based on these bus stops and public roads, taking into account factors such as road capacity, traffic congestion, and the travel needs of surrounding residents. This ensures that the mapped routes conform to actual traffic conditions and maximize passenger travel needs. Stations in the road traffic network map are paired to construct a set of bus terminal pairs. Each pair represents the starting and ending points of a potential bus route. Customer demand information for each bus route is collected through various channels, including bus card swipe records, online questionnaires, and on-site interviews. This customer demand data can be used to reflect passenger travel frequency, travel time, and travel purposes on different routes. Node expansion is performed for each starting station in the bus terminal pair set. Using the road network diagram, all adjacent stations directly connected to the starting station are identified to form an adjacent station list. This adjacent station list is then traversed, and each adjacent station is combined with the corresponding starting station to form a node list. This operation generates a series of possible route starting segments for each starting station. For each node list of starting stations, the nonlinearity coefficient and passenger flow intensity are calculated. The nonlinearity coefficient reflects the circuitousness of the route and is calculated as the ratio of the actual route length to the straight-line distance between the starting and ending stations. The passenger flow intensity reflects the passenger volume of the route and can be calculated based on previously collected customer demand data. Based on the preset nonlinearity coefficient and passenger flow intensity criteria, the node list is filtered. The initial route represents the potential starting segment of a bus route that meets basic operational requirements. Using the adjacent stations in the initial route list as new starting stations, the node expansion operation in step S12 is repeated to obtain a new adjacent station list. This new adjacent station list is filtered to generate a list of derivative routes that meet the construction requirements (nonlinearity coefficient less than a preset value and passenger flow intensity greater than a preset value). These derivative routes are further extended and expanded based on the initial route. During the process of generating the derived route list, the adjacent station is checked to see if it is the last stop of the bus's first or last stop. If it is, the node expansion operation is stopped, indicating that a complete bus route has been constructed. The initial route list and the derived route list are then integrated to form a complete bus route. Finally, all eligible bus routes are aggregated to form a bus route set. During the route selection process, indicators such as non-linearity coefficient and passenger flow intensity are used for evaluation to ensure that the generated bus routes have high operational efficiency and service quality. This avoids excessively circuitous routes, improves vehicle speed and punctuality, and ensures sufficient passenger flow along the routes, thereby enhancing operational efficiency.

[0091] Step S1 specifically operates as follows:

[0092] Preparation: The road traffic network graph G(V,E) consists of stations V and road segments E. Using road segments as indexes, construct a bus demand matrix DicOD in the form of (Link1, Link2, Vol), where the passenger demand from the bus station on link1 to the bus station on link2 is Vol, the nonlinear coefficient threshold NL between nodes, the bus line passenger flow intensity threshold D, and the set of bus terminal pairs Terminals.

[0093] Select a pair of bus terminal nodes (StartTerminal, EndTerminal) from the set of bus terminal pairs Terminals (excluding short-distance bus terminal pairs). Add the start node StartTerminal to the node list NodeList. Expand the start node StartTerminal and select an adjacent node Nodeadd from the start node's adjacent node list NextnodeList and add it to the node list NodeList. Calculate the nonlinear coefficient and the passenger flow intensity of the direct bus demand served by (StartTerminal, Nodeadd) in NodeList. If the nonlinear coefficient is less than the threshold NL and the passenger flow intensity is greater than the threshold D, add the node list NodeList as r0 to the line list LinesList. Select the next adjacent node Nodeadd from NextnodeList and repeat the above process until all adjacent nodes in NextnodeList are traversed, generating r1, r2, etc., and establishing the initial line list LinesList. Then, repeat steps S1b, S1c, and S1d for the tail node of each line r0, r1, r2, etc. in LinesList, performing node expansion and continuously generating new lines. If the end node of a line r is the terminal node EndTerminal, node expansion for that line stops. If the end nodes of all lines in LinesList are terminal nodes, node expansion for the start-end pair (StartTerminal, EndTerminal) stops and LinesList is added to the line pool LinePool. If the end nodes of all lines in LinesList are terminal nodes, the next start-end pair in Terminals is selected and the above process is repeated until all start-end pairs in Terminals are traversed, and the line pool LinePool is output.

[0094] Step S2 specifically includes the following steps:

[0095] Step S21: corresponding to the candidate bus route set, preset the maximum number of bus routes included in the local optimal bus network and the minimum service bus demand ratio threshold required to be met by the local optimal bus route set;

[0096] Step S22: pre-processing the candidate bus route set, sorting the routes according to the passenger flow intensity of each route, initializing the selected route set, and initializing the local optimal bus network set;

[0097] Step S23: Select the route with the largest passenger flow intensity in the candidate bus route set as the first route in the local optimal bus network set, add it to the selected route set, and remove the route from the candidate bus route set;

[0098] Step S24: Select the second route in the local optimal bus network set, add each bus route in the candidate bus route set to the selected route set, calculate the average passenger flow intensity of the selected route set after the different routes are added, obtain the route that maximizes the average passenger flow intensity of the selected route set, and add it as the second route to the local optimal bus network set. Calculate the proportion of total bus demand served by the current local optimal bus network; the total bus demand proportion includes the passenger flow intensity of direct bus demand and the passenger flow intensity of bus demand with a single transfer at the same platform.

