CN108831181A - A kind of method for establishing model and system for Forecasting of Travel Time for Public Transport Vehicles - Google Patents
A kind of method for establishing model and system for Forecasting of Travel Time for Public Transport Vehicles Download PDFInfo
- Publication number
- CN108831181A CN108831181A CN201810418625.8A CN201810418625A CN108831181A CN 108831181 A CN108831181 A CN 108831181A CN 201810418625 A CN201810418625 A CN 201810418625A CN 108831181 A CN108831181 A CN 108831181A
- Authority
- CN
- China
- Prior art keywords
- time
- model
- sequence
- public transport
- history journey
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/123—Traffic control systems for road vehicles indicating the position of vehicles, e.g. scheduled vehicles; Managing passenger vehicles circulating according to a fixed timetable, e.g. buses, trains, trams
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
- G08G1/0129—Traffic data processing for creating historical data or processing based on historical data
Abstract
The invention discloses a kind of method for establishing model and system for Forecasting of Travel Time for Public Transport Vehicles.This method includes:Obtain the public transport GPS data in default public bus network section;Determine the beginning and end in default public bus network section;Determine the start time for putting default public bus network section in different times;Determine that each time point presets the corresponding end of time of start time in public bus network section;Establish history journey time sequence;Determine parameter fitting sequence and model testing sequence;Autoregression integral moving average model is established according to parameter fitting sequence;Judge whether the parameter of autoregression integral moving average model needs to adjust using model testing sequence;If so, being adjusted, model adjusted is determined as Forecasting of Travel Time for Public Transport Vehicles model;If it is not, not adjusting then.Forecasting of Travel Time for Public Transport Vehicles model in the present invention, precision of prediction is high, can satisfy the public transport Forecasting of Travel Time accuracy requirement of signal-oriented control.
Description
Technical field
It is the invention belongs to public transport journey time technical field, in particular to a kind of for Forecasting of Travel Time for Public Transport Vehicles
Method for establishing model and system.
Background technique
Routine bus system is reached since many external factor such as website stop, intersection delay cause runing time uncertain
The randomness of time makes the optimization of public transport Real-Time Scheduling, priority acccess control technology be difficult to carry out and ensure, can not improve public transport entirety
Efficiency of operation.Therefore, using public transport GPS a large amount of historical datas prediction public transportation road section runing times can based on optimize bus signals
Control provides significant application value with Real-Time Scheduling.The prior art is the time for predicting that public transport reaches downstream website more,
Reach forecast using the vehicle for above laying particular emphasis on bus station, prediction error mostly at 1 minute or more, accuracy rate 80%~95% it
Between, it is unable to satisfy the public transport Forecasting of Travel Time accuracy requirement of signal-oriented control.
Summary of the invention
The object of the present invention is to provide a kind of method for establishing model and system for Forecasting of Travel Time for Public Transport Vehicles, with
The precision of prediction for improving Travel Time for Public Transport Vehicles meets the public transport Forecasting of Travel Time accuracy requirement of signal-oriented control.
The invention discloses a kind of method for establishing model for Forecasting of Travel Time for Public Transport Vehicles, include the following steps:
Obtain the public transport GPS data in default public bus network section;
Determine the beginning and end in the default public bus network section;
The start time for putting default public bus network section in different times is determined according to the public transport GPS data;
When determining that each time point presets the corresponding terminal of start time in public bus network section according to the start time
It carves;
History journey time sequence is established according to start time and end of time;The history journey time sequence includes multiple
The history journey time at different time points, the history journey time of each time point is by the corresponding start time of each time point
It is determined with end of time corresponding with the start time;
The history journey time of first preset ratio in the history journey time sequence is determined as parameter fitting sequence,
The history journey time of second preset ratio in the history journey time sequence is determined as model testing sequence;
Autoregression integral moving average model is established according to the parameter fitting sequence;
Judge whether the parameter of the autoregression integral moving average model needs to adjust using the model testing sequence;
The parameter includes model order, autoregressive coefficient and rolling average coefficient;
If so, being adjusted to the parameter of the Regression-Integral moving average model, by parameter autoregression adjusted
Integral moving average model is determined as Forecasting of Travel Time for Public Transport Vehicles model;The Forecasting of Travel Time for Public Transport Vehicles model is used
It is predicted in Travel Time for Public Transport Vehicles;
If it is not, autoregression integral moving average model is then determined as Forecasting of Travel Time for Public Transport Vehicles model.
Further, history journey time sequence is established according to start time and end of time, specifically included:
The history journey time of various time points is obtained, specially
yt=Tte-Tts (1)
Wherein, t indicates time point, TtsIndicate t-th of time point corresponding start time, TteIndicate t-th of time point pair
The end of time answered, ytFor the history journey time at t-th of time point;
History journey time according to various time points establishes history journey time sequence, specially
Yt={ y1,y2,y3,...}
YtIndicate the history journey time sequence in default public bus network section.
Further, after establishing history journey time sequence according to start time and end of time, further include:
Calculate the auto-correlation coefficient of history journey time sequence;
Judge whether history journey time sequence is steady according to auto-correlation coefficient;
If it is not, then carrying out difference processing to history journey time sequence, and determine difference order.
