Full-link travel time decision method and system for complex traffic scenarios

Abstract

The application discloses a kind of complex traffic scene-oriented full-link travel time decision method and system, belong to travel navigation technical field.The method is by fusing road network data, real-time traffic state data, historical congestion data, target station's passenger flow queue data and time characteristic data, forms travel time influence factor database;Based on the unified travel time influence factor database, construct integrated full-link time prediction model, according to integrated full-link time prediction model calculation travel full-link total time consumption;Obtain the expected arrival time input by user, by reverse time deduction algorithm, in combination with travel full-link total time consumption, calculate and output optimal departure time;In the process of travel, compare the deviation of real-time dynamic data and prediction data, when deviation meets preset condition, trigger dynamic updating mechanism, generate and push the updated travel decision.The application significantly improves the accuracy and stability of physical prediction.

Description

Technical Field

[0001] This invention relates to the field of travel navigation technology, and in particular to a method and system for making decisions on travel time across the entire journey in complex traffic scenarios. Background Technology

[0002] With the acceleration of urbanization, traffic congestion has become the norm, and users' demand for certainty in travel time is becoming increasingly urgent. In daily travel, especially in scenarios such as business trips, high-speed rail connections, and sightseeing, users not only need to know "which route is the fastest," but also hope to get clear answers to "when is the most suitable time to depart" and "whether they can arrive on time."

[0003] However, the core function of existing navigation tools (such as Gaode Maps and Baidu Maps) focuses on route planning, which provides users with the fastest or shortest route from point A to point B based on static or dynamic road information such as distance, speed limits, and real-time traffic conditions. When using these tools, users still need to deduce the departure time from the estimated travel time themselves, and they cannot know the time spent in non-road links such as airport security queues, high-speed rail station ticket checks, and scenic spot entry queues. This leads to significant deviations in the time estimation of the entire travel process. Furthermore, existing technologies can only output the predicted travel time, which is at the "information provision" level. They cannot directly output the "optimal departure time" based on the user's expected arrival time, failing to solve the core decision-making problem of "when to depart." At the same time, they lack a dynamic update mechanism for the entire travel process. When dynamic factors such as sudden traffic control, extreme weather, or a surge in passenger flow cause deviations between actual and predicted data, they cannot respond in real time and proactively push adjusted decision suggestions.

[0004] Therefore, there is an urgent need in this field for a method and system for making decisions on travel time across the entire transportation chain in complex traffic scenarios. Summary of the Invention

[0005] In view of this, the present invention provides a method for end-to-end travel time decision-making in complex traffic scenarios, in order to solve the problems of navigation tools being unable to provide accurate time decisions and the lack of end-to-end prediction in existing methods.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: On the one hand, this invention provides a method for end-to-end travel time decision-making in complex traffic scenarios, including: S1: Integrate road network data, real-time traffic status data, historical congestion data, passenger flow queuing data at target stations, and time characteristic data to form a database of factors affecting travel time; wherein, the road network data includes route geographic information and basic time consumption data for different modes of transportation; the real-time traffic status data includes real-time vehicle speed, traffic flow, traffic accidents, road construction, and temporary traffic control data for road segments; the passenger flow queuing data at target stations includes real-time passenger flow density, queue length, and dynamic waiting time data corresponding to airport security check queues, high-speed rail station ticket check queues, and scenic spot entrance verification queues; and the time characteristic data includes time period attribute data corresponding to weekdays, weekends, morning and evening peak hours, and holidays. S2: Based on the travel time influencing factor database, construct a full-link time prediction model, wherein the full-link time prediction model includes a path time calculation sub-model, a traffic congestion prediction sub-model, and a station passenger flow time prediction sub-model; S3: For the travel demand from the same origin to the destination terminal node, generate multiple candidate routes, and use the full-link time prediction model to calculate the total full-link time covering road traffic, congestion superposition, and waiting time within the terminal. S4: Obtain the expected arrival time of the target station's cutoff node input by the user, use the expected arrival time as the unique reference time for reverse calculation, automatically match the full-link key node sequence according to the target station type, and calculate the optimal departure time for each candidate path by combining the total full-link time corresponding to each candidate path; the key node sequence is in the forward order of travel as follows: user departure point, station drop-off point, security checkpoint or park entry verification point, and target station cutoff node; S5: Prioritize multiple candidate routes according to preset dimensions and output multiple travel plans for users, including the optimal departure time; wherein, the preset dimensions include the highest probability of on-time arrival, the shortest time, and the lowest cost.

[0007] Preferably, step S3 includes: The path time calculation sub-model, based on path geographic information and real-time traffic conditions, corrects the basic travel time for different time periods and different modes of transportation through a dynamic time evaluation algorithm, and calculates the basic travel time. The traffic congestion prediction sub-model, based on historical congestion data and real-time traffic conditions, uses time series analysis algorithms and spatiotemporal graph neural network algorithms to predict road congestion during the user's travel period and calculates the congestion increment time. The station passenger flow time prediction sub-model, based on the passenger flow queuing data and time characteristic data of the target station, predicts the passenger flow waiting status during the user's arrival time at the station through regression model and Bayesian prediction model, and calculates the waiting time of non-road links. The total time for the entire link is the sum of the basic travel time, the congestion increment time, and the waiting time for non-road links.

[0008] Preferably, in step S4, the optimal departure time for each candidate path is calculated based on the total end-to-end time corresponding to each candidate path, including: Based on the end-to-end time prediction model, the segmented time between adjacent key nodes is calculated respectively. Combined with the expected arrival time, the latest arrival time corresponding to each key node is calculated in reverse order of the key node sequence. The theoretical departure time is calculated using a preset formula, which is: Theoretical departure time = Expected arrival time at the target station's cutoff node - Total time across the entire link - Preset buffer time; By combining real-time traffic conditions, station passenger flow status, and users' historical travel data, the theoretical departure time is calibrated to generate the optimal departure time that finally matches the corresponding candidate path.

