A bus individual trip decision model

Abstract

The application discloses a bus individual trip decision model, obtains the trip rule of the bus individual based on the analysis of the bus individual trip data, and inputs the initial basic data of the model; the getting-off and getting-on station information of the bus individual is calculated and recognized through a time matching method, a trip chain theory and bus station characteristics, and the identification and division of the individual trip chain are completed in combination with the connotation and function level of the multi-mode bus, a virtual start and end point is defined for the departure or arrival station cluster of the individual, the analysis and processing of the individual trip data are completed, then the individual trip decision model about the departure time selection, the getting-on station selection and the bus line selection is established based on the individual historical trip information obtained through the above method, and the individual can reach the trip end point in the case that the total cost is as small as possible.

Description

Technical Field

[0001] This invention relates to the field of public transportation technology, specifically to a public transportation individual travel decision-making model. Background Technology

[0002] To alleviate the adverse effects of the rapidly increasing number of private vehicles on urban roads, meet the transportation needs of urban residents as much as possible, and ensure the fairness of their travel, multimodal public transport systems have quickly become a key component of the comprehensive transportation system of large cities, playing an irreplaceable role in undertaking high-capacity urban transportation services and alleviating urban traffic pressure.

[0003] Research on the origin-destination (OD) estimation method for public transport is fundamental to public transport passenger flow forecasting. Calculating the OD volume of public transport passengers can further determine the occurrence and attraction of each bus stop, playing a crucial role in judging the rationality of the current urban public transport network and subsequently optimizing and adjusting the network. Early urban public transport OD estimation primarily relied on manual surveys using questionnaires. With the widespread application of IC card technology in public transport, scholars began studying the passenger travel information hidden within IC cards to estimate passenger boarding and alighting points. However, this method mainly relied on the attractiveness of bus stops, combined with factors such as the surrounding land use, to estimate the number of passengers alighting at each stop. This method did not consider the differences in individual passenger travel characteristics and could not obtain the specific alighting points of passengers. Therefore, a public transport individual travel decision-making model is urgently needed to solve these problems. Summary of the Invention

[0004] This invention provides a method to establish a public transport individual travel decision-making model by calculating the individual's bus boarding and alighting points and dividing the travel chain, thereby realizing multi-modal public transport passenger flow prediction driven by individual travel data, thus solving the problems existing in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a public transport individual travel decision-making model, comprising:

[0006] The analysis of individual bus travel data reveals the travel patterns of individual bus users, which serve as the initial data input for the model.

[0007] Based on the definition of the connotation and functional hierarchy of multimodal public transport, the initial basic data is optimized: the travel chain of individual travelers is divided into multimodal public transport travel and single-modal public transport travel according to the transfer time threshold, and density clustering is performed according to the spatial location attributes of the individual's departure or arrival station in the road network to establish corresponding virtual OD pairs and identify the information of individual travel chains.

[0008] Based on the information of individual travel chains, a set of state spaces and action spaces for different travels is constructed, and the action set is continuously updated according to the incomplete information environment of the public transportation network. Combining the historical individual travel time cost, travel habit definition and labeling of the direct and future rewards of state-action pairs, as well as the state transition probability, an individual travel decision model is established on an individual basis, including departure time selection, boarding station selection and travel route selection.

[0009] Preferably, the individual bus travel data includes IC card swipe data, bus onboard GPS data, and bus route and station location information. Specifically, by combining the bus route and station location information, using time as the main connecting field and supplemented by the bus's GPS data, a vehicle arrival timetable is obtained. Sparse GPS data is then supplemented by interpolation, and matched with station latitude and longitude to obtain the vehicle arrival time. The time matching method is used to obtain the passenger's IC card swipe time, thus completing the process of identifying the boarding station of IC card holders.

[0010] Preferably, GPS data interpolation processing is used: in the selected acquired GPS data, the latitude and longitude of any two adjacent GPS records are (x... b,w,j ,y b,w,j ), (x b,w,j+1 ,y b,w,j+1 The recording times are RT. b,w,j RT b,w,j+1 With an interval of Λ, Λ data points are inserted between two GPS records at 1-second interpolation intervals. Calculate the latitude and longitude of the λth inserted GPS data point:

[0011]

[0012]

[0013] The corresponding recording time is RT. b,w,j +λ.

[0014] Preferably, based on the travel chain theory, individual public transport trips are divided into closed public transport travel chains and non-closed public transport travel chains, and the alighting stations for different types of public transport travel chains are estimated by combining the boarding station identification results and the characteristics of public transport stations.

[0015] Preferably, the calculation of the alighting station specifically includes:

[0016] Step 1: Based on the boarding station BS p,k Determine if the current card swipe record is the last one of the day. If k < K, proceed to step 2; otherwise, proceed to step 4.

[0017] Step 2: Determine the boarding station (BS) for the current train. p,k and the boarding station BS for the next trainp,k+1 Whether they belong to the same line or not, if so, the boarding station of the next train will be used as the alighting station for the current train, i.e., AS. p,k =BS p,k Otherwise, proceed to step 3;

[0018] Step 3: Determine BS p,k+1 CAS is the set of stations along the current train route from upstream to downstream within a circular area centered at D and with a maximum walking distance D as its radius. p,k ={cas m 0 ≤ m ≥ M; if the set is empty, then the travel chain belongs to a non-public transport travel chain; if the number of stops in the set is M = 1, then the unique stop in the set is the alighting stop AS of the current bus. p,k If M > 1, then calculate the values ​​starting from the upstream station cas1 and passing through cas1 respectively. p,k Any site in China CAS m , with BS p,k+1 Travel time t to the destination m Choose the station with the shortest travel time (CAS) * =argmin(t m () represents the current train's disembarkation station AS. p,k The calculation formula is as follows:

[0019]

[0020] Among them, v w d represents walking speed; m For the site cas m With BS p,k+1 The distance between them; L m-1,m For the site cas m-1 With CAS m The length of the bus routes between them; v b For bus operating speed;

[0021] Step 4: Determine the boarding station (BS) of the last train ride. p,k and the boarding station BS for the first train ride p,1 Whether they belong to the same line or not, if so, the boarding station of the first train will be used as the alighting station of the last train, i.e., AS. p,k =BS p,1 Otherwise, proceed to step 5;

[0022] Step 5: Determine the boarding station BS for your first train ride. p,1Within a circular area centered at D and with the maximum walking distance D as the radius, the set of stations along the route of the last bus ride from upstream to downstream is defined. If this set is empty, then the passenger's travel chain is a non-closed public transport travel chain; otherwise, proceed to step 3 to determine the alighting station of the last bus ride.