[0099] Step S25: updating the bus demand OD matrix of the bus terminal pair set, removing the bus demand already served by the current local optimal bus network from the bus demand OD matrix, and updating the candidate bus route set and the selected route set; updating the candidate bus route set includes deleting the newly selected route from the candidate bus route set, and deleting the routes in the candidate bus route set whose passenger flow intensity is lower than a preset passenger flow intensity threshold, to obtain an updated candidate bus route set; updating the selected route set includes keeping the selected route set consistent with the current local optimal bus network;

[0100] Step S26: Repeat the above steps to obtain the next route of the local optimal bus route set, update the bus demand OD matrix, the candidate bus route set and the selected route set, and calculate the proportion of the total bus demand served by the current local optimal bus network. This process continues until the proportion of the total bus demand served by the current local optimal bus network exceeds a preset minimum service bus demand proportion threshold, or the number of bus routes in the current local optimal bus route set is equal to the total number of bus routes in the preset first quantity. Then, the selected route set is output as the local optimal bus route set, and the passenger flow intensity of the local optimal bus network is calculated.

[0101] Specifically, for the set of candidate bus routes, the maximum number of bus routes included in the local optimal bus network and the minimum threshold for the proportion of bus demand served by the local optimal bus network are preset. The routes in the candidate bus route set are sorted by passenger flow intensity, and the selected route set and the local optimal bus network set are initialized. The route with the highest passenger flow intensity from the candidate bus route set is selected as the first route in the local optimal bus network set, added to the selected route set, and removed from the candidate bus route set. For the second route to be added, each route in the candidate bus route set is added to the selected route set. The average passenger flow intensity of the selected route set after each route is added is calculated. The route that maximizes the average passenger flow intensity of the selected route set is obtained and added as the second route to the local optimal bus network set. The proportion of total bus demand served by the current local optimal bus network (including the passenger flow intensity of direct bus demand and the passenger flow intensity of bus demand with a single transfer on the same platform) is calculated. Update the bus demand OD matrix for the set of bus terminal pairs, remove the bus demand already served by the current local optimal bus network, and update the set of candidate bus routes: remove the newly selected route from the set of candidate bus routes, and also remove routes in the set of candidate bus routes with passenger flow intensities below a preset passenger flow intensity threshold. Update the set of selected routes to align with the current local optimal bus network. Repeat steps 4 and 5 to continuously obtain the next route in the set of local optimal bus routes, update the bus demand OD matrix, the set of candidate bus routes, and the set of selected routes, and calculate the proportion of total bus demand served by the current local optimal bus network. The loop terminates if the proportion of total bus demand served by the current local optimal bus network exceeds the preset minimum service bus demand proportion threshold or if the number of bus routes in the current local optimal bus route set equals the preset first number of bus routes. At this point, the selected route set is output as the local optimal bus route set, and the passenger flow intensity of the local optimal bus network is calculated.

[0102] It can be understood that in the bus route set, two routes are randomly selected as the first route and the second route respectively. With the help of bus card swiping data, questionnaires, smart ticketing systems and other channels, passenger demand information from each bus stop on the first route to each bus stop on the second route is collected. The first route, the second route and the passenger demand are combined into a bus demand matrix. Each element in the matrix represents the number of passengers demanded from a station on the first route to a station on the second route. The minimum service bus demand ratio can be set according to the corresponding actual situation. According to the goals and actual situation of bus operation, the minimum service bus demand ratio that the local optimal bus route set needs to meet is pre-set as the criterion for subsequent screening and judgment. Based on the global passenger flow intensity, the route with the largest passenger flow intensity is selected and added to the selected route set. It is removed from the original set. The selected route set is traversed and combined with the candidate routes to calculate the passenger flow intensity (including direct and one same-platform transfer). The route with the largest intensity is selected and added to the real-time local optimal set. The current service total demand ratio is calculated. The served demand is removed from the demand matrix. The screened route set and candidate set are updated. The corresponding routes and low passenger flow routes are eliminated, and the candidate set is replaced with the current local optimal network. Repeat the above steps until the total service demand ratio exceeds the preset value, output the local optimal route set, and calculate its passenger flow intensity. By screening and combining bus routes, a local optimal bus route set can be obtained, so that the bus network achieves a relatively optimal passenger flow intensity while meeting a certain service demand ratio, improving the overall operational efficiency of the bus network. Route screening and optimization based on the bus demand matrix can ensure that the bus network covers passengers' travel needs to the greatest extent possible, improving the quality and satisfaction of bus services. During the screening process, route selection and updating based on passenger traffic intensity and passenger flow intensity help to rationally allocate bus resources and avoid waste and idle resources.

