CN111145537B - Travel generation amount prediction method and system - Google Patents

Travel generation amount prediction method and system Download PDF

Info

Publication number
CN111145537B
CN111145537B CN201911215218.8A CN201911215218A CN111145537B CN 111145537 B CN111145537 B CN 111145537B CN 201911215218 A CN201911215218 A CN 201911215218A CN 111145537 B CN111145537 B CN 111145537B
Authority
CN
China
Prior art keywords
travel
model
attraction
sequence
production
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.)
Active
Application number
CN201911215218.8A
Other languages
Chinese (zh)
Other versions
CN111145537A (en
Inventor
陆建
王成晨
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Southeast University
Original Assignee
Southeast University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Southeast University filed Critical Southeast University
Priority to CN201911215218.8A priority Critical patent/CN111145537B/en
Publication of CN111145537A publication Critical patent/CN111145537A/en
Application granted granted Critical
Publication of CN111145537B publication Critical patent/CN111145537B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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
    • 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
    • G08G1/0108—Measuring and analyzing of parameters relative to traffic conditions based on the source of data
    • G08G1/012—Measuring and analyzing of parameters relative to traffic conditions based on the source of data from other sources than vehicle or roadside beacons, e.g. mobile networks

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Theoretical Computer Science (AREA)
  • Chemical & Material Sciences (AREA)
  • Computational Mathematics (AREA)
  • Analytical Chemistry (AREA)
  • Mathematical Optimization (AREA)
  • Mathematical Analysis (AREA)
  • Evolutionary Biology (AREA)
  • Operations Research (AREA)
  • Probability & Statistics with Applications (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Algebra (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Databases & Information Systems (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a travel production prediction method and a travel production prediction system. The method comprises the following steps: acquiring user travel data from travel software; screening the travel production and the travel attraction at different time points from the travel data to generate a travel production time sequence and a travel attraction time sequence; dividing the travel production time sequence and the travel attraction time sequence into two sections according to time to respectively obtain a parameter fitting sequence and a model checking sequence; establishing an autoregressive moving average model by using a parameter fitting sequence; judging whether the autoregressive moving average model needs to be adjusted or not by using a model test sequence; if so, adjusting parameters of the autoregressive moving average model by using the model test sequence, and determining the adjusted autoregressive moving average model as a prediction model; and predicting the trip production amount and trip attraction amount at the future moment by using the prediction model to obtain the predicted trip generation amount. The method and the system can improve the convenience degree of travel generation amount prediction.