[0023] Preferably, individual travel also includes non-public transport travel chains, with the following drop-off points calculated:

[0024] Step 1: Determine the boarding station BS for your current train. p,k Downstream site collection DAS p,k ={das n |1≤n≤N};

[0025] Step 2: Calculate passenger das at any station n The probability PA of getting off at point p,n The calculation formula is:

[0026] PA p,n =f n,1 ·f n,2 / ∑(f η,1 ·f η,2 );

[0027] Among them, f n,1 f n,2 These represent the development intensity and public transport accessibility characteristics of the area surrounding the station, respectively; f n,1 =BF n / ∑BF η f n,2 =R n / ∑R η BF n For the site das n The average number of passengers arriving within a certain period indirectly reflects the development intensity around the station; R n For the stations along the route das n The number of bus routes.

[0028] Step 3: Based on the passenger's location at any station (das) n The probability PA of getting off at point p,n The roulette wheel method is used to calculate the current train's alighting station AS. p,k .

[0029] Preferably, individual travel chains are divided based on transfer time thresholds, specifically as follows:

[0030] Step 1: Determine the boarding station (BS) for the passenger's two boarding stops based on the passenger's boarding station identification results. p,k and BS p,k+1Whether they belong to the same bus route. If they belong to the same bus route, the passenger's trip on that day is a single-mode bus trip. If they do not belong to the same bus route, determine whether the two bus routes belong to the same functional level. If they belong to the same functional level, the passenger's trip on that day is still a single-mode bus trip. If they do not belong to the same functional level, it is a multi-mode bus trip.

[0031] Step 2: Based on the alighting station AS p,k Determine the site AS p,k Corresponding bus service schedule w b Obtain the arrival timetable for this train and transfer AS p,k The station name is matched with the station in the arrival timetable to obtain the vehicle's arrival time at station AS. p,k Arrival time AT b,w,j The passenger's alighting time Δt is calculated as 0.25 times the current bus trip time. p Then the passenger's disembarkation time AT b,w,j,l The calculation formula is as follows:

[0032] AT b,w,j,l =AT b,w,j +Δt p ;

[0033] Step 3: Calculate AT at a given disembarkation time b,w,j,l Next boarding station BS p,k+1 Corresponding boarding time TT p,k+1 The time interval μ p,l μ p,l =TT p,k+1 -AT b,w,p,l ;

[0034] Step 4: When a passenger transfers to a station within the same route, meaning the passenger's two boarding stations are on the same line, determine the maximum waiting time (WT) at the station based on the departure intervals of different functional levels of the multimodal bus system. p,k The maximum waiting time is used as the maximum transfer time threshold η for same-station transfers. p,k , that is, η p,k =WT p,k When a passenger transfers between different stations, meaning the passenger's boarding stations are not on the same line, the maximum acceptable walking distance D and the passenger's walking speed v are considered. w Calculate the maximum transfer time threshold;

[0035] Step 5: Set the travel time interval μ p,l With the maximum transfer time threshold η p,k Compare, if μ p,k ≤η p,kIf an individual bus user completes a single multi-modal or single-modal bus trip on that day, and μ p,l This refers to transfer time; conversely, if the transfer time is shorter, then an individual bus user completes multiple multi-modal or single-modal bus trips on the same day. p,l The event period is [time].

[0036] Preferably, the establishment of virtual OD pairs specifically involves:

[0037] Step 1: Identify and number all complete travel chains of individual buses, extract the starting and ending boarding stations of each travel chain, and form a set D. p,BS Each object in the set contains the number of each travel chain and the latitude and longitude coordinates of the corresponding boarding station;

[0038] Step 2: Set the neighborhood parameter ε according to the radiation range of bus stops at different functional levels, and adjust MinPts accordingly based on the specific research object; apply the neighborhood parameter (ε, MinPts) to the set D. p,BS Perform a core object search to obtain the core object set Ω p,BS ;

[0039] Step 3, from set Ω p,BS A core object is randomly selected as the seed to execute a clustering algorithm, searching for all sites reachable by its density, thus forming the first cluster. Then The core objects included are from Ω p,BS The process involves filtering out the weak clusters, then randomly selecting a seed from the updated set to generate the next cluster, repeating this process until the set is empty.

[0040] Step 4: According to each cluster For each object, the corresponding travel chain number is used to filter out the terminal drop-off station of the travel chain, forming various sets. Each object in the set also contains the number of each travel chain and the latitude and longitude coordinates of the corresponding drop-off point;

[0041] Step 5: Perform a set analysis based on the neighborhood parameters (ε, MinPts). Perform a core object search, remove outliers, and obtain a core object set.

[0042] Step 6: Calculate the set and The center point; the center points of each corresponding set together form multiple virtual OD point pairs for passenger p; the calculation formula is:

[0043]

[0044]

[0045] in, Represents the X-coordinate of the center point of each set; The x-coordinate represents the center point of each set; b,w,n The x-coordinate represents the coordinate of the object contained in each set; the y-coordinate represents the coordinate of the object contained in each set. b,w,n The y-coordinate represents the Y-coordinate of the objects contained in each set; n represents a specific object in each set; N represents all objects contained in each set.