[0103] The specific operation is as follows: Input the bus route set LinePool, the proportion of bus demand q that is served by the local optimal line network solution, and the bus demand matrix DicOD. Sort the routes in the bus route set LinePool in descending order according to passenger flow intensity, select the route with the largest passenger flow intensity as the first route of the local optimal bus network LocalPlan, add it to the candidate route set Paths, and remove it from the bus route set LinePool. Traverse all routes in the bus route set LinePool and calculate the passenger flow intensity of the candidate route set Paths if route 1 is added to the candidate route set Paths. Mainly calculate the increased service bus demand, including the passenger flow intensity of direct bus demand and the passenger flow intensity of one-time transfer bus demand. Add the route that maximizes the passenger flow intensity of the candidate route set Paths to the local optimal bus network LocalPlan. Calculate the total bus demand proportion Q served by the current local optimal bus network LocalPlan. Removing already served bus demands from the demand matrix DicOD will cause the passenger flow intensity of some routes in the bus route set LinePool to fall below the required level. Therefore, the bus route set LinePool is updated to remove routes whose passenger flow intensity is affected and falls below the threshold. The candidate route set Paths is updated and replaced with the locally optimal bus network LocalPlan. Repeat the above steps, searching for the currently optimal route each time. This route, which serves the greatest demand, is added to the locally optimal bus network LocalPlan until the proportion of total bus demand served by the network solution, Q, exceeds the minimum proportion of bus demand served, q. The passenger flow intensity LocalDensity of the locally optimal bus network LocalPlan is calculated and used as the input for the MCTS tree search.

[0104] The passenger flow intensity of direct bus demand and the passenger flow intensity of a single transfer bus are important components of calculating the passenger flow intensity of routes within a collection. The passenger flow intensity of a route within a collection reflects the overall passenger flow capacity of the route, encompassing both the passenger flow generated by direct trips and the passenger flow generated by a single transfer bus. Only by comprehensively considering these two components of passenger flow intensity can the passenger flow intensity of a route within a collection be accurately calculated, thereby fully understanding the actual passenger flow demand of the route. Calculating the passenger flow intensity of a route within a collection can help determine which routes require priority attention and optimization. The passenger flow intensity of direct bus demand and the passenger flow intensity of a single transfer bus further refine the optimization direction. For example, if the passenger flow intensity of a route within a collection is high and analysis reveals that the direct bus demand accounts for a large proportion, optimization may prioritize ensuring direct service on that route, such as by adding buses or increasing the frequency of buses. If the passenger flow intensity of a single transfer bus accounts for a large proportion, optimization may require optimizing transfer facilities and transfer plans to improve transfer efficiency.

[0105] Calculating the passenger flow intensity of direct bus demand in step S23 specifically includes the following steps:

[0106] Split the selected bus route into multiple continuous sections, select any two sections to construct a section combination, and obtain a section combination set;

[0107] Select any section combination from the section combination set and determine whether the section combination is included in the bus demand matrix. If so, allocate the passenger demand corresponding to the bus demand matrix to the selected bus route to obtain the passenger flow of the bus route. If not, select a new section combination and mark the loaded section combination.

[0108] Obtain the number of passengers on the bus route with the largest passenger flow section, sum it with the passenger flow of the bus route, and determine whether the sum is greater than the maximum cross-section passenger capacity of the bus route;

[0109] If the sum is greater than the maximum cross-section passenger capacity of the bus line, the passenger flow value of the bus line is the passenger flow intensity of the direct bus demand of the selected bus line;

[0110] If the sum is equal to or less than the maximum cross-sectional passenger capacity of the bus line, the passenger flow intensity of the direct bus demand of the selected bus line is the maximum cross-sectional passenger capacity of the selected bus line minus the passenger volume of the largest passenger flow section on the selected bus line.

[0111] Specifically, the passenger flow intensity of direct bus demand directly reflects the degree of passenger demand for bus services that allow passengers to travel directly from their starting point to their destination without transfers. A high direct bus demand intensity means that within the corresponding route section, a large number of passengers expect to travel directly to save time and energy. This provides a clear direction for bus network planning, namely, to prioritize these high-demand direct routes and improve passengers' travel experience and satisfaction. By analyzing the passenger flow intensity of direct bus demand, some potential important route directions can be identified. For areas with high direct demand intensity, bus operating departments can consider adding or optimizing bus routes to better cover these areas and improve the overall coverage and accessibility of the bus network.

[0112] Step S23 calculates the passenger flow intensity of a bus transfer demand at the same platform, specifically including the following steps:

[0113] Determine the bus routes and transfer sections that can be transferred on the same platform for the selected bus routes, and construct a transfer route set;

[0114] Select transfer routes that pass through the same road section as the selected bus route from the transfer route set to construct a transfer road section set;

[0115] Selecting a first transfer segment from the transfer segment set and combining it with the selected bus route to construct a first preceding segment set and a first succeeding segment set, wherein the first preceding segment set includes all segments between the starting segment of the bus route and the first transfer segment (including the starting segment), and the first succeeding segment set includes all segments between the first transfer segment and the ending segment of the bus route (including the ending segment);

[0116] Based on the first transfer segment, the second preceding segment set and the second succeeding segment set are combined with the transfer segments in the transfer segment set to construct the second preceding segment set, wherein the second preceding segment set includes all segments between the starting segment of the transfer segment and the first transfer segment (including the starting segment), and the second succeeding segment set includes all segments between the second transfer segment and the ending segment of the bus route (including the ending segment).

