CN103366224A - Bus-network-based system and method for predicting passenger requirements - Google Patents
Bus-network-based system and method for predicting passenger requirements Download PDFInfo
- Publication number
- CN103366224A CN103366224A CN2013102934073A CN201310293407A CN103366224A CN 103366224 A CN103366224 A CN 103366224A CN 2013102934073 A CN2013102934073 A CN 2013102934073A CN 201310293407 A CN201310293407 A CN 201310293407A CN 103366224 A CN103366224 A CN 103366224A
- Authority
- CN
- China
- Prior art keywords
- time
- bus
- passenger
- model
- passenger demand
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a bus-network-based system and method for predicting passenger requirements. According to the method, the unevenness, the abruptness, the periodicity and other factors are compressively taken into consideration, and predicting of the passenger requirements in a bus network is finally obtained through predicting models such as a Poisson model changing along with time, a Poisson model weighing time changing and a moving average model synthesizing autoregression and an integrating frame based on a sliding window. The demand of passengers obtained through prediction of the system and method can provide a more convenient, faster and more comfortable bus trip environment for the passengers, and therefore the bus waiting time of the passengers is shortened, and overcrowding or over loosening of buses is avoided.
Description
Technical field
The present invention relates to a kind of passenger demand prognoses system and method based on public traffic network.
Background technology
Be accompanied by rapid development of economy, transportation has obtained develop rapidly, but traffic but constantly worsens, produced a series of traffic problems: phenomenon congested in traffic and that a passage is blocked up is on the rise, frequent accidents occurs, and the energy resource consumption that thereupon brings and environmental pollution also more and more cause the common concern of society.These problems can be effectively alleviated in public transport, especially bus service.The public transport service can effectively utilize existing means of transportation, reduces traffic loading and environmental pollution, guarantees traffic safety, improves conevying efficiency, road improvement user's convenience and comfortableness.Bus has a very wide distribution in addition, low price, thereby day by day be subject to various countries people's favor.What on average take bus every day among Singapore 5,000,000 residents in 2011 just has more than 330 ten thousand.Yet the service quality of bus is still waiting to improve at present, and the passenger wishes to shorten as far as possible Waiting time when taking bus, take as far as possible not crowded bus.In fact, overcrowding bus may frighten a lot of passengers away, thereby makes them abandon taking bus.Therefore the rationally public transport service of equilibrium should be to satisfy the maximization of public transport company and passenger both sides' interests.If lose this equilibrium a kind of in following two kinds of situations will appear: 1) too much empty wagons and passenger demand seldom; 2) very long and overcrowding bus of passenger's stand-by period.As seen, accurately, the prediction of real-time passenger demand can help public transport company to determine the rational bus departure time interval, and can reduce passenger's Waiting time, this just people need in a hurry.
Yet because the existence of a lot of uncertain factors, the present invention need to face following three challenges: 1) heterogeneity.The passenger to the demand of public transport service between different websites, different operating in the daytime, and different time sections on the same day all there are differences; 2) sudden.The passenger demand amount of each bus station is different, and the passenger demand of a lot of bus stations is to have paroxysmally, can be subject to the impact of a lot of unscheduled events, such as traffic congestion, and Changes in weather etc.; 3) periodically.The passenger to the demand of public transport service on the same working days in different weeks, and morning on the same day and all have the very high degree of correlation at dusk.In order to solve well these challenge, the present invention proposes a kind of passenger demand prognoses system and method based on public traffic network.Based on gps data and the public transport service data (such as passenger's up/down station point) of history, set up a time series histogram of P minute for the passenger demand of each bus station.The present invention adopts famous time series forecasting technology, and the Poisson model that changes such as time dependent Poisson model, weighting time and ARIMA model etc. are predicted the passenger demand in the public traffic network.