Further, autoregression integral moving average model is established according to parameter fitting sequence, specifically included:
History journey time sequence is established about the linear function for working as time value and preceding time value;
History journey time sequence is established about the linear function for working as time value and preceding time value random error;
According to history journey time sequence about when the linear function of time value and preceding time value, history journey time sequence about
When linear function, difference order and the lag operator of time value and preceding time value random error establish prediction model;The lag is calculated
Son is to convert the preceding time value to the operator for working as time value;
Using trial and error procedure, the model order of the prediction model is determined;
The autoregressive coefficient and rolling average coefficient of the prediction model are determined using the parameter fitting sequence;
Prediction model after determining model order, autoregressive coefficient and rolling average coefficient is determined as autoregression integral
Moving average model.
Further, judge whether the parameter of autoregression integral moving average model needs to adjust using model testing sequence
It is whole, it specifically includes:
Using autoregression integral moving average model to the corresponding default public affairs of various time points in the model testing sequence
The Travel Time for Public Transport Vehicles in intersection road section is predicted, predicted value is obtained;
Maximum relative error prediction index is calculated according to the history journey time in predicted value and model testing sequence;
Average absolute percentage error prediction index is calculated according to maximum relative error prediction index;
Judge maximum relative error prediction index and average absolute percentage error prediction index whether in default progress model
In enclosing;
If so, not needing the parameter of adjustment autoregression integral moving average model;
If it is not, then needing to adjust the parameter of autoregression integral moving average model.
The invention also discloses a kind of model foundation systems for Forecasting of Travel Time for Public Transport Vehicles, including:
Data acquisition module, for obtaining the public transport GPS data in default public bus network section;
First determining module, for determining the beginning and end in default public bus network section;
Second determining module puts default public bus network section for determining according to public transport GPS data in different times
Start time;
Third determining module, for determining that each time point presets the start time in public bus network section according to start time
Corresponding end of time;
Time series establishes module, for establishing history journey time sequence according to start time and end of time;It is described
History journey time sequence includes the history journey time at multiple and different time points, when the history stroke of each time point
Between determined by the corresponding start time of each time point and end of time corresponding with the start time;
4th determining module, for the history journey time of the first preset ratio in the history journey time sequence is true
It is set to parameter fitting sequence, the history journey time of the second preset ratio in the history journey time sequence is determined as model
Checking sequence;
Model building module, for establishing autoregression integral moving average model according to the parameter fitting sequence;
Judgment module, for judging the parameter of the autoregression integral moving average model using the model testing sequence
Whether need to adjust;The parameter includes model order, autoregressive coefficient and rolling average coefficient;
Prediction model determining module, for if so, be adjusted to the parameter of the Regression-Integral moving average model,
Parameter autoregression integral moving average model adjusted is determined as Forecasting of Travel Time for Public Transport Vehicles model;The bus
Travel time prediction model is for predicting Travel Time for Public Transport Vehicles;It is slided if it is not, then integrating the autoregression
Averaging model is determined as Forecasting of Travel Time for Public Transport Vehicles model.
Further, time series establishes module, specifically includes:
Time acquisition unit, for obtaining the history journey time of various time points;
Sequence establishes unit, establishes history journey time sequence for the history journey time according to various time points.
Further, the invention also includes:
Computing module, for calculating the auto-correlation coefficient of the history journey time sequence;
Steady judgment module, for judging whether the history journey time sequence is steady according to the auto-correlation coefficient;
Differential processing module is used for if it is not, then carrying out difference processing to the history journey time sequence, and determine difference
Order.
Further, model building module specifically includes:
First function establishes unit, for establishing the history journey time sequence about linear when time value and preceding time value
Function;
Second function establishes unit, for establish the history journey time sequence about it is described when time value and preceding time value with
The linear function of chance error difference;
First model foundation unit, for according to the history journey time sequence about linear when time value and preceding time value
Function, the history journey time sequence are about linear function, the difference rank when time value and preceding time value random error
Several and lag operator establishes prediction model;The lag operator is to convert the preceding time value to the operator for working as time value;
Model order determination unit determines the model order of the prediction model for utilizing trial and error procedure;
Model coefficient determination unit, for determining the autoregressive coefficient of the prediction model using the parameter fitting sequence
With rolling average coefficient;
Second model foundation unit, for that will determine the model order, the autoregressive coefficient and the rolling average
Prediction model after coefficient is determined as autoregression integral moving average model.
Further, judgment module specifically includes:
Predicted value acquiring unit, for integrating moving average model to each in the model testing sequence using autoregression
The time point Travel Time for Public Transport Vehicles in the corresponding default public bus network section is predicted, predicted value is obtained;
First indicator calculating unit, for according to the history journey time in the predicted value and the model testing sequence
Calculate maximum relative error prediction index;
Second indicator calculating unit, for calculating average absolute percentage error according to the maximum relative error prediction index
Prediction index;
Index judging unit, for judging that the maximum relative error prediction index and the average absolute percentage error are pre-
Index is surveyed whether within the scope of default progress;If so, not needing to adjust the ginseng that the autoregression integrates moving average model
Number;If it is not, then needing to adjust the parameter of the autoregression integral moving average model.
