CN-122022048-A - Method and device for dynamically predicting passenger flow in rapid bus station

CN122022048ACN 122022048 ACN122022048 ACN 122022048ACN-122022048-A

Abstract

The invention relates to the technical field of data analysis, and discloses a method and a device for dynamically predicting passenger flow in a rapid bus station, wherein the method comprises the following steps: acquiring current inbound data and upstream inbound data of a current bus station, determining the lower bus passenger flow of the current inbound bus according to the upstream inbound data, determining the upper bus passenger flow and the inbound passenger flow of the current inbound bus according to the current inbound data and the lower bus passenger flow, determining the initial passenger flow of a platform, inputting the initial passenger flow, the lower bus passenger flow, the upper bus passenger flow and the inbound passenger flow of the platform into an outbound passenger flow prediction model, obtaining the optimal outbound passenger flow, and predicting the predicted passenger flow at the next moment according to the optimal outbound passenger flow. Therefore, by implementing the method and the device, the dynamic prediction of the passenger flow in the station platform can be realized, the prediction accuracy and reliability are improved, the prediction result directly reflects the number of people staying in the station platform, a real-time decision basis is provided for preventing and controlling the congestion treading risk and adjusting the bus operation schedule, and the bus station passenger flow management efficiency and the safety are further improved.

Inventors

Chen Qianci
ZHOU LONGTAO
ZENG YE
HUANG PENG

Assignees

广州羊城通有限公司

Dates

Publication Date: 20260512
Application Date: 20260202

Claims (10)