[0117] Selecting a first selected preceding segment and a second selected subsequent segment from the first preceding segment set and the second subsequent segment set respectively to form a selected segment combination, determining whether the selected segment combination is in the bus demand matrix, and obtaining a first passenger demand;

[0118] Allocate the passenger demand of the corresponding selected segment combination to the corresponding first transfer segment and transfer line, add the passenger volume of all segments in the first selected preceding segment and the passenger volume of all segments in the second selected subsequent segment by the first passenger demand, and obtain the passenger volume set of the first selected preceding segment and the passenger volume set of the second selected subsequent segment;

[0119] Obtain the maximum cross-sectional passenger capacity of the line corresponding to the first selected preceding section and the second selected subsequent section, obtain the maximum passenger flow section and maximum passenger flow of the first selected preceding section and the maximum passenger flow section and maximum passenger flow of the second selected subsequent section from the passenger volume set of the first selected preceding section and the passenger volume set of the second selected subsequent section, compare the maximum passenger flow of the first selected preceding section with the maximum cross-sectional passenger capacity of the line of the first selected preceding section, and compare the maximum passenger flow of the second selected subsequent section with the maximum cross-sectional passenger capacity of the line of the second selected subsequent section, when the maximum passenger flow of the first selected preceding section is the maximum passenger flow of the second selected subsequent section, When the maximum passenger flow is less than the maximum cross-sectional passenger capacity of the line, the passenger flow intensity of the bus transfer demand between the first selected preceding section and the second selected subsequent section is the first passenger demand; when the maximum passenger flow is greater than the maximum cross-sectional passenger capacity of the line, the passenger flow intensity of the bus transfer demand between the first selected preceding section and the second selected subsequent section is the difference between the maximum cross-sectional passenger capacity of the line and the passenger volume of all sections within the first selected preceding section, and the difference between the maximum cross-sectional passenger capacity of the line and the passenger volume of all sections within the second selected subsequent section;

[0120] Repeat the above process for the first set of subsequent road segments and the second set of subsequent road segments to calculate the passenger flow intensity of the corresponding one-time same-platform transfer bus demand.

[0121] Specifically, in actual bus travel, not all passengers can achieve direct travel. A single transfer on the same platform can meet the diverse travel needs of passengers to a certain extent. The passenger flow intensity of a single transfer on the same platform reflects the passengers' demand for traveling by transferring on the same platform. By analyzing this indicator, the bus operating department can understand which transfer nodes and transfer routes are in greater demand, thereby optimizing transfer facilities and transfer services, and improving the convenience and comfort of passenger transfers; by analyzing the passenger flow intensity of a single transfer on the same platform, some key transfer nodes and transfer routes can be identified, and these nodes and routes can be planned and laid out as important components of the bus network, which will help optimize the structure of the bus network.

[0122] Step S3 specifically includes the following steps:

[0123] Step S31: Preset the total number of Monte Carlo tree search runs and the total number of routes of the global optimal bus network solution; initialize the bus demand OD matrix, initialize the route probability selection weight set, and initialize the selected route set;

[0124] Step S32: Based on the route probability selection weight set, a bus route is randomly selected from the candidate route set, added to the selected route set, and the candidate route set is updated, removing the selected bus route; the route probability selection weight set is updated, and the selected bus route is removed;

[0125] Step S33: Determine whether the total number of routes in the selected route set is greater than the preset total number of routes in the global optimal bus network. If it is less than the preset value, select another bus route from the updated candidate bus route set, and then repeat the updating process until the total number of routes in the selected bus route set is equal to the preset value; calculate the average passenger flow intensity of the selected route set as the second passenger flow intensity.

[0126] Specifically, the total number of searches and the target number of routes are preset. The bus demand matrix (recording passenger flow demand between stops), the route weight set, and the selected route set are initialized. In the first iteration, routes are randomly selected based on the current weights and added to the selected set. The candidate routes and their weights are updated until the number of selected routes reaches the preset target. The average passenger flow intensity of the selected routes is calculated to obtain the second passenger flow intensity. Initially, all candidate routes are assigned equal weights. Routes are randomly selected to construct a candidate solution. Routes with superior performance (i.e., routes with increased passenger flow intensity) are weighted higher in subsequent iterations to increase their probability of selection. The search process is repeated, gradually focusing on more optimal route combinations and ultimately generating a globally optimal bus network. Furthermore, through random search and probabilistic weight adjustment, a bus route combination with a high proportion of bus demand coverage and optimal passenger flow intensity is sought. Initial random route selection ensures diverse exploration, while subsequent weight adjustment ensures efficient utilization of known information. This approach rapidly obtains a near-optimal solution through a limited number of iterations. This approach is suitable for large-scale transportation networks and can be adapted to bus systems of varying sizes by adjusting the number of searches and the number of routes, improving scalability. Furthermore, through the probabilistic iteration mechanism of Monte Carlo tree search, a balance between computational efficiency and optimization quality is achieved in bus network planning, which is particularly suitable for dealing with large-scale and highly complex transportation network optimization problems.

[0127] Step S4 specifically includes the following steps:

[0128] Step S41: When the second passenger flow intensity is greater than the first passenger flow intensity, the ratio of the two is used as the current route probability selection weight increase factor. For the route probability selection weight set, the probability selection weight of each route in the currently selected route set is multiplied by the weight increase factor, and the initial route probability weight set is updated.

