Real-time rescheduling method and device for solving train operation delay

Abstract

The present disclosure provides a real-time rescheduling method and device for solving train operation delays. According to one aspect of the present disclosure, a real-time rescheduling method for solving train operation delays is provided, which includes the following processes: collecting passenger information and operation information of the train, wherein the passenger information is information related to passengers entering and exiting at least one stop; using the passenger information and the operation information, calculating the number of passengers who got on and off the train first and generating a first data set including the number of passengers who got on and off the train first, wherein the number of passengers who got on and off the train first refers to the number of passengers who got on and off the train in at least one preceding train that stopped at the scheduled stop of the train; using the first data set, estimating the number of passengers who got on and off the train this time, wherein the number of passengers who got on and off the train this time refers to the number of passengers who got on and off the train at the scheduled stop; using a second data set including the number of passengers who got on and off the train this time, estimating the required stop time of the train at the scheduled stop; and comparing the required stop time with the set stop time and generating a timetable to update the operation timetable of the train.

Description

Technical Field

[0001] The present disclosure relates to a real-time rescheduling technology for estimating the number of people getting on and off a train and resolving operation delays. Background Art

[0002] The contents recorded in this section are only used to provide background information for the embodiments of the present invention and do not constitute prior art.

[0003] During peak hours, high-density urban rail lines in urban core areas experience increased passenger traffic, leading to congestion on platforms and trains. This can cause boarding and alighting delays at stops. These delays not only delay the train itself but also affect the operation of following trains, creating a root cause of cascading delays.

[0004] If a train's departure delay prevents it from departing from a station, the following train will be unable to enter the station. If the following train is unaware of the train's delay and continues to operate according to the original schedule, it will be stuck waiting in the interval between stations. As a result, the following train will repeatedly accelerate and decelerate in the interval between stations, degrading train service quality and reducing the scheduled speed.

[0005] In order to solve the interlocking delay of the current train and the following train, a technology is needed to sense the delay of the preceding train in advance and obtain additional parking time at the departure station. Summary of the Invention

[0006]

Technical Issues

[0007] According to one aspect of the present disclosure, a technology is provided for managing the schedule of subsequent trains in real time by using a machine learning model to predict in advance the passenger needs and required stop times of trains at scheduled arrival stations.

[0008]

Technical solution

[0009] According to one embodiment of the present disclosure, a real-time rescheduling method for resolving train operation delays is provided, comprising the following steps: collecting passenger information and train operation information, wherein the passenger information is information related to passengers entering and exiting at least one stop; using the passenger information and the operation information, calculating the number of passengers who boarded and alighted prior to the scheduled stop and generating a first data set including the number of passengers who boarded and alighted prior to the scheduled stop, wherein the number of passengers who boarded and alighted prior to the scheduled stop from at least one preceding train; using the first data set, estimating the number of passengers who boarded and alighted currently, wherein the number of passengers who boarded and alighted currently is the number of passengers who boarded and alighted currently at the scheduled stop; using a second data set including the number of passengers who boarded and alighted currently, estimating the required dwell duration of the train at the scheduled stop; and comparing the required dwell duration with a pre-set dwell duration to generate a schedule for updating the train operation schedule.

[0010] According to another embodiment of the present disclosure, a learning method for a dwell time estimation model is provided. The dwell time estimation model is executed by a server and is used to calculate the required dwell time for a train. The method includes the following processes: calculating the passenger occupancy of the train, the number of passengers entering and exiting between the platform of the scheduled stop and the train, namely the boarding and alighting quantity, and the actual dwell time of the train at the platform of the current stop, namely the recorded dwell duration, based on the stop that the train intends to enter, namely the current stop; generating a learning data set based on all or part of the passenger occupancy, the boarding and alighting quantity, the recorded dwell time and the stored station information; and training the dwell duration estimation model based on the learning data set.

[0011] According to another embodiment of the present disclosure, a real-time rescheduling device for resolving train operation delays includes: a data collection unit that collects passenger information and operation information of the train, wherein the passenger information is information related to passengers entering and exiting at least one stop; a data set generation unit that uses the passenger information and the operation information to calculate the number of passengers who boarded and alighted prior to the train and generates a first data set including the number of passengers who boarded and alighted prior to the train, wherein the number of passengers who boarded and alighted prior to the train's scheduled stop is the number of passengers who boarded and alighted at at least one preceding train that previously stopped at the train's scheduled stop; a passenger boarding and alighting number estimation unit that uses the first data set to estimate the number of passengers who boarded and alighted currently, wherein the number of passengers who boarded and alighted currently is the number of passengers who boarded and alighted at the scheduled stop; a dwell time estimation unit that uses a second data set including the number of passengers who boarded and alighted currently to estimate the required dwell time of the train at the scheduled stop; and a schedule generation unit that compares the required dwell time with a pre-set dwell time. duration), and generates a timetable for updating the train operations schedule (train operations chedule).