[0009] Preferably, the method further includes: During a user's travel, the deviation between actual traffic and station passenger flow data and predicted data is compared in real time. When the deviation exceeds a preset threshold, a dynamic update mechanism is triggered, recalculating the entire-link time prediction model and the reverse time extrapolation algorithm, and pushing the updated travel decision.

[0010] Preferably, the dynamic update mechanism continuously monitors changes in road congestion, sudden increases in passenger flow at stations, temporary traffic control, and sudden emergencies in temporary station management throughout the entire travel process through an event triggering mechanism and real-time data push technology; When the deviation exceeds the preset threshold, the system will automatically recalculate and push the updated departure time and route adjustment plan to the user through at least one of the following channels: in-app push, SMS, or instant messaging tools.

[0011] Preferably, the method further includes: User profiles are constructed based on users' historical travel data. These user profiles include users' transportation mode preferences, travel speed, fast passage permissions at stations, baggage carrying status, check-in or park entry habits, and time sensitivity. Based on the user profile, the time consumption calculation of each link in the entire process is personalized and calibrated, and corresponding travel reminder strategies are matched to optimize the calculation results of the reverse time extrapolation algorithm.

[0012] On the other hand, the present invention provides a full-link travel time decision-making system for complex traffic scenarios, the system comprising: The fusion module is used to integrate road network data, real-time traffic status data, historical congestion data, passenger flow queuing data at target stations, and time characteristic data to form a database of factors affecting travel time. The road network data includes route geographic information and basic time consumption data for different modes of transportation. The real-time traffic status data includes real-time vehicle speed, traffic volume, traffic accidents, road construction, and temporary traffic control data for road segments. The passenger flow queuing data at target stations includes real-time passenger flow density, queue length, and dynamic waiting time data for airport security queues, high-speed rail station ticket queues, and scenic area entrance verification queues. The time characteristic data includes time period attribute data corresponding to weekdays, weekends, morning and evening peak hours, and holidays. The construction module is used to build a full-link time prediction model based on the travel time influencing factor database. The full-link time prediction model includes a path time calculation sub-model, a traffic congestion prediction sub-model, and a station passenger flow time prediction sub-model. The calculation module is used to generate multiple candidate routes for travel demand from the same departure point to the destination terminal node, and use the full-link time prediction model to calculate the total time of the entire link covering road traffic, congestion superposition, and waiting in the entire process inside the terminal. The deduction module is used to obtain the expected arrival time of the target station's cutoff node input by the user, and use the expected arrival time as the unique reference time for reverse calculation. Based on the target station type, it automatically matches the sequence of key nodes in the entire link, and calculates the optimal departure time for each candidate path by combining the total time of the entire link corresponding to each candidate path. The sequence of key nodes, in the forward order of travel, is as follows: user departure point, station drop-off point, security checkpoint or park entry verification point, and target station cutoff node. The output module is used to prioritize multiple candidate routes according to preset dimensions and output multiple travel plans for users, including the optimal departure time. The preset dimensions include the highest probability of on-time arrival, the shortest time, and the lowest cost.

[0013] On the other hand, the present invention provides a computer device, including a memory and a processor, wherein the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory to implement the steps of any of the methods described in this specification.

[0014] On the other hand, the present invention provides a readable storage medium, characterized in that the storage medium stores a computer program, which, when executed by a processor, implements the steps of any of the methods described in the specification.

[0015] On the other hand, the present invention provides a computer program product, characterized in that it includes a computer program, which, when executed by a processor, implements the steps of any of the methods described in this specification. This invention provides a method for end-to-end travel time decision-making in complex traffic scenarios. First, it constructs a standardized database of travel time influencing factors through multi-source heterogeneous data cleaning, standardization, and spatiotemporal correlation matching, providing a complete and compliant input foundation for subsequent modeling. Then, it utilizes an end-to-end time prediction model including a path time calculation sub-model, a traffic congestion prediction sub-model, and a station passenger flow time prediction sub-model to convert discrete time data from multiple stages and dimensions, such as road travel, congestion accumulation, and waiting throughout the entire station process, into a unified, summable, and error-free total end-to-end time. This effectively addresses the core pain points of traditional navigation tools, such as lack of end-to-end modeling, blind spots in non-road information, incomparability of time across multiple stages, and distorted prediction results. Finally, it completes multiple candidate routes for travel demand from the same origin to the destination station's cutoff node. The generation of data and the calculation of end-to-end metrics are synchronized to obtain the total end-to-end time and on-time arrival probability for each path. Then, the path priority is sorted and multiple solutions are aggregated and output according to preset dimensions to ensure that the feature extraction of travel solutions is robust and unbiased and fully covers multi-dimensional needs. Next, using the expected arrival time of the target station's cutoff node input by the user as the sole benchmark, a multi-node hierarchical reverse time extrapolation algorithm is constructed to reversely decompose the latest arrival time of each key node in the end-to-end, separating the differential effects of road traffic, station connections, and time cycles, and realizing accurate spatiotemporal joint decision inference based on the target arrival time. Finally, through real-time data comparison of the entire journey and a dynamic update mechanism triggered by deviation thresholds, the optimal departure time with quantified uncertainty of on-time rate and dynamic adjustment plan are output, making travel time decisions credible, explainable, implementable, and able to support the user's full-process travel management. This invention achieves end-to-end, intelligent, spatiotemporal, and quantitative urban travel time decision-making through a collaborative closed-loop modeling process involving data fusion, model building, multi-scheme screening, reverse decision deduction, and dynamic updating and optimization. It fundamentally overcomes the inherent defects of existing technologies, such as misalignment of core requirements, lack of end-to-end modeling, insufficient prediction accuracy, lagging dynamic response, and lack of decision-making capabilities. It significantly improves the accuracy of travel time prediction, the reliability of travel decisions, the stability of adaptability to all scenarios, and practical value. Attached Figure Description