Beneficial effect:Compared with prior art, the present invention the present invention considers public transit vehicle during operation by random
The journey time fluctuation characteristic that factor is influenced and presented utilizes a large amount of GPS history number of public transit vehicle based on Time-series Theory
According to and by difference processing by after original time series tranquilization, preferentially controlled using ARIMA model realization towards bus signals
The high-precision link travel time prediction of system, and difference processing is utilized, fully consider public transport in the process of running, journey time
Sequence exists by public vehicles, passenger getting on/off station, the interference for stopping the fluctuation of the enchancement factors such as station, can significantly improve model
Precision of prediction in public transport Forecasting of Travel Time field.The journey time of model prediction was public transit vehicle from last node to downstream
The time of intersection faces intersection bus signals priority acccess control, it is intended to provide important data ginseng for the real-time control of signal
It examines, precision of prediction of the present invention is high, can satisfy the public transport Forecasting of Travel Time accuracy requirement of signal-oriented control, goes out for public transport
Row, traffic control provide important references, provide city bus operation Real-Time Scheduling, intersection bus signals priority acccess control etc.
Basic data support.
Detailed description of the invention
It in order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, below will be to institute in embodiment
Attached drawing to be used is needed to be briefly described;
Fig. 1 is a kind of flow chart of the method for establishing model for Forecasting of Travel Time for Public Transport Vehicles of the embodiment of the present invention;
Fig. 2 is a kind of structure chart of the model foundation system for Forecasting of Travel Time for Public Transport Vehicles of the embodiment of the present invention.
Specific embodiment
The inventive method is further described below with reference to embodiment.
The object of the present invention is to provide a kind of method for establishing model and system for Forecasting of Travel Time for Public Transport Vehicles, fill
Divide and consider public transport in the process of running, journey time is by enchancement factors waves such as public vehicles, passenger getting on/off station, stop stations
Dynamic interference significantly improves model in the precision of prediction in public transport Forecasting of Travel Time field.
In order to make the foregoing objectives, features and advantages of the present invention clearer and more comprehensible, with reference to the accompanying drawing and specific real
Applying mode, the present invention is described in further detail.
In order to export meet bus signals priority acccess control required precision Forecasting of Travel Time as a result, the present invention is based on
Time-series Theory integrates sliding average (Auto- using autoregression using a large amount of GPS historical datas of public transit vehicle
Regressive Integrated Moving Average, ARIMA) model, excavates the public transport hidden in GPS historical data and goes out
Row characteristic rule, the enchancement factor during consideration bus trip influences, and is adjusted by continuous parameter, optimal prediction model,
Output reaches precision, the prediction result of robustness requirement, it is subsequent further can acquire and handle by public transit vehicle GPS data divide
Analysis is set for optimization to prediction model and updates.
Fig. 1 is a kind of flow chart of the method for establishing model for Forecasting of Travel Time for Public Transport Vehicles of the embodiment of the present invention.
Referring to Fig. 1, the method for establishing model for Forecasting of Travel Time for Public Transport Vehicles of embodiment, including:
Step 101:Obtain the public transport GPS data in default public bus network section.
The public transport GPS data includes public transport reservation temporal information and public transport location information.
Step 102:Determine the beginning and end in the default public bus network section.
Step 103:It is determined according to the public transport GPS data and puts rising for the default public bus network section in different times
Point moment.
Step 104:The start time that public bus network section is preset described in each time point is determined according to the start time
Corresponding end of time.
Step 105:History journey time sequence is established according to the start time and the end of time.
The history journey time sequence includes the history journey time at multiple and different time points, and each time point is gone through
History journey time is determined by the corresponding start time of each time point and end of time corresponding with start time.
It specifically includes:
The history journey time of various time points is obtained, specially
yt=Tte-Tts (1)
Wherein, t indicates time point, TtsIndicate t-th of time point corresponding start time, TteIndicate t-th of time point pair
The end of time answered, ytFor the history journey time at t-th of time point;
History journey time according to various time points establishes history journey time sequence, specially
Yt={ y1,y2,y3,....}
YtIndicate the history journey time sequence in default public bus network section.
Step 106:The auto-correlation coefficient of the history journey time sequence is calculated, auto-correlation coefficient can be described with for the moment
Between sequence different time degree of correlation, specially:
Calculate history journey time sequence YtAuto-correlation coefficient ρ:
Wherein, r (t, t+k) is history journey time sequence YtDelay length k (k=1,2,3 ..., n) auto-covariance
Function,
R (t, t+k)=E (Yt-EYt)(Yt+k-EYt+k) (3)
ρ (t, t+k) is history journey time sequence YtDelay length k auto-correlation function, E, D are respectively YtThe mathematics phase
It hopes and variance, the auto-covariance function and auto-correlation function only depend on the delay length k of time, and the start-stop point with the time
It is unrelated.
Step 107:Judge whether the history journey time sequence is steady.Specially:
History journey time sequence Y is judged according to auto-correlation coefficient ρtStationarity, stationary sequence usually has short-term phase
Guan Xing, i.e. the auto-correlation coefficient ρ of stationary sequence can decay to rapidly 0 with the increase of delay length k, then the wave near 0
It is dynamic and gradually converge on 0, determine the threshold values of delay length k, determine any k, can use 5, with k from 1 rise to 5 during,
If the value of ρ (t, t+k) levels off to 0 quickly, then the time series YtFor stationary sequence, directly execution step 109;Otherwise, YtFor
Non-stationary series then follow the steps 108, after converting stationary sequence for non-stationary series, then execute step 109.