[0046] Preferably, the state space S:

[0047] S={Destination / Origin, Departure-time, Boarding-stion, En-route, Alighting-station};

[0048] Among them, Destination / Origin is the latitude and longitude coordinates of a pair of virtual OD points in an individual's history; Departure-time is the specific departure time information selected by the individual; Boarding-station is the latitude and longitude of the boarding station selected by the individual, and the information of the collateral lines included with the station; En-route is the vehicle, train number, and other information of the line corresponding to the individual's selected boarding station; Alighting-station is the line, vehicle, train number, latitude and longitude information of the alighting station selected by the individual.

[0049] Action Space A:

[0050] A = {Select-departure-time, Select-boarding-station, Select-bus-route, Select-alighting-station, To-the-destination}. Each action space is a set of candidate data. Among them, the Select-departure-time candidate set contains all departure time information of an individual's historical trips; the Select-boarding-station candidate set contains all boarding station information of an individual's historical trips; the Select-bus-route candidate set contains the information of the bus routes corresponding to all boarding stations of an individual's historical trips; the Select-alighting-station candidate set contains all historical alighting station information corresponding to an individual's "bus route selection"; and the To-the-destination candidate set contains the latitude and longitude coordinates of all established virtual OD points.

[0051] Preferably, in the individual travel decision model, the action selection for each state is determined based on the future reward to determine the state transition probability, thereby selecting the action with the highest state transition probability and transitioning to the next state. When transitioning to the next state, the direct reward of this action selection becomes known and the historical experience value set is updated.

[0052] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention identifies and calculates the boarding and alighting station information of individual bus travelers through time matching methods, travel chain theory, and bus stop feature identification. It also combines the connotation and functional hierarchy of multimodal public transport to complete the identification and division of individual travel chains, defining virtual origin and destination clusters for individual departures or arrivals, and completing the analysis and processing of individual travel data. Then, based on the individual's historical travel information obtained through the above methods, an individual travel decision model is established on an individual basis and based on Markov decision process theory, focusing on departure time selection, boarding station selection, and bus route selection. This allows individuals to reach their destination with the lowest possible total cost. Attached Figure Description

[0053] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0054] In the attached diagram:

[0055] Figure 1 This is a flowchart of the public transport individual travel decision-making model of the present invention;

[0056] Figure 2 This is a flowchart of the passenger boarding station identification process of the present invention;

[0057] Figure 3 This is a flowchart of the passenger alighting station calculation process of the present invention;

[0058] Figure 4 This is a distribution diagram of the number of passengers boarding at each stop of Route 108 in a specific embodiment of the present invention;

[0059] Figure 5 This is a state transition diagram of individual travelers in the multimodal public transport network of this invention. Detailed Implementation

[0060] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0061] Example: Figure 1 As shown, a public transport individual travel decision-making model includes:

[0062] The identification and estimation of boarding and alighting points for individual public transport trips, the identification of multimodal individual public transport travel chains and the establishment of virtual origin-destination (OD) pairs, and the establishment of a multimodal individual public transport travel decision-making model, including:

[0063] The analysis of individual bus travel data reveals the travel patterns of individual bus users, which serve as the initial data input for the model.

[0064] Based on the definition of the connotation and functional hierarchy of multimodal public transport, the initial basic data is optimized: the travel chain of individual travelers is divided into multimodal public transport travel and single-modal public transport travel according to the transfer time threshold, and density clustering is performed according to the spatial location attributes of the individual's departure or arrival station in the road network to establish corresponding virtual OD pairs and identify the information of individual travel chains.

[0065] Based on the information of individual travel chains, a set of state spaces and action spaces for different travels is constructed, and the action set is continuously updated according to the incomplete information environment of the public transportation network. Combining the historical individual travel time cost, travel habit definition and labeling of the direct and future rewards of state-action pairs, as well as the state transition probability, an individual travel decision model is established on an individual basis, including departure time selection, boarding station selection and travel route selection.

[0066] Individual bus travel data is the foundational data for bus passenger flow prediction. By analyzing and mining travel data, we can obtain information on various elements such as individual boarding and alighting points and times, thereby understanding the travel patterns of individual bus users and providing initial basic data input for building passenger flow prediction simulation models. Among them, individual bus travel data includes IC card swipe data, bus onboard GPS data, and bus route and stop location information.

[0067] refer to Figure 2 As shown, the boarding station identification is specifically as follows:

[0068] Step 1: Select research data.

[0069] Select the bus route r to be studied and obtain any of the stops s along the way. i latitude and longitude (x i ,y i Select a specific operating vehicle b and a specific operating train w on this route. b The GPS data and the IC card data of any passenger p riding in the vehicle.

[0070] Step 2: Clear redundant data: Calculate the time interval between each card swipe record and the two preceding and following card swipe records. p,k If 0 < IT p,k <L S,S+1 / v b (L S,S+1The distance between the current boarding station and its downstream adjacent station is calculated using the semi-versus formula; v b If the card swiping time interval is short, it can be assumed that passenger p has an accompanying person and they share an IC card. These two card swiping times are considered as one trip, and the earlier recorded time is taken as the card swiping time.