[0016] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0017] Figure 1 A flowchart of a full-link travel time decision-making method for complex traffic scenarios provided by an embodiment of the present invention; Figure 2 This is a hardware architecture diagram of a computer device provided in an embodiment of the present invention; Figure 3 This is a structural diagram of a full-link travel time decision system for complex traffic scenarios provided by an embodiment of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] The embodiments of the present invention are described below with reference to the figures.

[0020] like Figure 1 As shown, this invention provides a method for end-to-end travel time decision-making in complex traffic scenarios, including: Step S1: Integrate road network data, real-time traffic status data, historical congestion data, passenger flow queuing data at target stations, and time characteristic data to form a database of factors affecting travel time. The road network data includes route geographic information and basic travel time data for different modes of transportation. The real-time traffic status data includes real-time vehicle speed, traffic volume, traffic accidents, road construction, and temporary traffic control data for road segments. The passenger flow queuing data at target stations includes real-time passenger density, queue length, and dynamic waiting time data for airport security queues, high-speed rail station ticket queues, and scenic area entrance verification queues. The time characteristic data includes time period attribute data corresponding to weekdays, weekends, morning and evening peak hours, and holidays.

[0021] In a preferred embodiment, firstly, a multi-source data acquisition and fusion system is constructed. Road network data is obtained in real time by calling the application programming interface of map open platforms (such as Amap, Baidu Maps, etc.). This data includes at least the geographical information of the route (such as road topology, road segment length, direction of travel, location of points of interest, etc.) and basic time data of different modes of transportation (including driving, public transportation, walking, cycling, etc.), providing underlying support for subsequent route time calculation.

[0022] Real-time traffic status data is obtained by integrating real-time data sources opened by traffic management departments, real-time traffic interfaces provided by internet map service providers, and floating car data (such as GPS tracks uploaded by ride-hailing vehicles and taxis). Specifically, it includes dynamic information such as real-time vehicle speed, traffic flow, traffic accidents, road construction, and temporary traffic control. This data can capture instantaneous changes in traffic flow and is used to dynamically adjust basic travel times.

[0023] Historical congestion data is obtained through long-term accumulated road traffic records, including road congestion periods, congestion levels (such as traffic speed and congestion index) and durations for different time periods (e.g., weekday morning rush hour, evening rush hour, off-peak, and nighttime) and different date types (e.g., weekends and holidays). This data is used to train a traffic congestion prediction sub-model to learn the periodic and sudden patterns of congestion.

[0024] Passenger flow queuing data for target stations is obtained through dedicated passenger flow monitoring systems connected to travel scenarios such as airports, high-speed rail stations, and scenic spots. This includes real-time passenger flow density (number of people waiting per unit area), queue length (number of people or queuing time), and dynamic waiting time data for airport security queues, high-speed rail station ticket check queues, and scenic spot entrance verification queues. This data incorporates the time spent on non-road components into the overall prediction, compensating for the shortcomings of traditional navigation tools that only focus on the road travel phase.

[0025] Time feature data is automatically generated from calendar information, including time period attribute data corresponding to weekdays, weekends, morning and evening peak hours, and holidays, which is used to characterize the impact of travel time on traffic efficiency.

[0026] The aforementioned multi-source heterogeneous data underwent a series of preprocessing operations, including data cleaning (removing outliers and duplicates), deduplication, standardization (unifying data format, coordinate system, and timestamps), and association matching (aligning by key values ​​such as region, time, and route segment), before being integrated and stored in a unified database of travel time influencing factors. This database provides a complete, consistent, and reliable data foundation for the subsequent construction and calculation of the end-to-end time prediction model.

[0027] S2: Based on the database of factors affecting travel time, construct a full-link time prediction model, wherein the full-link time prediction model includes a path time calculation sub-model, a traffic congestion prediction sub-model, and a station passenger flow time prediction sub-model.

[0028] In a preferred embodiment, the end-to-end time prediction model is an integrated prediction framework for the entire travel process. Its core lies in the joint modeling of the time consumption of three parts: the road travel phase, the congestion-addition phase, and the off-road phase at the transportation hub. Each of the three sub-models is responsible for estimating the time of its respective stage, ultimately outputting a unified total travel time. Through this construction method, the end-to-end time prediction model proposed in this invention overcomes the limitation of existing technologies that only focus on the road travel phase, achieving complete time prediction for "road travel + congestion overlay + waiting throughout the entire process at the transportation hub," providing an accurate time benchmark for subsequent reverse time extrapolation and the calculation of the optimal departure time.

[0029] S3: For the travel demand from the same departure point to the destination terminal node, generate multiple candidate routes, and use the full-link time prediction model to calculate the total time of the entire link covering road traffic, congestion superposition, and waiting throughout the entire process inside the terminal.

[0030] Further, step S3 includes: The path time calculation sub-model, based on path geographic information and real-time traffic conditions, corrects the basic travel time for different time periods and different modes of transportation through a dynamic time evaluation algorithm, and calculates the basic travel time. The traffic congestion prediction sub-model, based on historical congestion data and real-time traffic conditions, uses time series analysis algorithms and spatiotemporal graph neural network algorithms to predict road congestion during the user's travel period and calculates the congestion increment time. The station passenger flow time prediction sub-model, based on the passenger flow queuing data and time characteristic data of the target station, predicts the passenger flow waiting status during the user's arrival time at the station through regression model and Bayesian prediction model, and calculates the waiting time of non-road links. The total time for the entire link is the sum of the basic travel time, the congestion increment time, and the waiting time for non-road links.