1. A method for dynamically predicting passenger flow in a rapid bus station, the method comprising: Acquiring inbound data corresponding to a current bus stop, wherein the inbound data comprises current inbound data of each current passenger at the current bus stop and upstream inbound data of each upstream passenger at each upstream bus stop corresponding to the current bus stop; Determining the lower bus passenger flow of the current bus stop at the current time according to the upstream arrival data, and determining the upper bus passenger flow and the lower bus passenger flow of the current bus stop at the current time according to the current arrival data and the lower bus passenger flow; Determining a platform initial passenger flow corresponding to the current bus platform at the current moment, and inputting the platform initial passenger flow, the getting-off passenger flow, the getting-on passenger flow and the getting-on passenger flow corresponding to the current moment into a preset getting-out passenger flow prediction model to obtain an optimal getting-out passenger flow corresponding to the current moment; and predicting the predicted passenger flow at the next moment corresponding to the current moment according to the optimal outbound passenger flow.
2. The rapid transit stop passenger flow dynamic prediction method according to claim 1, further comprising: determining model parameters of an outbound passenger flow prediction model, wherein the model parameters comprise the number of passengers in a platform, the outbound passenger flow and regularization parameters; Determining a first index penalty term according to the number of passengers in the platform, and determining a second index penalty term according to the outbound passenger flow, wherein the first index penalty term is used for restraining the number of passengers in the platform, and the second index penalty term is used for restraining the change amplitude of the outbound passenger flow; Constructing an outbound passenger flow objective function according to the first index penalty term, the second index penalty term and the regularization parameter; Determining the maximum outbound passenger flow of the current bus station, determining the model constraint condition of the outbound passenger flow prediction model according to the maximum outbound passenger flow, and constructing the outbound passenger flow prediction model according to the outbound passenger flow objective function and the model constraint condition.
3. The rapid transit station passenger flow dynamic prediction method according to claim 2, further comprising: collecting historical passenger flow data of the current bus station, and performing data cleaning on the historical passenger flow data to obtain target historical passenger flow data, wherein the historical passenger flow data comprises historical incoming passenger flow, historical outgoing passenger flow, historical incoming passenger flow and historical outgoing passenger flow; classifying the target historical passenger flow data according to a preset classification standard to obtain classified passenger flow data, and calculating the classified passenger flow data by adopting a regression fitting algorithm to obtain regularized parameters; and determining the maximum outbound passenger flow of the current bus station, comprising: collecting the exit parameters and the exit line structure of the current bus station, and determining the traffic efficiency influence factors according to the exit line structure, wherein the exit parameters comprise the exit width, the exit quantity and the exit channel length; and analyzing the maximum outbound passenger flow of the current bus stop according to the exit parameter and the traffic efficiency influence factor.
4. A method of dynamic prediction of passenger flow in a rapid transit stop according to any one of claims 1-3, wherein said determining the current boarding and disembarking passenger flow of the current inbound bus at the current time of the current transit stop based on the current inbound data and the disembarking passenger flow comprises: Predicting a departure station of each current passenger according to the current arrival data to obtain a departure station prediction result corresponding to the current arrival data, and counting the current arrival data to obtain arrival passenger flow; Determining downstream station information corresponding to the current bus entering station, matching the predicted result of the station entering station with the downstream station information, screening target boarding passengers, and determining the number of the target boarding passengers as the number of candidate boarding persons; Determining the maximum passenger carrying quantity of the current bus at the current moment, and determining the number of passengers in the current bus before the current bus arrives at the current bus station according to the upstream bus arrival data; calculating the residual passenger capacity of the current inbound bus according to the maximum passenger carrying quantity, the passenger quantity in the bus and the passenger flow of the getting-off bus; judging whether the number of the candidate boarding passengers is larger than the residual passenger capacity, when the number of the candidate boarding passengers is larger than the residual passenger capacity, determining the residual passenger capacity as the boarding passenger flow of the current boarding bus at the current moment, and when the number of the candidate boarding passengers is smaller than or equal to the residual passenger capacity, determining the number of the candidate boarding passengers as the boarding passenger flow of the current boarding bus at the current moment.
5. A method of dynamic prediction of passenger flow in a rapid transit station according to any one of claims 1 to 3, wherein the method further comprises: Acquiring preset passenger flow threshold ranges under different operation scenes, wherein the operation scenes comprise a peak time period scene, a flat peak time period scene and a low peak time period scene, and the passenger flow threshold range of each operation scene comprises a congestion threshold range, a normal threshold range and an idle threshold range; determining a target operation scene corresponding to the next moment, and comparing the predicted passenger flow of the next moment with the passenger flow threshold range in the target operation scene to obtain a comparison result; when the comparison result shows that the predicted passenger flow is in the congestion threshold range under the target operation scene, executing first adjustment operation on bus operation parameters according to a first difference value of a congestion low threshold corresponding to the predicted passenger flow and the congestion threshold range; And when the comparison result shows that the predicted passenger flow is in the idle threshold range under the target operation scene, executing a second adjustment operation on the bus operation parameters according to a second difference value of the idle high threshold corresponding to the predicted passenger flow and the idle threshold range.
6. A method of dynamic prediction of passenger flow in a rapid transit station according to any one of claims 1 to 3, wherein the method further comprises: Determining a plurality of functional areas of the current bus station, and collecting real passenger flow of each functional area at each moment; calculating a passenger flow difference value of each functional area at each moment according to the predicted passenger flow at each moment and the real passenger flow of each functional area at each moment; For each functional area, determining an average passenger flow difference value of the functional area in a unit period according to the passenger flow difference value of the functional area at each moment, and judging whether the average passenger flow difference value of the functional area is larger than a preset difference threshold value; For each functional area, when the average passenger flow difference value of the functional area is larger than the difference threshold value, determining the functional area as a candidate area; for each candidate area, if the average passenger flow difference value of the candidate area in the unit period of continuous preset times is larger than the difference threshold value, determining that the candidate area is a congestion high-occurrence area; And for each congestion high-rise area, acquiring area structure information of the congestion high-rise area, analyzing congestion reasons of the congestion high-rise area according to the area structure information, and generating a layout optimization scheme for the congestion high-rise area according to the congestion reasons and the area structure information.
7. A method for dynamically predicting passenger flow in a rapid transit stop according to any one of claims 1-3, wherein determining the passenger flow of the current bus stop at the current time of the current bus stop on the basis of the upstream arrival data comprises: acquiring historical trip data of each upstream passenger, wherein the historical trip data comprise historical arrival data, station position data and bus route data; For each upstream passenger, carrying out combined clustering on the historical travel data of the upstream passenger to obtain a travel clustering result of the upstream passenger, wherein the travel clustering result comprises at least one cluster, each cluster comprises a plurality of sample points, and each sample point comprises a travel station of the upstream passenger; for each upstream passenger, counting the riding frequency of the upstream passenger at each sample point, determining the weight factor of each cluster according to the riding frequency, and calculating the weight geometric center of each cluster according to the weight factor of each cluster; For each upstream passenger, calculating the space distance from the weighted geometric center of each cluster corresponding to the upstream passenger to each stop of the current bus, and screening the stop with the space distance smaller than the maximum transfer distance corresponding to the upstream passenger as a target station; for each upstream passenger, determining the historical getting-off frequency of the upstream passenger at each target station according to the historical trip data of the upstream passenger, and screening the target getting-off station of the upstream passenger in each target station according to the historical getting-off frequency; for each upstream passenger, if the target departure station of the upstream passenger is the current bus station, determining that the upstream passenger is at the current bus station; And counting the number of upstream passengers getting off at the current bus stop to obtain the get-off passenger flow of the current bus coming into the bus at the current moment at the current bus stop.
8. A rapid bus station passenger flow dynamic prediction device, the device comprising: the system comprises an acquisition module, a control module and a control module, wherein the acquisition module is used for acquiring inbound data corresponding to a current bus stop, and the inbound data comprises current inbound data of each current passenger at the current bus stop and upstream inbound data of each upstream passenger at each upstream bus stop corresponding to the current bus stop; The determining module is used for determining the lower bus passenger flow of the current bus stop at the current time according to the upstream arrival data and determining the upper bus passenger flow and the arrival passenger flow of the current bus stop at the current time according to the current arrival data and the lower bus passenger flow; the determining module is further configured to determine a platform initial passenger flow corresponding to the current bus platform at the current time, and input the platform initial passenger flow, the get-off passenger flow, the get-on passenger flow and the get-on passenger flow corresponding to the current time into a preset get-out passenger flow prediction model to obtain an optimal get-out passenger flow corresponding to the current time; And the prediction module is used for predicting the predicted passenger flow at the next moment corresponding to the current moment according to the optimal outbound passenger flow.
9. A rapid bus station passenger flow dynamic prediction device, the device comprising: a memory storing executable program code; a processor coupled to the memory; The processor invokes the executable program code stored in the memory to perform the rapid transit stop in-passenger flow dynamic prediction method of any one of claims 1-7.
10. A computer storage medium storing computer instructions which, when invoked, are operable to perform the rapid bus stop in-passenger flow dynamic prediction method of any one of claims 1 to 7.