Effectively the passenger demand prediction will become the important new feature that high-level service is provided in the public traffic network, and (Location Based Services, LBS) used in other location-based services also is very useful.But so far, the prediction of the passenger demand in the public traffic network never was considered.Vuchic (V. R. Vuchic.:Transit Operating Manual. Department of Transportation. Pennsylvania, USA. the method that 1976) proposes a kind of peak load district decides the interval at the time of departure of bus, thereby enough traffic capacities are provided.Daganzo (C. F. Daganzo.:A Headway-based Approach to Eliminate Bus Bunching:Systematic Analysis and Comparisons. Transp. Res. Part B 43:913-921. 2009) proposes a kind of adaptive control scheme and eliminates the bus pack, dynamically determines the time of departure of bus according to real-time time headway information.(the J. Zhao such as Zhao, M. Dessouky, S. Bukkapatnam.:Optimal Slack Time for Schedule-based Transit Operations. Transp. Sci. 40 (4): 529-539. 2006) studied the problem of optimum slack time, make passenger's stand-by period minimum by the control of bus haulage vehicle.Chen (H. Chen.:Stochastic Optimization in Computing Multiple Headways for a Single Bus Line. In Proceedings of the 35th Annual Simulation Symposium (ANSS-35). IEEE Computer Society, California, USA. 2002) considered the problem of many headway periods on the same public bus network, its Optimized model is in order to maximize the profit of public transport company.(the S. Y. Yan such as Yan, C. J. Chi, C. H. Tang.:Inter-city Bus Routing and Timetable Setting under Stochastic Demands. Transp. Res. Part A 40:572-586. 2006) studied the model that arranges of route and timetable for the random demand of intercity bus line, yet this model is not considered the city public bus network, and does not analyze the demand difference of different time sections.Above-mentioned these research work are not all considered the passenger to the real-time requirement of public transport service, and this emphasis of the present invention place just.
About bus prediction time of arrival, at present existing a lot of correlative studys.(the C. V. Hinsbergen such as Van Hinsbergen, J. V. Lint, H. J. Zuylen.: Bayesian Committee of Neural Networks to Predict Travel Times with Confidence Intervals. Transportation Research Part C, Vol. 17, pp. 498-509. 2009) with neural network fusion in the stand-by period forecast model, and utilize bayesian theory to solve to select the problem of optimal network.(the H. Chang such as Chang, D. Park, S. Lee, H. Lee, S. Baek.:Dynamic Multi-interval Bus Travel Time Prediction using Bus Transit Data. Transportmetrica Vol. 6, pp. 19-38. 2010) based on the arest neighbors non parametric regression mulitpath running time that a kind of dynamic model is predicted from the start site termination website has been proposed.(the B. Yu such as Yu, W. Lam, M. L. Tam.:Bus Arrival Time Prediction at Bus Stop with Multiple Routes. Transportation Research Part C Vol. 19, pp. 1157-1170. 2011) proposed to comprise support vector machine (SVM), artificial neural network (ANN), the several different methods such as k nearest neighbor algorithm (KNN) and linear regression (LR) are come prediction latency time, and the result proves that the SVM model is best to the stand-by period prediction of bus station that many circuits are arranged.About traffic jam relevant research work is arranged also at present.(A. Lakas and M. Chaqfeh.:A Novel Method for Reducing Road Traffic Congestion using Vehicular Communication. In Proceedings of the 6th International Wireless Communications and Mobile Computing Conference, IWCMC, ACM, pp. 16-20. 2010) Vehicular communication system has been proposed, by excavating and propagate detection and the early warning that road information is realized traffic jam.This system is comprised of two parts: one is based on the agreement that floods and is used for transmitting transport information, and one is that the Di Jiesitela algorithm is used for dynamically calculating the minimum route that blocks up of vehicle.(M. Ferreira, R. Fernandes, H. Conceicao, W. Viriyasitavat, and O. Tonguz.:Self-organized Traffic Control. In Proceedings of the Annual International Conference on Mobile Computing and Networking, MOBICOM, ACM, pp. 85-90. 2010) in designed a kind of virtual traffic lights agreement, in the situation of roadside infrastructure, the traffic flow of intersection is carried out dynamic optimization.Each car in the said method hypothesis road all participates in traffic administration voluntarily, and relevant information initiatively is provided.A lot of drivers only are ready to enjoy the convenience that transport information is brought, and are unwilling to share any information but in fact.Because the existence of this selfishness, these traffic control systems all are infeasible.Passenger demand of the present invention prediction can help public transport company to determine the rational bus departure time interval, reduces passenger's Waiting time, thereby reduces even avoid the jams.
In a word, existing methodical main deficiency is: 1, these present research work are not all considered the passenger to the real-time requirement amount of public transport service, and this is the principal element in the traffic administration; 2, present a lot of research work are all take the positive cooperation of supposing all drivers as prerequisite, and these hypothesis all are infeasible in most of practical applications; 3, the heterogeneity of passenger demand, sudden and periodicity are not all considered in existing research work, and these must be considered in passenger demand is analyzed, otherwise can affect the passenger demand result who finally dopes; 4, existing research work based on taxi, its data are incomplete.
Summary of the invention
Problem for prior art exists the object of the present invention is to provide a kind of passenger demand prognoses system and method based on public traffic network.The method considers the factors such as heterogeneity, sudden and periodicity, thus for public transport company and passenger provide a kind of accurately, passenger demand prognoses system and method in the real-time public traffic network.