[0031] In a preferred embodiment, firstly, after the user inputs the departure point (e.g., current location or specified starting point) and the destination station's closing time (e.g., airport gate closing time, high-speed rail station ticket checking time, scenic area last entry time), the system generates multiple candidate routes from the departure point to the destination station's drop-off point (e.g., airport departure level, high-speed rail station entrance, scenic area parking lot) based on the real-time road network topology by calling the route planning interface of the map open platform. The candidate route generation strategy may include conventional dimensions such as shortest distance, shortest time, highway priority, and avoiding congestion, to ensure that users have sufficient route selection space.

[0032] Then, for each candidate path, the system calls the end-to-end time prediction model built in step S2 above to calculate the total end-to-end time for that path.

[0033] Specifically, the route time calculation sub-model takes road network data and real-time traffic status data from the travel time influencing factors database as input. Specifically, it extracts route geographic information (including road segment length, road grade, travel direction, intersection information, etc.) and basic travel time data for different modes of transportation (such as driving, public transport, walking, and cycling) from the road network data; simultaneously, it obtains dynamic information such as real-time vehicle speed, traffic flow, traffic accidents, construction, and traffic control from the real-time traffic status data. The route time calculation sub-model employs a dynamic time evaluation algorithm, dynamically adjusting the basic travel time according to the travel period (e.g., morning peak, off-peak, evening peak). For example, during congested periods, the algorithm amplifies and corrects the travel time under free-flow conditions based on the ratio of historical speeds to the current real-time speed; during non-congested periods, the basic travel time is used. Through the above calculations, the basic travel time is obtained, which reflects the road travel time without considering additional congestion and waiting at stations.

[0034] The traffic congestion prediction sub-model takes historical congestion data and real-time traffic status data from a database of factors influencing travel time as input. Historical congestion data includes regular information such as road congestion index, congestion duration, and distribution of congested road segments under different date types (weekdays, weekends, holidays) and time periods (peak hours, off-peak hours); real-time traffic status data provides dynamic changes in the current road network. The traffic congestion prediction sub-model combines time series analysis algorithms (such as autoregressive moving average models and long short-term memory networks) with spatiotemporal graph neural network algorithms (such as T-GCN and DCRNN). Time series analysis algorithms capture the periodic evolution trend of congestion status on a single road segment, while spatiotemporal graph neural network algorithms further utilize the road network topology to model the mutual influence and propagation effect of congestion status between adjacent road segments. By predicting the road congestion status during the user's planned travel time, this sub-model outputs the congestion increment time, i.e., the additional time cost due to congestion compared to free-flowing travel time.

[0035] The station passenger flow time prediction sub-model takes target station passenger flow queuing data and time characteristic data from the travel time influencing factors database as input. Target stations include airports, high-speed rail stations, scenic spots, etc., and their passenger flow queuing data includes real-time passenger flow density (number of people waiting per unit area), queue length (number of people queuing or estimated queuing time), and dynamic waiting time (waiting time estimated based on the fusion of historical and real-time data); time characteristic data includes time period attributes such as weekdays, weekends, morning and evening peak hours, and holidays. The station passenger flow time prediction sub-model adopts a combination of regression models (such as linear regression and support vector regression) and Bayesian prediction models. The regression model is used to establish the functional relationship between passenger flow queuing data, time characteristics, and waiting time in non-road links; the Bayesian prediction model introduces prior distribution (e.g., based on historical passenger flow statistics for the same period) and uses real-time data to update the posterior probability, thereby quantifying the uncertainty of the prediction. Through the above model, this sub-model predicts the waiting time of non-road links such as security check queuing, ticket check queuing, or entrance verification during the time a user arrives at the station, and outputs the waiting time of non-road links.

[0036] Finally, the total time for the entire journey is obtained by adding the basic travel time, the congestion-induced time, and the waiting time for non-road components. This total time fully covers the three components of "road travel time + congestion-induced time + waiting time within the station", achieving precise time quantification for the entire travel journey.

[0037] It should be noted that for different candidate routes from the same origin to the same destination terminal, the basic travel time, congestion increment time, and off-road waiting time may vary significantly due to differences in route geographic information, real-time traffic conditions, and terminal passenger flow. Therefore, the system will calculate the independent total end-to-end time for each candidate route, providing an accurate time benchmark for calculating the optimal departure time for each candidate route in subsequent step S4.

[0038] Through the above method, step S3 of the present invention overcomes the limitation of the prior art that can only provide single-path travel time prediction, realizes the integrated time calculation of the entire link for multiple candidate paths, and provides users with a diversified basis for travel options.

[0039] S4: Obtain the expected arrival time of the target station's cutoff node input by the user, use the expected arrival time as the sole reference time for reverse calculation, automatically match the full-link key node sequence according to the target station type, and calculate the optimal departure time for each candidate path by combining the total full-link time corresponding to each candidate path; the key node sequence is in the forward order of travel as follows: user departure point, station drop-off point, security checkpoint or park entry verification point, and target station cutoff node.