[0071] Step 3: GPS Data Interpolation Processing: If the bus GPS data is a sparse sample with low-frequency returns and is not related to vehicle arrival / departure, then the GPS data will be supplemented using interpolation. In the selected GPS data, the latitude and longitude of any two adjacent GPS records will be (x...). b,w,j ,y b,w,j ), (x b,w,j+1 ,y b,w,j+1 The recording times are RT. b,w,j RT b,w,j+1 With an interval of Λ, Λ data points are inserted between two GPS records at 1-second interpolation intervals. Calculate the latitude and longitude of the λth inserted GPS data point:

[0072]

[0073]

[0074] The corresponding recording time is RT. b,w,j +λ;

[0075] Step 4: Determine the vehicle's arrival location, arrival time, and departure time: For any station s on line r... i The train number w, which was supplemented in step 2, is calculated using the semi-versus formula. b The distances between all GPS data and the station are used to select the latitude and longitude of the GPS data closest to the station. Record the corresponding time as the vehicle's arrival location. As of arrival time AT b,w,j Identify the time AT in GPS data b,w,j The moment when the speed changes from 0 to a non-zero state is the corresponding vehicle departure time DT. b,w,j ;

[0076] Step 5: Identify passenger boarding station: Match the station with the train number w b The associated IC card records are sorted according to the time of card swipe. It is known that the time of passenger p's kth card swipe is TT. p,k (1≤k≤K, where K is the number of times the passenger swipes their card that day), compare the swipe time of any card swipe record with the arrival time of the train at any station. When AT b,w,i ≤TT p,k <ATb,w,i+1 When, then AT b,w,i The corresponding site s i BS, the boarding station for this passenger's card swipe. p,k .

[0077] refer to Figure 4 As shown, based on the travel chain theory, individual public transport trips are divided into closed public transport travel chains, open public transport travel chains, and non-public transport travel chains. In open public transport travel chains, the method for estimating the alighting point for each trip other than the last is the same as for closed public transport travel chains. The alighting point for the last trip of the day needs to be estimated using historical travel data and by judging the characteristics of downstream stations, which is the same as the method for non-public transport travel chains.

[0078] Specifically, the calculation of drop-off points for non-closed public transport chains includes:

[0079] Step 1: Based on the boarding station BS p,k Determine if the current card swipe record is the last one of the day. If k < K, proceed to step 2; otherwise, proceed to step 4.

[0080] Step 2: Determine the boarding station (BS) for the current train. p,k and the boarding station BS for the next train p,k+1 Whether they belong to the same line or not, if so, the boarding station of the next train will be used as the alighting station for the current train, i.e., AS. p,k =BS p,k Otherwise, proceed to step 3;

[0081] Step 3: Determine BS p,k+1 CAS is the set of stations along the current train route from upstream to downstream within a circular area centered at D and with a maximum walking distance D as its radius. p,k ={cas m |0≤m≥M};If the set is empty, then the travel chain belongs to a non-public transport travel chain;If the number of stops in the set is M=1, then the unique stop in the set is the alighting stop AS of the current bus. p,k If M > 1, then calculate the values ​​starting from the upstream station cas1 and passing through cas1 respectively. p,k Any site in China CAS m , with BS p,k+1 Travel time t to the destination m Choose the station with the shortest travel time (CAS) * =argmin(t m () represents the current train's disembarkation station AS. p,k The calculation formula is as follows:

[0082]

[0083] Among them, v w d represents walking speed; m For the site cas m With BS p,k+1 The distance between them; L m-1,m For the site cas m-1 With CAS m The length of the bus routes between them; v b For bus operating speed;

[0084] Step 4: Determine the boarding station (BS) of the last train ride. p,k and the boarding station BS for the first train ride p,1 Whether they belong to the same line or not, if so, the boarding station of the first train will be used as the alighting station of the last train, i.e., AS. p,k =BS p,1 Otherwise, proceed to step 5;

[0085] Step 5: Determine the boarding station BS for your first train ride. p,1 Within a circular area centered at D and with the maximum walking distance D as the radius, the set of stations along the route of the last bus ride from upstream to downstream is defined. If this set is empty, then the passenger's travel chain is a non-closed public transport travel chain; otherwise, proceed to step 3 to determine the alighting station of the last bus ride.

[0086] Non-public transport stop calculations are as follows:

[0087] Step 1: Determine the boarding station BS for your current train. p,k Downstream site collection DAS p,k ={das n |1≤n≤N};

[0088] Step 2: Calculate passenger das at any station n The probability PA of getting off at point p,n The calculation formula is:

[0089] PA p,n =f n,1 ·f n,2 / ∑(f η,1 ·f η,2 );

[0090] Among them, f n,1 f n,2 These represent the development intensity and public transport accessibility characteristics of the area surrounding the station, respectively; f n,1 =BF n / ∑BF η f n,2 =R n / ∑Rη BF n For the site das n The average number of passengers arriving within a certain period indirectly reflects the development intensity around the station; R n For the stations along the route das n The number of bus routes.

[0091] Step 3: Based on the passenger's location at any station (das) n The probability PA of getting off at point p,n The roulette wheel method is used to calculate the current train's alighting station AS. p,k .

[0092] In one specific embodiment, bus route 108 in Rudong County is taken as the research object. The basic data sources are bus IC card data and vehicle GPS data obtained from the local public transportation system. Route 108 passes through 30 and 28 bus stops on its east and west sides, respectively, and the areas along the route are mainly residential and industrial land. Considering that the demand for public transportation in Rudong County is relatively small compared to large cities, and the number of trips a passenger makes in a day may not meet the needs of this analysis, data from the entire week of June 4th to June 10th, 2018, is selected for the study. A database is built using a MySQL program to complete preprocessing work such as key field filtering and cleaning of missing and redundant data, resulting in 1405 IC card swipe records to be analyzed within one week. At the same time, Python programming is used to complete the calculation of the subsequent boarding and alighting station estimation process.

[0093] Boarding stop identification: The GPS data in Rudong County's public transportation system has an interval of 1-3 minutes, making it sparse GPS data. The arrival times of each bus during the study period were determined using the method described above. The table below shows the arrival times of bus number 110286 at some stops. Based on the arrival times of all buses on Route 108 from June 4th to June 10th, 2018, the boarding stops for all passengers on that day were identified, and the distribution of passengers boarding at each stop was statistically obtained. Figure 4 As shown, matching the original GPS data and the supplemented GPS data with the IC card data reveals that 11.3% of the boarding stops on route 108 have different matching results, indicating that there is an error in the matching results of the original GPS data. The supplemented GPS data can correct the error and effectively improve the time matching accuracy.