For achieving the above object, passenger demand prognoses system and method based on public traffic network of the present invention, concrete steps are:
1) from the application of actual public traffic network, provides the summary description of passenger demand Forecasting Methodology;
2) consider the factors such as heterogeneity, sudden and periodicity, propose respectively three kinds of different passenger demand forecast models;
3) propose a kind of framework based on moving window and integrate three kinds of forecast models.
Further, the actual public traffic network in the step 1) is expressed as follows: suppose that certain bar public bus network comprises N(N
2) individual bus station S={ S
1, S
2..., S
N.Wherein first website is the website that sets out, and last is the termination website.Current follow specific route and the specific timetable of bus between website.D
b={ d
1, d
2..., d
jBe illustrated in j that website s gets on the bus, as to take b road bus passenger destination set.According to being divided into interval of service of bus 4 time periods at (5 of mornings are to point in the mornings 9, and at 9 in the morning is to point at noon 1, and at 1 at noon is to point in afternoons 5, in afternoon to point in evenings 9) working time at 5.The problem to be solved in the present invention is engraved in the number that bus station s will take the passenger of b road bus when being exactly prediction t.
Further, the forecast model to passenger demand step 2) comprises time dependent Poisson model, Poisson model and ARIMA model that the weighting time changes.
Further, time dependent Poisson model may further comprise the steps: certain bus station has the probability P (n) of n Public Transit Bus Stopping to satisfy Poisson distribution in preset time, is defined as
In the formula
Be illustrated in the interior passenger of set time section to the ratio of the average demand of public transport service, among the present invention
Value be not constant, but time dependent.Therefore we regard it as a function of time
Thereby, Poisson distribution is transformed into nonhomogeneous.
Be defined as
，
D in the formula (t) expression working day 1=Sunday, 2=Monday ...; H (t) be under the time t time period (for example, if per 30 minutes as a time period, then time 00:31 is contained in second time period).Need in addition to satisfy following two equatioies
D is the number of time period in one day in the formula;
It is the average ratio of all Poisson processes;
Represent i days relative variation (ratio such as Saturday is lower than Tuesday);
Represent the relative variation (such as the peak period) of j days i time periods;
Be a discrete function, be used for representing the upper time dependent passenger demand of bus station s.
Further, the Poisson model of weighting time variation may further comprise the steps: time dependent Poisson model has only been predicted the average passenger demand of time correlation, yet the passenger demand amount of each bus station is different.In fact, the passenger demand of a lot of bus stations is to have paroxysmally, can be subject to the impact of a lot of unscheduled events, such as traffic congestion, and Changes in weather etc.The Poisson model that the weighting time changes can solve sudden problem well.Its objective is the with it degree of correlation of former all passenger demand amounts of increase passenger demand amount last week.The weight w of the degree of correlation is with famous Time Series Method-exponential smoothing-calculate, and it is defined as
In the formula
The mean value of passenger demand in the in the past time period,
Be smoothing factor, its value is defined by the user, its magnitude range is 0＜
＜1.
Further, ARIMA model may further comprise the steps: two kinds of models suppose that all there is periodic law in the passenger to the demand of public transport service before, and in fact passenger's demand between different websites, different operating in the daytime, and different time sections on the same day all there are differences.The univariate time series data can be simulated and predict to ARIMA model well, such as traffic flow data and short-term forecasting problem etc.It is advantageous that to represent exactly dissimilar time serieses, such as auto-regressive time series, moving average time series, and the combination of the two.In ARIMA model, the predicted value of variable can be regarded the linear function of historical observation and stochastic error as.We take temperature the upper time dependent passenger demand of certain specific bus station s and do time series among the present invention, so forecasting process can be expressed as follows
In the formula
With
Respectively actual value and the stochastic error in moment t passenger demand amount;
With
Be parameter and the weights of model, wherein p and q are the positive integers on the rank of representation model.The rank of model and weights can utilize autocorrelation function and partial autocorrelation function to obtain from the historical time sequence.Whether these values can be used for detecting and exist periodically, with and periodic frequency.
Further, the conformable frame based on moving window may further comprise the steps in the step 3): three kinds of forecast models that propose step 2) are predicted for the historical data of long-term, mid-term and short-term respectively.Conformable frame based on moving window is intended to they are combined better prediction of realization.