[0040] Furthermore, in step S4, the optimal departure time for each candidate path is calculated based on the total end-to-end time for each candidate path, including: Based on the end-to-end time prediction model, the segmented time between adjacent key nodes is calculated respectively. Combined with the expected arrival time, the latest arrival time corresponding to each key node is calculated in reverse order of the key node sequence. The theoretical departure time is calculated using a preset formula, which is: Theoretical departure time = Expected arrival time at the target station's cutoff node - Total time across the entire link - Preset buffer time; By combining real-time traffic conditions, station passenger flow status, and users' historical travel data, the theoretical departure time is calibrated to generate the optimal departure time that finally matches the corresponding candidate path.

[0041] In a preferred embodiment, step S4 first obtains the expected arrival time of the target terminal node input by the user through the user interface, such as "I want to arrive at the airport gate before 9:00 AM". This expected arrival time serves as the sole reference time for all subsequent reverse time calculations, and all calculations are performed in reverse order around this time. Subsequently, the system automatically matches the corresponding full-link key node sequence based on the target terminal type selected by the user (such as airport, high-speed rail station, scenic spot, etc.). This sequence, in the forward travel order, is as follows: user's departure point, terminal drop-off point, security checkpoint or park entry verification point, and target terminal terminal node. The key node sequence may vary slightly for different terminal types; for example, airports may include check-in counter nodes, but the system automatically completes the matching through a preset terminal configuration library, without requiring manual setting by the user.

[0042] For each candidate path, the system executes a reverse time extrapolation algorithm. First, based on the end-to-end time prediction model, it calculates the segmented time between adjacent key nodes, such as road travel time from the departure point to the drop-off point, walking and queuing time from the drop-off point to the security checkpoint, and internal commuting time from the security checkpoint to the cutoff point. Then, starting with the expected arrival time of the target station's cutoff point, it calculates the latest arrival time for each key node in reverse order of the key node sequence. Based on this, it calculates the latest arrival time for each key node according to the preset formula: "Theoretical departure time = Expected arrival time". Total time of the entire link The system calculates the theoretical departure time using a "preset buffer time," which can be set by the system default or adjusted by the user based on their own time sensitivity. Finally, by combining real-time traffic conditions, station passenger flow, and the user's historical travel data (such as the user's typical early arrival time at the station and walking speed), the theoretical departure time is optimized and calibrated to generate the optimal departure time for the candidate route. For each candidate route, the system independently performs the above reverse time calculation and calibration process, thereby generating a corresponding optimal departure time for each route. Users can choose a suitable travel plan according to their preferences.

[0043] S5: Prioritize multiple candidate routes according to preset dimensions and output multiple travel plans for users, including the optimal departure time; wherein, the preset dimensions include the highest probability of on-time arrival, the shortest time, and the lowest cost.

[0044] In a preferred embodiment, step S5 prioritizes the multiple candidate routes with the optimal departure time calculated in step S4 according to preset dimensions, and outputs multiple travel plans containing the optimal departure time to the user. The preset dimensions include the highest probability of on-time arrival, the shortest time, and the lowest cost. The user can choose any dimension as the sorting criterion according to their own preferences, and the system can also simultaneously display the results sorted by different dimensions for the user's reference.

[0045] Specifically, for the "shortest time" dimension, the system directly compares the total time of each candidate path across the entire link, with the shortest path ranked first. For the "lowest cost" dimension, the system estimates the travel cost based on the path mileage and mode of transportation (such as whether it includes toll roads, parking fees, etc.), with the lowest cost path ranked first. For the "highest probability of on-time arrival" dimension, the system calculates the probability that each candidate path can ensure that the user arrives at the target station's cutoff node before the expected arrival time, based on the uncertainty measure of the time spent in each segment output by the end-to-end time prediction model (such as through historical prediction deviation distribution, real-time passenger flow fluctuation range, etc.), combined with the remaining margin of the preset buffer time, with the path with the highest probability ranked first.

[0046] In the output phase, the system presents multiple travel options to the user through a mobile application interface or web page. Each option includes at least: the corresponding candidate route (such as route name or map illustration), the calculated optimal departure time, the estimated total travel time and its breakdown (basic transit time, congestion increment time, and off-road waiting time), the probability of on-time arrival, and the estimated cost. Users can choose one option to execute based on their needs, or they can adopt the system's recommended "optimal" option (e.g., the one with the highest probability of on-time arrival and acceptable time) with a single click. Through this method, the present invention not only provides a single optimal departure time but also offers users diverse and comparable travel decision options, enhancing the system's practicality and user experience.

[0047] The method further includes: During a user's travel, the deviation between actual traffic and station passenger flow data and predicted data is compared in real time. When the deviation exceeds a preset threshold, a dynamic update mechanism is triggered, recalculating the entire-link time prediction model and the reverse time extrapolation algorithm, and pushing the updated travel decision.

[0048] Furthermore, the dynamic update mechanism continuously monitors changes in road congestion, sudden increases in passenger flow at stations, temporary traffic control, and sudden emergencies in temporary station management throughout the entire travel process through event triggering mechanism and real-time data push technology; When the deviation exceeds the preset threshold, the system will automatically recalculate and push the updated departure time and route adjustment plan to the user through at least one of the following channels: in-app push, SMS, or instant messaging tools.

[0049] In a preferred embodiment, during a user's journey, the system continuously acquires real-time traffic status data (such as real-time vehicle speed, accidents, and traffic control on road segments) and station passenger flow data (such as security check queue length and passenger density) through a real-time data interface, and compares these data item by item with the predicted data on which the end-to-end time prediction model is based. When the deviation between the actual data and the predicted data exceeds a preset threshold (for example, the actual travel time on a road segment is more than 20% longer than the predicted value, or the security check queue length is more than 50% longer than the predicted value), the system automatically triggers a dynamic update mechanism. This mechanism reuses the already constructed end-to-end time prediction model and reverse time extrapolation algorithm, using the current moment as a new starting point, and recalculates the total end-to-end time and the corresponding optimal departure time based on the latest real-time data, generating an updated travel decision and pushing it to the user, thereby achieving a dynamic closed loop of "decision-action-feedback-re-decision".