[0094]

[0095]

[0096] Disembarkation station estimation: Based on the above method, Python code was used for calculation; there were 1405 IC card swipe records to be analyzed for Route 108, and 1183 disembarkation stations were successfully estimated, with a recognition success rate of 84.2%.

[0097] Among them, the transfer behavior of individuals using urban public transport can be divided based on two methods: transfer mode of transportation and transfer spatial distance. Individual travel chains are further divided according to transfer time thresholds, specifically as follows:

[0098] Step 1: Determine the boarding station (BS) for the passenger's two boarding stops based on the passenger's boarding station identification results. p,k and BS p,k+1 Whether they belong to the same bus route. If they belong to the same bus route, the passenger's trip on that day is a single-mode bus trip. If they do not belong to the same bus route, determine whether the two bus routes belong to the same functional level. If they belong to the same functional level, the passenger's trip on that day is still a single-mode bus trip. If they do not belong to the same functional level, it is a multi-mode bus trip.

[0099] Step 2: Based on the alighting station AS p,k Determine the site AS p,k Corresponding bus service schedule w b Obtain the arrival timetable for this train and transfer AS p,k The station name is matched with the station in the arrival timetable to obtain the vehicle's arrival time at station AS. p,k Arrival time AT b,w,j The passenger's alighting time Δt is calculated as 0.25 times the current bus trip time. p Then the passenger's disembarkation time AT b,w,j,l The calculation formula is as follows:

[0100] AT b,w,j,l =AT b,w,j +Δt p ;

[0101] Step 3: Calculate AT at a given disembarkation time b,w,j,l Next boarding station BS p,k+1 Corresponding boarding time TT p,k+1 The time interval μ p,l μ p,l =TT p,k+1 -AT b,w,p,l ;

[0102] Step 4: When a passenger transfers to a station within the same route, meaning the passenger's two boarding stations are on the same line, determine the maximum waiting time (WT) at the station based on the departure intervals of different functional levels of the multimodal bus system. p,k The maximum waiting time is used as the maximum transfer time threshold η for same-station transfers. p,k , that is, η p,k =WT p,kWhen a passenger transfers between different stations, meaning the passenger's boarding stations are not on the same line, the maximum acceptable walking distance D and the passenger's walking speed v are considered. w Calculate the maximum transfer time threshold;

[0103] Step 5: Set the travel time interval μ p,l With the maximum transfer time threshold η p,k Compare, if μ p,k ≤η p,k If an individual bus user completes a single multi-modal or single-modal bus trip on that day, and μ p,l This refers to transfer time; conversely, if the transfer time is shorter, then an individual bus user completes multiple multi-modal or single-modal bus trips on the same day. p,l The event period is [time].

[0104] After identifying and segmenting individual travel chains, virtual O and D points need to be defined for each individual in the multimodal public transport network according to their historical travel habits. Based on the spatial location attributes of the individual's departure (arrival) station in the road network, the Manhattan distance between different departure (arrival) stations is calculated, and departure (arrival) stations are clustered. A virtual O (D) point is defined for each departure (arrival) station cluster. DBSCAN (Density-Based Spatial Clustering of Applications with Noise) is the most classic density clustering algorithm. It can determine the clustering structure of the research object based on the density of the sample data distribution. By setting a set of "neighborhood" parameters, one or more core objects in a set of samples can be obtained. Moreover, it can discover some outliers while clustering and is not sensitive to outliers in the dataset. Finally, the virtual OD pairs of the core objects (identifying individual travelers) are determined.

[0105] The establishment of virtual OD pairs is specifically as follows:

[0106] Step 1: Identify and number all complete travel chains of individual buses, extract the starting and ending boarding stations of each travel chain, and form a set D. p,BS Each object in the set contains the number of each travel chain and the latitude and longitude coordinates of the corresponding boarding station;

[0107] Step 2: Set the neighborhood parameter ε according to the radiation range of bus stops at different functional levels, and adjust MinPts accordingly based on the specific research object; apply the neighborhood parameter (ε, MinPts) to the set D. p,BS Perform a core object search to obtain the core object set Ω p,BS ;

[0108] Step 3, from set Ω p,BSA core object is randomly selected as the seed to execute a clustering algorithm, searching for all sites reachable by its density, thus forming the first cluster. Then The core objects included are from Ω p,BS The process involves filtering out the weak clusters, then randomly selecting a seed from the updated set to generate the next cluster, repeating this process until the set is empty.

[0109] Step 4: According to each cluster For each object, the corresponding travel chain number is used to filter out the terminal drop-off station of the travel chain, forming various sets. Each object in the set also contains the number of each travel chain and the latitude and longitude coordinates of the corresponding drop-off point;

[0110] Step 5: Perform a set analysis based on the neighborhood parameters (ε, MinPts). Perform a core object search, remove outliers, and obtain a core object set.

[0111] Step 6: Calculate the set and The center point; the center points of each corresponding set together form multiple virtual OD point pairs for passenger p; the calculation formula is:

[0112]

[0113]

[0114] in, Represents the X-coordinate of the center point of each set; The x-coordinate represents the center point of each set; b,w,n The x-coordinate represents the coordinate of the object contained in each set; the y-coordinate represents the coordinate of the object contained in each set. b,w,n The y-coordinate represents the Y-coordinate of the objects contained in each set; n represents a specific object in each set; N represents all objects contained in each set.

[0115] After identifying and segmenting individual travel chains and establishing virtual origin-destination (OD) pairs, individuals form complete travel chains. Considering the continuity of travel habits, individuals will have multiple travel records on the same travel chain. These travel record data form the basic data source for the multimodal public transport passenger flow prediction model. To ensure accurate data input for the individual travel decision-making model, it is necessary to identify key information of the individual travel chain. This information includes the individual's origin and destination, time to reach the boarding station, boarding station, waiting time at the station, vehicle stop time, boarding time, alighting station, alighting time, transfer time, and time to reach the destination.