Expression to one preset time modeling time series the set of z model;
Represent that these models are in the set of moment t to the predicted value of next time period.The consolidated forecast value
Can be calculated by following formula
Wherein
It is model
The predicted value that certain time period in time window [t-H, t] is made.H is the size of moving window defined by the user.Because the public transport data message is successively to arrive in the follow-up time period, so time window also will constantly slide, thereby guarantees that these models are in next H normally operation in the time period.In order to estimate better the accuracy of prediction, what we adopted is famous time series forecasting error metrics mechanism-symmetrical average percent error (sMAPE).
Passenger demand prognoses system in the public traffic network of the present invention comprises data storage layer and data analysis layer, and data storage layer is used for storing the public transport data; The data analysis layer is used for the public transport data according to the data storage layer storage, by the time dependent Poisson model in the data analysis layer, the Poisson model that the weighting time changes, ARIMA model and process the passenger demand amount that obtains in the public traffic network based on the conformable frame of moving window.
Further, described public transport data comprise five property values: 1) public transport state value, wherein busy represents passenger's quantity greater than the capacity of bus, and free represents passenger's quantity less than the capacity of bus, park represent bus just resting in initial or the termination website on; 2) ID of bus station; 3) time of data generation; 4) the bus trade mark; 5) longitude and latitude of gps data correspondence position.
Further, time dependent Poisson model, the Poisson model that the weighting time changes and ARIMA model can solve respectively the problems such as heterogeneity, sudden and periodicity.
The present invention has considered the influence factor of public traffic network and these two keys of passengers quantity, thus can be for public transport company and the passenger provides more accurately, based on passenger demand prognoses system and the method for public traffic network.The method considers the factors such as heterogeneity, sudden and periodicity, thus for the passenger provide a kind of accurately, passenger demand prediction in the real-time public traffic network.Passenger demand prognoses system and method based on public traffic network that the present invention proposes can provide more convenient comfortable bus trip environment for the passenger, such as the Waiting time that reduces the passenger, avoid the overcrowding or excessive loose situation of bus.
Description of drawings
Fig. 1 is the synoptic diagram of the passenger demand prognoses system based on public traffic network of the present invention;
Fig. 2 is the as a result comparison diagram that passenger demand of the present invention is affected by time and Changes in weather.
Embodiment
Below in conjunction with accompanying drawing and example the present invention is described in detail.
As shown in Figure 1, the passenger demand prognoses system comprises in the public traffic network: public transport data Layer, data pre-service and analysis layer, passenger demand prediction interval, knowledge base, service layer.
Data Layer is used for storing public transport gps data source, and its function is to carry out collection and the arrangement of data, for the upper strata provides abundant source data.In actual applications can be according to the data customization of different application scene suitable memory module.In the present invention, use XML file storage sample data.
Data pre-service and analysis layer are responsible for the source data in the data Layer is carried out pre-service, eliminate noise, the redundant data that filtering and passenger demand prediction are irrelevant, and the method such as utilization statistics is presorted to the public transport data.
Just in order to obtain the applicable public transport data source of processing, its preprocess method adopts conventional method to get final product, and does not do too much explanation at this for above-mentioned data Layer and data pre-service and analysis layer.
The passenger demand prediction interval is absorbed in the realization of logic function, is the core of whole system, uses Java to write.It is time dependent Poisson model, the Poisson model that the weighting time changes, and ARIMA model is based on the set of each class methods such as conformable frame of moving window.And diverse ways is divided into little module, each module can be finished a kind of specific demand forecast.
What store in the knowledge base is to analyze each rule-like and the knowledge that obtains afterwards through demand forecast.Before this, we at first will assess detected result in the demand forecast analytical procedure.After user or machine assessment, may find wherein to exist redundancy or irrelevant result, should reject it this moment.Only keep in the knowledge base those through the assessment and the checking after, can truly reflect Repository passenger demand, useful.
Service layer presents to public transport company and passenger with being responsible for the visual result that will dope, provides the certain operations interface to them simultaneously, is used for sending query requests to the passenger demand prediction interval, thereby can provides convenient comfortable service for the passenger better.The design object of service layer is user friendly, complete function, light and compatible good.
Passenger demand prognoses system and method based on public traffic network of the present invention, concrete steps are:
1) from the application of actual public traffic network, provides the summary description of passenger demand Forecasting Methodology;
2) consider the factors such as heterogeneity, sudden and periodicity, propose respectively three kinds of different passenger demand forecast models;
3) propose a kind of framework based on moving window and integrate three kinds of forecast models.