[0129] Step S42: Compare the number of Monte Carlo tree searches with a preset total number of runs. If the number of Monte Carlo tree searches is less than the preset total number of runs, save the selected route set to the global optimal bus network solution set and rerun the Monte Carlo tree search.

[0130] Step S43: If the number of Monte Carlo tree searches is greater than the preset total number of runs, the set of selected routes with the second largest passenger flow intensity is output as the global optimal bus network.

[0131] It can be understood that comparing the passenger flow intensity of the initial optimal bus network set with the passenger flow intensity of the local optimal bus network obtained in step S3 to determine the relative merits of the initial optimal bus network set and the local optimal bus network in terms of passenger flow carrying capacity can continuously optimize the bus network's route combination and weight distribution, ensuring that the final bus network output achieves the optimal passenger flow carrying capacity, thereby improving the overall service quality and operational efficiency of the bus network. Based on the comparison of passenger flow intensities, the route search weights are adjusted. If the passenger flow intensity of the initial optimal bus network set is greater than that of the local optimal bus network, it indicates that the initial optimal bus network set performs better in terms of passenger flow. If the passenger flow intensity of the initial optimal bus network set is less than that of the local optimal bus network, the route weight settings need to be reviewed. During this adjustment process, a weight increase factor is obtained for subsequent specific adjustments to the route weights. When the passenger flow intensity of the initial optimal bus network set is greater than that of the local optimal bus network, a weight increase factor is calculated using a specific method to obtain the initial weight increase factor. This calculation method may consider factors such as the difference and ratio of passenger flow intensities between the two. The initial optimal bus network set is copied to the final bus network as the current candidate final network. When the passenger flow intensity of the initial optimal bus network set is less than that of the local optimal bus network, the Monte Carlo tree search is rerun. During the bus network optimization process, both the initial optimal bus network set obtained through the Monte Carlo tree search model on a global scale and the results of the local optimal bus network are considered. By comparing the two and adjusting the weights, a balance is achieved between global and local optima, avoiding the pursuit of local optima while ignoring the global optimality, making the bus network more rational and scientific.

[0132] It can be understood that the specific operations of steps S3 and S4 are as follows:

[0133] Initialize the total number of Monte Carlo tree search runs to Runs, and the total number of lines in the public transportation network solution to LinesNum; initialize the bus demand matrix to DicOD, and initialize the line counter LinesCount of the public transportation network target solution to 0; initialize the Monte Carlo tree search space, obtain the model line set AllChoices, which is all lines in the bus line set LinePool, initialize the search weight set of all lines in the search space, and obtain the spatial line search weight set Allwts=[100,100,..,...,100], and the current weight increase factor Curfactor=0;

[0134] Copy the model route set AllChoices to the public transportation network CurChoices, and copy all route search weight sets Allwts to the current route search weight set Curwts; based on the current route selection weight set Curwts, randomly select a public transportation route LineChoiced from the public transportation network CurChoices and add it to the optimal public transportation network set GlobePlan; increase the route counter LinesCount by 1;

[0135] Calculate the additional bus demand served by the optimal bus network set GlobePlan and update the bus demand matrix DicOD, deleting the bus demand already served by the network. Update the public transportation network CurChoices, delete the selected bus line LineChoiced, update the current route selection weight set to obtain the updated spatial route search weight set Curwts, and delete the weight of the selected line LineChoiced. Determine whether the line counter LinesCount is greater than LinesNum. If not, return to step S3c and select the next route. If the line counter LinesCount is greater than LinesNum, evaluate the passenger flow intensity of the generated optimal bus network set GlobePlan. Calculate the average passenger flow intensity GlobeDensity of the optimal bus network set GlobePlan, and use the passenger flow intensity LocalDensity of the local optimal network plan LocalPlan as a reference for adjusting route search weights. When the passenger flow intensity GlobeDensity is greater than LocalDensity, calculate the weight increase factor factor = GlobeDensity / LocalDensity. Update the route search weight set Allwts by multiplying the weight of each bus route in the optimal bus network set GlobePlan by the weight increase factor factor; otherwise, factor = 0. If factor > Curfactor, then Curfactor = factor, and copy the bus network GlobePlan to the final bus network LinePlan. Determine whether the number of Monte Carlo tree search runs is greater than the total number of runs Runs. If so, terminate the search and output the final bus network LinePlan. Otherwise, return to step S3.

[0136] In a second aspect, the present invention discloses a bus network optimization system based on reinforcement learning technology, which includes a bus network optimization method based on reinforcement learning technology.

[0137] Specifically, the system adopts the bus network optimization method based on reinforcement learning technology disclosed in the first aspect, constructs a local optimal bus route based on passenger flow intensity, obtains the first passenger flow intensity, constructs a Monte Carlo tree search model, constructs the optimal bus network for the bus route set, calculates its average passenger flow intensity, obtains the second passenger flow intensity, compares and judges the first passenger flow intensity and the second passenger flow intensity to obtain the final bus network, plans bus routes based on passenger flow intensity, can accurately match passenger flow, reasonably allocate resources to hot spots according to passenger flow intensity in different regions and time periods, and flexibly respond to passenger flow changes. The Monte Carlo tree search model can quickly generate plans, calculate search weights using multiple attributes of routes, and comprehensively analyze routes to make route selection and network construction more scientific, based on data and algorithm decisions, and reduce human experience bias.