[0012] According to another embodiment of the present disclosure, a server for training a dwell time estimation model is provided, wherein the dwell time estimation model is used to calculate the required dwell time, i.e., the time a train needs to stop. The server includes: a data calculation unit, which calculates the passenger occupancy of the train, the number of passengers entering and exiting between the platform of the current stop and the train, i.e., the boarding and alighting quantity, and the actual dwell time of the train at the platform of the current stop, i.e., the recorded dwell duration, based on the stop that the train intends to enter, i.e., the current stop; a data set generation unit, which generates a learning data set based on all or part of the passenger occupancy, the boarding and alighting quantity, the recorded dwell time, and the stored station information; and a learning unit, which trains the dwell time estimation model based on the learning data set.

[0013] According to another embodiment of the present disclosure, a computer program is provided, which is stored in a computer-readable recording medium and is used to implement each process of the real-time rescheduling method.

[0014] Beneficial effects

[0015] According to one embodiment of the present disclosure, based on the required stop time estimated using a machine learning model, the schedule of the following train is adjusted in real time, which has the effect of minimizing the chain delay of the train operation.

[0016] According to another embodiment of the present disclosure, by obtaining additional parking time at a stop, the congestion level of the stop is alleviated and the delay of the following train is reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic diagram for illustrating a real-time rescheduling system for resolving train operation delays according to an embodiment of the present disclosure.

[0018] Figure 2 It is a structural block diagram for illustrating various components including a real-time rescheduling device for solving train operation delays according to an embodiment of the present disclosure.

[0019] 3 is an example diagram of the number of passengers boarding a train and the number of passengers getting off a train at each stop in a real-time rescheduling method according to an embodiment of the present disclosure.

[0020] FIG4 is an example diagram of the number of passengers getting on and off the preceding train and the current train when entering a scheduled stop in the real-time rescheduling method according to an embodiment of the present disclosure.

[0021] Figure 5 It is a structural block diagram for illustrating various components included in a server according to an embodiment of the present disclosure.

[0022] Figure 6 1 is a flowchart for illustrating various processes of a real-time rescheduling method according to an embodiment of the present disclosure.

[0023] Figure 7 It is a flowchart for illustrating various processes included in the process of generating a first data set in a real-time rescheduling method according to an embodiment of the present disclosure.

[0024] Figure 8 It is a flowchart for illustrating the various processes included in the process of estimating the number of people getting on and off the bus in the real-time rescheduling method according to one embodiment of the present disclosure.

[0025] Figure 9 1 is a flowchart for explaining each process of a method for learning a parking time estimation model according to an embodiment of the present disclosure.

[0026] [Description of Reference Numerals]

[0027] 10: Real-time rescheduling system 100: This train

[0028] 102: First leading train 104: Second leading train

[0029] 106: Rear train 108: Server

[0030] 20: Real-time rescheduling device 200: Data collection unit

[0031] 202: Dataset generation unit 204: Number of people getting on and off the bus estimated unit

[0032] 206: Parking time estimation unit 208: Schedule generation unit

[0033] 500: Calculation unit 502: Learning data generation unit

[0034] 504: Learning Unit DETAILED DESCRIPTION

[0035] Below, embodiments of the present invention are described in detail with reference to the accompanying drawings. When annotating the figures, even if the same components appear in different figures, the same reference numerals are used whenever possible. It should also be noted that when describing the embodiments of the present invention, if a detailed description of related known components and functions is deemed to obscure the subject matter of the present invention, such detailed description will be omitted.

[0036] In addition, in the process of describing the components of the present invention, terms such as first, second, A, B, (a), (b) and the like may be used. These terms are only used to distinguish the corresponding components from other components, and do not limit their nature, order or sequence, etc. Throughout the entire specification, if a component "includes" or "has" another component, unless there is a clear statement to the contrary, it can be understood that a component also includes the other component, rather than being understood as a component excluding the other component. In addition, terms such as "... unit" and "module" recorded in the specification represent units for processing at least one function or action, which can be implemented by hardware, software or a combination thereof.

[0037] The present disclosure provides a real-time rescheduling technology for solving train operation delays. Specifically, the real-time rescheduling device calculates the number of people getting on and off the preceding train, and estimates the number of people getting on and off and the required dwell duration of the current train based on machine learning. The real-time rescheduling device pre-calculates the number of people getting on and off and the required dwell duration before entering the scheduled dwell station. Therefore, according to the real-time rescheduling technology of the present disclosure, the operation schedule of the subsequent train can be updated by pre-judging the platform congestion of the scheduled dwell station. The real-time rescheduling device causes the train to enter the scheduled dwell station and stop within the estimated required dwell time so that passengers on the platform can get on and off. As a result, the number of people getting on and off the subsequent train can be reduced, and thus the platform congestion of the dwell station can be reduced according to the real-time rescheduling technology of the present disclosure.

[0038] As used herein, a "scheduled stop" refers to the station at which the train is scheduled to stop, based on the current time, among the multiple stops at which the train is scheduled to stop on the train's schedule. As used herein, a "departure station" refers to the first stop at which the train begins its journey on the train's route. As used herein, a "prior station" refers to the station at which the train stops immediately before the scheduled stop.

[0039] In the present disclosure, a preceding train refers to at least one train that runs before the current train. In the present disclosure, a following train refers to at least one train that runs after the current train.

[0040] As used herein, the number of passengers boarding and alighting on the preceding train refers to the number of passengers who boarded and alighted on the preceding train at the scheduled stop. The number of passengers boarding and alighting on the current train refers to the number of passengers expected to board and alight on the current train at the scheduled stop. The required stop time refers to the time the current train should stop at the scheduled stop to mitigate train delays.