Actual public traffic network in the step 1) is expressed as follows: suppose that certain bar public bus network comprises N(N
2) individual bus station S={ S
1, S
2..., S
N.Wherein first website is the website that sets out, and last is the termination website.Current follow specific route and the specific timetable of bus between website.D
b={ d
1, d
2..., d
jBe illustrated in j that website s gets on the bus, as to take b road bus passenger destination set.According to being divided into interval of service of bus 4 time periods at (5 of mornings are to point in the mornings 9, and at 9 in the morning is to point at noon 1, and at 1 at noon is to point in afternoons 5, in afternoon to point in evenings 9) working time at 5.The problem to be solved in the present invention is engraved in the number that bus station s will take the passenger of b road bus when being exactly prediction t.
Step 2) forecast model to passenger demand in comprises time dependent Poisson model, Poisson model and ARIMA model that the weighting time changes.Time dependent Poisson model may further comprise the steps: certain bus station has the probability P (n) of n Public Transit Bus Stopping to satisfy Poisson distribution in preset time, is defined as
In the formula
Be illustrated in the interior passenger of set time section to the ratio of the average demand of public transport service, among the present invention
Value be not constant, but time dependent.Therefore we regard it as a function of time
Thereby, Poisson distribution is transformed into nonhomogeneous.
Be defined as
D in the formula (t) expression working day 1=Sunday, 2=Monday ...; H (t) be under the time t time period (for example, if per 30 minutes as a time period, then time 00:31 is contained in second time period).Need in addition to satisfy following two equatioies
D is the number of time period in one day in the formula;
It is the average ratio of all Poisson processes;
Represent i days relative variation (ratio such as Saturday is lower than Tuesday);
Represent the relative variation (such as the peak period) of j days i time periods;
Be a discrete function, be used for representing the upper time dependent passenger demand of bus station s.
The Poisson model that the weighting time changes may further comprise the steps: time dependent Poisson model has only been predicted the average passenger demand of time correlation, yet the passenger demand amount of each bus station is different.In fact, the passenger demand of a lot of bus stations is to have paroxysmally, can be subject to the impact of a lot of unscheduled events, such as traffic congestion, and Changes in weather etc.Fig. 2 shows is that time and Changes in weather are on the impact of passenger demand.As shown in Figure 2, certain, the passenger demand difference of different time sections was obvious on working day, and the passenger is in 7 o'clock to the 9 o'clock morning (working peak period), 4 o'clock to 6 o'clock afternoon (peak period of coming off duty) having the call to the public transport service.And rainy day, the passenger will lack than usual to the demand of public transport service, hired a car or the trip mode more efficiently such as private car because a lot of passengers on rainy day can select.The Poisson model that the weighting time changes can solve sudden problem well.Its objective is the with it degree of correlation of former all passenger demand amounts of increase passenger demand amount last week.The weight w of the degree of correlation is with famous Time Series Method-exponential smoothing-calculate, and it is defined as
In the formula
The mean value of passenger demand in the in the past time period,
Be smoothing factor, its value is defined by the user, its magnitude range is 0＜
＜1.
ARIMA model may further comprise the steps: two kinds of models suppose that all there is periodic law in the passenger to the demand of public transport service before, and in fact passenger's demand between different websites, different operating in the daytime, and different time sections on the same day all there are differences.The univariate time series data can be simulated and predict to ARIMA model well, such as traffic flow data and short-term forecasting problem etc.It is advantageous that to represent exactly dissimilar time serieses, such as auto-regressive time series, moving average time series, and the combination of the two.In ARIMA model, the predicted value of variable can be regarded the linear function of historical observation and stochastic error as.We take temperature the upper time dependent passenger demand of certain specific bus station s and do time series among the present invention, so forecasting process can be expressed as follows
In the formula
With
Respectively actual value and the stochastic error in moment t passenger demand amount;
With
Be parameter and the weights of model, wherein p and q are the positive integers on the rank of representation model.The rank of model and weights can utilize autocorrelation function and partial autocorrelation function to obtain from the historical time sequence.Whether these values can be used for detecting and exist periodically, with and periodic frequency.
Conformable frame based on moving window in the step 3) may further comprise the steps: three kinds of forecast models that propose step 2) are predicted for the historical data of long-term, mid-term and short-term respectively.Conformable frame based on moving window is intended to they are combined better prediction of realization.
Expression to one preset time modeling time series the set of z model;
Represent that these models are in the set of moment t to the predicted value of next time period.The consolidated forecast value
Can be calculated by following formula
Wherein
It is model
The predicted value that certain time period in time window [t-H, t] is made.H is the size of moving window defined by the user.Because the public transport data message is successively to arrive in the follow-up time period, so time window also will constantly slide, thereby guarantees that these models are in next H normally operation in the time period.In order to estimate better the accuracy of prediction, what we adopted is famous time series forecasting error metrics mechanism-symmetrical average percent error (sMAPE).