[0138] In a third aspect, the present invention discloses an electronic device comprising a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus;

[0139] Memory for storing computer programs;

[0140] The processor is used to implement the steps of the bus network optimization method based on reinforcement learning technology when executing the program stored in the memory.

[0141] Specifically, the electronic device can implement the steps of the bus network optimization method based on reinforcement learning technology disclosed in the first aspect through a processor, construct a local optimal bus route based on passenger flow intensity, obtain a first passenger flow intensity, construct a Monte Carlo tree search model, construct an optimal bus network for the bus route set, calculate its average passenger flow intensity, obtain a second passenger flow intensity, compare and judge the first passenger flow intensity and the second passenger flow intensity to obtain the final bus network, plan bus routes based on passenger flow intensity, accurately match passenger flow, reasonably allocate resources to hot spots according to passenger flow intensity in different regions and time periods, and flexibly respond to passenger flow changes. The Monte Carlo tree search model can be used to quickly generate plans, use route multi-attributes to calculate search weights, and comprehensively analyze routes to make route selection and network construction more scientific, make decisions based on data and algorithms, and reduce human experience bias.

[0142] In a fourth aspect, the present invention discloses a storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of a bus network optimization method based on reinforcement learning technology.

[0143] Specifically, the storage medium stores a computer program related to the bus network optimization method based on reinforcement learning technology disclosed in the first aspect, which can be implemented by running on a computer, constructing a local optimal bus route based on passenger flow intensity, obtaining a first passenger flow intensity, constructing a Monte Carlo tree search model, constructing an optimal bus network for a set of bus routes, calculating its average passenger flow intensity, obtaining a second passenger flow intensity, comparing and judging the first passenger flow intensity and the second passenger flow intensity to obtain the final bus network, planning bus routes based on passenger flow intensity, accurately matching passenger flow, and rationally allocating resources to hot spots according to passenger flow intensity in different regions and time periods, flexibly responding to passenger flow changes, using a Monte Carlo tree search model to quickly generate plans, using multiple attributes of routes to calculate search weights, and comprehensively analyzing routes to make route selection and network construction more scientific, making decisions based on data and algorithms, and reducing human experience bias.

[0144] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0145] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise" and the like to indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0146] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature identified as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0147] In the present invention, unless otherwise expressly specified or limited, terms such as "mounted," "connected," "connect," and "fixed" should be understood broadly. For example, they may refer to connection, detachable connection, or integration; mechanical connection or electrical connection; direct connection or indirect connection through an intermediate medium; and internal communication between two components or interaction between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0148] In the present invention, unless otherwise expressly specified or limited, a first feature being "above" or "below" a second feature may include the first and second features being in direct contact, or may include the first and second features being in contact not directly but through another feature between them. Furthermore, a first feature being "above," "above," and "above" a second feature may include the first feature being directly above or obliquely above the second feature, or may simply mean that the first feature is higher in level than the second feature. A first feature being "below," "below," and "below" a second feature may include the first feature being directly below or obliquely below the second feature, or may simply mean that the first feature is lower in level than the second feature.

[0149] In the description of this specification, the reference terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" mean that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic expressions of the above terms should not be understood as necessarily referring to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification.

[0150] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, to the extent such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to encompass such changes and modifications.

[0151] The above description is a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and such modifications or substitutions are intended to be within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A bus network optimization method based on reinforcement learning technology, characterized in that: Specifically include: S1: Obtain bus stops, draw bus routes between the bus stops, construct a set of candidate bus routes, and calculate the passenger flow intensity of each bus route in the set of candidate bus routes. The set of candidate bus routes must meet a preset non-linear coefficient threshold and a preset passenger flow intensity threshold. S2 includes S21-S26, as follows: S21: For the set of alternative bus routes, preset the maximum number of bus routes included in the local optimal bus network and the minimum service bus demand ratio threshold that the local optimal bus route set needs to meet; S22: Preprocess the routes of the candidate bus route set, sort them according to the passenger flow intensity of each route, initialize the selected route set, and initialize the local optimal bus network set; S23: Selecting the route with the largest passenger flow intensity in the set of candidate bus routes as the first route in the local optimal bus network set, adding it to the set of selected routes, and removing the route from the set of candidate bus routes; S24: Selecting the second route of the local optimal bus network set, adding each bus route in the candidate bus route set to the selected route set, calculating the average passenger flow intensity of the selected route set after the different routes are added, obtaining the route that maximizes the average passenger flow intensity of the selected route set, and adding it as the second route to the local optimal bus network set, and calculating the proportion of total bus demand served by the current local optimal bus network; the total bus demand proportion includes the passenger flow intensity of direct bus demand and the passenger flow intensity of bus demand with a single transfer at the same platform; S25: updating the bus demand OD matrix of the bus terminal pair set, removing the bus demand already served by the current local optimal bus network from the bus demand OD matrix, and updating the candidate bus route set and the selected route set; updating the candidate bus route set includes deleting the newly selected route from the candidate bus route set, and deleting the routes in the candidate bus route set whose passenger flow intensity is lower than a preset passenger flow intensity threshold, to obtain an updated candidate bus route set; updating the selected route set includes keeping the selected route set consistent with the current local optimal bus network; S26: Repeat the steps S21-25 above to obtain the next route of the local optimal bus route set, update the bus demand OD matrix, the candidate bus route set and the selected route set, and calculate the total bus demand ratio served by the current local optimal bus network until the total bus demand ratio served by the current local optimal bus network exceeds the preset minimum service bus demand ratio threshold, or the number of bus routes in the current local optimal bus route set is equal to the total number of bus routes in the preset first number. Then, the selected route set is output as the local optimal bus route set, and the passenger flow intensity of the local optimal bus network is calculated, and the passenger flow intensity of the local optimal bus network is used as the first passenger flow intensity. S3: Constructing a Monte Carlo tree search model, assigning the same initial probability selection weight to each bus route in the candidate bus route set, randomly searching for selected bus routes consisting of a preset first number of different bus routes, calculating the average passenger flow intensity, and obtaining a second passenger flow intensity; S4: Compare the second passenger flow intensity with the first passenger flow intensity. If it is greater than the first passenger flow intensity, increase the probability selection weights of the selected preset first number of bus routes to obtain updated probability selection weights. Based on the updated probability selection weights, re-run the Monte Carlo tree search to screen the selected bus routes. After repeatedly iterating the updated probability selection weights and the selected bus routes, finally generate a route combination plan with the optimal passenger flow intensity to obtain the global optimal bus network.