[0041] The following detailed description and attached Figure 1 It is used to illustrate exemplary embodiments of the present disclosure, and does not represent the only embodiment of the present disclosure.

[0042] Figure 1 It is a schematic diagram for illustrating a real-time rescheduling system for resolving train operation delays according to an embodiment of the present disclosure.

[0043] A real-time rescheduling system 10 according to an embodiment of the present disclosure includes a train 100 , a first preceding train 102 , a second preceding train 104 , a following train 106 , and a server 108 .

[0044] In the present disclosure, the train 100 refers to the train that serves as the benchmark of the real-time rescheduling system 10. The real-time rescheduling device included in the train 100 calculates the number of passengers getting on and off the train and the required parking time of at least one preceding train that precedes the train 100. The real-time rescheduling device uses a machine learning model to estimate the number of passengers getting on and off the train and the required parking time. The machine learning model includes a trained passenger number calculation model and a parking time estimation model. The real-time rescheduling device estimates the number of passengers getting on and off the train 100 and the required parking time based on the number of passengers getting on and off the train, thereby determining in real time whether congestion occurs in a specific section of the route.

[0045] In one embodiment, the leading trains include a first leading train 102 and a second leading train 104. The first leading train 102 refers to the leading train that stops at a scheduled stop immediately before the current train arrives at the scheduled stop and then departs from the scheduled stop. The second leading train 104 refers to the leading train that stops at a scheduled stop immediately before the first leading train 102 arrives at the scheduled stop and then departs from the scheduled stop.

[0046] The following train 106 refers to at least one train that travels after the current train 100. The real-time rescheduling device updates the operation schedule of the current train and / or at least one of the following trains 106 based on the required stop time. The updated operation schedule is transmitted to each of the following trains 106 using wireless communication. Before entering the scheduled stop, the real-time rescheduling device included in the following train 106 estimates the number of passengers getting on and off at the scheduled stop and the required stop time. In other words, the real-time rescheduling device included in the following train 106 can operate in accordance with the real-time rescheduling device included in the current train 100, thereby determining in real time whether the train is delayed, thereby resolving the problem of train delays.

[0047] Server 108 uses the learning data to train the machine learning model of the real-time rescheduling device. Server 108 uses the accumulated learning data set to retrain the stop time estimation model. Server 108 transmits the trained machine learning model to at least one train traveling on the route. Server 108 periodically retrains the machine learning model using the accumulated learning data set, thereby improving the accuracy of the real-time train rescheduling device.

[0048] Figure 2 It is a structural block diagram for illustrating various components including a real-time rescheduling device for solving train operation delays according to an embodiment of the present disclosure.

[0049] According to one embodiment of the present disclosure, the real-time rescheduling device 20 includes all or part of a data collection unit 200 (data collection unit), a data set generation unit 202 (data set generation unit), a boarding and alighting quantity estimation unit 204 (boarding and alighting quantity estimation unit), a dwell duration estimation unit 206 (dwell duration estimation unit) and a schedule generation unit 208 (schedule generation unit). Figure 2 The real-time rescheduling device 20 shown in FIG is an embodiment of the present disclosure. Figure 2 Not all blocks shown in the figure are necessary components. In another embodiment, some blocks included in the real-time rescheduling device 20 may be added, changed or deleted.

[0050] Below, refer to Figure 2 The components of the real-time rescheduling device 20 are described.

[0051] The data collection unit 200 collects passenger information and train operation information. Passenger information refers to information related to passengers entering and exiting at least one stop. In one embodiment, passenger information includes transportation card tag information of passengers entering (tag-in) or exiting (tag-out) at at least one stop. In another embodiment, passenger information refers to the number of passengers waiting at the platform of the stop. The number of passengers on the platform can be calculated based on 5G communication (5th generation communications) between the terminal held by the passenger and the train. The number of passengers on the platform can also be calculated through ultra-wideband communication (UWB communications: UltraWide-Band) between the tag device held by the passenger and the train. The method for calculating the number of passengers waiting on the platform is not limited to this embodiment. Operation information includes all or part of the position of at least one train, the arrival time of the train at at least one stop, and the departure time of the train. In one embodiment, the data collection unit 200 can collect data by wirelessly communicating with at least one of an Automatic Train Supervision (ATS) control system, a station server, and other trains. Passenger information and / or operational information is transmitted via V2V (Vehicle-to-Vehicle) communication and / or V2I (Vehicle-to-Infrastructure) communication. Specific methods for inter-train communication or communication between a train and ground equipment are well known in the relevant technical field, and therefore detailed descriptions are omitted.