[0050] Specifically, the event triggering mechanism monitors various unforeseen circumstances throughout the trip, including but not limited to changes in road congestion (such as a sudden traffic accident exacerbating congestion), sudden surges in passenger flow at stations (such as a surge in security check queues due to concentrated flight arrivals), temporary traffic control (such as road closures and restrictions), and temporary station management (such as security upgrades and changes in ticket gates). Once the system detects that the deviation between actual and predicted data exceeds a preset threshold due to any of these events, it automatically recalculates without user intervention. The updated decisions after recalculation (including adjusted optimal departure times, recommended route adjustments, and risk warnings) are pushed to users in real time through multiple channels, specifically in-app push notifications (such as app pop-up notifications), SMS, and instant messaging tools (such as WeChat and DingTalk). This design ensures that even if users encounter unforeseen circumstances during their trip, they can receive timely adjustments to their travel plans, guaranteeing overall time controllability and the probability of on-time arrival.

[0051] It should be noted that the preset threshold can be configured according to different scenarios and user sensitivity. For example, a stricter threshold (e.g., triggering with a 5% deviation) can be set for highly time-sensitive scenarios such as business trips to catch flights, while a more lenient threshold (e.g., triggering with a 20% deviation) can be set for ordinary travel scenarios. Users can also adjust the threshold in the system settings to suit their personal preferences. Through the above dynamic update mechanism, this invention overcomes the shortcomings of existing technologies that cannot respond to emergencies and push updated decisions in real time, significantly improving the safety and reliability of the travel process.

[0052] The method further includes: User profiles are constructed based on users' historical travel data. These user profiles include users' transportation mode preferences, travel speed, fast passage permissions at stations, baggage carrying status, check-in or park entry habits, and time sensitivity. Based on the user profile, the time consumption calculation of each link in the entire process is personalized and calibrated, and corresponding travel reminder strategies are matched to optimize the calculation results of the reverse time extrapolation algorithm.

[0053] In a preferred embodiment, the system continuously collects users' historical travel data to construct a user profile corresponding to each user. The user profile includes at least the following dimensions of information: preferred mode of transportation (e.g., driving, public transportation, walking), travel speed (e.g., walking speed, driving speed), whether the user has fast-track access privileges at stations (e.g., airport VIP lanes, high-speed rail business class lanes, skip-the-line privileges at scenic spots), baggage carrying status (e.g., checked baggage, large carry-on baggage affecting walking speed and security check time), check-in or park entry habits (e.g., whether online check-in is done in advance, whether e-tickets are used), and time sensitivity (e.g., how far in advance the user usually arrives at the station, and their tolerance for delays). This data can be obtained through user registration, historical travel behavior analysis, and authorized integration with third-party service platforms (e.g., airlines, China Railway 12306, scenic spot ticketing systems).

[0054] Based on the user profile, the system performs personalized calibration of the time consumption calculation for each link in the entire chain. For example, for the route time consumption calculation sub-model, if the user profile shows that their driving speed is usually faster than the average vehicle speed, the system can multiply the basic passage time by a correction coefficient less than 1; for the station passenger flow time prediction sub-model, if the user has fast passage permission, the system can correspondingly shorten the security check queuing time; if the user is carrying large luggage, the system can increase the walking time from the drop-off point to the security checkpoint. At the same time, the system matches corresponding travel reminder strategies according to the user profile. For example, for users with high time sensitivity, the system can set earlier departure reminder times and more frequent dynamic update pushes; for users who are used to arriving early, the system can appropriately reduce the preset buffer time to avoid waiting due to departing too early.

[0055] In addition, user profiles can also be used to optimize the calculation results of reverse time extrapolation algorithms.

[0056] Specifically, after calculating the theoretical departure time, the system performs a secondary calibration of the optimal departure time by combining user profile characteristics such as time sensitivity and habitual advance travel. For example, if historical data shows that users tend to depart 10 minutes earlier than the system's suggested time, the system will automatically incorporate this offset into the calculation, making the output optimal departure time more consistent with the user's actual behavior patterns. Through this user profile-driven personalized calibration mechanism, this invention can provide personalized and accurate travel decision-making services for different users, further improving user experience and decision adoption rates.

[0057] It should be noted that the collection and use of user profile data comply with relevant privacy protection regulations and are carried out only after obtaining explicit authorization from the user.

[0058] like Figure 2 , Figure 3As shown, this embodiment of the invention provides a full-link travel time decision-making system for complex traffic scenarios. The system can be implemented through software, hardware, or a combination of both. From a hardware perspective, as... Figure 2 The diagram shown is a hardware architecture diagram of a computing device for a full-link travel time decision system for complex traffic scenarios provided in an embodiment of the present invention. Besides... Figure 2 In addition to the processor, memory, network interface, and non-volatile memory shown, the computing device housing the system in this embodiment may also include other hardware, such as a forwarding chip responsible for processing packets. Taking software implementation as an example, such as... Figure 3 As shown, a system in a logical sense is formed by the CPU of the computing device in which it resides reading the corresponding computer program from the non-volatile memory into the main memory for execution.