2. The method according to claim 1, characterized in that The method of obtaining bus stops, drawing bus routes between the bus stops, constructing a set of candidate bus routes, and calculating the passenger flow intensity of each bus route in the set of candidate bus routes, wherein the set of candidate bus routes needs to meet a preset non-linear coefficient threshold and a preset passenger flow intensity threshold, specifically includes the following steps: Obtain a road traffic network diagram, draw bus routes based on the stations and road sections in the diagram, establish a set of bus terminal pairs based on the stations in the road traffic network diagram, establish a bus demand OD matrix indexed by road sections, and preset non-linear coefficient thresholds and passenger flow intensity thresholds; Expand the nodes of the first station of the bus terminal pair, expand the nodes of any first station, obtain the adjacent station list, traverse the adjacent station list, and combine the adjacent stations in the adjacent station list with the corresponding first station to form a node list, and obtain the node list of all the first stations; Calculating the non-linear coefficient and passenger flow intensity of the node lists of all first stations to determine whether they meet the construction requirements, and constructing an initial route list from the node lists of all first stations that meet the construction requirements to obtain the initial route list. The construction requirements include that the non-linear coefficient of the node list is less than a preset non-linear coefficient threshold and the passenger flow intensity is greater than a preset passenger flow intensity threshold. The node list of the first station is the initial route. The adjacent sites in the initial route list are used as the first sites to expand the nodes again, obtain the corresponding adjacent site list, and generate a derived route list that meets the construction requirements; When the adjacent station in the derived route list is the last station of the bus terminal, stop the node expansion, integrate the initial route list and the derived route list into a bus route, and obtain the set of alternative bus routes.

3. The method according to claim 1, characterized in that The Monte Carlo tree search model is constructed, and the same initial probability selection weight is assigned to each bus route in the candidate bus route set. The selected bus routes consisting of a preset first number of different bus routes are randomly searched, and the average passenger flow intensity is calculated to obtain the second passenger flow intensity. Specifically, the following steps are included: Preset the total number of Monte Carlo tree search runs and the total number of routes in the global optimal bus network solution; initialize the bus demand OD matrix, initialize the route probability selection weight set, and initialize the selected route set; According to the route probability selection weight set, a bus route is randomly selected from the candidate route set, added to the selected route set, the candidate route set is updated, and the selected bus route is removed; the route probability selection weight set is updated, and the selected bus route is removed; Determine whether the total number of routes in the selected route set is greater than the preset total number of routes in the global optimal bus network. If it is less than the preset value, select a bus route from the updated alternative bus route set again, and then repeat the update process until the total number of routes in the selected bus route set is equal to the preset value; calculate the average passenger flow intensity of the selected route set as the second passenger flow intensity.

4. The method according to claim 1, wherein The second passenger flow intensity is compared with the first passenger flow intensity. If the second passenger flow intensity is greater than the first passenger flow intensity, the probability selection weights of the selected preset first number of bus routes are increased to obtain updated probability selection weights. Based on the updated probability selection weights, the Monte Carlo tree search is re-run to screen the selected bus routes. After repeatedly iterating the updated probability selection weights and the selected bus routes, a route combination plan with the optimal passenger flow intensity is finally generated to obtain the global optimal bus network. Specifically, the following steps are included: When the second passenger flow intensity is greater than the first passenger flow intensity, the ratio of the two is used as the factor increasing the probability selection weight of this route. For the route probability selection weight set, the probability selection weight of each route in the currently selected route set is multiplied by the weight increase factor, and the initial route probability weight set is updated. Compare the number of Monte Carlo tree searches with the preset total number of runs. If the number of Monte Carlo tree searches is less than the preset total number of runs, save the selected route set to the global optimal bus network solution set and rerun the Monte Carlo tree search. If the number of Monte Carlo tree searches is greater than the preset total number of runs, the set of selected routes with the second largest passenger flow intensity is output as the global optimal bus network.