Description

Travel generation amount prediction method and system
Technical Field
The invention relates to the field of traffic prediction, in particular to a travel production prediction method and system.
Background
The resident travel generation amount prediction is the first step of traffic demand prediction, the prediction result directly restricts the prediction precision of each subsequent stage, and the urban traffic planning scheme is significantly influenced.
At present, traffic survey is the basis of generation and prediction of resident trip amount, and traditional resident trip survey methods are mature, but the methods generally have the defects of high survey cost, high working strength, low data accuracy and the like. In the process of urbanization and motorized rapid propulsion in China, the traveling behaviors of urban residents are often influenced by various novel traffic modes, and local and even wider changes occur in a short period, so that the data of traffic investigation does not have the requirement of real-time property. The passive survey, namely survey travelers carry equipment such as a GPS and the like to passively record travel tracks, although survey data meet the real-time requirement, the survey data can be completed only by depending on the cooperation of the equipment such as the GPS and the travelers, and the convenience degree of the travel output prediction of residents is greatly reduced.
Disclosure of Invention
The invention aims to provide a travel generation amount prediction method and a travel generation amount prediction system, which improve the convenience degree of travel generation amount prediction.
In order to achieve the purpose, the invention provides the following scheme:
a travel production prediction method comprises the following steps:
step 101: acquiring user travel data from travel software;
step 102: screening the travel production and travel attraction at different time points from the travel data to generate a travel production time sequence and a travel attraction time sequence; the travel generation amount comprises travel production amount and travel attraction amount;
the travel production is the sum of all home endpoints of home travel and all starting points of non-home travel and cargo travel; the trip attraction amount is the sum of all non-family endpoints of the home trip and all endpoints of the non-home trip and the cargo trip; wherein
The trip production time series is represented as:
Xt＝{x1，x2，x3，...}
the trip attraction time series is represented as:
Yt＝{y1，y2，，y3，...}
wherein t represents a time point, xtIs the travel production at the t-th time point, XtRepresenting a time series of travel production, ytIs the travel attraction at the t-th time point, YtRepresenting a trip attraction time sequence;
step 103: dividing the travel production time sequence and the travel attraction time sequence into two sections according to time to respectively obtain a parameter fitting sequence and a model checking sequence;
step 104: calculating autocorrelation coefficients of the travel production time series and the travel attraction time series;
step 105: judging whether the travel production time sequence and the travel attraction time sequence are stable or not according to the autocorrelation coefficient to obtain a second judgment result;
step 106: performing at least one difference processing on the travel production time series and the travel attraction time series until the travel production time series and the travel attraction time series become stationary sequences; determining a difference order after difference processing;
step 107: establishing an autoregressive moving average model by using the parameter fitting sequence; the travel generation amount time sequence has certain periodicity, continuity and stability, and the future moment generation amount time sequence can be predicted by fitting the earlier moment generation amount time sequence;
step 108: judging whether the parameters of the autoregressive moving average model need to be adjusted or not by using the model test sequence to obtain a first judgment result;
step 109: adjusting parameters of the autoregressive moving average model by using the model test sequence, and determining the adjusted autoregressive moving average model as a prediction model;
step 110: determining the autoregressive moving average model as a prediction model;
step 111: and predicting the trip production amount and trip attraction amount at the future moment by using the prediction model to obtain the predicted trip generation amount.
The invention is further improved in that: in step 104, the autocorrelation coefficient calculation formula is:
Figure BDA0002299314100000031
Figure BDA0002299314100000032
wherein r isx(t，t+k)＝E(Xt-EXt)(Xt+k-EXt+k) Is XtIs the delay k (k ═ 1, 2, 3.., n) autocovariance function, ρ, ofx(t, t + k) is XtIs the delay k autocorrelation function, ry(t，t+k)＝E(Yt-EYt)(Yt+k-EYt+k) Is YtIs the delay k (k ═ 1, 2, 3.., n) autocovariance function, ρ, ofy(t, t + k) is YtE, D are the mathematical expectation and variance, respectively, and the autocovariance function and the autocorrelation function depend only on the delay length k of time, and are independent of the start and end points of time.
The invention is further improved in that: in the step 105, X is determined according to the autocorrelation coefficienttAnd YtThe stationary sequence usually has short-term correlation, that is, the autocorrelation coefficient of the stationary sequence rapidly decays to 0 with the increase of the delay period number k, then fluctuates around 0 and gradually converges to 0, the threshold value of the delay period number k is determined, any k is determined, and 5 can be taken, and if the value of ρ (t, t + k) approaches to 0 quickly in the process of increasing k from 1 to 5, the time sequence X can be taken astAnd YtFor the stationary sequence, step 107 is performed directly; otherwise, XtAnd YtIf the sequence is a non-stationary sequence, step 106 is executed, and step 107 is executed after the non-stationary sequence is converted into a stationary sequence.
The invention is further improved in that: the step 107 specifically includes:
a: establishing a linear function between any time point and the early value of the time point;
determining an autoregressive term of the model, wherein the autoregressive term is a linear function of the trip production amount time series and the trip attraction amount time series with respect to a current period value and a previous period value;
b: introducing a difference order and a hysteresis operator, and establishing an autoregressive moving average model on the basis of the linear function;
c: calculating the model order of the autoregressive moving average model by using a trial algorithm;
d: and determining an autoregressive coefficient and a moving average coefficient of the autoregressive moving average model by using the parameter fitting sequence to obtain the well-established autoregressive moving average model.
The invention is further improved in that: the step 108 specifically includes:
a: calculating the model test according to the autoregressive moving average modelTravel generation prediction quantity of each time point in test sequence
Figure BDA0002299314100000033
And travel attraction prediction
Figure BDA0002299314100000034
B: calculating a maximum relative error prediction index of the output and an average absolute percentage error prediction index RE of the output according to the output prediction quantity and the output in the model test sequence; in particular to
Figure 1
Figure 2
Wherein x'TAnd y'TTo examine the trip production and attraction at the T-th time point in the sequence,
Figure BDA0002299314100000043
and
Figure BDA0002299314100000044
generating and predicting values of the attraction amount for the trip corresponding to the T-th time point, wherein N represents the number of predicted samples;
c: calculating a maximum relative error prediction index of the attraction amount and an average absolute percentage error prediction index MAPE of the attraction amount according to the travel attraction prediction amount and the travel attraction amount in the model test sequence, which specifically comprises the following steps:
Figure 3
Figure BDA0002299314100000046
d: judging whether the maximum generated quantity relative error prediction index, the average generated quantity absolute percentage error prediction index, the maximum attraction quantity relative error prediction index and the average attraction quantity absolute percentage error prediction index are all in corresponding preset ranges to obtain a third judgment result; the method specifically comprises the following steps:
judging whether the average absolute percentage error prediction index RE meets RE < 15% and whether the average absolute percentage error prediction index MAPE meets MAPE < 15%;
e: if the third judgment result shows that the parameter of the autoregressive moving average model does not need to be adjusted; step 110 is executed;
f: if the third judgment result shows no, determining that the parameters of the autoregressive moving average model need to be adjusted; step 109 is performed.
The invention is further improved in that: the trip software is mainly mobile phone trip software and can be one of a dripping trip APP, a clicking trip APP or other mobile phone trip software, and the user trip data mainly comprises trip time and position information; the travel time and location information may be used to determine the time of each travel start and end.
A travel generation amount prediction system comprising the following:
a data obtaining module 201, configured to obtain user travel data from travel software;
a screening module 202, configured to screen travel production amounts and travel attraction amounts at different time points from the travel data, and generate a travel production amount time series and a travel attraction amount time series;
the dividing module 203 is configured to divide the trip production amount time sequence and the trip attraction amount time sequence into two segments according to time, so as to obtain a parameter fitting sequence and a model checking sequence respectively;
a correlation coefficient calculating module 204, configured to calculate an autocorrelation coefficient of the travel production time series and the travel attraction time series;
a second judging module 205, configured to judge whether the travel production time series and the travel attraction time series are stable according to the autocorrelation coefficient, so as to obtain a second judgment result;
a difference processing module 206, configured to, if the second determination result indicates no, perform at least one difference processing on the travel production time series and the travel attraction time series until the travel production time series and the travel attraction time series become stationary series.
A model establishing module 207, configured to establish an autoregressive moving average model using the parameter fitting sequence;
a first judging module 208, configured to judge whether a parameter of the autoregressive moving average model needs to be adjusted by using the model checking sequence, so as to obtain a first judgment result;
a first prediction model determining module 209, configured to, if the first determination result indicates yes, adjust parameters of the autoregressive moving average model by using the model check sequence, and determine the adjusted autoregressive moving average model as a prediction model;
a second prediction model determining module 210, configured to determine the autoregressive moving average model as the prediction model if the first determination result indicates no;
the prediction module 211 is configured to predict the trip production amount and trip attraction amount at a future time by using the prediction model, so as to obtain a predicted trip production amount.
The invention is further improved in that: the model building module 207 includes: the linear function establishing unit is used for establishing a linear function between any time point and the early value of the time point;
the model initial building unit is used for introducing a difference order and a hysteresis operator and building an autoregressive moving average model on the basis of the linear function;
the model order calculation unit is used for calculating the model order of the autoregressive moving average model by using a trial algorithm;
a model coefficient calculation unit for determining the autoregressive sliding horizon using the parameter fit sequence according to a further improvement of the present invention is: the first determining module 208 includes: the prediction quantity calculation unit is used for calculating the travel generation prediction quantity and the travel attraction prediction quantity of each time point in the model test sequence according to the autoregressive moving average model;
the production index calculation unit is used for calculating a maximum relative error prediction index of the production and an average absolute percentage error prediction index of the production according to the trip production prediction quantity and the trip production quantity in the model test sequence;
the attraction amount index calculation unit is used for calculating an attraction amount maximum relative error prediction index and an attraction amount average absolute percentage error prediction index according to the trip attraction prediction amount and the trip attraction amount in the model test sequence;
the third judging unit is used for judging whether the maximum relative error prediction index of the production amount, the average absolute percentage error prediction index of the production amount, the maximum relative error prediction index of the suction amount and the average absolute percentage error prediction index of the suction amount are all in corresponding preset ranges to obtain a third judging result;
a first result determining unit, configured to determine that the parameter of the autoregressive moving average model does not need to be adjusted if the third determination result indicates yes;
and the second result determining unit is used for determining that the parameters of the autoregressive moving average model need to be adjusted if the third judgment result shows no.
The invention discloses the following technical effects: according to the travel generation amount prediction method and system disclosed by the invention, the travel generation amount time sequence and the travel attraction amount time sequence are obtained by utilizing data in travel software, and an autoregressive moving average model is established on the basis, so that the future travel generation amount prediction is directly finished. The method can realize the prediction of the travel generation amount by directly utilizing the data in the travel software without depending on specific GPS equipment, thereby improving the convenience degree of the travel generation amount prediction.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
Fig. 1 is a flowchart of a method of an embodiment of a trip generation amount prediction method according to the present invention;
fig. 2 is a system configuration diagram of the trip generation amount prediction system according to the embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention aims to provide a travel generation amount prediction method and a travel generation amount prediction system, which improve the convenience degree of travel generation amount prediction.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Fig. 1 is a flowchart of a method of predicting trip generation amount according to an embodiment of the present invention.
Referring to fig. 1, the trip generation amount prediction method includes:
step 101: and acquiring user travel data from the travel software. The trip software is mainly mobile phone trip software. Such as drip travel APP, click travel APP, etc. The user trip data mainly comprises trip time and position information. The travel time and location information may be used to determine the time of each travel start and end.
Step 102: and screening the travel production and the travel attraction at different time points from the travel data to generate a travel production time sequence and a travel attraction time sequence. The travel production is the sum of all home endpoints of home travel and all starting points of non-home travel and cargo travel; in other words, the travel production of a certain partition in a unit time is equal to the sum of the number of home trips of the home endpoint in the partition and the number of non-home trips and cargo trips of the starting point in the partition. The trip attraction amount is the sum of all non-family endpoints of the home trip and all endpoints of the non-home trip and the cargo trip; or, the travel attraction amount of a partition in a unit time is equal to the sum of the number of home-trip rows of the non-home terminal in the partition and the number of non-home-trip rows and the number of cargo trip rows of the terminal in the partition.
The trip production time series is represented as:
Xt＝{x1，x2，x3，}
the trip attraction time series is represented as:
Yt＝{y1，y2，，y3，...}
wherein t represents a time point, xtIs the travel production at the t-th time point, XtRepresenting a time series of travel production, ytIs the travel attraction at the t-th time point, YtRepresenting a trip attraction time series.