[0052] The data set generation unit 202 uses the passenger information and operation information to calculate the number of passengers who boarded and disembarked on at least one preceding train and generates a first data set including the number of passengers who boarded and disembarked on the preceding train. In one embodiment, the first data set includes the number of passengers on the preceding train, the number of passengers on the current train, and the number of passengers who boarded and disembarked on the preceding train. The number of passengers on the preceding train refers to the number of passengers on board the preceding train when the preceding train arrives at the scheduled stop. The number of passengers on the current train is the number of passengers on board the preceding train when the preceding train arrives at the scheduled stop. Refers to the number of passengers on board the train 100 when the train 100 arrives at the scheduled stop. In this embodiment, the data set generating unit 202 generates the first data set on the premise that the preceding trains include the first preceding train 102 and the second preceding train 104. The first data set includes the number of passengers on board the first preceding train 102. Number of passengers on the second preceding train 104 The number of people on this train is 100 Number of passengers boarding and alighting on the first preceding train 102 and the number of passengers who boarded and alighted on the second preceding train 104 In another embodiment, the data set generating unit 202 may consider generating information related to other leading trains other than the first leading train 102 and the second leading train 104 defined in this embodiment.

[0053] In order to calculate the number of passengers on the previous bus or the number of passengers on the current bus, the data set generation unit 202 can estimate the number of passengers at each bus stop based on the number of passengers entering each bus stop, and can estimate the number of passengers getting off at each bus stop based on the number of passengers leaving each bus stop. The number of passengers entering the bus stop or the number of passengers leaving the bus stop can be calculated using passenger information. In one embodiment, the number of passengers entering and exiting the bus stop can be calculated using transportation card tag information collected from the ticket gate of the bus stop. The specific method by which the data set generation unit 202 estimates the number of passengers at each bus stop or the number of passengers getting off at each bus stop based on the number of passengers entering and exiting the bus stop will be described in detail in the following sections. Figure 5 . In another embodiment, the server 108 may also estimate the number of passengers at each stop or the number of passengers alighting at each stop based on the number of passengers entering the stop. The data set generation unit 202 calculates the total boarding quantity by accumulating the number of passengers from the departure station to the previous stop, and calculates the total alighting quantity by accumulating the number of alightings from the departure station to the previous stop. The data set generation unit 202 deducts the total alighting quantity from the total boarding quantity as the number of passengers on the previous vehicle or the number of passengers on the current vehicle.

[0054] 3 is an example diagram of the number of passengers boarding a train and the number of passengers getting off a train at each stop in a real-time rescheduling method according to an embodiment of the present disclosure.

[0055] FIG3a illustrates boarding the train TR at the platform of the departure station STA1. x The number of passengers on board and the number of passengers getting off the train FIG. 3 b shows the number of passengers boarding the train at the platform of the second stop STA2. and the number of passengers getting off the train FIG3c illustrates the previous station STA j-1 The number of passengers boarding the train on the platform and the number of passengers getting off the train The data set generating unit 202 accumulates the number of passengers at the departure station Number of passengers to the previous station Calculate the total number of passengers The number of people getting off at the cumulative departure station Number of people getting off at the previous station Calculate the total number of people getting off the bus The data set generation unit 202 deducts the total number of passengers from the total number of people who have gotten off the vehicle as the number of people on board when the vehicle enters the scheduled stop. The data set generation unit 202 calculates the number of passengers on the first preceding train using the method shown in FIG3 Number of passengers on the second preceding train and the number of passengers on this train

[0056] In order to calculate the number of passengers who boarded or alighted the bus first, the data set generation unit 202 estimates the number of passengers who boarded the bus first based on the number of passengers who entered the scheduled stop, and estimates the number of passengers who alighted the bus first based on the number of passengers who left the scheduled stop. The specific method for estimating the number of passengers who boarded or alighted the bus first based on the number of passengers who entered or left the scheduled stop will be described in detail in the following sections. Figure 5 The data set generation unit 202 takes the sum of the number of people who boarded the vehicle before and the number of people who got off the vehicle before as the number of people who got on and off the vehicle before. and

[0057] FIG4 is an example diagram of the number of passengers getting on and off the preceding train and the current train when entering a scheduled stop in the real-time rescheduling method according to an embodiment of the present disclosure.

[0058] 4a, at the scheduled parking station STA j On the platform, passengers get on and off the second preceding train 104. The number of passengers who boarded the second preceding train 104 is calculated by The number of passengers who got off the second preceding train 104 Can calculate the number of people getting on and off the second preceding train 4b, at the scheduled parking station STA j On the platform, passengers get on and off the first preceding train 102. The number of passengers who boarded the first preceding train 102 is calculated by The number of passengers who got off the first preceding train 102 Can calculate the number of people getting on and off the first preceding train

[0059] 4c, the train 100 leaves the previous station STA j-1 Departure to the scheduled stop STA j The number of passengers getting on and off the bus can be estimated using the passenger number calculation model. Next, Figure 2 The number of people getting on and off the bus is estimated by the first data set using the unit 204 in the middle figure. The method is described.

[0060] The number of passengers getting on and off the train estimating unit 204 uses a trained passenger number calculation model to estimate the number of passengers getting on and off the train at the stop. The passenger number calculation model is a model that performs machine learning based on at least one train operation record and a record data set (record data set) that accumulates and stores the number of passengers getting on and off the train of at least one preceding train. The record data set includes the record of the number of passengers on board the first preceding train 102, the record of the number of passengers on board the second preceding train 104, the record of the number of passengers on board the current train 100, the record of the number of passengers getting on and off the train of the first preceding train 102, and the record of the number of passengers getting on and off the train of the second preceding train 104. In one embodiment, the machine learning method for training the passenger number calculation model may include the K-NN (K-Nearest Neighbors) method. Next, the number of passengers getting on and off the train is estimated by the K-NN method using the K-NN method for the number of passengers getting on and off the train. The method is described.