[0059] This invention provides a full-link travel time decision-making system for complex traffic scenarios, comprising: The fusion module 100 is used to fuse road network data, real-time traffic status data, historical congestion data, passenger flow queuing data at target stations, and time characteristic data to form a database of factors affecting travel time. The road network data includes route geographic information and basic time consumption data for different modes of transportation. The real-time traffic status data includes real-time vehicle speed, traffic flow, traffic accidents, road construction, and temporary traffic control data for road segments. The passenger flow queuing data at target stations includes real-time passenger flow density, queue length, and dynamic waiting time data for airport security queues, high-speed rail station ticket queues, and scenic area entrance verification queues. The time characteristic data includes time period attribute data corresponding to weekdays, weekends, morning and evening peak hours, and holidays. The construction module 200 is used to construct a full-link time prediction model based on the travel time influencing factor database, wherein the full-link time prediction model includes a path time calculation sub-model, a traffic congestion prediction sub-model, and a station passenger flow time prediction sub-model. The calculation module 300 is used to generate multiple candidate routes for travel demand from the same departure point to the destination terminal node, and use the full-link time prediction model to calculate the total full-link time covering road traffic, congestion superposition, and waiting time within the terminal. The deduction module 400 is used to obtain the expected arrival time of the target station's cutoff node input by the user, and use the expected arrival time as the unique reference time for reverse calculation. Based on the target station type, it automatically matches the sequence of key nodes in the entire link, and calculates the optimal departure time for each candidate path by combining the total time of the entire link corresponding to each candidate path. The sequence of key nodes, in the forward order of travel, is as follows: user departure point, station drop-off point, security checkpoint or park entry verification point, and target station cutoff node. The output module 500 is used to prioritize multiple candidate routes according to preset dimensions, and output multiple travel plans for the user, each including the optimal departure time. The preset dimensions include highest probability of on-time arrival, shortest travel time, and lowest cost.

[0060] In a preferred system architecture implementation, the present invention uses a modular and decoupled microservice architecture to build the above system.

[0061] Specifically, the fusion module 100, construction module 200, calculation module 300, inference module 400, and output module 500 are deployed as independent microservices, with each module interacting with the others through standardized interfaces (such as RESTful APIs or message queues). This architecture allows each module to be developed, deployed, and extended independently. When new travel scenarios (such as scenic spots or sports venues) are added, only the corresponding passenger flow data adaptation interface and lightweight prediction model extension need to be developed, without requiring large-scale modifications to the existing system. The development cost of adding a new travel scenario can be controlled within a preset percentage of the initial investment (such as 20%). In addition, the system can also set up an external interface layer to encapsulate map API interfaces, traffic data interfaces, station data interfaces, and push service interfaces, enabling standardized data interaction with external data sources and user terminals.

[0062] It is understood that the structures illustrated in the embodiments of the present invention do not constitute a specific limitation on the end-to-end travel time decision-making system for complex traffic scenarios. In other embodiments of the present invention, the end-to-end travel time decision-making system for complex traffic scenarios may include more or fewer components than illustrated, or combine some components, split some components, or have different component arrangements. The illustrated components may be implemented in hardware, software, or a combination of software and hardware.

[0063] The information interaction and execution process between the modules in the above system are based on the same concept as the method embodiment of the present invention, and the specific details can be found in the description in the method embodiment of the present invention, and will not be repeated here.

[0064] This invention also provides a computing device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the method in any embodiment of this invention.

[0065] This invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the method in any embodiment of this invention.

[0066] Embodiments of this application also provide a computer program product, which includes a computer program. A processor of a computer device reads the computer program from a computer-readable storage medium and executes the computer program, causing the computer device to perform any of the test methods described in the above embodiments.

[0067] Specifically, a system or apparatus equipped with a storage medium may be provided, on which software program code implementing the functions of any of the embodiments described above is stored, and the computer (or CPU or MPU) of the system or apparatus may read and execute the program code stored in the storage medium.

[0068] In this case, the program code read from the storage medium can itself implement the function of any of the above embodiments, and therefore the program code and the storage medium storing the program code constitute part of the present invention.

[0069] Storage media embodiments for providing program code include floppy disks, hard disks, magneto-optical disks, optical disks (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD+RW), magnetic tapes, non-volatile memory cards, and ROMs. Alternatively, program code can be downloaded from a server computer via a communication network.

[0070] Furthermore, it should be clear that not only can the program code read by the computer be executed, but also the operating system or other components operating on the computer can be instructed based on the program code to perform some or all of the actual operations, thereby realizing the function of any of the embodiments described above.

[0071] Furthermore, it is understood that the program code read from the storage medium is written to the memory set in the expansion board inserted into the computer or to the memory set in the expansion module connected to the computer. Then, based on the instructions of the program code, the CPU or other components installed on the expansion board or expansion module execute some and all of the actual operations, thereby realizing the function of any of the above embodiments.

[0072] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0073] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media that can store program code, such as ROM, RAM, magnetic disk, or optical disk.

[0074] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for end-to-end travel time decision-making in complex traffic scenarios, characterized in that, include: S1: Integrate road network data, real-time traffic status data, historical congestion data, passenger flow queuing data at target stations, and time characteristic data to form a database of factors affecting travel time; wherein, the road network data includes route geographic information and basic time consumption data for different modes of transportation; the real-time traffic status data includes real-time vehicle speed, traffic flow, traffic accidents, road construction, and temporary traffic control data for road segments; the passenger flow queuing data at target stations includes real-time passenger flow density, queue length, and dynamic waiting time data corresponding to airport security check queues, high-speed rail station ticket check queues, and scenic spot entrance verification queues; and the time characteristic data includes time period attribute data corresponding to weekdays, weekends, morning and evening peak hours, and holidays. S2: Based on the database of factors affecting travel time, construct a full-link time prediction model; wherein, the full-link time prediction model includes a path time calculation sub-model, a traffic congestion prediction sub-model, and a station passenger flow time prediction sub-model; S3: For the travel demand from the same origin to the destination terminal node, generate multiple candidate routes, and use the full-link time prediction model to calculate the total full-link time covering road traffic, congestion superposition, and waiting time within the terminal. S4: Obtain the expected arrival time of the target station's cutoff node input by the user, use the expected arrival time as the unique reference time for reverse calculation, automatically match the full-link key node sequence according to the target station type, and calculate the optimal departure time for each candidate path by combining the total full-link time corresponding to each candidate path; the key node sequence is in the forward order of travel as follows: user departure point, station drop-off point, security checkpoint or park entry verification point, and target station cutoff node; S5: Prioritize multiple candidate routes according to preset dimensions and output multiple travel plans for users, including the optimal departure time; wherein, the preset dimensions include the highest probability of on-time arrival, the shortest time, and the lowest cost.