5. The method according to claim 1, wherein Calculating the passenger flow intensity of direct bus demand includes the following steps: Split the selected bus route into multiple continuous sections, select any two sections to construct a section combination, and obtain a section combination set; Select any section combination from the section combination set and determine whether the section combination is included in the bus demand matrix. If so, allocate the passenger demand corresponding to the bus demand matrix to the selected bus route to obtain the passenger flow of the bus route. If not, select a new section combination and mark the loaded section combination. Obtain the number of passengers on the bus route with the largest passenger flow section, sum it with the passenger flow of the bus route, and determine whether the sum is greater than the maximum cross-section passenger capacity of the bus route; If the sum is greater than the maximum cross-section passenger capacity of the bus line, the passenger flow value of the bus line is the passenger flow intensity of the direct bus demand of the selected bus line; If the sum is equal to or less than the maximum cross-sectional passenger capacity of the bus line, the passenger flow intensity of the direct bus demand of the selected bus line is the maximum cross-sectional passenger capacity of the selected bus line minus the passenger volume of the largest passenger flow section on the selected bus line.

6. The method according to claim 1, characterized in that Calculating the passenger flow intensity of a bus transfer demand on the same platform includes the following steps: Determine the bus routes and transfer sections that can be transferred on the same platform for the selected bus routes, and construct a transfer route set; Select transfer routes that pass through the same road section as the selected bus route from the transfer route set to construct a transfer road section set; Selecting a first transfer segment from the transfer segment set and combining it with the selected bus route to construct a first preceding segment set and a first succeeding segment set, wherein the first preceding segment set includes all segments between the starting segment of the bus route and the first transfer segment, and the first succeeding segment set includes all segments between the first transfer segment and the ending segment of the bus route; Based on the first transfer segment, the second preceding segment set and the second subsequent segment set are combined with the transfer segments in the transfer segment set to construct the second preceding segment set, wherein the second preceding segment set includes all segments between the starting segment of the transfer segment and the first transfer segment, and the second subsequent segment set includes all segments between the second transfer segment and the ending segment of the bus route; Selecting a first selected preceding segment and a second selected subsequent segment from the first preceding segment set and the second subsequent segment set respectively to form a selected segment combination, determining whether the selected segment combination is in the bus demand matrix, and obtaining a first passenger demand; Allocate the passenger demand of the corresponding selected segment combination to the corresponding first transfer segment and transfer line, add the passenger volume of all segments in the first selected preceding segment and the passenger volume of all segments in the second selected subsequent segment by the first passenger demand, and obtain the passenger volume set of the first selected preceding segment and the passenger volume set of the second selected subsequent segment; Obtain the maximum cross-sectional passenger capacity of the line corresponding to the first selected preceding section and the second selected subsequent section, obtain the maximum passenger flow section and maximum passenger flow of the first selected preceding section and the maximum passenger flow section and maximum passenger flow of the second selected subsequent section from the passenger volume set of the first selected preceding section and the passenger volume set of the second selected subsequent section, compare the maximum passenger flow of the first selected preceding section with the maximum cross-sectional passenger capacity of the line of the first selected preceding section, and compare the maximum passenger flow of the second selected subsequent section with the maximum cross-sectional passenger capacity of the line of the second selected subsequent section, when the maximum passenger flow of the first selected preceding section is the maximum passenger flow of the second selected subsequent section, When the maximum passenger flow is less than the maximum cross-sectional passenger capacity of the line, the passenger flow intensity of the bus transfer demand between the first selected preceding section and the second selected subsequent section is the first passenger demand; when the maximum passenger flow is greater than the maximum cross-sectional passenger capacity of the line, the passenger flow intensity of the bus transfer demand between the first selected preceding section and the second selected subsequent section is the difference between the maximum cross-sectional passenger capacity of the line and the passenger volume of all sections within the first selected preceding section, and the difference between the maximum cross-sectional passenger capacity of the line and the passenger volume of all sections within the second selected subsequent section; Repeat the above process for the first set of subsequent road segments and the second set of subsequent road segments to calculate the passenger flow intensity of the corresponding one-time same-platform transfer bus demand.

7. The method according to claim 1, characterized in that The following steps are also included: For the final bus network, obtain the passenger flow of each station in real time; Develop differentiated bus departure time adjustment strategies based on the travel characteristics and needs of different groups of people; For routes with high passenger flow determined through data analysis and field investigations, the frequency of routes will be increased during peak hours. At the same time, shortened routes will be designed and added.

8. A bus network optimization system based on reinforcement learning technology, characterized in that: A bus network optimization method based on reinforcement learning technology comprising the method described in any one of claims 1-7.

9. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus; Memory for storing computer programs; The processor is configured to implement the steps of the bus network optimization method based on reinforcement learning technology as described in any one of claims 1 to 7 when executing the program stored in the memory.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the bus network optimization method based on reinforcement learning technology as described in any one of claims 1 to 7 are implemented.

Images