[0061] The boarding and alighting number estimation unit 204 extracts at least one data corresponding to the first data set based on the time period from the recorded data set as a candidate data set. The candidate data set includes the candidate number of passengers on the first preceding train 102. Number of passengers on candidate leading cars of the second leading train 104 The number of people on board this train 100 candidates Number of candidates for getting on and off the first preceding train 102 and the number of candidates for getting on and off the second preceding train 104 The number of people getting on and off the bus estimating unit 204 respectively calculates the Euclidean distance between the data of the first data set and the data of the candidate data set. Among them, the Euclidean distance is a method used to evaluate the similarity between data. The method of calculating the Euclidean distance is a conventional method in this technical field, so its detailed description is omitted. The number of people getting on and off the bus estimating unit 204 determines the data of the candidate data set corresponding to the upper k (k is a natural number greater than or equal to 1) with a shorter Euclidean distance as the data of the analogous data set. Among them, the k value for determining the number of similar data sets can be set to an appropriate value to improve the classification accuracy of the model. For example, if the k value is too small, over-fitting may occur. If the k value is too large, under-fitting may occur. Therefore, the k value can be changed in various ways according to the embodiments of the present disclosure. The number of people getting on and off the bus estimating unit 204 calculates the number of people getting on and off the bus this time by calculating the weighted average of the data included in the similar data set.

[0062] The parking time estimation unit 206 uses the estimated number of people getting on and off the bus The parking time estimation unit 206 uses the trained required parking time estimation model to estimate the required parking time. Estimated number of people getting on and off this bus and stored station information SI j All or part of the input parameters of the parking time estimation model are used to estimate the required parking time. Passengers board and alight from the train 100 at the scheduled stops within the required stop time calculated by the stop time estimation unit 206. Therefore, during peak hours when traffic surges, the extended required stop time ensures sufficient time for passengers to board and alight. Furthermore, when estimating the required stop time, it can be difficult to identify specific factors that cause delays. To address this issue, a Random Forest (RF)-based importance measure analysis method can be used to determine the weighted values ​​of factors that have a significant impact on delays. Subsequently, when rescheduling subsequent trains, these factors can be taken into account, thereby adjusting the departure times of multiple subsequent trains at various stops.

[0063] The schedule generation unit 208 compares the required parking time and pre-set dwell duration To update the train operation schedule. For example, when the required stop time When the dwell duration is greater than the scheduled dwell time on the operation schedule, the scheduling generation unit 208 updates the operation schedule based on the dwell duration difference between the required dwell time and the planned dwell time. The train operation schedule updated by the scheduling generation unit 208 includes the operation schedule of the current train 100 and the operation schedule of at least one subsequent train that will travel after the current train 100. In one embodiment, the scheduling generation unit 208 may update the schedule of the subsequent train 106 so that the departure time of the subsequent train 106 at the stop is delayed by 5 to 10 seconds. The subsequent train 106, as the object of the schedule update, refers to the train from the departure station STA1 to the previous station STA j-1 At least one following train at at least one stopping station between them.

[0064] Figure 5 It is a structural block diagram for illustrating various components of a server according to an embodiment of the present disclosure.

[0065] According to an embodiment of the present disclosure, the server 108 includes all or part of a calculation unit 500 (calculation unit), a learning data generation unit 502 (learning data generation unit), and a learning unit 504 (learning unit). Figure 5 The server 108 shown in FIG is an embodiment of the present disclosure. Figure 1 Not all blocks shown in the figure are necessary components. In another embodiment, some blocks included in the server 108 may be added, changed or deleted.

[0066] Below, refer to Figure 5 Each component of the server 108 will be described.

[0067] The calculation unit 500 uses the current parking area STA cBased on the passenger information, the calculation unit 500 calculates the number of passengers on board the train traveling on the train route, the number of passengers entering and exiting the platform at the current stop (i.e., the number of passengers getting on and off), and the actual time the train stopped at the platform at the current stop (i.e., the recorded dwell duration). The calculation unit 500 uses passenger information and at least one train's operational information to calculate the number of passengers on board, the number of passengers getting on and off, and the recorded dwell duration. In one embodiment, the passenger information may include the transportation card tag information of passengers entering and exiting the stop.

[0068] The calculation unit 500 matches passengers entering and exiting each stop with the trains that stop there based on the transit card tag information, thereby estimating the number of passengers and alighters at each stop. To calculate the number of passengers getting on and off at each stop, the calculation unit 500 individually matches passengers with trains. Using the transit card tag information of at least one passenger entering the current stop, the calculation unit 500 calculates the first passenger's entry station (tag-in station) and entry time (tag-in time), and the first passenger's exit station (tag-out station) and exit time (tag-out station). The calculation unit 500 determines whether the exit station matches the current stop. If so, the calculation unit 500 matches the train that arrived last at the current stop platform before the exit time as the train that the first passenger alighted. If the entry station is the current stop, the calculation unit 500 matches the train that arrived first at the current stop platform after the entry time as the train that the first passenger boarded. The calculation unit 500 uses the above-mentioned separate matching method to match the passengers entering and exiting the current stop with the trains on which the passengers get on and off, thereby respectively calculating the number of passengers and the number of people getting off the train.