Step 108:Difference processing is carried out to history journey time sequence, and determines difference order.
The difference order d is that may make Y after d differencetIt is converted into stationary sequence by non-stationary series, according to reality
After border experience, d desirable 1 or 2, difference processing and determining difference order d, then execute step 109.
In time series analysis, it is desirable that studied time series is smoothly, otherwise will to generate " shadowing property ", lead
Cause prediction result insincere.However time series in practice be all often it is jiggly, difference processing be obtain stationary time
The common method of sequence, some sequences can be steady by first difference processing, and some sequences are but needed by multi-difference
Reason just becomes steady.
Step 109:Determine parameter fitting sequence and model testing sequence.
The history journey time of first preset ratio in history journey time sequence is determined as parameter fitting sequence, will be gone through
The history journey time of the second preset ratio is determined as model testing sequence in history journey time sequence, and parameter fitting sequence is used for
Determine model parameter, reliability of the model testing sequence for testing model output result.The first default ratio in the present embodiment
Example is 90%, and the second preset ratio is 10%.
Step 110:Autoregression integral moving average model is established according to parameter fitting sequence.It specifically includes:
Determine the autoregression item of model, autoregression item is history journey time sequence about linear when time value and preceding time value
Function, i.e., the history journey time y at t-th time pointtPreceding time value y can be usedt-1、yt-2... it indicates, then
Real parameterFor autoregressive coefficient, and parameter to be estimated, utFor stochastic error, p indicates first
Model order;
Determine the rolling average item of model, rolling average item be history journey time sequence about when time value and preceding time value with
The linear function of chance error difference, i.e., the history journey time y at t-th time pointtIt is represented by:
yt=ut-θ1ut-1+θ2ut-2+...+θqut-q (5)
Real parameter θ1,θ2,...,θqIt is the parameter to be estimated of model for rolling average coefficient, q indicates the second model order;
Lag operator B is introduced, which is to switch to the preceding time value of Travel Time for Public Transport Vehicles to work as time value, i.e., by the
The journey time at t-k time point switchs to the journey time y at t-th of time pointt,
Bkyt=yt-k (6)
Then, autoregression item can be abbreviated as
Rolling average item can be abbreviated as
yt=θ (B) ut (9)
θ (B)=1- θ1B-θ2B2-...-θqBq (10)
Prediction model is established according to autoregression item, rolling average item, difference order d and lag operator B, specially:
Determine the linear function of prediction model, the Forecasting of Travel Time value at the T time point may also indicate that for its with
The linear function is determined as the linear letter of prediction model by linear function composed by the random error of preceding time value and early period
Number, i.e.,
The linear function of prediction model is the ARMA model of (p, q) rank, according to the definition of model, is determined pre-
Model is surveyed, lag operator B and difference order d is introduced on the basis of the linear function of prediction model, then predicts the T time point
Journey time yTPrediction model can be abbreviated as
Using trial and error procedure, determine that the model order of prediction model, model order include the first model order p and the second model
Order q, specially:P=n, q=n-1 are enabled, using trial and error procedure, the first model order p and the second model order are determined by n=1
Number q;
The autoregressive coefficient of prediction model is determined using parameter fitting sequenceWith rolling average coefficient θ, specially:It is based on
Parameter fitting sequence, and model parameter to be estimated is determined using MATLAB or SPSS softwareAnd θ;
It will determine model order d, autoregressive coefficientIt is determined as autoregression product with the prediction model after rolling average coefficient θ
Divide moving average model.
Step 111:Judge whether the parameter of autoregression integral moving average model needs to adjust.
In the step, judge whether the parameter of autoregression integral moving average model needs to adjust using model testing sequence
Whole, if so, thening follow the steps 112, if it is not, thening follow the steps 113, the parameter includes model order, autoregressive coefficient and shifting
Dynamic mean coefficient.
Step 111 specifically includes:
Utilize autoregression integral moving average model default public transport line corresponding to various time points in model testing sequence
The Travel Time for Public Transport Vehicles in road section is predicted, predicted value is obtained
Maximum relative error prediction is calculated according to the history journey time in the predicted value and the model testing sequence
Index RE, specially
Wherein, y'TFor the history journey time at the T time point in checking sequence,For the row at corresponding the T time point
Journey temporal predictive value, N indicate forecast sample number;
Average absolute percentage error prediction index MAPE is calculated according to maximum relative error prediction index, specially
Judge maximum relative error prediction index and average absolute percentage error prediction index whether in default progress model
In enclosing, specially:Judge whether average absolute percentage error prediction index RE meets RE<15% and average absolute percentage error it is pre-
Survey whether index MAPE meets MAPE<15%;
If so, not needing the parameter of adjustment autoregression integral moving average model, directly execution step 113;
If it is not, then needing to adjust the parameter of autoregression integral moving average model, step 112 is executed.
Step 112:The parameter of Regression-Integral moving average model is adjusted, parameter autoregression adjusted is integrated
Moving average model is determined as Forecasting of Travel Time for Public Transport Vehicles model, specially:Readjust the first model order p, second
Model order q enables n=n+1, updates prediction model.When Forecasting of Travel Time for Public Transport Vehicles model is used for public transit vehicle stroke
Between predicted.