In sum, in the passenger demand prognoses system and method based on public traffic network of the present invention, utilized to comprise 1,326 bus, the public transport data of 806,257 up/down station points have covered Bus informations whole Yantai city, 22 weeks all sidedly; The factors such as heterogeneity, sudden and periodicity have been considered; The applying practical application has been analyzed the passenger demand forecasting problem at public traffic network, thus can for public transport company and passenger provide a kind of accurately, passenger demand Forecasting Methodology in the real-time public traffic network, its predictablity rate reaches 96%.
Above-mentioned each example only is used for explanation the present invention, and wherein the structure of each parts, connected mode etc. all can change to some extent, and every equivalents and improvement of carrying out on the basis of technical solution of the present invention all should do not got rid of outside protection scope of the present invention.
Claims (10)
1. based on passenger demand prognoses system and the method for public traffic network, concrete steps are:
1) from the application of actual public traffic network, provides the summary description of passenger demand Forecasting Methodology;
2) consider the factors such as heterogeneity, sudden and periodicity, propose respectively three kinds of different passenger demand forecast models;
3) propose a kind of framework based on moving window and integrate three kinds of forecast models.
2. passenger demand prognoses system and the method in the public traffic network as claimed in claim 1 is characterized in that, the actual public traffic network in the step 1) is expressed as follows: suppose that certain bar public bus network comprises N(N
2) individual bus station S={ S
1, S
2..., S
N; Wherein first website is the website that sets out, and last is the termination website; Current follow specific route and the specific timetable of bus between website; D
b={ d
1, d
2..., d
jBe illustrated in j that website s gets on the bus, as to take b road bus passenger destination set; According to being divided into interval of service of bus 4 time periods at (5 of mornings are to point in the mornings 9, and at 9 in the morning is to point at noon 1, and at 1 at noon is to point in afternoons 5, in afternoon to point in evenings 9) working time at 5; The problem to be solved in the present invention is engraved in the number that bus station s will take the passenger of b road bus when being exactly prediction t.
3. passenger demand prognoses system and the method in the public traffic network as claimed in claim 1, it is characterized in that, step 2) forecast model to passenger demand in comprises time dependent Poisson model, Poisson model and ARIMA model that the weighting time changes.
4. passenger demand prognoses system and the method in the public traffic network as claimed in claim 3, it is characterized in that, time dependent Poisson model may further comprise the steps: certain bus station has the probability P (n) of n Public Transit Bus Stopping to satisfy Poisson distribution in preset time, is defined as
, in the formula
Be illustrated in the interior passenger of set time section to the ratio of the average demand of public transport service, among the present invention
Value be not constant, but time dependent; Therefore we regard it as a function of time
Thereby, Poisson distribution is transformed into nonhomogeneous;
Be defined as
, d in the formula (t) expression working day 1=Sunday, 2=Monday ..., h (t) be under the time t time period (for example, if per 30 minutes as a time period, then time 00:31 is contained in second time period); Need in addition to satisfy following two equatioies
With
, D is the number of time period in one day in the formula,
The average ratio of all Poisson processes,
Represent i days relative variation (ratio such as Saturday is lower than Tuesday),
Represent the relative variation (such as the peak period) of j days i time periods,
Be a discrete function, be used for representing the upper time dependent passenger demand of bus station s.
5. passenger demand prognoses system and the method in the public traffic network as claimed in claim 3 is characterized in that, the Poisson model that the weighting time changes may further comprise the steps: increase passenger demand amount last week is the degree of correlation of former all passenger demand amounts with it; The weight w of the degree of correlation is with famous Time Series Method-exponential smoothing-calculate, and it is defined as
, in the formula
The mean value of passenger demand in the in the past time period,
Be smoothing factor, its value is defined by the user, its magnitude range is 0＜
＜1.
6. passenger demand prognoses system and the method in the public traffic network as claimed in claim 3, it is characterized in that, ARIMA model may further comprise the steps: simulate and predict dissimilar univariate time series data, the predicted value of variable can be regarded the linear function of historical observation and stochastic error as; We take temperature the upper time dependent passenger demand of certain specific bus station s and do time series among the present invention, so forecasting process can be expressed as follows
, in the formula
With
Respectively actual value and the stochastic error in moment t passenger demand amount,
With
Be parameter and the weights of model, wherein p and q are the positive integers on the rank of representation model; The rank of model and weights can utilize autocorrelation function and partial autocorrelation function to obtain from the historical time sequence; Whether these values can be used for detecting and exist periodically, with and periodic frequency.