[0116] like Figure 5As shown, in a multimodal public transport network, the individual traveler, as the decision-maker, first selects their departure time in the "Destination (i.e., the starting point of the next trip)" state, and then transitions to the "Departure Time Selected" state. Subsequently, they select a boarding stop based on their travel principles and transition to the "Boarding Stop Selected" state. At the boarding stop, the individual selects a bus route between their origin and destination, and after selection, reaches the "On the Way" state. In the "Bus Route Selection" state, the individual has determined whether to take the bus directly to their destination or disembark at certain stops to transfer to another route. If no transfer is needed, the individual transitions from the "On the Way" state to the "Disembarkation Stop Selected" state and proceeds to their destination, returning to the "Destination" state. If a transfer is required, the individual selects a boarding stop in the "Disembarkation Stop Selected" state, beginning a new round of action selection decisions and state transitions. This illustrates the state transitions of the individual's travel process within the public transport network.

[0117] Wherein, the state space S:

[0118] S={Destination / Origin, Departure-time, Boarding-stion, En-route, Alighting-station};

[0119] Among them, Destination / Origin is the latitude and longitude coordinates of a pair of virtual OD points in an individual's history; Departure-time is the specific departure time information selected by the individual; Boarding-station is the latitude and longitude of the boarding station selected by the individual, and the information of the collateral lines included with the station; En-route is the vehicle, train number, and other information of the line corresponding to the individual's selected boarding station; Alighting-station is the line, vehicle, train number, latitude and longitude information of the alighting station selected by the individual.

[0120] Action Space A:

[0121] A = {Select-departure-time, Select-boarding-station, Select-bus-route, Select-alighting-station, To-the-destination}. Each action space is a set of candidate data. Among them, the Select-departure-time candidate set contains all departure time information of an individual's historical trips; the Select-boarding-station candidate set contains all boarding station information of an individual's historical trips; the Select-bus-route candidate set contains the information of the bus routes corresponding to all boarding stations of an individual's historical trips; the Select-alighting-station candidate set contains all historical alighting station information corresponding to an individual's "bus route selection"; and the To-the-destination candidate set contains the latitude and longitude coordinates of all established virtual OD points.

[0122] The travel decision-making model includes direct reward R, future reward F, and state transition probability P. The action selection for each state is based on the future reward F, which determines the state transition probability P. The action with the highest P is selected, and the user transitions to the next state. When transitioning to the next state, the direct reward R of that action selection becomes known, and the historical experience value set is updated. First, the corresponding R, F, and P values ​​in the state and action spaces are defined as shown in the table below:

[0123]

[0124] In this context, the direct reward value is represented by the travel time cost. However, the time cost of an individual during the travel process is actually a penalty value, which contradicts the definition of reward value. Therefore, the direct reward corresponding to each action is represented by a negative number of time cost.

[0125] (1)R SDT The direct reward of departure time selection is determined by two factors: an individual's expected arrival time and the total historical travel time. When selecting a departure time, an individual first determines their expected arrival time. Based on the difference Δt between the actual arrival time and the expected arrival time, they determine whether the trip is early (negative value), on time (0), or delayed (positive value). The individual's actual arrival time is determined by the sum of the departure time and the actual total travel time. The formula for calculating Δt is: Δt = t depart +t trip -t desire , where t depart Indicates the departure time, t trip Indicates the actual travel time, t desire Indicates the expected arrival time.

[0126] Since individuals prioritize punctuality in their actual travels, to align with real-life situations, the direct reward is defined as positive for both early arrival and delay, minimizing the reward for punctuality; R SDT It can be represented as: Where, Δt early The difference in time between early arrivals, Δt late This represents the time difference of the delay.

[0127] (2)R SBS Individual travelers will determine their boarding station at their final destination, then R SBS It can be defined as the travel time t of an individual from their destination to their boarding station. D / O,BS Negative numbers: R SBS =-t D / O,BS ;

[0128] Individuals can reach their boarding station by walking, cycling, or car, but walking is the most common method. When an individual needs to transfer at a stop to another stop, R... SBS Defined as the travel time t of an individual from one exit station to another boarding station. BS,BS' Negative numbers: R SBS =-t BS,BS' ;

[0129] (3)R SBR Individuals choose bus routes at boarding stops. The direct reward for each bus route is reflected in the total time required for that route, including the individual's waiting time t at the current stop. wait The stopping time t of the bus at this stop stop The travel time t between the boarding station and the final destination station travel When an individual needs to transfer midway, R SBR It also includes individual transfer travel time t BS,BS' . t wait Specifically, it refers to the difference between the time an individual arrives at the station and the time a vehicle arrives at the current station; t stop Specifically, it refers to the difference between the time a vehicle arrives at a station and the time it leaves that station; t travel Specifically, it refers to the total time an individual spends inside the vehicle en route; when the individual reaches their final destination, R... SBR Represented as: R SBR =-(t) wait +t stop +t travel );

[0130] When an individual needs to transfer midway to reach the final station, R SBR Represented as: RSBR =-(t) wait +t stop +t travel +t BS,BS' );

[0131] (4)R TD The individual's drop-off point and destination have been determined in the aforementioned action selection; therefore, the direct reward for reaching the destination is the travel time t between the individual's drop-off point and destination. TD,D / O Negative numbers: R TD =-t TD,D / O ;

[0132] 2. Future return F

[0133] The future reward F is the basis for calculating the state transition probability. The direct reward value of an individual when making action choices in a dynamic public transportation information environment is unknown. The direct reward value can only be accurately known after the individual actually completes the action. Therefore, based on the continuity of individual travel habits, the future reward can be estimated through historical experience values. Each action choice of a travel individual has a corresponding action choice set, and each action in the set will be recorded multiple times in the historical travel. These recorded values ​​are the direct reward values ​​R corresponding to the individual after completing the action at that time. The future reward F is calculated based on the average of the direct reward values ​​R of the historical records of each action.

[0134] (1)F SDT The future return of departure time selection is the average difference between the actual arrival time and the expected arrival time of a selected departure time based on an individual's historical travel history. Confirmed; F SDT Represented as:

[0135]

[0136] in, This represents the average difference in early arrival times across historical trips. This represents the average difference in time spent on delays throughout the history of travel.

[0137] (2)F SBS :F SBS This refers to the average historical travel time of an individual from their destination to their boarding station. The negative number; if an individual is transferring from an alighting station to another boarding station, then F SBS The average historical travel time for an individual from one exit station to another boarding station. negative numbers; F SBS They are represented as follows:

[0138]

[0139]

[0140] (3)F SBR Individuals need to select a bus route after selecting a boarding stop. Considering that individuals are always in a dynamic travel environment and will adjust and reselect the travel route that minimizes their travel time cost based on the arrival status of vehicles at the stop, when waiting at the stop, individuals first consider the average waiting time at each stop for each route in historical travel data. Negative numbers, average stopping time of buses at this station Negative numbers and the average travel time between the boarding station and the final destination station The sum of negative numbers determines the optimal route for this trip; when an individual needs to transfer midway, the average transfer travel time of that individual also needs to be considered. If F is a negative number, then F SBR It can be represented as:

[0141]

[0142]

[0143] At the same time, based on the principle of minimizing travel time costs, individuals also need to consider that the first vehicle arriving during the waiting process may not be the one they intend to board, in which case the future return F needs to be recalculated. SBR ', F SBR The difference is that individuals do not consider the current site. F SBR ' is represented as:

[0144]

[0145]

[0146] So, after waiting for a period of time at the boarding point, if the first bus arriving is on the route the individual planned to board, the individual boards the bus and completes the route selection. If the bus is not on the route the individual planned to board, the individual needs to determine the route based on the current bus's current path and the other routes. SBR 'Compare the state transition probabilities and then decide whether to board the vehicle to complete the current action selection.

[0147] It should be noted that since individuals have already considered the entire travel process when choosing a bus route and determined the transition probabilities of each action here, there is no need to define future reward calculations for the selection of the drop-off point and the journey to the destination. The action decisions involved in the process from "choosing a bus route" to "going to the destination" can be regarded as a whole.

[0148] 3. State transition probability P

[0149] Each time an individual makes a choice of action, the state transition probability must be calculated to determine the final action to be performed. This paper uses a logistic regression expression to calculate the state transition probability P; the action with the largest P is the optimal choice for the current action decision, and the individual prioritizes this action. The various state transition probabilities P are expressed as follows:

[0150]

[0151]

[0152]

[0153] An individual traveling in a multimodal public transport network has 5 states. Therefore, the state transition probability matrix M is... (5) Defined as:

[0154]

[0155] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A public transit individual trip decision model, characterized in that, include: The analysis of individual bus travel data reveals the travel patterns of individual bus users, which serve as the initial data input for the model. Individual public transport travel data includes IC card swipe data, bus-mounted GPS data, and bus route and stop location information; Based on the definition of the connotation and functional hierarchy of multimodal public transport, the initial basic data is optimized: the travel chain of individual travelers is divided into multimodal public transport travel and single-modal public transport travel according to the transfer time threshold, and density clustering is performed according to the spatial location attributes of the individual's departure or arrival station in the road network to establish corresponding virtual OD pairs and identify the information of individual travel chains. Based on the information of individual travel chains, construct sets of state spaces and action spaces for different travels, and continuously update the action sets according to the incomplete information environment of the public transport network; combine historical individual travel time costs, travel habit definitions and labeling of direct and future rewards of state-action pairs, as well as state transition probabilities, to establish an individual travel decision model that includes departure time selection, boarding station selection and travel route selection on an individual basis. Individual travel chains are divided based on transfer time thresholds, specifically as follows: Step 1: judging the previous and next boarding station according to the boarding station identification result of the passenger and whether they belong to the same line, if they belong to the same line, the passenger's travel on the day belongs to single-mode public transport travel, if they do not belong to the same line, it is judged whether the two public transport lines belong to the same functional level, if they belong to the same functional level, the passenger's travel on the day still belongs to single-mode public transport travel, if they do not belong to the same functional level, it belongs to multi-mode public transport travel; Step 2: determining the station according to the off-site corresponding to the bus vehicle operation shift , obtaining the arrival time table of the shift vehicle, matching the station name of the off-site with the station in the arrival time table, obtaining the arrival time of the vehicle at the station ; taking 0.25 of the current shift travel time of the bus vehicle as the off-time of the passenger, then the off-time of the passenger is calculated as follows:​ ； Step 3: Calculate a certain alighting time with a next alighting station corresponding alighting time time interval , ; Step 4: When the passenger transfers belongs to the same station transfer, according to the different functional levels of the multi-mode bus line interval, the maximum waiting time of the passenger at the station is determined , the maximum waiting time is taken as the maximum transfer time threshold of the same station transfer , that is ; When the passenger transfers between different stations, i.e., the boarding stations of the passenger before and after do not belong to the same line, the maximum walking distance that the passenger can accept is determined The walking speed of the passenger The maximum transfer time threshold is calculated; Step 5: Intervals of travel time are determined compared to the maximum transfer time threshold , if then the bus individual completes a single multi-modal or single-mode bus trip on the day, and is the transfer time; otherwise, the bus individual completes a multiple multi-modal or single-mode bus trip on the day, is the activity time. 2.The bus individual trip decision model according to claim 1, characterized in that: in, By combining bus route and station location information, using time as the main connection field and supplemented by GPS data of buses, a vehicle arrival timetable is obtained. Sparse GPS data is supplemented by interpolation and matched with station latitude and longitude to obtain vehicle arrival times. The time matching method is used to obtain the IC card swipe time of passengers, thus completing the process of identifying the boarding station of IC card holders.

3. The bus individual trip decision model according to claim 2, wherein: GPS data interpolation processing: In the selected acquired GPS data, the latitude and longitude of any two adjacent GPS records are respectively... , The recording times are respectively , The interval is Interpolation was performed between two GPS records at 1-second intervals. The number of data points is calculated, and the inserted data point is... Latitude and longitude of each GPS data point: ； ； wherein the corresponding recording time is .

4. The bus individual trip decision model according to claim 3, wherein: Based on the travel chain theory, individual public transport trips are divided into closed and non-closed public transport travel chains. Combining the boarding station identification results and bus station characteristics, the alighting stations for different types of public transport travel chains are calculated.

5. The bus individual trip decision model according to claim 4, wherein: The estimated drop-off point is as follows: Step 1, according to the boarding station , determine whether the current card swiping record is the last one for the day, if , then execute Step 2, otherwise execute Step 4; Step 2, judging whether the boarding station of the current ride train and the boarding station of the next ride train belong to the same line, if yes, taking the boarding station of the next ride train as the arrival station of the current ride train, i.e. , if not, executing Step 3; Step 3, judging whether the arrival station of the current ride train and the boarding station of the next ride train belong to the same line, if yes, taking the boarding station of the next ride train as the arrival station of the current ride train, i.e. , if not, executing Step 4. Step 3, determine the Centered on the circle, maximum walking distance Within a circular area with radius , the set of stations along the current train route, from upstream to downstream. If the set is empty, then the travel chain belongs to a non-public transport travel chain; if the number of stations in the set... Then the unique station in this set is the alighting station for the current train. ;like Then calculate the calculations based on the upstream station. Starting point, route any of the stations ,by Travel time to the destination Choose the station with the shortest travel time. This is the stop where you will get off the current train. The calculation formula is as follows: ； wherein, is the walking speed; is the distance between the stations and is the bus route length between the stations and is the bus operating speed;​​ Step 4, judging the boarding station of the last train and the boarding station of the first train whether they belong to the same line, if yes, taking the boarding station of the first train as the alighting station of the last train, namely , otherwise, executing Step 5; Step 5, determining the boarding station of the first boarding bus The set of stations from upstream to downstream that the last boarding bus passes through in the circular region with the center as the boarding station and the maximum walking distance D as the radius. If the set is empty, the passenger trip chain belongs to a non-closed public transport trip chain. Otherwise, the alighting station of the last boarding bus is determined according to step 3.

6. The bus individual trip decision model according to claim 4, wherein: Individual travel also includes non-public transport chains, with the following drop-off points estimated: Step 1, determining the boarding station of the current ride Downstream station set ; Step 2, calculating the probability of the passenger getting off at any station Step 2, calculating the probability of the passenger getting off at any station , the formula is: ； wherein, , respectively represent the site surrounding development intensity and the bus accessibility characteristic parameter; , , is the average number of passengers of the site in a certain period of time, which indirectly reflects the site surrounding development intensity; is the number of bus lines passing through the site ; Step 3: Based on the passenger's location at any station The probability of getting off the bus The roulette wheel method is used to predict the alighting station for the current train. .

7. The individual trip decision model of a bus according to claim 5 or 6, characterized in that: The establishment of virtual OD pairs is specifically as follows: Step 1, identify all complete trip chains of public transport individual and number them, extract the starting boarding station of each trip chain, and form a set Each object in the set contains the number of each trip chain and the latitude and longitude coordinates of the corresponding boarding station; Step 2, set neighborhood parameters according to the range of radiation of bus stations of different function levels , Then, set according to the specific research object; according to the neighborhood parameters , search the core objects on the set , and obtain the core object set ; Step 3, randomly select a core object from the set as a seed to execute the clustering cluster generation algorithm, search out all the density reachable sites from it, thereby forming the first cluster ; then filter out the core objects contained in the set from the set , and then randomly select a seed from the updated set to generate the next cluster, repeat until the set is empty ​​ Step 4, according to each cluster The travel chain number corresponding to each object is screened out, and the terminal drop-off site corresponding to the number is formed into each set Each object in the set also contains the number of each travel chain and the latitude and longitude coordinates of the corresponding drop-off site; Step 5, neighborhood parameter On set Core object search, remove outliers, get core object set ; Step 6, calculating the set of center points and The center points of each group of corresponding sets of center points collectively form a plurality of groups of virtual OD point pairs for the passengers ​ The calculation formula is: ； ； wherein, represents the X coordinate of the center point of each cluster; represents the Y coordinate of the center point of each cluster; represents the X coordinate of the objects contained in each cluster; represents the Y coordinate of the objects contained in each cluster; represents a certain object in each cluster; represents all the objects contained in each cluster.

8. The bus individual trip decision model of claim 1, wherein: The state space S: ； wherein, is the longitude and latitude coordinate of a certain pair of virtual OD points in the individual history; is the specific departure time information selected by the individual; is the longitude and latitude of the pick-up station selected by the individual, and the information of the co-linear lines included in the station; is the vehicle and operating train information of the line corresponding to the pick-up station selected by the individual; is the line, vehicle, operating train, longitude and latitude information of the drop-off station selected by the individual; Action Space A: Each action space is a candidate data set, wherein, The candidate set contains all departure time information of individual historical trips; The candidate set contains all pick-up station information of individual historical trips; The candidate set contains all co-line bus route information corresponding to the pick-up station of individual historical trips; The candidate set contains all historical drop-off station information corresponding to the individual "bus route selection"; The candidate set contains the latitude and longitude coordinates of all established virtual OD points.

9. The bus individual trip decision model according to claim 8, wherein: In the individual travel decision-making model, the action selection for each state is based on the probability of state transition determined by the future reward. The action with the highest state transition probability is selected and the user transitions to the next state. When transitioning to the next state, the direct reward of this action selection becomes known and the set of historical experience values ​​is updated.

Images