Step 113:Autoregression integral moving average model is then determined as Forecasting of Travel Time for Public Transport Vehicles model.Public transport
Vehicle travel time prediction model directly exports the first model order p, the second model order q and predicted value.
In this implementation, after the public transport GPS data for obtaining default public bus network section, further include:It is default public to obtaining
The public transport GPS data in intersection road section is pre-processed, and pretreatment includes data rejecting, map match and projection correction, in this way
It ensure that the accuracy of information of the vehicle in default public bus network section on map, including be accurately located temporal information and standard
True location information.
The method for establishing model for Forecasting of Travel Time for Public Transport Vehicles in the present embodiment, it is contemplated that public transit vehicle is being transported
The journey time fluctuation characteristic for being influenced and being presented by enchancement factor during battalion is based on Time-series Theory, utilizes public transit vehicle
A large amount of GPS historical data and by difference processing by after original time series tranquilization, using ARIMA model realization towards
The high-precision link travel time prediction of bus signals priority acccess control, precision of prediction is high, can satisfy the public affairs of signal-oriented control
Forecasting of Travel Time accuracy requirement is handed over, provides important references for bus trip, traffic control.
The present invention also provides a kind of model foundation system for Forecasting of Travel Time for Public Transport Vehicles, Fig. 2 is the present invention
A kind of model foundation system for Forecasting of Travel Time for Public Transport Vehicles structure chart, which includes:
Data acquisition module 201, for obtaining the public transport GPS data in default public bus network section;The public transport GPS data
Including public transport reservation temporal information and public transport location information.
First determining module 202, for determining the beginning and end in the default public bus network section.
Second determining module 203, for putting the default public transport in different times according to public transport GPS data determination
The start time in route section.
Third determining module 204 presets public bus network road for determining according to the start time described in each time point
The corresponding end of time of start time of section.
Time series establishes module 205, when for establishing history stroke according to the start time and the end of time
Between sequence;The history journey time sequence includes the history journey time at multiple and different time points, the institute of each time point
History journey time is stated to be determined by the corresponding start time of each time point and end of time corresponding with the start time.
The time series establishes module 205, specifically includes:
Time acquisition unit, for obtaining the history journey time of various time points, specially
yt=Tte-Tts (1)
Wherein, t indicates time point, TtsIndicate t-th of time point corresponding start time, TteIndicate t-th of time point pair
The end of time answered, ytFor the history journey time at t-th of time point;
Sequence establishes unit, establishes history journey time sequence for the history journey time according to various time points, has
Body is:
Yt={ y1,y2,y3,...}
YtIndicate the history journey time sequence in default public bus network section.
Computing module 206, for calculating the auto-correlation coefficient of the history journey time sequence.
Steady judgment module 207, for judging whether the history journey time sequence puts down according to the auto-correlation coefficient
Surely.
Differential processing module 208 is used for if it is not, then carrying out difference processing to the history journey time sequence, and determine
Difference order.
4th determining module 209, for will be in the history journey time sequence when history stroke of the first preset ratio
Between be determined as parameter fitting sequence, the history journey time of the second preset ratio in the history journey time sequence is determined as
Model testing sequence.
Model building module 210, for establishing autoregression integral moving average model according to the parameter fitting sequence.
The model building module 210, specifically includes:
First function establishes unit, for establishing the history journey time sequence about linear when time value and preceding time value
Function;
Second function establishes unit, for establish the history journey time sequence about it is described when time value and preceding time value with
The linear function of chance error difference;
First model foundation unit, for according to the history journey time sequence about linear when time value and preceding time value
Function, the history journey time sequence are about linear function, the difference rank when time value and preceding time value random error
Several and lag operator establishes prediction model;The lag operator is to convert the preceding time value to the operator for working as time value;
Model order determination unit determines the model order of the prediction model for utilizing trial and error procedure;
Model coefficient determination unit, for determining the autoregressive coefficient of the prediction model using the parameter fitting sequence
With rolling average coefficient;
Second model foundation unit, for that will determine the model order, the autoregressive coefficient and the rolling average
Prediction model after coefficient is determined as autoregression integral moving average model.
Judgment module 211, for judging the autoregression integral moving average model using the model testing sequence
Whether parameter, which needs, adjusts;The parameter includes model order, autoregressive coefficient and rolling average coefficient.
The judgment module 211, specifically includes:
Predicted value acquiring unit, for integrating moving average model to each in the model testing sequence using autoregression
The time point Travel Time for Public Transport Vehicles in the corresponding default public bus network section is predicted, predicted value is obtained;
First indicator calculating unit, for according to the history journey time in the predicted value and the model testing sequence
Calculate maximum relative error prediction index;
Second indicator calculating unit, for calculating average absolute percentage error according to the maximum relative error prediction index
Prediction index;
Index judging unit, for judging that the maximum relative error prediction index and the average absolute percentage error are pre-
Index is surveyed whether within the scope of default progress;If so, not needing to adjust the ginseng that the autoregression integrates moving average model
Number;If it is not, then needing to adjust the parameter of the autoregression integral moving average model.