The travel production amount includes a travel production amount and a travel attraction amount.
Step 103: and dividing the travel production time sequence and the travel attraction time sequence into two sections according to time to respectively obtain a parameter fitting sequence and a model checking sequence. In this embodiment, the travel production amount and the travel suction amount in the first preset proportion are divided into a parameter fitting sequence, and the travel production amount and the travel suction amount in the second preset proportion are divided into a model checking sequence. The first preset proportion is greater than or equal to 3 times of the second preset proportion, and the second preset proportion is the proportion occupied by data except the first preset proportion.
Preferably, the first predetermined proportion is 90% and the second predetermined proportion is 10%.
Step 104: and calculating the autocorrelation coefficients of the travel production time series and the travel attraction time series. The autocorrelation coefficient calculation formula is as follows:
Figure BDA0002299314100000081
Figure BDA0002299314100000082
wherein r isx(t，t+k)＝E(Xt-EXt)(Xt+k-EXt+k) Is XtIs the delay k (k ═ 1, 2, 3.., n) autocovariance function, ρ, ofx(t, t + k) is XtIs the delay k autocorrelation function, ry(t，t+k)＝E(Yt-EYt)(Yt+k-EYt+k) Is YtIs the delay k (k ═ 1, 2, 3.., n) autocovariance function, ρ, ofy(t, t + k) is YtE, D are the mathematical expectation and variance, respectively, and the autocovariance function and the autocorrelation function depend only on the delay length k of time, and are independent of the start and end points of time.
Step 105: and judging whether the travel production time sequence and the travel attraction time sequence are stable or not according to the autocorrelation coefficient to obtain a second judgment result.
Determining X according to the autocorrelation coefficienttAnd YtThe stationary sequence usually has short-term correlation, that is, the autocorrelation coefficient of the stationary sequence rapidly decays to 0 with the increase of the delay period number k, then fluctuates around 0 and gradually converges to 0, the threshold value of the delay period number k is determined, any k is determined, and 5 can be taken, and if the value of ρ (t, t + k) approaches to 0 quickly in the process of increasing k from 1 to 5, the time sequence X can be taken astAnd YtFor the stationary sequence, step 107 is performed directly; otherwise, XtAnd YtIf the sequence is a non-stationary sequence, step 106 is executed, and step 107 is executed after the non-stationary sequence is converted into a stationary sequence.
Step 106: and performing at least one difference processing on the travel production time series and the travel attraction time series until the travel production time series and the travel attraction time series become a steady sequence. And determining a difference order after difference processing.
The difference order d is that X is obtained after d differencestAnd YtConverting the non-stationary sequence into a stationary sequence, taking 1 or 2 as d according to practical experience, carrying out differential processing and determining a differential order d, and then executing step 107.
Step 107: establishing an autoregressive moving average model by using the parameter fitting sequence; the travel generation amount time sequence has certain periodicity, continuity and stability, and the future time generation amount time sequence can be predicted by fitting the earlier-stage time generation amount time sequence. Step 107 specifically includes:
a: a linear function is established between an arbitrary time point and the previous value for that time point.
Determining an autoregressive term of the model, wherein the autoregressive term is a linear function of the trip production amount time series and the trip attraction amount time series with respect to the current period value and the previous period value, namely the trip production amount x of the t time pointtAvailable future value xt-1、xt-2Means, then
Figure BDA0002299314100000091
Similarly, the travel attraction y at the t-th time pointtAvailable future value yt-1、yt-2Means, then
Figure BDA0002299314100000101
Real parameters
Figure BDA0002299314100000102
For the autoregressive coefficients, which are also parameters to be estimated, p represents the first model order.
Introducing a lag operator B, wherein the lag operator converts the previous values of the travel production and the travel attraction into current values, namely converts the travel production and the travel attraction at the t-k time point into the travel production x at the t time pointtAnd travel attraction yt，Bkxt＝xt-k， Bkyt＝yt-kThen the autoregressive term can be abbreviated as
Figure BDA0002299314100000103
Wherein
Figure BDA0002299314100000104
Moving average term can be abbreviated as xt=θ(B)uxt，yt＝θ(B)uytWherein θ (B) ═ 1- θ1B-θ2B2-...-θqBq(ii) a Wherein q represents the second model order, BkRepresenting a delay back by a factor of time k,
Figure BDA0002299314100000105
is represented by BkTheta (B) represents a moving average coefficient, and a real parameter theta1～θqCoefficients in the form of a theta (B) expansion.
Establishing a prediction model according to the autoregressive term, the moving average term, the difference order d and the lag operator B, wherein the method specifically comprises the following steps: determining a linear function of the prediction model, wherein the predicted value of the trip generation amount at the Tth time point can be also expressed as a linear function consisting of the predicted value, an earlier value and an earlier random error, and the linear function is determined as the linear function of the prediction model, namely
Figure BDA0002299314100000106
Figure BDA0002299314100000107
B: and introducing a difference order and a hysteresis operator, and establishing an autoregressive moving average model on the basis of the linear function.
The linear function of the prediction model is an autoregressive moving average model of (p, q) order, the prediction model is determined according to the definition of the model, and the linear function of the prediction model isIntroducing a lag operator B and a difference order d on the basis of the linear function, and predicting the travel production x of the T-th time pointtAnd travel attraction ytCan be abbreviated as
Figure BDA0002299314100000108
And
Figure BDA0002299314100000109
c: and calculating the model order of the autoregressive moving average model by using a trial algorithm.
And determining the model order of the model by using a trial algorithm. The model order comprises a first model order p and a second model order q, wherein p is equal to n, and q is equal to n-1.
D: and determining an autoregressive coefficient and a moving average coefficient of the autoregressive moving average model by using the parameter fitting sequence to obtain the well-established autoregressive moving average model.
Determining autoregressive coefficients for a predictive model using a parameter fit sequence
Figure BDA0002299314100000111
And a moving average coefficient θ, specifically: determining the parameters to be estimated of the model based on the parameter fitting sequence by using MATLAB software
Figure BDA0002299314100000112
And theta.
Determining the model orders p and q, the difference order d, the autoregressive coefficient
Figure BDA0002299314100000113
And determining the prediction model after the moving average coefficient theta as an autoregressive moving average model.
Step 108: and judging whether the parameters of the autoregressive moving average model need to be adjusted or not by using the model test sequence to obtain a first judgment result. The step 108 specifically includes:
a: calculating the travel generation prediction quantity of each time point in the model test sequence according to the autoregressive moving average model
Figure BDA0002299314100000114
And travel attraction prediction
Figure BDA0002299314100000115
B: calculating a maximum relative error prediction index of the output and an average absolute percentage error prediction index RE of the output according to the output prediction quantity and the output in the model test sequence; in particular to
Figure 4
Figure BDA0002299314100000117
Wherein x'TAnd y'TTo examine the trip production and attraction at the T-th time point in the sequence,
Figure BDA0002299314100000118
and
Figure BDA0002299314100000119
and generating and predicting values of the attraction amount for the trip corresponding to the Tth time point, wherein N represents the number of predicted samples.
C: calculating a maximum relative error prediction index of the attraction amount and an average absolute percentage error prediction index MAPE of the attraction amount according to the travel attraction prediction amount and the travel attraction amount in the model test sequence, which specifically comprises the following steps:
Figure 5
Figure BDA00022993141000001111
d: judging whether the maximum generated quantity relative error prediction index, the average generated quantity absolute percentage error prediction index, the maximum attraction quantity relative error prediction index and the average attraction quantity absolute percentage error prediction index are all in corresponding preset ranges to obtain a third judgment result; the method specifically comprises the following steps:
the method specifically comprises the following steps: and judging whether the average absolute percentage error prediction index RE meets RE < 15% and whether the average absolute percentage error prediction index MAPE meets MAPE < 15%.
E: if the third judgment result shows that the parameter of the autoregressive moving average model does not need to be adjusted;
f: and if the third judgment result shows that the parameter of the autoregressive moving average model needs to be adjusted, determining whether the parameter of the autoregressive moving average model is required to be adjusted.
If no adjustment is required, go to step 110. If so, go to step 109.
Step 109: adjusting parameters of the autoregressive moving average model by using the model test sequence, and determining the adjusted autoregressive moving average model as a prediction model; the method specifically comprises the following steps: readjusting the order p of the first model and the order q of the second model, adding 1 to the value of n, and updating the prediction model until the RE and MAPE requirements are met.
Step 110: and determining the autoregressive moving average model as a prediction model.
Step 111: and predicting the trip production amount and trip attraction amount at the future moment by using the prediction model to obtain the predicted trip generation amount.
In the travel generation amount prediction method in this embodiment, in consideration of a time-space change rule of travel data of a mobile phone APP, based on a time sequence theory, after an original time sequence is stabilized by using a large amount of location service data of the mobile phone APP and by differential processing, travel generation amount prediction is realized by using an autoregressive moving average model (ARMA), short-term prediction of time-space change of the travel data of the mobile phone APP can be satisfied, good prediction accuracy is obtained, and experience and reference are provided for formulation of policies such as traffic planning and traffic management.
Fig. 2 is a system configuration diagram of the trip generation amount prediction system according to the embodiment of the present invention.
Referring to fig. 2, the travel generation amount prediction system includes:
a data obtaining module 201, configured to obtain user travel data from travel software;
a screening module 202, configured to screen travel production amounts and travel attraction amounts at different time points from the travel data, and generate a travel production amount time series and a travel attraction amount time series;
the dividing module 203 is configured to divide the trip production amount time sequence and the trip attraction amount time sequence into two segments according to time, so as to obtain a parameter fitting sequence and a model checking sequence respectively;
an autocorrelation coefficient calculation module 204, configured to calculate autocorrelation coefficients of the travel production time series and the travel attraction time series;
a second judging module 205, configured to judge whether the travel production time series and the travel attraction time series are stable according to the autocorrelation coefficient, so as to obtain a second judgment result;
a difference processing module 206, configured to, if the second determination result indicates no, perform at least one difference processing on the travel production time series and the travel attraction time series until the travel production time series and the travel attraction time series become stationary series.
A model establishing module 207, configured to establish an autoregressive moving average model using the parameter fitting sequence;
a first judging module 208, configured to judge whether a parameter of the autoregressive moving average model needs to be adjusted by using the model checking sequence, so as to obtain a first judgment result;
a first prediction model determining module 209, configured to, if the first determination result indicates yes, adjust parameters of the autoregressive moving average model by using the model check sequence, and determine the adjusted autoregressive moving average model as a prediction model;
a second prediction model determining module 210, configured to determine the autoregressive moving average model as the prediction model if the first determination result indicates no;
the prediction module 211 is configured to predict the trip production amount and trip attraction amount at a future time by using the prediction model, so as to obtain a predicted trip production amount.
Optionally, the model building module 207 includes:
the linear function establishing unit is used for establishing a linear function between any time point and the early value of the time point;
the model initial building unit is used for introducing a difference order and a hysteresis operator and building an autoregressive moving average model on the basis of the linear function;
the model order calculation unit is used for calculating the model order of the autoregressive moving average model by using a trial algorithm;
and the model coefficient calculation unit is used for determining the autoregressive coefficient and the moving average coefficient of the autoregressive moving average model by using the parameter fitting sequence to obtain the well-established autoregressive moving average model.
Optionally, the first determining module 208 includes:
the prediction quantity calculation unit is used for calculating the travel generation prediction quantity and the travel attraction prediction quantity of each time point in the model test sequence according to the autoregressive moving average model;
the production index calculation unit is used for calculating a maximum relative error prediction index of the production and an average absolute percentage error prediction index of the production according to the trip production prediction quantity and the trip production quantity in the model test sequence;
the attraction amount index calculation unit is used for calculating an attraction amount maximum relative error prediction index and an attraction amount average absolute percentage error prediction index according to the trip attraction prediction amount and the trip attraction amount in the model test sequence;
the third judging unit is used for judging whether the maximum relative error prediction index of the production amount, the average absolute percentage error prediction index of the production amount, the maximum relative error prediction index of the suction amount and the average absolute percentage error prediction index of the suction amount are all in corresponding preset ranges to obtain a third judging result;
a first result determining unit, configured to determine that the parameter of the autoregressive moving average model does not need to be adjusted if the third determination result indicates yes;
and the second result determining unit is used for determining that the parameters of the autoregressive moving average model need to be adjusted if the third judgment result shows no.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects: according to the travel generation amount prediction method and system disclosed by the invention, the travel generation amount time sequence and the travel attraction amount time sequence are obtained by utilizing data in travel software, and an autoregressive moving average model is established on the basis, so that the future travel generation amount prediction is directly finished. The method can realize the prediction of the travel generation amount by directly utilizing the data in the travel software (app for travel in the mobile phone, such as dripping), and does not need to depend on specific GPS equipment, thereby improving the convenience degree of the travel generation amount prediction.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The principles and embodiments of this invention have been described herein using specific examples,
the above description of the embodiments is only intended to facilitate the understanding of the method of the invention and its core ideas;
meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.