7. passenger demand prognoses system and the method in the public traffic network as claimed in claim 1 is characterized in that, the conformable frame based on moving window in the step 3) may further comprise the steps: three kinds of models couplings in the claim 3 are got up to realize better prediction;
Expression to one preset time modeling time series the set of z model;
Represent that these models are in the set of moment t to the predicted value of next time period; The consolidated forecast value
Can be calculated by following formula
,
, wherein
It is model
The predicted value that certain time period in the time window [t-H, t] is made, H is the size of moving window defined by the user; Because the public transport data message is successively to arrive in the follow-up time period, so time window also will constantly slide, thereby guarantees that these models are in next H normally operation in the time period; In order to estimate better the accuracy of prediction, what we adopted is famous time series forecasting error metrics mechanism-symmetrical average percent error (sMAPE).
8. passenger demand prognoses system in an employing such as each described public traffic network of claim 1-7 is characterized in that comprise data storage layer and data analysis layer, data storage layer is used for storing the public transport data; The data analysis layer is used for the public transport data according to the data storage layer storage, by the time dependent Poisson model in the data analysis layer, the Poisson model that the weighting time changes, ARIMA model and process the passenger demand amount that obtains in the public traffic network based on the conformable frame of moving window.
9. system as claimed in claim 8, it is characterized in that, described public transport data comprise five property values: 1) public transport state value, wherein busy represents that passenger's quantity is greater than the capacity of bus, free represents passenger's quantity less than the capacity of bus, park represent bus just resting in initial or the termination website on; 2) ID of bus station; 3) time of data generation; 4) the bus trade mark; 5) longitude and latitude of gps data correspondence position.
10. system as claimed in claim 8 is characterized in that, time dependent Poisson model, and the Poisson model that the weighting time changes and ARIMA model can solve respectively the problems such as heterogeneity, sudden and periodicity.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310293407.3A CN103366224B (en) | 2013-07-15 | 2013-07-15 | Passenger demand prediction system and method based on public transport network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310293407.3A CN103366224B (en) | 2013-07-15 | 2013-07-15 | Passenger demand prediction system and method based on public transport network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103366224A true CN103366224A (en) | 2013-10-23 |
| CN103366224B CN103366224B (en) | 2021-09-17 |
Family
ID=49367517
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310293407.3A Active CN103366224B (en) | 2013-07-15 | 2013-07-15 | Passenger demand prediction system and method based on public transport network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103366224B (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103955973A (en) * | 2014-04-15 | 2014-07-30 | 浙江工业大学 | Method for detecting and acquiring passenger capacity of bus |
| CN105489000A (en) * | 2015-09-08 | 2016-04-13 | 同济大学 | Night-shift bus stop and path selection method |
| CN106529787A (en) * | 2016-11-02 | 2017-03-22 | 南京国电南自轨道交通工程有限公司 | Subway energy management system based on face detection technology |
| CN108199888A (en) * | 2018-01-09 | 2018-06-22 | 东北大学 | A kind of method predicted based on historical information and alleviate public traffic network congestion |
| CN109146113A (en) * | 2017-06-16 | 2019-01-04 | 宏碁股份有限公司 | Riding demand prediction method and system |
| CN110853156A (en) * | 2019-11-18 | 2020-02-28 | 西南交通大学 | Passenger OD identification method integrating bus GPS track and IC card data |
| CN111723871A (en) * | 2020-07-09 | 2020-09-29 | 广州市公共交通数据管理中心 | Estimation method for real-time carriage full load rate of bus |
| CN113554878A (en) * | 2021-09-18 | 2021-10-26 | 深圳市城市交通规划设计研究中心股份有限公司 | Road section impedance function determination method, calculation device and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101620781A (en) * | 2008-06-30 | 2010-01-06 | 株式会社查纳位资讯情报 | System and method for forecasting passenger information and searching the same |
| US20100185486A1 (en) * | 2009-01-21 | 2010-07-22 | Disney Enterprises, Inc. | Determining demand associated with origin-destination pairs for bus ridership forecasting |
| CN102436603A (en) * | 2011-08-29 | 2012-05-02 | 北京航空航天大学 | Rail transit full-road-network passenger flow prediction method based on probability tree destination (D) prediction |
| CN103310287A (en) * | 2013-07-02 | 2013-09-18 | 北京航空航天大学 | Rail transit passenger flow predicting method for predicting passenger travel probability and based on support vector machine (SVM) |