Prediction model determining module 212, for if so, being adjusted to the parameter of the Regression-Integral moving average model
It is whole, parameter autoregression integral moving average model adjusted is determined as Forecasting of Travel Time for Public Transport Vehicles model;The public affairs
Hand over vehicle travel time prediction model for predicting Travel Time for Public Transport Vehicles;If it is not, then the autoregression is integrated
Moving average model is determined as Forecasting of Travel Time for Public Transport Vehicles model.
The model foundation system for Forecasting of Travel Time for Public Transport Vehicles in the present embodiment is based on Time-series Theory,
Using ARIMA model, and difference processing is utilized, fully considered public transport in the process of running, journey time sequence exists by society
Meeting vehicle, passenger getting on/off station, the interference for stopping the fluctuation of the enchancement factors such as station, can significantly improve model in public transport stroke
Between predict field precision of prediction.The journey time of model prediction be public transit vehicle from last node to downstream intersection when
Between, face intersection bus signals priority acccess control, it is intended to provide important data reference for the real-time control of signal.
For the system disclosed in the embodiment, since it is corresponded to the methods disclosed in the examples, so the ratio of description
Relatively simple, reference may be made to the description of the method.
Claims (10)
1. a kind of method for establishing model for Forecasting of Travel Time for Public Transport Vehicles, which is characterized in that include the following steps:
Obtain the public transport GPS data in default public bus network section;
Determine the beginning and end in the default public bus network section;
The start time for putting default public bus network section in different times is determined according to the public transport GPS data;
Determine that each time point presets the corresponding end of time of start time in public bus network section according to the start time;
History journey time sequence is established according to start time and end of time;The history journey time sequence includes multiple and different
Time point history journey time, the history journey time of each time point by the corresponding start time of each time point and with
The corresponding end of time of the start time determines;
The history journey time of first preset ratio in the history journey time sequence is determined as parameter fitting sequence, by institute
The history journey time for stating the second preset ratio in history journey time sequence is determined as model testing sequence;
Autoregression integral moving average model is established according to the parameter fitting sequence;
Judge whether the parameter of the autoregression integral moving average model needs to adjust using the model testing sequence;It is described
Parameter includes model order, autoregressive coefficient and rolling average coefficient;
If so, being adjusted to the parameter of the Regression-Integral moving average model, parameter autoregression adjusted is integrated
Moving average model is determined as Forecasting of Travel Time for Public Transport Vehicles model;The Forecasting of Travel Time for Public Transport Vehicles model for pair
Travel Time for Public Transport Vehicles is predicted;
If it is not, autoregression integral moving average model is then determined as Forecasting of Travel Time for Public Transport Vehicles model.
2. a kind of method for establishing model for Forecasting of Travel Time for Public Transport Vehicles according to claim 1, feature exist
In foundation start time and end of time establish history journey time sequence, specifically include:
The history journey time of various time points is obtained, specially
yt=Tte-Tts (1)
Wherein, t indicates time point, TtsIndicate t-th of time point corresponding start time, TteIndicate that t-th of time point is corresponding
End of time, ytFor the history journey time at t-th of time point;
History journey time according to various time points establishes history journey time sequence, specially
Yt={ y1,y2,y3,...}
YtIndicate the history journey time sequence in default public bus network section.
3. a kind of method for establishing model for Forecasting of Travel Time for Public Transport Vehicles according to claim 1 or 2, feature
It is, after establishing history journey time sequence according to start time and end of time, further includes:
Calculate the auto-correlation coefficient of history journey time sequence;
Judge whether history journey time sequence is steady according to auto-correlation coefficient;
If it is not, then carrying out difference processing to history journey time sequence, and determine difference order.
4. a kind of method for establishing model for Forecasting of Travel Time for Public Transport Vehicles according to claim 3, feature exist
In foundation parameter fitting sequence establishes autoregression integral moving average model, specifically includes:
History journey time sequence is established about the linear function for working as time value and preceding time value;
History journey time sequence is established about the linear function for working as time value and preceding time value random error;
According to history journey time sequence about when the linear function of time value and preceding time value, history journey time sequence are about current
Linear function, difference order and the lag operator of value and preceding time value random error establish prediction model;The lag operator is
The operator for working as time value is converted by the preceding time value;
Using trial and error procedure, the model order of the prediction model is determined;
The autoregressive coefficient and rolling average coefficient of the prediction model are determined using the parameter fitting sequence;
Prediction model after determining model order, autoregressive coefficient and rolling average coefficient is determined as autoregression integral sliding
Averaging model.
5. a kind of method for establishing model for Forecasting of Travel Time for Public Transport Vehicles according to claim 1 or 4, feature
It is, judges whether the parameter of autoregression integral moving average model needs to adjust using model testing sequence, specifically include:
Using autoregression integral moving average model to the corresponding default public transport line of various time points in the model testing sequence
The Travel Time for Public Transport Vehicles in road section is predicted, predicted value is obtained;
Maximum relative error prediction index is calculated according to the history journey time in predicted value and model testing sequence;
Average absolute percentage error prediction index is calculated according to maximum relative error prediction index;
Judge maximum relative error prediction index and average absolute percentage error prediction index whether within the scope of default progress;
If so, not needing the parameter of adjustment autoregression integral moving average model;
If it is not, then needing to adjust the parameter of autoregression integral moving average model.