Claims (9)

1. A travel production prediction method is characterized by comprising the following steps:
step 101: acquiring user travel data from travel software;
step 102: screening the travel production and travel attraction at different time points from the travel data to generate a travel production time sequence and a travel attraction time sequence; the travel generation amount comprises travel production amount and travel attraction amount;
the travel production is the sum of all home endpoints of home travel and all starting points of non-home travel and cargo travel; the trip attraction amount is the sum of all non-family endpoints of the home trip and all endpoints of the non-home trip and the cargo trip; wherein
The trip production time series is represented as:
Xt＝{x1，x2，x3，...}
the trip attraction time series is represented as:
Yt＝{y1，y2，y3，...}
wherein t represents a time point, xtIs the travel production at the t-th time point, XtRepresenting a time series of travel production, ytIs the travel attraction at the t-th time point, YtRepresenting a trip attraction time sequence;
step 103: dividing the travel production time sequence and the travel attraction time sequence into two sections according to time to respectively obtain a parameter fitting sequence and a model checking sequence;
step 104: calculating autocorrelation coefficients of the travel production time series and the travel attraction time series;
step 105: judging whether the travel production time sequence and the travel attraction time sequence are stable or not according to the autocorrelation coefficient to obtain a second judgment result;
step 106: performing at least one difference processing on the travel production time series and the travel attraction time series until the travel production time series and the travel attraction time series become stationary sequences; determining a difference order after difference processing;
step 107: establishing an autoregressive moving average model by using the parameter fitting sequence; the travel generation amount time sequence has certain periodicity, continuity and stability, and the future moment generation amount time sequence can be predicted by fitting the earlier moment generation amount time sequence;
step 108: judging whether the parameters of the autoregressive moving average model need to be adjusted or not by using the model test sequence to obtain a first judgment result;
step 109: adjusting parameters of the autoregressive moving average model by using the model test sequence, and determining the adjusted autoregressive moving average model as a prediction model;
step 110: determining the autoregressive moving average model as a prediction model;
step 111: and predicting the trip production amount and trip attraction amount at the future moment by using the prediction model to obtain the predicted trip generation amount.
2. A travel production prediction method according to claim 1, characterized in that in step 104, the autocorrelation coefficient calculation formula is:
Figure FDA0003032858420000021
Figure FDA0003032858420000022
wherein r isx(t，t+k)＝E(Xt-EXt)(Xt+k-EXt+k) Is XtIs the delay k (k ═ 1, 2, 3.., n) autocovariance function, ρ, ofx(t, t + k) is XtIs the delay k autocorrelation function, ry(t，t+k)＝E(Yt-EYt)(Yt+k-EYt+k) Is YtIs the delay k (k ═ 1, 2, 3.., n) autocovariance function, ρ, ofy(t, t + k) is YtE, D are the mathematical expectation and variance, respectively, and the autocovariance function and the autocorrelation function depend only on the delay length k of time, and are independent of the start and end points of time.
3. A travel production prediction method according to claim 1,
in the step 105, X is determined according to the autocorrelation coefficienttAnd YtThe stationary sequence usually has short-term correlation, that is, the autocorrelation coefficient of the stationary sequence rapidly decays to 0 with the increase of the delay period number k, then fluctuates around 0 and gradually converges to 0, the threshold value of the delay period number k is determined, any k is determined, and 5 can be taken, and if the value of ρ (t, t + k) approaches to 0 quickly in the process of increasing k from 1 to 5, the time sequence X can be taken astAnd YtFor the stationary sequence, step 107 is performed directly; otherwise, XtAnd YtIf the sequence is a non-stationary sequence, step 106 is executed, and step 107 is executed after the non-stationary sequence is converted into a stationary sequence.
4. A travel production prediction method according to claim 1,
the step 107 specifically includes:
a1: establishing a linear function between any time point and the early value of the time point;
determining an autoregressive term of the model, wherein the autoregressive term is a linear function of the trip production amount time series and the trip attraction amount time series with respect to a current period value and a previous period value;
b1: introducing a difference order and a hysteresis operator, and establishing an autoregressive moving average model on the basis of the linear function;
c1: calculating the model order of the autoregressive moving average model by using a trial algorithm;
d1: and determining an autoregressive coefficient and a moving average coefficient of the autoregressive moving average model by using the parameter fitting sequence to obtain the well-established autoregressive moving average model.
5. A travel production prediction method according to claim 1,
the step 108 specifically includes:
a: calculating the travel generation prediction quantity of each time point in the model test sequence according to the autoregressive moving average model
Figure FDA0003032858420000031
And travel attraction prediction
Figure FDA0003032858420000032
B: respectively calculating a maximum relative percentage error prediction index RE of the generated quantity and the attracted quantity according to the predicted quantity of the generated and attracted travel quantity and the generated and attracted travel quantity in the model test sequence; in particular to
Figure FDA0003032858420000033
Figure FDA0003032858420000034
Wherein x'TAnd y'TTo examine the trip production and attraction at the T-th time point in the sequence,
Figure FDA0003032858420000035
and
Figure FDA0003032858420000036
generating and predicting values of the attraction amount for the trip corresponding to the T-th time point, wherein N represents the number of predicted samples;
respectively calculating average relative percentage error prediction indexes MAPE of the production amount and the attraction amount according to the prediction amount of the production amount and the attraction amount of the travel and the production amount and the attraction amount of the travel in the model test sequence; in particular to
Figure FDA0003032858420000041
Figure FDA0003032858420000042
D: judging whether the maximum generation amount relative percentage error prediction index, the average generation amount relative percentage error prediction index, the maximum attraction amount relative percentage error prediction index and the average attraction amount relative percentage error prediction index are all in corresponding preset ranges to obtain a third judgment result; the method specifically comprises the following steps:
judging whether the maximum relative percentage error prediction index RE meets RE < 15% and whether the average relative percentage error prediction index MAPE meets MAPE < 15%;
e: if the third judgment result shows that the parameter of the autoregressive moving average model does not need to be adjusted; step 110 is executed;
f: if the third judgment result shows no, determining that the parameters of the autoregressive moving average model need to be adjusted; step 109 is performed.
6. A travel generation amount prediction method according to claim 1, wherein the travel software is mobile phone travel software, and the user travel data includes travel time and location information; the travel time and location information may be used to determine the time of each travel start and end.
7. A travel generation amount prediction system characterized by comprising:
the data acquisition module (201) is used for acquiring user trip data from trip software;
the screening module (202) is used for screening the travel production and the travel attraction at different time points from the travel data to generate a travel production time sequence and a travel attraction time sequence;
a dividing module (203) for dividing the travel production amount time sequence and the travel attraction amount time sequence into two sections according to time to respectively obtain a parameter fitting sequence and a model checking sequence;
a correlation coefficient calculation module (204) for calculating autocorrelation coefficients of the travel production amount time series and the travel attraction amount time series;
a second judging module (205) for judging whether the travel production time series and the travel attraction time series are stable according to the autocorrelation coefficient to obtain a second judgment result;
a difference processing module (206) configured to, if the second determination result indicates no, perform at least one difference processing on the travel production time series and the travel attraction time series until the travel production time series and the travel attraction time series become stationary series;
a model building module (207) for building an autoregressive moving average model using the parameter fitting sequence;
a first judging module (208) for judging whether the parameters of the autoregressive moving average model need to be adjusted by using the model test sequence to obtain a first judgment result;
a first prediction model determining module (209) for adjusting parameters of the autoregressive moving average model by using the model checking sequence and determining the adjusted autoregressive moving average model as a prediction model if the first judgment result indicates yes;
a second prediction model determination module (210) for determining the autoregressive moving average model as a prediction model if the first judgment result indicates no;
and the prediction module (211) is used for predicting the trip production and the trip attraction at the future time by using the prediction model to obtain the predicted trip production.
8. A travel generation amount prediction system according to claim 7,
the model building module (207) comprises: the linear function establishing unit is used for establishing a linear function between any time point and the early value of the time point;
the model initial building unit is used for introducing a difference order and a hysteresis operator and building an autoregressive moving average model on the basis of the linear function;
the model order calculation unit is used for calculating the model order of the autoregressive moving average model by using a trial algorithm;
and the model coefficient calculation unit is used for determining the autoregressive coefficient and the moving average coefficient of the autoregressive moving average model by using the parameter fitting sequence to obtain the well-established autoregressive moving average model.
9. A travel generation amount prediction system according to claim 7, wherein the first judgment module (208) comprises: the prediction quantity calculation unit is used for calculating the travel generation prediction quantity and the travel attraction prediction quantity of each time point in the model test sequence according to the autoregressive moving average model;
the production index calculation unit is used for calculating a maximum relative error prediction index of the production and an average absolute percentage error prediction index of the production according to the trip production prediction quantity and the trip production quantity in the model test sequence;
the attraction amount index calculation unit is used for calculating an attraction amount maximum relative error prediction index and an attraction amount average absolute percentage error prediction index according to the trip attraction prediction amount and the trip attraction amount in the model test sequence;
the third judging unit is used for judging whether the maximum relative error prediction index of the production amount, the average absolute percentage error prediction index of the production amount, the maximum relative error prediction index of the suction amount and the average absolute percentage error prediction index of the suction amount are all in corresponding preset ranges to obtain a third judging result;
a first result determining unit, configured to determine that the parameter of the autoregressive moving average model does not need to be adjusted if the third determination result indicates yes;
and the second result determining unit is used for determining that the parameters of the autoregressive moving average model need to be adjusted if the third judgment result shows no.
CN201911215218.8A 2019-12-02 2019-12-02 Travel generation amount prediction method and system Active CN111145537B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911215218.8A CN111145537B (en) 2019-12-02 2019-12-02 Travel generation amount prediction method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911215218.8A CN111145537B (en) 2019-12-02 2019-12-02 Travel generation amount prediction method and system

Publications (2)

Publication Number Publication Date
CN111145537A CN111145537A (en) 2020-05-12
CN111145537B true CN111145537B (en) 2021-06-15

Family

ID=70517494

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911215218.8A Active CN111145537B (en) 2019-12-02 2019-12-02 Travel generation amount prediction method and system

Country Status (1)

Country Link
CN (1) CN111145537B (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105679021A (en) * 2016-02-02 2016-06-15 重庆云途交通科技有限公司 Travel time fusion prediction and query method based on traffic big data
CN107767659A (en) * 2017-10-13 2018-03-06 东南大学 Shared bicycle traffic attraction and prediction of emergence size method based on ARIMA models
CN108874748A (en) * 2018-06-01 2018-11-23 交通运输部路网监测与应急处置中心 A kind of ETC operation indicator prediction technique
JP2019016081A (en) * 2017-07-05 2019-01-31 国立大学法人 東京大学 Congestion prediction device and congestion prediction method therefor
CN110491124A (en) * 2019-08-19 2019-11-22 上海新共赢信息科技有限公司 A kind of vehicle flow prediction technique, device, equipment and storage medium

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104899443B (en) * 2015-06-05 2018-03-06 陆化普 For assessing the method and system of current trip requirements and the following trip requirements of prediction
CN105070042B (en) * 2015-07-22 2017-10-10 济南市市政工程设计研究院（集团）有限责任公司 A kind of modeling method of traffic forecast
JP7082338B2 (en) * 2017-11-15 2022-06-08 リンクスウェア株式会社 A method for supporting taxi dispatch and a taxi dispatch support system using it

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105679021A (en) * 2016-02-02 2016-06-15 重庆云途交通科技有限公司 Travel time fusion prediction and query method based on traffic big data
JP2019016081A (en) * 2017-07-05 2019-01-31 国立大学法人 東京大学 Congestion prediction device and congestion prediction method therefor
CN107767659A (en) * 2017-10-13 2018-03-06 东南大学 Shared bicycle traffic attraction and prediction of emergence size method based on ARIMA models
CN108874748A (en) * 2018-06-01 2018-11-23 交通运输部路网监测与应急处置中心 A kind of ETC operation indicator prediction technique
CN110491124A (en) * 2019-08-19 2019-11-22 上海新共赢信息科技有限公司 A kind of vehicle flow prediction technique, device, equipment and storage medium

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
Time series forecasting using a moving average model for extrapolation of number of tourist;Ivanovski,Z.等;《UTMS Journal of Economics》;20180505;第9卷(第2期);第121-132页 *

Also Published As

Publication number Publication date
CN111145537A (en) 2020-05-12

Similar Documents

Publication Publication Date Title
WO2017028632A1 (en) Method of predicting distribution network operation reliability
CN104318334B (en) A kind of Short-Term Load Forecasting Method based on correlation FARIMA models long
CN107547269B (en) Method for constructing intelligent substation communication flow threshold model based on FARIMA
CN106921158B (en) demand coefficient analysis method for historical collected data based on time sequence of distribution transformer
CN108155648A (en) Method for estimating state based on the infinite Extended Kalman filter of adaptive H
CN104239982A (en) Method for predicting energy consumption of buildings during holidays and festivals on basis of time series and neural networks
CN106714301B (en) Carrier optimization method in wireless positioning network
CN113114497B (en) Low-voltage distribution area topology automatic identification method
CN109711662B (en) Power grid robust state estimation method based on multi-source data fusion
CN111145537B (en) Travel generation amount prediction method and system
CN115877312A (en) Electric energy meter informatization evaluation calibration model based on station area electric energy conservation
CN104537233B (en) A kind of power distribution network puppet based on Density Estimator measures generation method
CN112911497A (en) Real-time prediction method and system for track of cooperative unmanned aerial vehicle
CN110707693A (en) Ensemble Kalman filtering dynamic state estimation method based on AMI full-scale measuring point partition
CN111443718B (en) High-speed train state feedback prediction control method and system based on prediction error method
CN113935244A (en) Method and system for predicting short-term power load of urban rural distribution transformer
CN112993988A (en) Power grid line loss analysis method
CN111553080B (en) Closed-loop identification method for load dynamic equivalent non-mechanism model parameters of distribution substation
Hayes et al. Short-term operational planning and state estimation in power distribution networks
Niu et al. Study of ionospheric TEC short-term forecast model based on combination method
Liu et al. A multivariate forecasting method for short‐term load using chaotic features and RBF neural network
CN116317094B (en) Low-voltage distribution network topology identification method based on Internet of things equipment measurement data
CN107767060B (en) Theoretical line loss calculation system and method for distribution network line
CN110865328B (en) Intelligent electric meter phase identification, topology identification and impedance estimation method based on AMI
CN104979828A (en) Analysis method and analysis device for reconstructing power distribution network after access of distributed power source to power distribution network

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
GR01 Patent grant
GR01 Patent grant