[0069] The calculation unit 500 calculates the total number of passengers by accumulating the number of passengers at each stop from the departure station to the station immediately before the current stop, i.e., the previous stop. It also calculates the total number of alighters by accumulating the number of alighters at each stop from the departure station to the previous stop. The calculation unit 500 calculates the number of passengers on board by subtracting the total number of alighters from the total number of passengers. The calculation unit 500 calculates the number of alightings and alightings by summing the number of passengers and alighters at the current stop. Based on the operational information, the calculation unit 500 calculates the difference between the departure time and the arrival time (i.e., the time the train departs from the current stop and the time it arrives at the current stop), thereby estimating and recording the stop time.

[0070] The number of passengers on board, the number of passengers getting on and off the train, and the recorded stop time calculated by the calculation unit 500 can be cumulatively saved as a record data set. Specifically, at least one train operation record and the number of passengers getting on and off the train running before the train can be cumulatively saved as a record data set. The record data set includes the number of passengers on board the same type of train, the number of passengers on board the first preceding train, the number of passengers on board the second preceding train, the number of passengers getting on and off the first preceding train, and the number of passengers getting on and off the second preceding train. The same type of train refers to at least one other train that has entered the current stop during the time period corresponding to the time period when the train enters the current stop. The first preceding train refers to the preceding train that entered the current stop just before the same type of train entered the current stop. The second preceding train refers to the preceding train that entered the current stop just before the first preceding train entered the current stop. In order to use the passenger number calculation model trained as described above to infer the scheduled stop STA j The server 108 transmits the record data set accumulated and stored by the computing unit to at least one train traveling on the train route.

[0071] The learning data generating unit 502 generates a learning data set based on all or part of the number of people on board, the number of people getting on and off the vehicle, the recorded stop time, and the stored station information.

[0072] The learning unit 504 trains the parking time estimation model based on the learning data set. The learning unit 504 can use the accumulated learning data set to train the parking time estimation model. The accumulated learning data set refers to the learning data set accumulated in the learning data set based on the newly added number of passengers on board, the newly added number of passengers getting on and off the train, and the newly added recorded parking time. The newly added number of passengers on board, the newly added number of passengers getting on and off the train, and the newly added recorded parking time can be calculated based on the newly collected new passenger information and new operational information. The method by which the learning unit trains the estimation model using the learning data is a conventional method in the art, and therefore a detailed description thereof is omitted. The server 108 periodically transmits the retrained parking time estimation model to at least one train.

[0073] Figure 6 1 is a flowchart for illustrating various processes of a real-time rescheduling method according to an embodiment of the present disclosure.

[0074] Below, refer to Figure 6 The various processes of the real-time rescheduling method will be described. Figures 1 to 5 Duplicate contents will be omitted.

[0075] The data collection unit 200 collects passenger information and train operation information (S600). The specific data included in the passenger information and train operation information are as follows: Figure 2 Therefore, the detailed description is omitted.

[0076] The data set generating unit 202 calculates the number of passengers who have boarded and disembarked on at least one preceding train that stops at a scheduled train stop using the passenger information and operation information, and generates a first data set including the number of passengers who have boarded and disembarked on the preceding train (S602). The specific method for calculating the information included in the first data set will be described in detail in Figure 7 is described in .

[0077] The number of passengers getting on and off the train is estimated by the first data set and the trained passenger number calculation model 204 to estimate the number of passengers getting on and off the train at the scheduled stop (S604). Figure 8 is described in .

[0078] The parking time estimation unit 206 estimates the required parking time of the train at the scheduled stop by using the second data set including the estimated number of people getting on and off the train this time and the trained required parking time estimation model (S606). The specific input parameters used by the parking time estimation unit 206 are as follows: Figure 2 Therefore, the detailed description is omitted.

[0079] The schedule generation unit 208 compares the required stop time with the set stop time and updates the train's operation schedule. Figure 2 Therefore, the detailed description is omitted.

[0080] Figure 7 It is a flowchart for illustrating various processes included in the process of generating a first data set in a real-time rescheduling method according to an embodiment of the present disclosure.

[0081] The data set generation unit 202 calculates the number of passengers on the preceding train when at least one preceding train enters a scheduled stop (S700). The specific method for the data set generation unit 202 to calculate the number of passengers on the preceding train is as follows: Figure 2 Therefore, the detailed description is omitted.

[0082] The data set generating unit 202 calculates the number of passengers on the train when the train arrives at the scheduled stop (S702). The specific method of the data set generating unit 202 calculating the number of passengers on the train is as shown in FIG3, so its detailed description is omitted.

[0083] The data set generating unit 202 estimates the number of passengers who boarded and disembarked before the scheduled stop based on the number of passengers entering and leaving the scheduled stop, and calculates the number of passengers who boarded and disembarked before the scheduled stop (S704). The specific method for calculating the number of passengers who boarded and disembarked before the scheduled stop is as shown in FIG4 , so its detailed description is omitted.

[0084] Figure 8 It is a flowchart for illustrating the various processes included in the process of estimating the number of people getting on and off the bus in the real-time rescheduling method according to one embodiment of the present disclosure.

[0085] The number of passengers getting on and off the train is estimated by the unit 204. The unit 204 extracts data corresponding to the first data set based on the time period from the cumulatively stored data set of the number of passengers getting on and off the train at each stop (S800). The specific data included in the candidate data set is as described above, and a detailed description thereof is omitted.

[0086] The number of people getting on and off the bus estimating unit 204 calculates the Euclidean distance between the data of the first dataset and the data of the candidate dataset (S802). The method of calculating the Euclidean distance between data based on data values ​​is a conventional method in the art, so its detailed description is omitted.

[0087] The number of people getting on and off the bus estimating unit 204 determines the data of the candidate dataset corresponding to the upper level k with the shortest Euclidean distance as the data of the similar dataset (S804). The value of k used as a reference for distinguishing similar datasets can be variously changed according to the embodiments of the present disclosure.

[0088] The number of people getting on and off the bus is estimated by calculating the weighted average of the data of the similar data set (S806). The operation of the number of people getting on and off the bus in advance by the number of people getting on and off the bus 204 can predict the congestion at the scheduled stop in advance.

[0089] Figure 9 1 is a flowchart for explaining each process of a method for learning a parking time estimation model according to an embodiment of the present disclosure.

[0090] Below, refer to Figure 9 Explain each process of the learning method. Figures 1 to 8 Duplicate contents will be omitted.

[0091] The calculation unit 500 calculates the number of passengers on the train, the number of passengers getting on and off the train at the current stop, and the recorded stop time of the train at the current stop based on the current stop ( S900 ).

[0092] The learning data generating unit 502 generates a learning data set based on all or part of the number of people on board, the number of people getting on and off the vehicle, the recorded stop time, and the stored station information ( S902 ).

[0093] The learning unit 504 trains the parking time estimation model based on the learning data set ( S904 ).

[0094] Although Figures 6 to 9 The various processes are described as being performed sequentially, but this is only used to exemplify the technical ideas of some embodiments of the present invention. In other words, for ordinary technicians in the technical field to which some embodiments of the present invention belong, they can modify or change the following within the scope of the essence of some embodiments of the present invention. Figures 6 to 9 The processes described in the above or at least one process in each process is performed in parallel. These various modifications or variations are applicable to the present invention. Figures 6 to 9 Not limited to clockwise order.

[0095] Various implementations of the apparatus and methods described in this specification may be implemented by digital electronic circuits, integrated circuits, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include implementations by one or more computer programs on a programmable system. The programmable system includes at least one programmable processor (which may be a special-purpose processor or a general-purpose processor) coupled to receive data and instructions from a storage system and to send data and instructions to the storage system; at least one input device; and at least one output device. A computer program (also referred to as a program, software, software application, or code) contains instructions for a programmable processor and is stored in a "computer-readable recording medium."

[0096] Computer-readable recording media include all types of recording devices that store data readable by a computer system. Such computer-readable recording media may also include non-volatile or non-transitory media such as ROMs, CD-ROMs, magnetic tapes, disks, memory cards, hard disks, optical disks, storage devices, etc., or transitory media such as data transmission media. Furthermore, computer-readable media may be distributed across computer systems connected via a network and may store and execute computer-readable code in a distributed manner.

[0097] Various embodiments of the apparatus and methods described herein may be implemented using a programmable computer. The computer includes a programmable processor, a data storage system (including volatile memory, non-volatile memory, or other types of storage systems, or a combination thereof), and at least one communication interface. For example, the programmable computer may be a server, a network appliance, a set-top box, an embedded device, a computer expansion module, a personal computer, a laptop computer, a personal data assistant (PDA), a cloud computing system, or a mobile device.

[0098] The above description is merely an illustrative description of the technical concept of the present invention. The embodiments of the present invention are intended to illustrate the technical concept of the embodiments and are not intended to limit the scope of the technical concept of the present invention. The scope of protection of the present invention should be interpreted in accordance with the following claims, and all technical concepts within the scope of equivalents should be understood to fall within the scope of rights of the present invention.

Claims

1. A real-time rescheduling method for resolving train operation delays, comprising the following steps: Collecting passenger information and train operation information, wherein: The passenger information is information related to passengers entering or exiting at least one stop; Calculating the number of passengers who boarded and alighted prior to the train using the passenger information and the operation information and generating a first data set including the number of passengers who boarded and alighted prior to the train, wherein the number of passengers who boarded and alighted prior to the train refers to the number of passengers who boarded and alighted prior to the train at least one of the preceding trains that stopped at the scheduled stop of the train; Using the first data set, estimating the number of passengers getting on and off the train, wherein the number of passengers getting on and off the train refers to the number of passengers getting on and off the train at the scheduled stop; Calculating the required stop time of the train at the scheduled stop using the second data set including the number of people getting on and off the train this time; and Comparing the required parking time with the set parking time and generating a timetable for updating the operation timetable of the train; The process of estimating the number of passengers getting on and off the train utilizes a trained passenger number calculation model to estimate the number of passengers getting on and off the train at the stop. The passenger number calculation model is obtained through machine learning based on the operation records of at least one train and a dataset of accumulated records of the number of passengers getting on and off the train preceding the at least one train. The machine learning method includes a K-NN method, and the process of estimating the number of people getting on and off the bus includes the following steps: extracting data corresponding to the first data set based on a time period from the record data set as a candidate data set; respectively calculating the Euclidean distance between the data of the first dataset and the data of the candidate dataset; Determining the data of the candidate data sets corresponding to the k superiors with the shorter Euclidean distance as the data of the similar data sets, where k is a natural number greater than or equal to 1; and The number of people getting on and off the bus this time is estimated by calculating the weighted average of the data in the similar data set.