6. a kind of model foundation system for Forecasting of Travel Time for Public Transport Vehicles, which is characterized in that including:
Data acquisition module, for obtaining the public transport GPS data in default public bus network section;
First determining module, for determining the beginning and end in default public bus network section;
Second determining module, for determining the starting point for putting default public bus network section in different times according to public transport GPS data
Moment;
Third determining module, for determining that the start time in the default public bus network section of each time point is corresponding according to start time
End of time;
Time series establishes module, for establishing history journey time sequence according to start time and end of time;The history
Journey time sequence includes the history journey time at multiple and different time points, the history journey time of each time point by
The corresponding start time of each time point and end of time corresponding with the start time determine;
4th determining module, for the history journey time of the first preset ratio in the history journey time sequence to be determined as
The history journey time of second preset ratio in the history journey time sequence is determined as model testing by parameter fitting sequence
Sequence;
Model building module, for establishing autoregression integral moving average model according to the parameter fitting sequence;
Judgment module, for judged using the model testing sequence autoregression integral moving average model parameter whether
It needs to adjust;The parameter includes model order, autoregressive coefficient and rolling average coefficient;
Prediction model determining module, for will join if so, be adjusted to the parameter of the Regression-Integral moving average model
Number autoregression integral moving average model adjusted is determined as Forecasting of Travel Time for Public Transport Vehicles model;The public transit vehicle row
Journey time prediction model is for predicting Travel Time for Public Transport Vehicles;If it is not, the autoregression is then integrated sliding average
Model is determined as Forecasting of Travel Time for Public Transport Vehicles model.
7. a kind of model foundation system for Forecasting of Travel Time for Public Transport Vehicles according to claim 6, feature exist
In the time series establishes module, specifically includes:
Time acquisition unit, for obtaining the history journey time of various time points;
Sequence establishes unit, establishes history journey time sequence for the history journey time according to various time points.
8. a kind of model foundation system for Forecasting of Travel Time for Public Transport Vehicles according to claim 6, feature exist
In further including:
Computing module, for calculating the auto-correlation coefficient of the history journey time sequence;
Steady judgment module, for judging whether the history journey time sequence is steady according to the auto-correlation coefficient;
Differential processing module is used for if it is not, then carrying out difference processing to the history journey time sequence, and determine difference rank
Number.
9. a kind of model foundation system for Forecasting of Travel Time for Public Transport Vehicles according to claim 6, feature exist
In the model building module specifically includes:
First function establishes unit, for establishing the history journey time sequence about the linear letter for working as time value and preceding time value
Number;
Second function establishes unit, for establishing the history journey time sequence about described when time value and preceding time value miss at random
The linear function of difference;
First model foundation unit, for according to the history journey time sequence about work as time value and preceding time value linear letter
Several, the described history journey time sequence is about linear function, the difference order when time value and preceding time value random error
And lag operator establishes prediction model;The lag operator is to convert the preceding time value to the operator for working as time value;
Model order determination unit determines the model order of the prediction model for utilizing trial and error procedure;
Model coefficient determination unit, for determining autoregressive coefficient and the shifting of the prediction model using the parameter fitting sequence
Dynamic mean coefficient;
Second model foundation unit, for that will determine the model order, the autoregressive coefficient and the rolling average coefficient
Prediction model afterwards is determined as autoregression integral moving average model.
10. a kind of model foundation system for Forecasting of Travel Time for Public Transport Vehicles according to claim 6, feature exist
In the judgment module specifically includes:
Predicted value acquiring unit, for integrating moving average model to each time in the model testing sequence using autoregression
The Travel Time for Public Transport Vehicles in the corresponding default public bus network section of point is predicted, predicted value is obtained;
First indicator calculating unit, for being calculated according to the history journey time in the predicted value and the model testing sequence
Maximum relative error prediction index;
Second indicator calculating unit, for calculating average absolute percentage error prediction according to the maximum relative error prediction index
Index;
Index judging unit, for judging that the maximum relative error prediction index and average absolute percentage error prediction refer to
Whether mark is within the scope of default progress;If so, not needing to adjust the parameter that the autoregression integrates moving average model;If
It is no, then need to adjust the parameter of the autoregression integral moving average model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810418625.8A CN108831181A (en) | 2018-05-04 | 2018-05-04 | A kind of method for establishing model and system for Forecasting of Travel Time for Public Transport Vehicles |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810418625.8A CN108831181A (en) | 2018-05-04 | 2018-05-04 | A kind of method for establishing model and system for Forecasting of Travel Time for Public Transport Vehicles |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108831181A true CN108831181A (en) | 2018-11-16 |