2. The method according to claim 1, characterized in that, Step S3 includes: The path time calculation sub-model, based on path geographic information and real-time traffic conditions, corrects the basic travel time for different time periods and different modes of transportation through a dynamic time evaluation algorithm, and calculates the basic travel time. The traffic congestion prediction sub-model, based on historical congestion data and real-time traffic conditions, uses time series analysis algorithms and spatiotemporal graph neural network algorithms to predict road congestion during the user's travel period and calculates the congestion increment time. The station passenger flow time prediction sub-model, based on the passenger flow queuing data and time characteristic data of the target station, predicts the passenger flow waiting status during the user's arrival time at the station through regression model and Bayesian prediction model, and calculates the waiting time of non-road links. The total time for the entire link is the sum of the basic travel time, the congestion increment time, and the waiting time for non-road links.

3. The method according to claim 1, characterized in that, In step S4, the optimal departure time for each candidate path is calculated based on the total end-to-end time for each candidate path, including: Based on the end-to-end time prediction model, the segmented time between adjacent key nodes is calculated respectively. Combined with the expected arrival time, the latest arrival time corresponding to each key node is calculated in reverse order of the key node sequence. The theoretical departure time is calculated using a preset formula, which is: Theoretical departure time = Expected arrival time at the target station's cutoff node - Total time for the entire link - Preset buffer time; By combining real-time traffic conditions, station passenger flow status, and users' historical travel data, the theoretical departure time is calibrated to generate the optimal departure time that finally matches the corresponding candidate path.

4. The method according to claim 1, characterized in that, The method further includes: During a user's travel, the deviation between actual traffic and station passenger flow data and predicted data is compared in real time. When the deviation exceeds a preset threshold, a dynamic update mechanism is triggered, recalculating the entire-link time prediction model and the reverse time extrapolation algorithm, and pushing the updated travel decision.

5. The method according to claim 4, characterized in that, The dynamic update mechanism continuously monitors changes in road congestion, sudden increases in passenger flow at stations, temporary traffic control, and sudden emergencies in temporary station management throughout the entire travel process through event triggering mechanism and real-time data push technology; When the deviation exceeds the preset threshold, the system will automatically recalculate and push the updated departure time and route adjustment plan to the user through at least one of the following channels: in-app push, SMS, or instant messaging tools.

6. The method according to claim 1, characterized in that, The method further includes: User profiles are constructed based on users' historical travel data. These user profiles include users' transportation mode preferences, travel speed, fast passage permissions at stations, baggage carrying status, check-in or park entry habits, and time sensitivity. Based on the user profile, the time consumption calculation of each link in the entire process is calibrated in a personalized manner, and corresponding travel reminder strategies are matched to optimize the calculation results of the reverse time extrapolation algorithm.

7. A full-link travel time decision-making system for complex traffic scenarios, characterized in that, include: The fusion module is used to integrate road network data, real-time traffic status data, historical congestion data, passenger flow queuing data at target stations, and time characteristic data to form a database of factors affecting travel time. The road network data includes route geographic information and basic time consumption data for different modes of transportation. The real-time traffic status data includes real-time vehicle speed, traffic volume, traffic accidents, road construction, and temporary traffic control data for road segments. The passenger flow queuing data at target stations includes real-time passenger flow density, queue length, and dynamic waiting time data for airport security queues, high-speed rail station ticket queues, and scenic area entrance verification queues. The time characteristic data includes time period attribute data corresponding to weekdays, weekends, morning and evening peak hours, and holidays. The construction module is used to build a full-link time prediction model based on the travel time influencing factor database. The full-link time prediction model includes a path time calculation sub-model, a traffic congestion prediction sub-model, and a station passenger flow time prediction sub-model. The calculation module is used to generate multiple candidate routes for travel demand from the same departure point to the destination terminal node, and use the full-link time prediction model to calculate the total full-link time covering road traffic, congestion superposition, and waiting time within the terminal. The deduction module is used to obtain the expected arrival time of the target station's cutoff node input by the user, and use the expected arrival time as the unique reference time for reverse calculation. Based on the target station type, it automatically matches the sequence of key nodes in the entire link, and calculates the optimal departure time for each candidate path by combining the total time of the entire link corresponding to each candidate path. The sequence of key nodes, in the forward order of travel, is as follows: user departure point, station drop-off point, security checkpoint or park entry verification point, and target station cutoff node. The output module is used to prioritize multiple candidate routes according to preset dimensions and output multiple travel plans for users, including the optimal departure time. The preset dimensions include the highest probability of on-time arrival, the shortest time, and the lowest cost.

8. A computer device, characterized in that, The computer device includes a memory and a processor. The memory is used to store computer programs, and the processor is used to execute the computer programs stored in the memory to implement the steps of the method according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the steps of the method described in any one of claims 1-6.

10. A computer program product, characterized in that, Includes a computer program, which, when executed by a processor, implements the steps of the method according to any one of claims 1-6.

Images