2. The real-time rescheduling method according to claim 1, wherein: The passenger information includes transportation card tag information of passengers entering or exiting the at least one stop, and the operation information includes all or part of the position of at least one train, the arrival time of the at least one train at the at least one stop, and the departure time of the at least one train from the at least one stop.

3. The real-time rescheduling method according to claim 1, wherein: The process of generating the first data set includes: When the at least one preceding train enters the scheduled stop, calculating the number of passengers on the preceding train, that is, the number of passengers on the preceding train; When the train arrives at the scheduled stop, the number of passengers on board is calculated, i.e., the number of passengers on board the train; and Calculate the number of people who get on and off the bus first.

4. The real-time rescheduling method according to claim 3, wherein: The at least one preceding train includes a first preceding train that is a preceding train that stops at the scheduled stop just before the train and a second preceding train that is a preceding train that stops at the scheduled stop just before the first preceding train.

5. The real-time rescheduling method according to claim 3, wherein: The process of calculating the number of people on the previous vehicle or the process of calculating the number of people on the current vehicle is calculated in the following manner: The number of passengers at each stop is estimated based on the number of passengers entering each station, and the number of passengers getting off at each stop is estimated based on the number of passengers exiting each station. The total number of passengers is calculated by accumulating the number of passengers from the departure station where the train starts to the station where the train stops just before the scheduled stop, i.e., the previous station. The total number of people getting off is calculated by accumulating the number of people getting off from the departure station to the previous station, and the value subtracting the total number of people getting off from the total number of passengers is used as the number of people on the preceding vehicle or the number of people on this vehicle.

6. The real-time rescheduling method according to claim 3, wherein: The process of calculating the number of people who get on and off the bus first is calculated in the following way: The number of passengers who boarded the bus first is calculated based on the number of passengers who entered the scheduled bus stop, and the number of passengers who got off the bus first is calculated based on the number of passengers who left the scheduled bus stop. The total of the number of passengers who boarded the bus first and the number of passengers who got off the bus first is used as the number of passengers who got on and off the bus first.

7. The real-time rescheduling method according to claim 3, wherein: The process of estimating the required parking time uses a trained required parking time estimation model to estimate the required parking time, and uses all or part of the number of people getting on and off the vehicle, the number of people on board the vehicle, and the saved station information as input parameter values ​​of the parking time estimation model to estimate the required parking time.

8. The real-time rescheduling method according to claim 1, wherein: During the timetable generation process, when the required stop time is greater than the planned stop time on the operation timetable, the timetable of at least one subsequent train of the train is further updated based on the stop time difference between the required stop time and the planned stop time.

9. The real-time rescheduling method according to claim 8, wherein: During the process of generating the timetable, the train is located at at least one stop between the starting station where the train starts running and the station that stops just before the scheduled stop of the train, i.e., the previous station. In order to delay the departure time of at least one subsequent train that follows the train by 5 seconds to 10 seconds, the timetable of the subsequent train is further updated.

10. A real-time rescheduling device for resolving train operation delays, comprising: a data collection unit for collecting passenger information and train operation information, wherein the passenger information is information related to passengers entering and exiting at least one stop; a data set generating unit, which calculates the number of passengers who boarded and alighted prior to the train using the passenger information and the operation information and generates a first data set including the number of passengers who boarded and alighted prior to the train, wherein the number of passengers who boarded and alighted prior to the train refers to the number of passengers who boarded and alighted prior to the train at the scheduled stop of the train; a number of passengers getting on and off the train estimating unit, which estimates the number of passengers getting on and off the train using the first data set, wherein the number of passengers getting on and off the train refers to the number of passengers getting on and off the train at the scheduled stop; a stop time estimating unit, which estimates the required stop time of the train at the scheduled stop using a second data set including the number of people getting on and off the train this time; and a schedule generating unit, which compares the required stop time with the set stop time and generates a schedule for updating the operation schedule of the train; The process of estimating the number of passengers getting on and off the train utilizes a trained passenger number calculation model to estimate the number of passengers getting on and off the train at the stop. The passenger number calculation model is obtained through machine learning based on the operation records of at least one train and a dataset of accumulated records of the number of passengers getting on and off the train preceding the at least one train. The machine learning method includes a K-NN method, and the process of estimating the number of people getting on and off the bus includes the following steps: extracting data corresponding to the first data set based on a time period from the record data set as a candidate data set; respectively calculating the Euclidean distance between the data of the first dataset and the data of the candidate dataset; Determining the data of the candidate data sets corresponding to the k superiors with the shorter Euclidean distance as the data of the similar data sets, where k is a natural number greater than or equal to 1; and The number of people getting on and off the bus this time is estimated by calculating the weighted average of the data in the similar data set.

11. A storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 9 is implemented.

Images