Family
ID=64148342
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810418625.8A Pending CN108831181A (en) | 2018-05-04 | 2018-05-04 | A kind of method for establishing model and system for Forecasting of Travel Time for Public Transport Vehicles |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108831181A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109584552A (en) * | 2018-11-28 | 2019-04-05 | 青岛大学 | A kind of public transport arrival time prediction technique based on network vector autoregression model |
CN112132485A (en) * | 2020-09-30 | 2020-12-25 | 上海众源网络有限公司 | Index data processing method and device, electronic equipment and storage medium |
CN112489418A (en) * | 2020-10-22 | 2021-03-12 | 浙江交通职业技术学院 | Road section travel time dynamic error correction method based on road section travel time prediction model |
CN113393663A (en) * | 2020-10-22 | 2021-09-14 | 浙江交通职业技术学院 | Rolling prediction method for road path travel time |
CN117272847A (en) * | 2023-11-22 | 2023-12-22 | 中控技术股份有限公司 | Method and device for predicting temperature data of heating rod |
CN117787568A (en) * | 2024-02-23 | 2024-03-29 | 深圳市南粤建工有限公司 | Intelligent monitoring and early warning method for road construction progress |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102708701A (en) * | 2012-05-18 | 2012-10-03 | 中国科学院信息工程研究所 | System and method for predicting arrival time of buses in real time |
CN103903429A (en) * | 2014-03-26 | 2014-07-02 | 东南大学 | Fast bus stop delay time combination predicting method |
CN106652497A (en) * | 2017-02-17 | 2017-05-10 | 华东师范大学 | Scheduling method for intelligent transportation by fusing self-regression forecast model |
-
2018
- 2018-05-04 CN CN201810418625.8A patent/CN108831181A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102708701A (en) * | 2012-05-18 | 2012-10-03 | 中国科学院信息工程研究所 | System and method for predicting arrival time of buses in real time |
CN103903429A (en) * | 2014-03-26 | 2014-07-02 | 东南大学 | Fast bus stop delay time combination predicting method |
CN106652497A (en) * | 2017-02-17 | 2017-05-10 | 华东师范大学 | Scheduling method for intelligent transportation by fusing self-regression forecast model |
Non-Patent Citations (1)
Title |
---|
高金喆: "基于公交GPS数据的交叉口公交优先信号控制关键技术研究", 《工程科技Ⅱ辑》 * |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109584552A (en) * | 2018-11-28 | 2019-04-05 | 青岛大学 | A kind of public transport arrival time prediction technique based on network vector autoregression model |
CN109584552B (en) * | 2018-11-28 | 2021-04-30 | 青岛大学 | Bus arrival time prediction method based on network vector autoregressive model |
CN112132485A (en) * | 2020-09-30 | 2020-12-25 | 上海众源网络有限公司 | Index data processing method and device, electronic equipment and storage medium |
CN112489418A (en) * | 2020-10-22 | 2021-03-12 | 浙江交通职业技术学院 | Road section travel time dynamic error correction method based on road section travel time prediction model |
CN113393663A (en) * | 2020-10-22 | 2021-09-14 | 浙江交通职业技术学院 | Rolling prediction method for road path travel time |
CN117272847A (en) * | 2023-11-22 | 2023-12-22 | 中控技术股份有限公司 | Method and device for predicting temperature data of heating rod |
CN117787568A (en) * | 2024-02-23 | 2024-03-29 | 深圳市南粤建工有限公司 | Intelligent monitoring and early warning method for road construction progress |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108831181A (en) | A kind of method for establishing model and system for Forecasting of Travel Time for Public Transport Vehicles | |
CN107610464A (en) | A kind of trajectory predictions method based on Gaussian Mixture time series models | |
CN104215249B (en) | Smoothening method of driving track | |
CN109544908B (en) | Method for predicting bus arrival time in real time, electronic equipment and storage medium | |
Hofleitner et al. | Learning the dynamics of arterial traffic from probe data using a dynamic Bayesian network | |
CN109191845B (en) | Bus arrival time prediction method | |
CN102087788B (en) | Method for estimating traffic state parameter based on confidence of speed of float car | |
US7953544B2 (en) | Method and structure for vehicular traffic prediction with link interactions | |
CN104573849A (en) | Bus dispatch optimization method for predicting passenger flow based on ARIMA model | |
CN107284452B (en) | Merge the hybrid vehicle future operating mode forecasting system of intelligent communication information | |
CN107346460A (en) | Following operating mode Forecasting Methodology based on the lower front truck operation information of intelligent network contact system | |
CN103903430A (en) | Dynamic fusion type travel time predicting method with multi-source and isomorphic data adopted | |
CN105427594A (en) | Bidirectional-getting-on-passenger-flow-based obtaining method and system for bus section passenger flow volume | |
CN105512741A (en) | Bus passenger traffic combined prediction method | |
CN108615391A (en) | A kind of intelligent bus dispatching method, system and medium | |
CN104318759B (en) | Bus stop station time real-time estimation method based on self-learning algorithm and system | |
CN111746538B (en) | Strict collision avoidance vehicle queue following control method and control system | |
CN104679989A (en) | Hydrogen atom clock error predicting method based on modified BP (back propagation) neural network | |
CN104900061A (en) | Road section travel time monitoring method and device | |
CN108263400B (en) | High-speed rail train running speed control method, device, storage medium and high-speed rail train | |
CN104599500A (en) | Grey entropy analysis and Bayes fusion improvement based traffic flow prediction method | |
CN109859515A (en) | GPS positioning compensation data method and electronic equipment in public transit system | |
CN116007647A (en) | Routing inspection path planning method and device, electronic equipment and storage medium | |
CN103685014B (en) | Time series predicting model is utilized to strengthen the system and method for router-level topology reliability | |
CN105303835B (en) | A kind of Forecasting Approach for Short-term of road traffic stream mode |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181116 |