Description

Method and device for dynamically predicting passenger flow in rapid bus station Technical Field The invention relates to the technical field of data analysis, in particular to a method and a device for dynamically predicting passenger flow in a rapid bus station. Background Rapid Transit (BRT) is an important component of urban public transportation systems, and becomes a key transportation way for relieving urban traffic jams by virtue of the advantages of high efficiency and large traffic. With the acceleration of the urban process, BRT passenger flow volume continuously increases, and dynamic real-time prediction of passenger flow volume in a platform becomes a core requirement for guaranteeing operation scheduling efficiency, preventing congestion from treading on safety risks and improving passenger travel experience. However, BRT platform passenger flow is dynamically and crosswise influenced by multiple links such as incoming passenger flow, bus boarding and disembarking passenger flow, outgoing passenger flow and the like, under the conditions of centralized passenger flow in peak time, complex platform layout (such as multiple berthing ports and multiple exiting ports) and the like, the passenger flow change rule is complex, and the traditional prediction method is difficult to accurately capture real-time passenger flow situation, so that the problems of delayed operation scheduling response, insufficient security risk prevention and control, poor passenger waiting experience and the like are caused In the prior art, aiming at BRT station outbound passenger flow and in-station passenger flow prediction, a traditional exponential decay model is mostly adopted. The model fits the outbound process only through a single decay rate constant, does not consider actual influencing factors such as platform key region distribution, passenger walking speed difference, exit capacity limitation and the like, cannot adapt to complex platform environments and dynamically-changed passenger flow scenes, has low prediction result precision and poor robustness, and is difficult to meet the actual requirements of BRT refined operation scheduling and safety management. Therefore, it is important to provide a technical scheme capable of improving accuracy and reliability of predicting passenger flow dynamics in a station platform Disclosure of Invention The invention provides a method and a device for predicting the dynamic state of the passenger flow in a rapid bus station, which can be beneficial to improving the accuracy and the reliability of predicting the dynamic state of the passenger flow in the station. In order to solve the technical problems, the first aspect of the invention discloses a method for dynamically predicting passenger flow in a rapid bus station, which comprises the following steps: Acquiring inbound data corresponding to a current bus stop, wherein the inbound data comprises current inbound data of each current passenger at the current bus stop and upstream inbound data of each upstream passenger at each upstream bus stop corresponding to the current bus stop; Determining the lower bus passenger flow of the current bus stop at the current time according to the upstream arrival data, and determining the upper bus passenger flow and the lower bus passenger flow of the current bus stop at the current time according to the current arrival data and the lower bus passenger flow; Determining a platform initial passenger flow corresponding to the current bus platform at the current moment, and inputting the platform initial passenger flow, the getting-off passenger flow, the getting-on passenger flow and the getting-on passenger flow corresponding to the current moment into a preset getting-out passenger flow prediction model to obtain an optimal getting-out passenger flow corresponding to the current moment; and predicting the predicted passenger flow at the next moment corresponding to the current moment according to the optimal outbound passenger flow. As an alternative embodiment, in the first aspect of the present invention, the method further includes: determining model parameters of an outbound passenger flow prediction model, wherein the model parameters comprise the number of passengers in a platform, the outbound passenger flow and regularization parameters; Determining a first index penalty term according to the number of passengers in the platform, and determining a second index penalty term according to the outbound passenger flow, wherein the first index penalty term is used for restraining the number of passengers in the platform, and the second index penalty term is used for restraining the change amplitude of the outbound passenger flow; Constructing an outbound passenger flow objective function according to the first index penalty term, the second index penalty term and the regularization parameter; Determining the maximum outbound passenger flow of the current bus station, determini