-
2013
- 2013-07-15 CN CN201310293407.3A patent/CN103366224B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101620781A (en) * | 2008-06-30 | 2010-01-06 | 株式会社查纳位资讯情报 | System and method for forecasting passenger information and searching the same |
| US20100185486A1 (en) * | 2009-01-21 | 2010-07-22 | Disney Enterprises, Inc. | Determining demand associated with origin-destination pairs for bus ridership forecasting |
| CN102436603A (en) * | 2011-08-29 | 2012-05-02 | 北京航空航天大学 | Rail transit full-road-network passenger flow prediction method based on probability tree destination (D) prediction |
| CN103310287A (en) * | 2013-07-02 | 2013-09-18 | 北京航空航天大学 | Rail transit passenger flow predicting method for predicting passenger travel probability and based on support vector machine (SVM) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103955973A (en) * | 2014-04-15 | 2014-07-30 | 浙江工业大学 | Method for detecting and acquiring passenger capacity of bus |
| CN103955973B (en) * | 2014-04-15 | 2016-02-24 | 浙江工业大学 | A kind of detection of the bus volume of passenger traffic and acquisition methods |
| CN105489000A (en) * | 2015-09-08 | 2016-04-13 | 同济大学 | Night-shift bus stop and path selection method |
| CN106529787A (en) * | 2016-11-02 | 2017-03-22 | 南京国电南自轨道交通工程有限公司 | Subway energy management system based on face detection technology |
| CN109146113A (en) * | 2017-06-16 | 2019-01-04 | 宏碁股份有限公司 | Riding demand prediction method and system |
| CN108199888A (en) * | 2018-01-09 | 2018-06-22 | 东北大学 | A kind of method predicted based on historical information and alleviate public traffic network congestion |
| CN108199888B (en) * | 2018-01-09 | 2021-05-04 | 东北大学 | Method for predicting and relieving public transport network congestion based on historical information |
| CN110853156A (en) * | 2019-11-18 | 2020-02-28 | 西南交通大学 | Passenger OD identification method integrating bus GPS track and IC card data |
| CN110853156B (en) * | 2019-11-18 | 2020-11-17 | 西南交通大学 | Passenger OD identification method integrating bus GPS track and IC card data |
| CN111723871A (en) * | 2020-07-09 | 2020-09-29 | 广州市公共交通数据管理中心 | Estimation method for real-time carriage full load rate of bus |
| CN113554878A (en) * | 2021-09-18 | 2021-10-26 | 深圳市城市交通规划设计研究中心股份有限公司 | Road section impedance function determination method, calculation device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103366224B (en) | 2021-09-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103366224A (en) | Bus-network-based system and method for predicting passenger requirements | |
| Amirgholy et al. | Optimal design of sustainable transit systems in congested urban networks: A macroscopic approach | |
| Luo et al. | A new framework of intelligent public transportation system based on the internet of things | |
| Mohapatra et al. | An IoT based efficient multi-objective real-time smart parking system | |
| Guidoni et al. | Vehicular traffic management based on traffic engineering for vehicular ad hoc networks | |
| Gentile et al. | Route choice on transit networks with online information at stops | |
| CN102081859B (en) | Control method of bus arrival time prediction model | |
| CN107358319A (en) | Flow Prediction in Urban Mass Transit method, apparatus, storage medium and computer equipment | |
| CN109612488B (en) | Big data micro-service-based mixed travel mode path planning system and method | |
| CN101123551B (en) | An intelligent bus system based on communication and grid computing technology | |
| CN104123833A (en) | Road condition planning method and device thereof | |
| CN105070044A (en) | Dynamic scheduling method for customized buses and car pooling based on passenger appointments | |
| Zhou et al. | The passenger demand prediction model on bus networks | |
| CN102044149A (en) | City bus operation coordinating method and device based on time variant passenger flows | |
| Zhou et al. | Predicting the passenger demand on bus services for mobile users | |
| Dai et al. | A balanced assignment mechanism for online taxi recommendation | |
| CN111667114A (en) | Intelligent route planning recommendation method based on space-time big data fusion | |
| Raiyn | Road traffic congestion management based on a search-allocation approach | |
| Wang et al. | Large-scale intelligent taxicab scheduling: A distributed and future-aware approach | |
| Miao et al. | Robust taxi dispatch under model uncertainties | |
| CN117546218A (en) | Distributed multitasking machine learning for traffic prediction | |
| Zhang et al. | Analysis of spatial-temporal characteristics of operations in public transport networks based on multisource data | |
| Wang et al. | Towards accessible shared autonomous electric mobility with dynamic deadlines | |
| Kamiński et al. | Multiagent routing simulation with partial smart vehicles penetration | |
| Yuan et al. | Dynamic integration of heterogeneous transportation modes under disruptive events |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |