CN105913664A - Traffic flow monitoring and predicting system - Google Patents

Abstract

The invention provides a traffic flow monitoring and predicting system which comprises a monitoring device and a predicting device connected with the monitoring device, wherein the predicting device comprises a collection module, a data preprocessing module, a data classifying module, a stability check module, a correlation coefficient calculation module, a threshold setting module, a space-time correlation coefficient matrix generation module, a historical correlation coefficient matrix generation module, a prediction factor selection module and a prediction model construction module which are sequentially connected. The traffic flow monitoring and predicting system is relatively high in prediction accuracy and the constructed prediction model has pertinence.

Description

A kind of traffic flow monitoring and controlling forecast system

Technical field

The present invention relates to intelligent transportation field, be specifically related to a kind of traffic flow monitoring and controlling forecast system.

Background technology

Traffic flow passes through the actual vehicle number of a certain section of road in referring to the unit interval, be the key character describing traffic behavior Parameter.The change of traffic flow is again real-time, higher-dimension, non-linear a, stochastic process for non-stationary, the change of correlative factor All may affect the traffic flow of subsequent time.In correlation technique, strong about prediction means limitation in short-term, it was predicted that precision is relatively Low, real-time estimate fails to achieve satisfactory results, and fails the Real-time Road to people and selects to provide effectively suggestion, thus hands over Through-current capacity prediction major part rests on the medium-and long-term forecasting of traffic flow.

Summary of the invention

For the problems referred to above, the present invention provides a kind of traffic flow monitoring and controlling forecast system.

The purpose of the present invention realizes by the following technical solutions:

A kind of traffic flow monitoring and controlling forecast system, including supervising device and the prediction means that is connected with supervising device, described monitoring dress Put and include:

Being sequentially arranged at the photographic head of vehicle front, rear and the left and right sides, each photographic head connects monitoring processor, described Monitoring processor connects the watch-dog being positioned at driver's cabin, and described monitoring processor includes:

For gathering the acquisition module of the video image of each photographic head；

For at least one width video image gathered being zoomed in and out the image conversion module of process；

Each video image after processing exports the image display of watch-dog；

It is characterized in that: described dynamic vehicle monitoring device also includes that the camera lens for driving left and right photographic head is along front and back To the angle controller rotated, described angle controller is arranged on the arranged on left and right sides of vehicle, described left and right photographic head with Described angle controller connects, and described angle controller connects described monitoring processor, and described monitoring processor also includes: For starting angle controller, and the Dynamic control module of the anglec of rotation in direction before and after left and right photographic head is set, described Dynamic control module connects described angle controller.

Preferably, described angle controller is controlled electro-motor, the output shaft of described controlled electro-motor and described photographic head Base connect.

Preferably, described output shaft and the central axis of vehicle fore-and-aft direction.

Preferably, it is characterized in that, it was predicted that acquisition module that device includes being sequentially connected with, data preprocessing module, data classification mould Block, stationary test module, Calculation of correlation factor module, threshold value setting module, temporal and spatial correlations coefficient matrix generation module, go through History correlation matrix generation module, predictor choose module and forecast model construction module:

(1) acquisition module, is used for gathering observation section S in road network S_i, prediction section S_jThe traffic flow of corresponding each time period Data and passage situation；

(2) data preprocessing module, for described traffic flow data carries out data prediction, and rejecting does not meets traffic in fact The data of border situation；

(3) data categorization module, for carrying out classification of type, described type bag to the traffic flow data through data prediction Include traffic flow data festivals or holidays, traffic flow data at weekend and traffic flow data on working day；

(4) stationary test module, for being in same type of observation section S_iTraffic flow sequence X_iWith prediction road Section S_jTraffic flow sequence X_jCarrying out stationary test respectively, the auto-correlation function of inspection stationarity is:

$P\left(\mathrm{\τ}\right)=\frac{E\[(X_x-{\mathrm{\ν}}_i)(X_{x+\mathrm{\τ}}-{\mathrm{\ν}}_{x+\mathrm{\τ}})\]}{{\mathrm{\σ}}^2}$

Wherein, X_xRepresent traffic flow sequence to be tested, ν_iRepresent the average of traffic flow sequence to be tested, X_x+τRepresent X_x? Traffic flow sequence after time delay τ, ν_x+τFor X_x+τAverage, σ²For X_xWith X_x+τBetween variance；

When auto-correlation function P (τ) can level off to 0 or fluctuate 0 near by rapid decay, and the most described traffic flow sequence to be tested is logical Cross stationary test；When auto-correlation function P (τ) can not rapid decay level off to 0 or near 0 fluctuate, then to described to be tested Traffic flow sequence proceeds stationary test after carrying out calm disposing；

(5) Calculation of correlation factor module, for calculating the observation section S by stationary test_iTraffic flow sequence X_iWith Prediction section S_jTraffic flow sequence X_jTime correlation coefficient ρ under time delay τ_ij(τ) with space correlation coefficient ρ_ij(w), If having N number of section, traffic flow sequence X in road network S_i=[x_i(1),x_i(2),...,x_i(n)], traffic flow sequencex_iT () represents observation section S_iAt the flow of t, x_jT () represents prediction section S_jIn t Flow, t=1,2 ... n, time correlation coefficient ρ_ij(τ) computing formula is:

${\mathrm{\ρ}}_{ij}\left(\mathrm{\τ}\right)=\frac{{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_i\left(t\right)x_j(t+\mathrm{\τ})-\frac{1}{n-\mathrm{\τ}}{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_i\left(t\right){\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_j(t+\mathrm{\τ})}{{\sqrt{{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}\[x_i\left(t\right)-\frac{1}{n-\mathrm{\τ}}{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_i\left(t\right)\]}}^2\×{\sqrt{{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}\[x_j(t+\mathrm{\τ})-\frac{1}{n-\mathrm{\τ}}{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_j(t+\mathrm{\τ})\]}}^2}$

Space correlation coefficient ρ_ijW the computing formula of () is:

Preferably, it is characterized in that, it was predicted that device also includes:

(6) threshold value setting module, for setting time delay maximum L, the temporal and spatial correlations coefficient threshold T between each section₁With History correlation coefficient threshold T₂；

(7) temporal and spatial correlations coefficient matrix generation module, for the time correlation coefficient ρ according to each section_ij(τ) with space correlation system Number ρ_ijW () builds each observation section S_iWith prediction section S_jTemporal and spatial correlations coefficient matrix ρ (τ) ' under different time postpones τ, and Calculate the temporal and spatial correlations coefficient ρ in each section_ij(τ) ', wherein i ∈ [1, N] and τ ∈ [0, L], the span of L is [8,12], time Kongxiang The computing formula closing coefficient matrix ρ (τ) ' is:

$\mathrm{\ρ}{\left(\mathrm{\τ}\right)}^{\′}=\left[\begin{array}{cccc}{\mathrm{\ρ}}_{1j}{\left(0\right)}^{\′}& {\mathrm{\ρ}}_{1j}{\left(0\right)}^{\′}& \mathrm{...}& {\mathrm{\ρ}}_{Nj}{\left(0\right)}^{\′}\\ {\mathrm{\ρ}}_{1j}{\left(1\right)}^{\′}& {\mathrm{\ρ}}_{2j}{\left(1\right)}^{\′}& \mathrm{...}& {\mathrm{\ρ}}_{Nj}{\left(1\right)}^{\′}\\ \mathrm{...}& \mathrm{...}& \mathrm{...}& \mathrm{...}\\ {\mathrm{\ρ}}_{1j}{\left(L\right)}^{\′}& {\mathrm{\ρ}}_{2j}{\left(L\right)}^{\′}& \mathrm{...}& {\mathrm{\ρ}}_{Nj}{\left(L\right)}^{\′}\end{array}\right];$

Temporal and spatial correlations coefficient ρ_ij(τ) ' computing formula be:

ρ_ij(τ) '=ρ_ij(τ)ρ_ij(w)；

(8) history correlation matrix generation module, is used for generating prediction section S_jHistory correlation matrix ρ (t):

Wherein, the same period all for nearly M and same type of historical traffic are chosen as traffic flow sequence X_jHistory correlated series, It is designated asM=1,2 ... the span of M, M is [3,5], described history correlation coefficient ρ_jmT the computing formula of () is:

${\mathrm{\ρ}}_{jm}\left(t\right)=\frac{{\mathrm{\Σ}}_{t=1}^n{x}_j\left(t\right)x_{jm}\left(t\right)-\frac{1}{n}{\mathrm{\Σ}}_{t=1}^n{x}_j\left(t\right){\mathrm{\Σ}}_{t=1}^n{x}_{jm}\left(t\right)}{\sqrt{{\mathrm{\Σ}}_{t=1}^n{\[x_j\left(t\right)-\frac{1}{n}{\mathrm{\Σ}}_{t=1}^n{x}_j\left(t\right)\]}^2}\×\sqrt{{\mathrm{\Σ}}_{t=1}^n{\[x_{jm}\left(t\right)-\frac{1}{n}{\mathrm{\Σ}}_{t=1}^n{x}_{jm}\left(t\right)\]}^2}}$

(9) predictor chooses module, for according to described temporal and spatial correlations coefficient threshold T₁With history correlation coefficient threshold T₂Choose The predictor relevant to prediction impact point, and carry out matrix reconstruction according to locus j selected by it with time delay τ, choose Principle is:

If ρ_ij(τ) ' ＞ T₁, then will observation section S_iTraffic flow sequence X_iIn meet the traffic flow new sequence of composition of condition And as the first predictor, it is denoted as X', X_'=(x₁',x₂',...,x_p'), wherein p is the described traffic flow number meeting condition, If L₁It is the maximum of time delay, L in the first predictor₁=max{ τ | τ ∈ [0, L] and ρ_ij(τ) ' ＞ T₁, then the first prediction Factor X' can state following matrix form as:

$X^{\′}=\left[\begin{array}{cccc}{x_1}^{\′}\left(1\right)& {x_2}^{\′}\left(1\right)& \mathrm{...}& {x_p}^{\′}\left(1\right)\\ {x_1}^{\′}\left(2\right)& {x_2}^{\′}\left(2\right)& \mathrm{...}& {x_p}^{\′}\left(2\right)\\ \mathrm{...}& \mathrm{...}& \mathrm{...}& \mathrm{...}\\ {x_1}^{\′}(n-L_1)& {x_2}^{\′}(n-L_1)& \mathrm{...}& {x_p}^{\′}(n-L_1)\end{array}\right];$

If ρ_jm(t) ＞ T₂, then by all history correlated series X meeting condition_jmT (), as the second predictor, is denoted as Y', Y'={y₁',y₂',...,y_q', wherein q is the historical traffic number meeting condition, and the second predictor Y' can state following rectangular as Formula:

(10) forecast model construction module, it is by constructing the first predictor and the second predictor as training sample Measurable section is at the forecast model of the traffic flow of subsequent time.

Wherein, in described data preprocessing module, the rule of the data not meeting traffic practical situation described in rejecting is: at one In the data update cycle, set the threshold range of total traffic flow data in each section respectively, if total friendship in certain section collected Through-current capacity data fall in corresponding threshold range, then show that these group data are reliable, retain this group data；If certain road collected Total traffic flow data of section falls not in corresponding threshold range, then show that these group data are unreliable, and rejected.

Wherein, described stationary test module includes following submodule:

(1) syndrome module, for the traffic flow to the traffic flow sequence with prediction section being in same type of observation section Amount sequence carries out stationary test respectively；

(2) continuity check submodule, is connected with syndrome module, for not by the traffic flow to be tested of stationary test Amount sequence carries out continuity check, if the seriality of not meeting, described continuity check submodule uses average interpolation method to enter data Row polishing；

(3) misarrangement submodule, is connected with continuity check submodule, for deleting the data of apparent error, uses average simultaneously Interpolation method carries out polishing to data；

(4) difference processing submodule, connects misarrangement submodule and syndrome module, for carrying out the data after polishing at difference Reason, and the data after difference processing are sent to syndrome module.

The invention have the benefit that

1, data categorization module and stationarity inspection module are set, add the accuracy of data, and make the forecast model of structure more Targetedly；

2, arrange Calculation of correlation factor module, temporal and spatial correlations coefficient matrix generation module, history correlation matrix generation module, Predictor chooses module and forecast model construction module, and wherein predictor directly affects precision of prediction, and correlation coefficient is to measure The index of stochastic variable dependency, it is possible to help to choose the variable closely-related with the future position training sample as forecast model, Choose multiple correlation coefficient as predictor, eliminate the subjectivity that initial predictor is chosen, by increasing capacitance it is possible to increase precision of prediction, Make forecast model construction module more stable and accurate；

3, the space correlation coefficient in Calculation of correlation factor module reflects the accessibility impact on forecast model of road network, time phase Close coefficient and can express the time sequencing of flow sequence, reflect the cause effect relation on two sequence times, thus improve predictor choosing The efficiency taken；Due to the Weekly similarity of traffic flow, introduce the history correlation coefficient of history correlation matrix generation module, with Time correlation coefficient and space correlation coefficient with the use of, provide more data support for Accurate Prediction.

Accompanying drawing explanation

The invention will be further described to utilize accompanying drawing, but the embodiment in accompanying drawing does not constitute any limitation of the invention, for Those of ordinary skill in the art, on the premise of not paying creative work, it is also possible to obtains the attached of other according to the following drawings Figure.

Fig. 1 is the connection diagram of each module of prediction means of the present invention.

Fig. 2 is left side or the scheme of installation of right side photographic head.

Detailed description of the invention

The invention will be further described with the following Examples.

Embodiment 1

See Fig. 1, Fig. 2, the present embodiment one traffic flow monitoring and controlling forecast system, be connected including supervising device with supervising device Prediction means, described supervising device includes:

Being sequentially arranged at the photographic head of vehicle front, rear and the left and right sides, each photographic head connects monitoring processor, described Monitoring processor connects the watch-dog being positioned at driver's cabin, and described monitoring processor includes:

For gathering the acquisition module of the video image of each photographic head；

For at least one width video image gathered being zoomed in and out the image conversion module of process；

Each video image after processing exports the image display of watch-dog；

It is characterized in that: described dynamic vehicle monitoring device also includes that the camera lens for driving left and right photographic head is along front and back To the angle controller rotated, described angle controller is arranged on the arranged on left and right sides of vehicle, described left and right photographic head with Described angle controller connects, and described angle controller connects described monitoring processor, and described monitoring processor also includes: For starting angle controller, and the Dynamic control module of the anglec of rotation in direction before and after left and right photographic head is set, described Dynamic control module connects described angle controller.

Preferably, described angle controller is controlled electro-motor, the output shaft of described controlled electro-motor and described photographic head Base connect.

Preferably, described output shaft and the central axis of vehicle fore-and-aft direction.

Preferably, it is characterized in that, it was predicted that acquisition module that device includes being sequentially connected with, data preprocessing module, data classification mould Block, stationary test module, Calculation of correlation factor module, threshold value setting module, temporal and spatial correlations coefficient matrix generation module, go through History correlation matrix generation module, predictor choose module and forecast model construction module:

(1) acquisition module, is used for gathering observation section S in road network S_i, prediction section S_jThe traffic flow of corresponding each time period Data and passage situation；

(2) data preprocessing module, for described traffic flow data carries out data prediction, and rejecting does not meets traffic in fact The data of border situation；

(3) data categorization module, for carrying out classification of type, described type bag to the traffic flow data through data prediction Include traffic flow data festivals or holidays, traffic flow data at weekend and traffic flow data on working day；

(4) stationary test module, for being in same type of observation section S_iTraffic flow sequence X_iWith prediction road Section S_jTraffic flow sequence X_jCarrying out stationary test respectively, the auto-correlation function of inspection stationarity is:

$P\left(\mathrm{\τ}\right)=\frac{E\[(X_x-{\mathrm{\ν}}_i)(X_{x+\mathrm{\τ}}-{\mathrm{\ν}}_{x+\mathrm{\τ}})\]}{{\mathrm{\σ}}^2}$

Wherein, X_xRepresent traffic flow sequence to be tested, ν_iRepresent the average of traffic flow sequence to be tested, X_x+τRepresent X_x? Traffic flow sequence after time delay τ, ν_x+τFor X_x+τAverage, σ²For X_xWith X_x+τBetween variance；

When auto-correlation function P (τ) can level off to 0 or fluctuate 0 near by rapid decay, and the most described traffic flow sequence to be tested is logical Cross stationary test；When auto-correlation function P (τ) can not rapid decay level off to 0 or near 0 fluctuate, then to described to be tested Traffic flow sequence proceeds stationary test after carrying out calm disposing；

(5) Calculation of correlation factor module, for calculating the observation section S by stationary test_iTraffic flow sequence X_iWith Prediction section S_jTraffic flow sequence X_jTime correlation coefficient ρ under time delay τ_ij(τ) with space correlation coefficient ρ_ij(w), If having N number of section, traffic flow sequence X in road network S_i=[x_i(1),x_i(2),...,x_i(n)], traffic flow sequencex_iT () represents observation section S_iAt the flow of t, x_jT () represents prediction section S_jIn t Flow, t=1,2 ... n, time correlation coefficient ρ_ij(τ) computing formula is:

${\mathrm{\ρ}}_{ij}\left(\mathrm{\τ}\right)=\frac{{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_i\left(t\right)x_j(t+\mathrm{\τ})-\frac{1}{n-\mathrm{\τ}}{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_i\left(t\right){\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_j(t+\mathrm{\τ})}{{\sqrt{{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}\[x_i\left(t\right)-\frac{1}{n-\mathrm{\τ}}{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_i\left(t\right)\]}}^2\×{\sqrt{{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}\[x_j(t+\mathrm{\τ})-\frac{1}{n-\mathrm{\τ}}{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_j(t+\mathrm{\τ})\]}}^2}$

Space correlation coefficient ρ_ijW the computing formula of () is:

Preferably, it is characterized in that, it was predicted that device also includes:

(6) threshold value setting module, for setting time delay maximum L, the temporal and spatial correlations coefficient threshold T between each section₁With History correlation coefficient threshold T₂；

(7) temporal and spatial correlations coefficient matrix generation module, for the time correlation coefficient ρ according to each section_ij(τ) with space correlation system Number ρ_ijW () builds each observation section S_iWith prediction section S_jTemporal and spatial correlations coefficient matrix ρ (τ) ' under different time postpones τ, and Calculate the temporal and spatial correlations coefficient ρ in each section_ij(τ) ', wherein i ∈ [1, N] and τ ∈ [0, L], the span of L is [8,12], time Kongxiang The computing formula closing coefficient matrix ρ (τ) ' is:

$\mathrm{\ρ}{\left(\mathrm{\τ}\right)}^{\′}=\left[\begin{array}{cccc}{\mathrm{\ρ}}_{1j}{\left(0\right)}^{\′}& {\mathrm{\ρ}}_{1j}{\left(0\right)}^{\′}& \mathrm{...}& {\mathrm{\ρ}}_{Nj}{\left(0\right)}^{\′}\\ {\mathrm{\ρ}}_{1j}{\left(1\right)}^{\′}& {\mathrm{\ρ}}_{2j}{\left(1\right)}^{\′}& \mathrm{...}& {\mathrm{\ρ}}_{Nj}{\left(1\right)}^{\′}\\ \mathrm{...}& \mathrm{...}& \mathrm{...}& \mathrm{...}\\ {\mathrm{\ρ}}_{1j}{\left(L\right)}^{\′}& {\mathrm{\ρ}}_{2j}{\left(L\right)}^{\′}& \mathrm{...}& {\mathrm{\ρ}}_{Nj}{\left(L\right)}^{\′}\end{array}\right];$

Temporal and spatial correlations coefficient ρ_ij(τ) ' computing formula be:

ρ_ij(τ) '=ρ_ij(τ)ρ_ij(w)；

(8) history correlation matrix generation module, is used for generating prediction section S_jHistory correlation matrix ρ (t):

Wherein, the same period all for nearly M and same type of historical traffic are chosen as traffic flow sequence X_jHistory correlated series, It is designated asM=1,2 ... the span of M, M is [3,5], described history correlation coefficient ρ_jmT the computing formula of () is:

${\mathrm{\ρ}}_{jm}\left(t\right)=\frac{{\mathrm{\Σ}}_{t=1}^n{x}_j\left(t\right)x_{jm}\left(t\right)-\frac{1}{n}{\mathrm{\Σ}}_{t=1}^n{x}_j\left(t\right){\mathrm{\Σ}}_{t=1}^n{x}_{jm}\left(t\right)}{\sqrt{{\mathrm{\Σ}}_{t=1}^n{\[x_j\left(t\right)-\frac{1}{n}{\mathrm{\Σ}}_{t=1}^n{x}_j\left(t\right)\]}^2}\×\sqrt{{\mathrm{\Σ}}_{t=1}^n{\[x_{jm}\left(t\right)-\frac{1}{n}{\mathrm{\Σ}}_{t=1}^n{x}_{jm}\left(t\right)\]}^2}}$

(9) predictor chooses module, for according to described temporal and spatial correlations coefficient threshold T₁With history correlation coefficient threshold T₂Choose The predictor relevant to prediction impact point, and carry out matrix reconstruction according to locus j selected by it with time delay τ, choose Principle is:

If ρ_ij(τ) ' ＞ T₁, then will observation section S_iTraffic flow sequence X_iIn meet the traffic flow new sequence of composition of condition And as the first predictor, it is denoted as X', X'=(x₁',x₂',...,x_p'), wherein p is the described traffic flow number meeting condition, If L₁It is the maximum of time delay, L in the first predictor₁=max{ τ | τ ∈ [0, L] and ρ_ij(τ) ' ＞ T₁, then the first prediction Factor X' can state following matrix form as:

$X^{\′}=\left[\begin{array}{cccc}{x_1}^{\′}\left(1\right)& {x_2}^{\′}\left(1\right)& \mathrm{...}& {x_p}^{\′}\left(1\right)\\ {x_1}^{\′}\left(2\right)& {x_2}^{\′}\left(2\right)& \mathrm{...}& {x_p}^{\′}\left(2\right)\\ \mathrm{...}& \mathrm{...}& \mathrm{...}& \mathrm{...}\\ {x_1}^{\′}(n-L_1)& {x_2}^{\′}(n-L_1)& \mathrm{...}& {x_p}^{\′}(n-L_1)\end{array}\right];$

If ρ_jm(t) ＞ T₂, then by all history correlated series X meeting condition_jmT (), as the second predictor, is denoted as Y', Y'={y₁',y₂',...,y_q', wherein q is the historical traffic number meeting condition, and the second predictor Y' can state following rectangular as Formula:

(10) forecast model construction module, it is by constructing the first predictor and the second predictor as training sample Measurable section is at the forecast model of the traffic flow of subsequent time.

Wherein, in described data preprocessing module, the rule of the data not meeting traffic practical situation described in rejecting is: at one In the data update cycle, set the threshold range of total traffic flow data in each section respectively, if total friendship in certain section collected Through-current capacity data fall in corresponding threshold range, then show that these group data are reliable, retain this group data；If certain road collected Total traffic flow data of section falls not in corresponding threshold range, then show that these group data are unreliable, and rejected.

Wherein, described stationary test module includes following submodule:

(1) syndrome module, for the traffic flow to the traffic flow sequence with prediction section being in same type of observation section Amount sequence carries out stationary test respectively；

(2) continuity check submodule, is connected with syndrome module, for not by the traffic flow to be tested of stationary test Amount sequence carries out continuity check, if the seriality of not meeting, described continuity check submodule uses average interpolation method to enter data Row polishing；

(3) misarrangement submodule, is connected with continuity check submodule, for deleting the data of apparent error, uses average simultaneously Interpolation method carries out polishing to data；

(4) difference processing submodule, connects misarrangement submodule and syndrome module, for carrying out the data after polishing at difference Reason, and the data after difference processing are sent to syndrome module.

The present embodiment arranges data categorization module and stationarity inspection module, adds the accuracy of data, and makes the prediction of structure Model is more targeted；Calculation of correlation factor module, temporal and spatial correlations coefficient matrix generation module, history correlation matrix are set Generation module, predictor choose module and forecast model construction module, eliminate the subjectivity that initial predictor is chosen, energy Enough increase precision of prediction, make forecast model construction module more stable and accurate；The present embodiment value L=8, M=3, it was predicted that Precision improves 1.5% relative to correlation technique.

Embodiment 2

See Fig. 1, Fig. 2, the present embodiment one traffic flow monitoring and controlling forecast system, be connected including supervising device with supervising device Prediction means, described supervising device includes:

Being sequentially arranged at the photographic head of vehicle front, rear and the left and right sides, each photographic head connects monitoring processor, described Monitoring processor connects the watch-dog being positioned at driver's cabin, and described monitoring processor includes:

For gathering the acquisition module of the video image of each photographic head；

For at least one width video image gathered being zoomed in and out the image conversion module of process；

Each video image after processing exports the image display of watch-dog；

It is characterized in that: described dynamic vehicle monitoring device also includes that the camera lens for driving left and right photographic head is along front and back To the angle controller rotated, described angle controller is arranged on the arranged on left and right sides of vehicle, described left and right photographic head with Described angle controller connects, and described angle controller connects described monitoring processor, and described monitoring processor also includes: For starting angle controller, and the Dynamic control module of the anglec of rotation in direction before and after left and right photographic head is set, described Dynamic control module connects described angle controller.

Preferably, described angle controller is controlled electro-motor, the output shaft of described controlled electro-motor and described photographic head Base connect.

Preferably, described output shaft and the central axis of vehicle fore-and-aft direction.

Preferably, it is characterized in that, it was predicted that acquisition module that device includes being sequentially connected with, data preprocessing module, data classification mould Block, stationary test module, Calculation of correlation factor module, threshold value setting module, temporal and spatial correlations coefficient matrix generation module, go through History correlation matrix generation module, predictor choose module and forecast model construction module:

(1) acquisition module, is used for gathering observation section S in road network S_i, prediction section S_jThe traffic flow of corresponding each time period Data and passage situation；

(2) data preprocessing module, for described traffic flow data carries out data prediction, and rejecting does not meets traffic in fact The data of border situation；

(3) data categorization module, for carrying out classification of type, described type bag to the traffic flow data through data prediction Include traffic flow data festivals or holidays, traffic flow data at weekend and traffic flow data on working day；

(4) stationary test module, for being in same type of observation section S_iTraffic flow sequence X_iWith prediction road Section S_jTraffic flow sequence X_jCarrying out stationary test respectively, the auto-correlation function of inspection stationarity is:

$P\left(\mathrm{\τ}\right)=\frac{E\[(X_x-{\mathrm{\ν}}_i)(X_{x+\mathrm{\τ}}-{\mathrm{\ν}}_{x+\mathrm{\τ}})\]}{{\mathrm{\σ}}^2}$

Wherein, X_xRepresent traffic flow sequence to be tested, ν_iRepresent the average of traffic flow sequence to be tested, X_x+τRepresent X_x? Traffic flow sequence after time delay τ, ν_x+τFor X_x+τAverage, σ²For X_xWith X_x+τBetween variance；

When auto-correlation function P (τ) can level off to 0 or fluctuate 0 near by rapid decay, and the most described traffic flow sequence to be tested is logical Cross stationary test；When auto-correlation function P (τ) can not rapid decay level off to 0 or near 0 fluctuate, then to described to be tested Traffic flow sequence proceeds stationary test after carrying out calm disposing；

(5) Calculation of correlation factor module, for calculating the observation section S by stationary test_iTraffic flow sequence X_iWith Prediction section S_jTraffic flow sequence X_jTime correlation coefficient ρ under time delay τ_ij(τ) with space correlation coefficient ρ_ij(w), If having N number of section, traffic flow sequence X in road network S_i=[x_i(1),x_i(2),...,x_i(n)], traffic flow sequencex_iT () represents observation section S_iAt the flow of t, x_jT () represents prediction section S_jIn t Flow, t=1,2 ... n, time correlation coefficient ρ_ij(τ) computing formula is:

${\mathrm{\ρ}}_{ij}\left(\mathrm{\τ}\right)=\frac{{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_i\left(t\right)x_j(t+\mathrm{\τ})-\frac{1}{n-\mathrm{\τ}}{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_i\left(t\right){\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_j(t+\mathrm{\τ})}{{\sqrt{{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}\[x_i\left(t\right)-\frac{1}{n-\mathrm{\τ}}{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_i\left(t\right)\]}}^2\×{\sqrt{{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}\[x_j(t+\mathrm{\τ})-\frac{1}{n-\mathrm{\τ}}{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_j(t+\mathrm{\τ})\]}}^2}$

Space correlation coefficient ρ_ijW the computing formula of () is:

Preferably, it is characterized in that, it was predicted that device also includes:

(6) threshold value setting module, for setting time delay maximum L, the temporal and spatial correlations coefficient threshold T between each section₁With History correlation coefficient threshold T₂；

(7) temporal and spatial correlations coefficient matrix generation module, for the time correlation coefficient ρ according to each section_ij(τ) with space correlation system Number ρ_ijW () builds each observation section S_iWith prediction section S_jTemporal and spatial correlations coefficient matrix ρ (τ) ' under different time postpones τ, and Calculate the temporal and spatial correlations coefficient ρ in each section_ij(τ) ', wherein i ∈ [1, N] and τ ∈ [0, L], the span of L is [8,12], time Kongxiang The computing formula closing coefficient matrix ρ (τ) ' is:

$\mathrm{\ρ}{\left(\mathrm{\τ}\right)}^{\′}=\left[\begin{array}{cccc}{\mathrm{\ρ}}_{1j}{\left(0\right)}^{\′}& {\mathrm{\ρ}}_{1j}{\left(0\right)}^{\′}& \mathrm{...}& {\mathrm{\ρ}}_{Nj}{\left(0\right)}^{\′}\\ {\mathrm{\ρ}}_{1j}{\left(1\right)}^{\′}& {\mathrm{\ρ}}_{2j}{\left(1\right)}^{\′}& \mathrm{...}& {\mathrm{\ρ}}_{Nj}{\left(1\right)}^{\′}\\ \mathrm{...}& \mathrm{...}& \mathrm{...}& \mathrm{...}\\ {\mathrm{\ρ}}_{1j}{\left(L\right)}^{\′}& {\mathrm{\ρ}}_{2j}{\left(L\right)}^{\′}& \mathrm{...}& {\mathrm{\ρ}}_{Nj}{\left(L\right)}^{\′}\end{array}\right];$

Temporal and spatial correlations coefficient ρ_ij(τ) ' computing formula be:

ρ_ij(τ) '=ρ_ij(τ)ρ_ij(w)；

(8) history correlation matrix generation module, is used for generating prediction section S_jHistory correlation matrix ρ (t):

Wherein, the same period all for nearly M and same type of historical traffic are chosen as traffic flow sequence X_jHistory correlated series, It is designated asM=1,2 ... the span of M, M is [3,5], described history correlation coefficient ρ_jmT the computing formula of () is:

${\mathrm{\ρ}}_{jm}\left(t\right)=\frac{{\mathrm{\Σ}}_{t=1}^n{x}_j\left(t\right)x_{jm}\left(t\right)-\frac{1}{n}{\mathrm{\Σ}}_{t=1}^n{x}_j\left(t\right){\mathrm{\Σ}}_{t=1}^n{x}_{jm}\left(t\right)}{\sqrt{{\mathrm{\Σ}}_{t=1}^n{\[x_j\left(t\right)-\frac{1}{n}{\mathrm{\Σ}}_{t=1}^n{x}_j\left(t\right)\]}^2}\×\sqrt{{\mathrm{\Σ}}_{t=1}^n{\[x_{jm}\left(t\right)-\frac{1}{n}{\mathrm{\Σ}}_{t=1}^n{x}_{jm}\left(t\right)\]}^2}}$

(9) predictor chooses module, for according to described temporal and spatial correlations coefficient threshold T₁With history correlation coefficient threshold T₂Choose The predictor relevant to prediction impact point, and carry out matrix reconstruction according to locus j selected by it with time delay τ, choose Principle is:

If ρ_ij(τ) ' ＞ T₁, then will observation section S_iTraffic flow sequence X_iIn meet the traffic flow new sequence of composition of condition And as the first predictor, it is denoted as X', X'=(x₁',x₂',...,x_p'), wherein p is the described traffic flow number meeting condition, If L₁It is the maximum of time delay, L in the first predictor₁=max{ τ | τ ∈ [0, L] and ρ_ij(τ) ' ＞ T₁, then the first prediction Factor X' can state following matrix form as:

$X^{\′}=\left[\begin{array}{cccc}{x_1}^{\′}\left(1\right)& {x_2}^{\′}\left(1\right)& \mathrm{...}& {x_p}^{\′}\left(1\right)\\ {x_1}^{\′}\left(2\right)& {x_2}^{\′}\left(2\right)& \mathrm{...}& {x_p}^{\′}\left(2\right)\\ \mathrm{...}& \mathrm{...}& \mathrm{...}& \mathrm{...}\\ {x_1}^{\′}(n-L_1)& {x_2}^{\′}(n-L_1)& \mathrm{...}& {x_p}^{\′}(n-L_1)\end{array}\right];$

If ρ_jm(t) ＞ T₂, then by all history correlated series X meeting condition_jmT (), as the second predictor, is denoted as Y', Y'={y₁',y₂',...,y_q', wherein q is the historical traffic number meeting condition, and the second predictor Y' can state following rectangular as Formula:

(10) forecast model construction module, it is by constructing the first predictor and the second predictor as training sample Measurable section is at the forecast model of the traffic flow of subsequent time.

Wherein, in described data preprocessing module, the rule of the data not meeting traffic practical situation described in rejecting is: at one In the data update cycle, set the threshold range of total traffic flow data in each section respectively, if total friendship in certain section collected Through-current capacity data fall in corresponding threshold range, then show that these group data are reliable, retain this group data；If certain road collected Total traffic flow data of section falls not in corresponding threshold range, then show that these group data are unreliable, and rejected.

Wherein, described stationary test module includes following submodule:

(1) syndrome module, for the traffic flow to the traffic flow sequence with prediction section being in same type of observation section Amount sequence carries out stationary test respectively；

(2) continuity check submodule, is connected with syndrome module, for not by the traffic flow to be tested of stationary test Amount sequence carries out continuity check, if the seriality of not meeting, described continuity check submodule uses average interpolation method to enter data Row polishing；

(3) misarrangement submodule, is connected with continuity check submodule, for deleting the data of apparent error, uses average simultaneously Interpolation method carries out polishing to data；

(4) difference processing submodule, connects misarrangement submodule and syndrome module, for carrying out the data after polishing at difference Reason, and the data after difference processing are sent to syndrome module.

The present embodiment arranges data categorization module and stationarity inspection module, adds the accuracy of data, and makes the prediction of structure Model is more targeted；Calculation of correlation factor module, temporal and spatial correlations coefficient matrix generation module, history correlation matrix are set Generation module, predictor choose module and forecast model construction module, eliminate the subjectivity that initial predictor is chosen, energy Enough increase precision of prediction, make forecast model construction module more stable and accurate；The present embodiment value L=9, M=3, it was predicted that Precision improves 2% relative to correlation technique.

Embodiment 3

See Fig. 1, Fig. 2, the present embodiment one traffic flow monitoring and controlling forecast system, be connected including supervising device with supervising device Prediction means, described supervising device includes:

Being sequentially arranged at the photographic head of vehicle front, rear and the left and right sides, each photographic head connects monitoring processor, described Monitoring processor connects the watch-dog being positioned at driver's cabin, and described monitoring processor includes:

For gathering the acquisition module of the video image of each photographic head；

For at least one width video image gathered being zoomed in and out the image conversion module of process；

Each video image after processing exports the image display of watch-dog；

It is characterized in that: described dynamic vehicle monitoring device also includes that the camera lens for driving left and right photographic head is along front and back To the angle controller rotated, described angle controller is arranged on the arranged on left and right sides of vehicle, described left and right photographic head with Described angle controller connects, and described angle controller connects described monitoring processor, and described monitoring processor also includes: For starting angle controller, and the Dynamic control module of the anglec of rotation in direction before and after left and right photographic head is set, described Dynamic control module connects described angle controller.

Preferably, described angle controller is controlled electro-motor, the output shaft of described controlled electro-motor and described photographic head Base connect.

Preferably, described output shaft and the central axis of vehicle fore-and-aft direction.

Preferably, it is characterized in that, it was predicted that acquisition module that device includes being sequentially connected with, data preprocessing module, data classification mould Block, stationary test module, Calculation of correlation factor module, threshold value setting module, temporal and spatial correlations coefficient matrix generation module, go through History correlation matrix generation module, predictor choose module and forecast model construction module:

(1) acquisition module, is used for gathering observation section S in road network S_i, prediction section S_jThe traffic flow of corresponding each time period Data and passage situation；

(2) data preprocessing module, for described traffic flow data carries out data prediction, and rejecting does not meets traffic in fact The data of border situation；

(3) data categorization module, for carrying out classification of type, described type bag to the traffic flow data through data prediction Include traffic flow data festivals or holidays, traffic flow data at weekend and traffic flow data on working day；

(4) stationary test module, for being in same type of observation section S_iTraffic flow sequence X_iWith prediction road Section S_jTraffic flow sequence X_jCarrying out stationary test respectively, the auto-correlation function of inspection stationarity is:

$P\left(\mathrm{\τ}\right)=\frac{E\[(X_x-{\mathrm{\ν}}_i)(X_{x+\mathrm{\τ}}-{\mathrm{\ν}}_{x+\mathrm{\τ}})\]}{{\mathrm{\σ}}^2}$

Wherein, X_xRepresent traffic flow sequence to be tested, ν_iRepresent the average of traffic flow sequence to be tested, X_x+τRepresent X_x? Traffic flow sequence after time delay τ, ν_x+τFor X_x+τAverage, σ²For X_xWith X_x+τBetween variance；

When auto-correlation function P (τ) can level off to 0 or fluctuate 0 near by rapid decay, and the most described traffic flow sequence to be tested is logical Cross stationary test；When auto-correlation function P (τ) can not rapid decay level off to 0 or near 0 fluctuate, then to described to be tested Traffic flow sequence proceeds stationary test after carrying out calm disposing；

(5) Calculation of correlation factor module, for calculating the observation section S by stationary test_iTraffic flow sequence X_iWith Prediction section S_jTraffic flow sequence X_jTime correlation coefficient ρ under time delay τ_ij(τ) with space correlation coefficient ρ_ij(w), If road network_SInside have_NIndividual section, traffic flow sequence X_i=[x_i(1),x_i(2),...,x_i(n)], traffic flow sequencex_iT () represents observation section S_iAt the flow of t, x_jT () represents prediction section S_jIn t Flow, t=1,2 ... n, time correlation coefficient ρ_ij(τ) computing formula is:

${\mathrm{\ρ}}_{ij}\left(\mathrm{\τ}\right)=\frac{{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_i\left(t\right)x_j(t+\mathrm{\τ})-\frac{1}{n-\mathrm{\τ}}{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_i\left(t\right){\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_j(t+\mathrm{\τ})}{{\sqrt{{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}\[x_i\left(t\right)-\frac{1}{n-\mathrm{\τ}}{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_i\left(t\right)\]}}^2\×{\sqrt{{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}\[x_j(t+\mathrm{\τ})-\frac{1}{n-\mathrm{\τ}}{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_j(t+\mathrm{\τ})\]}}^2}$

Space correlation coefficient ρ_ijW the computing formula of () is:

Preferably, it is characterized in that, it was predicted that device also includes:

(6) threshold value setting module, for setting time delay maximum L, the temporal and spatial correlations coefficient threshold T between each section₁With History correlation coefficient threshold T₂；

(7) temporal and spatial correlations coefficient matrix generation module, for the time correlation coefficient ρ according to each section_ij(τ) with space correlation system Number ρ_ijW () builds each observation section S_iWith prediction section S_jTemporal and spatial correlations coefficient matrix ρ (τ) ' under different time postpones τ, and Calculate the temporal and spatial correlations coefficient ρ in each section_ij(τ) ', wherein i ∈ [1, N] and τ ∈ [0, L], the span of L is [8,12], time Kongxiang The computing formula closing coefficient matrix ρ (τ) ' is:

$\mathrm{\ρ}{\left(\mathrm{\τ}\right)}^{\′}=\left[\begin{array}{cccc}{\mathrm{\ρ}}_{1j}{\left(0\right)}^{\′}& {\mathrm{\ρ}}_{1j}{\left(0\right)}^{\′}& \mathrm{...}& {\mathrm{\ρ}}_{Nj}{\left(0\right)}^{\′}\\ {\mathrm{\ρ}}_{1j}{\left(1\right)}^{\′}& {\mathrm{\ρ}}_{2j}{\left(1\right)}^{\′}& \mathrm{...}& {\mathrm{\ρ}}_{Nj}{\left(1\right)}^{\′}\\ \mathrm{...}& \mathrm{...}& \mathrm{...}& \mathrm{...}\\ {\mathrm{\ρ}}_{1j}{\left(L\right)}^{\′}& {\mathrm{\ρ}}_{2j}{\left(L\right)}^{\′}& \mathrm{...}& {\mathrm{\ρ}}_{Nj}{\left(L\right)}^{\′}\end{array}\right];$

Temporal and spatial correlations coefficient ρ_ij(τ) ' computing formula be:

ρ_ij(τ) '=ρ_ij(τ)ρ_ij(w)；

(8) history correlation matrix generation module, is used for generating prediction section S_jHistory correlation matrix ρ (t):

Wherein, the same period all for nearly M and same type of historical traffic are chosen as traffic flow sequence X_jHistory correlated series, It is designated asM=1,2 ... the span of M, M is [3,5], described history correlation coefficient ρ_jmT the computing formula of () is:

${\mathrm{\ρ}}_{jm}\left(t\right)=\frac{{\mathrm{\Σ}}_{t=1}^n{x}_j\left(t\right)x_{jm}\left(t\right)-\frac{1}{n}{\mathrm{\Σ}}_{t=1}^n{x}_j\left(t\right){\mathrm{\Σ}}_{t=1}^n{x}_{jm}\left(t\right)}{\sqrt{{\mathrm{\Σ}}_{t=1}^n{\[x_j\left(t\right)-\frac{1}{n}{\mathrm{\Σ}}_{t=1}^n{x}_j\left(t\right)\]}^2}\×\sqrt{{\mathrm{\Σ}}_{t=1}^n{\[x_{jm}\left(t\right)-\frac{1}{n}{\mathrm{\Σ}}_{t=1}^n{x}_{jm}\left(t\right)\]}^2}}$

(9) predictor chooses module, for according to described temporal and spatial correlations coefficient threshold T₁With history correlation coefficient threshold T₂Choose The predictor relevant to prediction impact point, and carry out matrix reconstruction according to locus j selected by it with time delay τ, choose Principle is:

If ρ_ij(τ) ' ＞ T₁, then will observation section S_iTraffic flow sequence X_iIn meet the traffic flow new sequence of composition of condition And as the first predictor, it is denoted as X', X'=(x₁',x₂',...,x_p'), wherein p is the described traffic flow number meeting condition, If L₁It is the maximum of time delay, L in the first predictor₁=max{ τ | τ ∈ [0, L] and ρ_ij(τ) ' ＞ T₁, then the first prediction Factor X' can state following matrix form as:

$X^{\′}=\left[\begin{array}{cccc}{x_1}^{\′}\left(1\right)& {x_2}^{\′}\left(1\right)& \mathrm{...}& {x_p}^{\′}\left(1\right)\\ {x_1}^{\′}\left(2\right)& {x_2}^{\′}\left(2\right)& \mathrm{...}& {x_p}^{\′}\left(2\right)\\ \mathrm{...}& \mathrm{...}& \mathrm{...}& \mathrm{...}\\ {x_1}^{\′}(n-L_1)& {x_2}^{\′}(n-L_1)& \mathrm{...}& {x_p}^{\′}(n-L_1)\end{array}\right];$

If ρ_jm(t) ＞ T₂, then by all history correlated series X meeting condition_jmT (), as the second predictor, is denoted as Y', Y'={y₁',y₂',...,y_q', wherein q is the historical traffic number meeting condition, and the second predictor Y' can state following rectangular as Formula:

(10) forecast model construction module, it is by constructing the first predictor and the second predictor as training sample Measurable section is at the forecast model of the traffic flow of subsequent time.

Wherein, in described data preprocessing module, the rule of the data not meeting traffic practical situation described in rejecting is: at one In the data update cycle, set the threshold range of total traffic flow data in each section respectively, if total friendship in certain section collected Through-current capacity data fall in corresponding threshold range, then show that these group data are reliable, retain this group data；If certain road collected Total traffic flow data of section falls not in corresponding threshold range, then show that these group data are unreliable, and rejected.

Wherein, described stationary test module includes following submodule:

(1) syndrome module, for the traffic flow to the traffic flow sequence with prediction section being in same type of observation section Amount sequence carries out stationary test respectively；

(2) continuity check submodule, is connected with syndrome module, for not by the traffic flow to be tested of stationary test Amount sequence carries out continuity check, if the seriality of not meeting, described continuity check submodule uses average interpolation method to enter data Row polishing；

(3) misarrangement submodule, is connected with continuity check submodule, for deleting the data of apparent error, uses average simultaneously Interpolation method carries out polishing to data；

(4) difference processing submodule, connects misarrangement submodule and syndrome module, for carrying out the data after polishing at difference Reason, and the data after difference processing are sent to syndrome module.

The present embodiment arranges data categorization module and stationarity inspection module, adds the accuracy of data, and makes the prediction of structure Model is more targeted；Calculation of correlation factor module, temporal and spatial correlations coefficient matrix generation module, history correlation matrix are set Generation module, predictor choose module and forecast model construction module, eliminate the subjectivity that initial predictor is chosen, energy Enough increase precision of prediction, make forecast model construction module more stable and accurate；The present embodiment value L=10, M=4, in advance Survey precision and improve 2.6% relative to correlation technique.

Embodiment 4

See Fig. 1, Fig. 2, the present embodiment one traffic flow monitoring and controlling forecast system, be connected including supervising device with supervising device Prediction means, described supervising device includes:

Being sequentially arranged at the photographic head of vehicle front, rear and the left and right sides, each photographic head connects monitoring processor, described Monitoring processor connects the watch-dog being positioned at driver's cabin, and described monitoring processor includes:

For gathering the acquisition module of the video image of each photographic head；

For at least one width video image gathered being zoomed in and out the image conversion module of process；

Each video image after processing exports the image display of watch-dog；

It is characterized in that: described dynamic vehicle monitoring device also includes that the camera lens for driving left and right photographic head is along front and back To the angle controller rotated, described angle controller is arranged on the arranged on left and right sides of vehicle, described left and right photographic head with Described angle controller connects, and described angle controller connects described monitoring processor, and described monitoring processor also includes: For starting angle controller, and the Dynamic control module of the anglec of rotation in direction before and after left and right photographic head is set, described Dynamic control module connects described angle controller.

Preferably, described angle controller is controlled electro-motor, the output shaft of described controlled electro-motor and described photographic head Base connect.

Preferably, described output shaft and the central axis of vehicle fore-and-aft direction.

Preferably, it is characterized in that, it was predicted that acquisition module that device includes being sequentially connected with, data preprocessing module, data classification mould Block, stationary test module, Calculation of correlation factor module, threshold value setting module, temporal and spatial correlations coefficient matrix generation module, go through History correlation matrix generation module, predictor choose module and forecast model construction module:

(1) acquisition module, is used for gathering observation section S in road network S_i, prediction section S_jThe traffic flow of corresponding each time period Data and passage situation；

(2) data preprocessing module, for described traffic flow data carries out data prediction, and rejecting does not meets traffic in fact The data of border situation；

(3) data categorization module, for carrying out classification of type, described type bag to the traffic flow data through data prediction Include traffic flow data festivals or holidays, traffic flow data at weekend and traffic flow data on working day；

(4) stationary test module, for being in same type of observation section S_iTraffic flow sequence X_iWith prediction road Section S_jTraffic flow sequence X_jCarrying out stationary test respectively, the auto-correlation function of inspection stationarity is:

$P\left(\mathrm{\τ}\right)=\frac{E\[(X_x-{\mathrm{\ν}}_i)(X_{x+\mathrm{\τ}}-{\mathrm{\ν}}_{x+\mathrm{\τ}})\]}{{\mathrm{\σ}}^2}$

Wherein, X_xRepresent traffic flow sequence to be tested, ν_iRepresent the average of traffic flow sequence to be tested, X_x+τRepresent X_x? Traffic flow sequence after time delay τ, ν_x+τFor X_x+τAverage, σ²For X_xWith X_x+τBetween variance；

When auto-correlation function P (τ) can level off to 0 or fluctuate 0 near by rapid decay, and the most described traffic flow sequence to be tested is logical Cross stationary test；When auto-correlation function P (τ) can not rapid decay level off to 0 or near 0 fluctuate, then to described to be tested Traffic flow sequence proceeds stationary test after carrying out calm disposing；

(5) Calculation of correlation factor module, for calculating the observation section S by stationary test_iTraffic flow sequence X_iWith Prediction section S_jTraffic flow sequence X_jTime correlation coefficient ρ under time delay τ_ij(τ) with space correlation coefficient ρ_ij(w), If having N number of section, traffic flow sequence X in road network S_i=[x_i(1),x_i(2),...,x_i(n)], traffic flow sequencex_iT () represents observation section S_iAt the flow of t, x_jT () represents prediction section S_jIn t Flow, t=1,2 ... n, time correlation coefficient ρ_ij(τ) computing formula is:

${\mathrm{\ρ}}_{ij}\left(\mathrm{\τ}\right)=\frac{{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_i\left(t\right)x_j(t+\mathrm{\τ})-\frac{1}{n-\mathrm{\τ}}{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_i\left(t\right){\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_j(t+\mathrm{\τ})}{{\sqrt{{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}\[x_i\left(t\right)-\frac{1}{n-\mathrm{\τ}}{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_i\left(t\right)\]}}^2\×{\sqrt{{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}\[x_j(t+\mathrm{\τ})-\frac{1}{n-\mathrm{\τ}}{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_j(t+\mathrm{\τ})\]}}^2}$

Space correlation coefficient ρ_ijW the computing formula of () is:

Preferably, it is characterized in that, it was predicted that device also includes:

(6) threshold value setting module, for setting time delay maximum L, the temporal and spatial correlations coefficient threshold T between each section₁With History correlation coefficient threshold T₂；

(7) temporal and spatial correlations coefficient matrix generation module, for the time correlation coefficient ρ according to each section_ij(τ) with space correlation system Number ρ_ijW () builds each observation section S_iWith prediction section S_jTemporal and spatial correlations coefficient matrix ρ (τ) ' under different time postpones τ, and Calculate the temporal and spatial correlations coefficient ρ in each section_ij(τ) ', wherein i ∈ [1, N] and τ ∈ [0, L], the span of L is [8,12], time Kongxiang The computing formula closing coefficient matrix ρ (τ) ' is:

$\mathrm{\ρ}{\left(\mathrm{\τ}\right)}^{\′}=\left[\begin{array}{cccc}{\mathrm{\ρ}}_{1j}{\left(0\right)}^{\′}& {\mathrm{\ρ}}_{1j}{\left(0\right)}^{\′}& \mathrm{...}& {\mathrm{\ρ}}_{Nj}{\left(0\right)}^{\′}\\ {\mathrm{\ρ}}_{1j}{\left(1\right)}^{\′}& {\mathrm{\ρ}}_{2j}{\left(1\right)}^{\′}& \mathrm{...}& {\mathrm{\ρ}}_{Nj}{\left(1\right)}^{\′}\\ \mathrm{...}& \mathrm{...}& \mathrm{...}& \mathrm{...}\\ {\mathrm{\ρ}}_{1j}{\left(L\right)}^{\′}& {\mathrm{\ρ}}_{2j}{\left(L\right)}^{\′}& \mathrm{...}& {\mathrm{\ρ}}_{Nj}{\left(L\right)}^{\′}\end{array}\right];$

Temporal and spatial correlations coefficient ρ_ij(τ) ' computing formula be:

ρ_ij(τ) '=ρ_ij(τ)ρ_ij(w)；

(8) history correlation matrix generation module, is used for generating prediction section S_jHistory correlation matrix ρ (t):

Wherein, the same period all for nearly M and same type of historical traffic are chosen as traffic flow sequence X_jHistory correlated series, It is designated asM=1,2 ... the span of M, M is [3,5], described history correlation coefficient ρ_jmT the computing formula of () is:

${\mathrm{\ρ}}_{jm}\left(t\right)=\frac{{\mathrm{\Σ}}_{t=1}^n{x}_j\left(t\right)x_{jm}\left(t\right)-\frac{1}{n}{\mathrm{\Σ}}_{t=1}^n{x}_j\left(t\right){\mathrm{\Σ}}_{t=1}^n{x}_{jm}\left(t\right)}{\sqrt{{\mathrm{\Σ}}_{t=1}^n{\[x_j\left(t\right)-\frac{1}{n}{\mathrm{\Σ}}_{t=1}^n{x}_j\left(t\right)\]}^2}\×\sqrt{{\mathrm{\Σ}}_{t=1}^n{\[x_{jm}\left(t\right)-\frac{1}{n}{\mathrm{\Σ}}_{t=1}^n{x}_{jm}\left(t\right)\]}^2}}$

(9) predictor chooses module, for according to described temporal and spatial correlations coefficient threshold T₁With history correlation coefficient threshold T₂Choose The predictor relevant to prediction impact point, and carry out matrix reconstruction according to locus j selected by it with time delay τ, choose Principle is:

If ρ_ij(τ) ' ＞ T₁, then will observation section S_iTraffic flow sequence X_iIn meet the traffic flow new sequence of composition of condition And as the first predictor, it is denoted as X', X'=(x₁',x₂',...,x_p'), wherein p is the described traffic flow number meeting condition, If L₁It is the maximum of time delay, L in the first predictor₁=max{ τ | τ ∈ [0, L] and ρ_ij(τ) ' ＞ T₁, then the first prediction Factor X' can state following matrix form as:

$X^{\′}=\left[\begin{array}{cccc}{x_1}^{\′}\left(1\right)& {x_2}^{\′}\left(1\right)& \mathrm{...}& {x_p}^{\′}\left(1\right)\\ {x_1}^{\′}\left(2\right)& {x_2}^{\′}\left(2\right)& \mathrm{...}& {x_p}^{\′}\left(2\right)\\ \mathrm{...}& \mathrm{...}& \mathrm{...}& \mathrm{...}\\ {x_1}^{\′}(n-L_1)& {x_2}^{\′}(n-L_1)& \mathrm{...}& {x_p}^{\′}(n-L_1)\end{array}\right];$

If ρ_jm(t) ＞ T₂, then by all history correlated series X meeting condition_jmT (), as the second predictor, is denoted as Y', Y'={y₁',y₂',...,y_q', wherein q is the historical traffic number meeting condition, and the second predictor Y' can state following rectangular as Formula:

(10) forecast model construction module, it is by constructing the first predictor and the second predictor as training sample Measurable section is at the forecast model of the traffic flow of subsequent time.

Wherein, in described data preprocessing module, the rule of the data not meeting traffic practical situation described in rejecting is: at one In the data update cycle, set the threshold range of total traffic flow data in each section respectively, if total friendship in certain section collected Through-current capacity data fall in corresponding threshold range, then show that these group data are reliable, retain this group data；If certain road collected Total traffic flow data of section falls not in corresponding threshold range, then show that these group data are unreliable, and rejected.

Wherein, described stationary test module includes following submodule:

(1) syndrome module, for the traffic flow to the traffic flow sequence with prediction section being in same type of observation section Amount sequence carries out stationary test respectively；

(2) continuity check submodule, is connected with syndrome module, for not by the traffic flow to be tested of stationary test Amount sequence carries out continuity check, if the seriality of not meeting, described continuity check submodule uses average interpolation method to enter data Row polishing；

(3) misarrangement submodule, is connected with continuity check submodule, for deleting the data of apparent error, uses average simultaneously Interpolation method carries out polishing to data；

(4) difference processing submodule, connects misarrangement submodule and syndrome module, for carrying out the data after polishing at difference Reason, and the data after difference processing are sent to syndrome module.

The present embodiment arranges data categorization module and stationarity inspection module, adds the accuracy of data, and makes the prediction of structure Model is more targeted；Calculation of correlation factor module, temporal and spatial correlations coefficient matrix generation module, history correlation matrix are set Generation module, predictor choose module and forecast model construction module, eliminate the subjectivity that initial predictor is chosen, energy Enough increase precision of prediction, make forecast model construction module more stable and accurate；The present embodiment value L=11, M=5, it was predicted that Precision improves 3.2% relative to correlation technique.

Embodiment 5

See Fig. 1, Fig. 2, the present embodiment one traffic flow monitoring and controlling forecast system, be connected including supervising device with supervising device Prediction means, described supervising device includes:

Being sequentially arranged at the photographic head of vehicle front, rear and the left and right sides, each photographic head connects monitoring processor, described Monitoring processor connects the watch-dog being positioned at driver's cabin, and described monitoring processor includes:

For gathering the acquisition module of the video image of each photographic head；

For at least one width video image gathered being zoomed in and out the image conversion module of process；

Each video image after processing exports the image display of watch-dog；

It is characterized in that: described dynamic vehicle monitoring device also includes that the camera lens for driving left and right photographic head is along front and back To the angle controller rotated, described angle controller is arranged on the arranged on left and right sides of vehicle, described left and right photographic head with Described angle controller connects, and described angle controller connects described monitoring processor, and described monitoring processor also includes: For starting angle controller, and the Dynamic control module of the anglec of rotation in direction before and after left and right photographic head is set, described Dynamic control module connects described angle controller.

Preferably, described angle controller is controlled electro-motor, the output shaft of described controlled electro-motor and described photographic head Base connect.

Preferably, described output shaft and the central axis of vehicle fore-and-aft direction.

Preferably, it is characterized in that, it was predicted that acquisition module that device includes being sequentially connected with, data preprocessing module, data classification mould Block, stationary test module, Calculation of correlation factor module, threshold value setting module, temporal and spatial correlations coefficient matrix generation module, go through History correlation matrix generation module, predictor choose module and forecast model construction module:

(1) acquisition module, is used for gathering observation section S in road network S_i, prediction section S_jThe traffic flow of corresponding each time period Data and passage situation；

(2) data preprocessing module, for described traffic flow data carries out data prediction, and rejecting does not meets traffic in fact The data of border situation；

(3) data categorization module, for carrying out classification of type, described type bag to the traffic flow data through data prediction Include traffic flow data festivals or holidays, traffic flow data at weekend and traffic flow data on working day；

(4) stationary test module, for being in same type of observation section S_iTraffic flow sequence X_iWith prediction road Section S_jTraffic flow sequence X_jCarrying out stationary test respectively, the auto-correlation function of inspection stationarity is:

$P\left(\mathrm{\τ}\right)=\frac{E\[(X_x-{\mathrm{\ν}}_i)(X_{x+\mathrm{\τ}}-{\mathrm{\ν}}_{x+\mathrm{\τ}})\]}{{\mathrm{\σ}}^2}$

Wherein, X_xRepresent traffic flow sequence to be tested, ν_iRepresent the average of traffic flow sequence to be tested, X_x+τRepresent X_x? Traffic flow sequence after time delay τ, ν_x+τFor X_x+τAverage, σ²For X_xWith X_x+τBetween variance；

When auto-correlation function P (τ) can level off to 0 or fluctuate 0 near by rapid decay, and the most described traffic flow sequence to be tested is logical Cross stationary test；When auto-correlation function P (τ) can not rapid decay level off to 0 or near 0 fluctuate, then to described to be tested Traffic flow sequence proceeds stationary test after carrying out calm disposing；

(5) Calculation of correlation factor module, for calculating the observation section S by stationary test_iTraffic flow sequence X_iWith Prediction section S_jTraffic flow sequence X_jTime correlation coefficient ρ under time delay τ_ij(τ) with space correlation coefficient ρ_ij(w), If having N number of section, traffic flow sequence X in road network S_i=[x_i(1),x_i(2),...,x_i(n)], traffic flow sequencex_iT () represents observation section S_iAt the flow of t, x_jT () represents prediction section S_jIn t Flow, t=1,2 ... n, time correlation coefficient ρ_ij(τ) computing formula is:

${\mathrm{\ρ}}_{ij}\left(\mathrm{\τ}\right)=\frac{{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_i\left(t\right)x_j(t+\mathrm{\τ})-\frac{1}{n-\mathrm{\τ}}{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_i\left(t\right){\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_j(t+\mathrm{\τ})}{{\sqrt{{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}\[x_i\left(t\right)-\frac{1}{n-\mathrm{\τ}}{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_i\left(t\right)\]}}^2\×{\sqrt{{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}\[x_j(t+\mathrm{\τ})-\frac{1}{n-\mathrm{\τ}}{\mathrm{\Σ}}_{t=1}^{n-\mathrm{\τ}}{x}_j(t+\mathrm{\τ})\]}}^2}$

Space correlation coefficient ρ_ijW the computing formula of () is:

Preferably, it is characterized in that, it was predicted that device also includes:

(6) threshold value setting module, for setting time delay maximum L, the temporal and spatial correlations coefficient threshold T between each section₁With History correlation coefficient threshold T₂；

(7) temporal and spatial correlations coefficient matrix generation module, for the time correlation coefficient ρ according to each section_ij(τ) with space correlation system Number ρ_ijW () builds each observation section S_iWith prediction section S_jTemporal and spatial correlations coefficient matrix ρ (τ) ' under different time postpones τ, and Calculate the temporal and spatial correlations coefficient ρ in each section_ij(τ) ', wherein i ∈ [1, N] and τ ∈ [0, L], the span of L is [8,12], time Kongxiang The computing formula closing coefficient matrix ρ (τ) ' is:

$\mathrm{\ρ}{\left(\mathrm{\τ}\right)}^{\′}=\left[\begin{array}{cccc}{\mathrm{\ρ}}_{1j}{\left(0\right)}^{\′}& {\mathrm{\ρ}}_{1j}{\left(0\right)}^{\′}& \mathrm{...}& {\mathrm{\ρ}}_{Nj}{\left(0\right)}^{\′}\\ {\mathrm{\ρ}}_{1j}{\left(1\right)}^{\′}& {\mathrm{\ρ}}_{2j}{\left(1\right)}^{\′}& \mathrm{...}& {\mathrm{\ρ}}_{Nj}{\left(1\right)}^{\′}\\ \mathrm{...}& \mathrm{...}& \mathrm{...}& \mathrm{...}\\ {\mathrm{\ρ}}_{1j}{\left(L\right)}^{\′}& {\mathrm{\ρ}}_{2j}{\left(L\right)}^{\′}& \mathrm{...}& {\mathrm{\ρ}}_{Nj}{\left(L\right)}^{\′}\end{array}\right];$

Temporal and spatial correlations coefficient ρ_ij(τ) ' computing formula be:

ρ_ij(τ) '=ρ_ij(τ)ρ_ij(w)；

(8) history correlation matrix generation module, is used for generating prediction section S_jHistory correlation matrix ρ (t):

Wherein, the same period all for nearly M and same type of historical traffic are chosen as traffic flow sequence X_jHistory correlated series, It is designated asM=1,2 ... the span of M, M is [3,5], described history correlation coefficient ρ_jmT the computing formula of () is:

${\mathrm{\ρ}}_{jm}\left(t\right)=\frac{{\mathrm{\Σ}}_{t=1}^n{x}_j\left(t\right)x_{jm}\left(t\right)-\frac{1}{n}{\mathrm{\Σ}}_{t=1}^n{x}_j\left(t\right){\mathrm{\Σ}}_{t=1}^n{x}_{jm}\left(t\right)}{\sqrt{{\mathrm{\Σ}}_{t=1}^n{\[x_j\left(t\right)-\frac{1}{n}{\mathrm{\Σ}}_{t=1}^n{x}_j\left(t\right)\]}^2}\×\sqrt{{\mathrm{\Σ}}_{t=1}^n{\[x_{jm}\left(t\right)-\frac{1}{n}{\mathrm{\Σ}}_{t=1}^n{x}_{jm}\left(t\right)\]}^2}}$

(9) predictor chooses module, for according to described temporal and spatial correlations coefficient threshold T₁With history correlation coefficient threshold T₂Choose The predictor relevant to prediction impact point, and carry out matrix reconstruction according to locus j selected by it with time delay τ, choose Principle is:

If ρ_ij(τ) ' ＞ T₁, then will observation section S_iTraffic flow sequence X_iIn meet the traffic flow new sequence of composition of condition And as the first predictor, it is denoted as X', X'=(x₁',x₂',...,x_p'), wherein p is the described traffic flow number meeting condition, If L₁It is the maximum of time delay, L in the first predictor₁=max{ τ | τ ∈ [0, L] and ρ_ij(τ) ' ＞ T₁, then the first prediction Factor X' can state following matrix form as:

$X^{\′}=\left[\begin{array}{cccc}{x_1}^{\′}\left(1\right)& {x_2}^{\′}\left(1\right)& \mathrm{...}& {x_p}^{\′}\left(1\right)\\ {x_1}^{\′}\left(2\right)& {x_2}^{\′}\left(2\right)& \mathrm{...}& {x_p}^{\′}\left(2\right)\\ \mathrm{...}& \mathrm{...}& \mathrm{...}& \mathrm{...}\\ {x_1}^{\′}(n-L_1)& {x_2}^{\′}(n-L_1)& \mathrm{...}& {x_p}^{\′}(n-L_1)\end{array}\right];$

If ρ_jm(t) ＞ T₂, then by all history correlated series X meeting condition_jmT (), as the second predictor, is denoted as Y', Y'={y₁',y₂',...,y_q', wherein q is the historical traffic number meeting condition, and the second predictor Y' can state following rectangular as Formula:

(10) forecast model construction module, it is by constructing the first predictor and the second predictor as training sample Measurable section is at the forecast model of the traffic flow of subsequent time.

Wherein, in described data preprocessing module, the rule of the data not meeting traffic practical situation described in rejecting is: at one In the data update cycle, set the threshold range of total traffic flow data in each section respectively, if total friendship in certain section collected Through-current capacity data fall in corresponding threshold range, then show that these group data are reliable, retain this group data；If certain road collected Total traffic flow data of section falls not in corresponding threshold range, then show that these group data are unreliable, and rejected.

Wherein, described stationary test module includes following submodule:

(1) syndrome module, for the traffic flow to the traffic flow sequence with prediction section being in same type of observation section Amount sequence carries out stationary test respectively；

(2) continuity check submodule, is connected with syndrome module, for not by the traffic flow to be tested of stationary test Amount sequence carries out continuity check, if the seriality of not meeting, described continuity check submodule uses average interpolation method to enter data Row polishing；

(3) misarrangement submodule, is connected with continuity check submodule, for deleting the data of apparent error, uses average simultaneously Interpolation method carries out polishing to data；

(4) difference processing submodule, connects misarrangement submodule and syndrome module, for carrying out the data after polishing at difference Reason, and the data after difference processing are sent to syndrome module.

The present embodiment arranges data categorization module and stationarity inspection module, adds the accuracy of data, and makes the prediction of structure Model is more targeted；Calculation of correlation factor module, temporal and spatial correlations coefficient matrix generation module, history correlation matrix are set Generation module, predictor choose module and forecast model construction module, eliminate the subjectivity that initial predictor is chosen, energy Enough increase precision of prediction, make forecast model construction module more stable and accurate；The present embodiment value L=12, M=5, it was predicted that Precision improves 3.5% relative to correlation technique.

Last it should be noted that, above example is only in order to illustrate technical scheme, rather than to scope Restriction, although having made to explain to the present invention with reference to preferred embodiment, it will be understood by those within the art that, Technical scheme can be modified or equivalent, without deviating from the spirit and scope of technical solution of the present invention.

Claims (7)

1. a traffic flow monitoring and controlling forecast system, including supervising device and the prediction means that is connected with supervising device, described supervising device includes:

Being sequentially arranged at the photographic head of vehicle front, rear and the left and right sides, each photographic head connects monitoring processor, and described monitoring processor connects the watch-dog being positioned at driver's cabin, and described monitoring processor includes:

For gathering the acquisition module of the video image of each photographic head；

For at least one width video image gathered being zoomed in and out the image conversion module of process；

Each video image after processing exports the image display of watch-dog；

It is characterized in that: described dynamic vehicle monitoring device also includes for driving a left side, the angle controller that the camera lens of right photographic head rotates along fore-and-aft direction, described angle controller is arranged on a left side for vehicle, right both sides, a described left side, right photographic head is connected with described angle controller, described angle controller connects described monitoring processor, described monitoring processor also includes: be used for starting angle controller, and a left side is set, the Dynamic control module of the anglec of rotation in direction before and after right photographic head, described Dynamic control module connects described angle controller.

A kind of traffic flow monitoring and controlling forecast system the most according to claim 1, is characterized in that, described angle controller is controlled electro-motor, and the output shaft of described controlled electro-motor is connected with the base of described photographic head.

A kind of traffic flow monitoring and controlling forecast system the most according to claim 2, is characterized in that, described output shaft and the central axis of vehicle fore-and-aft direction.

A kind of traffic flow monitoring and controlling forecast system the most according to claim 3, it is characterized in that, described prediction means includes that the acquisition module being sequentially connected with, data preprocessing module, data categorization module, stationary test module, Calculation of correlation factor module, threshold value setting module, temporal and spatial correlations coefficient matrix generation module, history correlation matrix generation module, predictor choose module and forecast model construction module:

(1) acquisition module, is used for gathering observation section S in road network S_i, prediction section S_jThe traffic flow data of corresponding each time period and passage situation；

(2) data preprocessing module, for described traffic flow data carries out data prediction, and rejects the data not meeting traffic practical situation；

(3) data categorization module, for the traffic flow data through data prediction is carried out classification of type, described type includes traffic flow data festivals or holidays, traffic flow data at weekend and traffic flow data on working day；

(4) stationary test module, for being in same type of observation section S_iTraffic flow sequence X_iWith prediction section S_jTraffic flow sequence X_jCarrying out stationary test respectively, the auto-correlation function of inspection stationarity is:

Wherein, X_xRepresent traffic flow sequence to be tested, ν_iRepresent the average of traffic flow sequence to be tested, X_{x+ τ}Represent X_xTraffic flow sequence after time delay τ, ν_{x+ τ}For X_{x+ τ}Average, σ²For X_xWith X_{x+ τ}Between variance；

When auto-correlation function P (τ) can rapid decay level off to 0 or 0 near fluctuation, the most described traffic flow sequence to be tested passes through stationary test；When auto-correlation function P (τ) can not rapid decay level off to 0 or near 0 fluctuate, then proceed stationary test after described traffic flow sequence to be tested being carried out calm disposing；

(5) Calculation of correlation factor module, for calculating the observation section S by stationary test_iTraffic flow sequence X_iWith prediction section S_jTraffic flow sequence X_jTime correlation coefficient ρ under time delay τ_ij(τ) with space correlation coefficient ρ_ijW (), if having N number of section, traffic flow sequence X in road network S_i=[x_i(1),x_i(2),...,x_i(n)], traffic flow sequencex_iT () represents observation section S_iAt the flow of t, x_jT () represents prediction section S_jAt the flow of t, t=1,2 ... n, time correlation coefficient ρ_ij(τ) computing formula is:

Space correlation coefficient ρ_ijW the computing formula of () is:

。

A kind of traffic flow monitoring and controlling forecast system the most according to claim 4, is characterized in that,

(6) threshold value setting module, for setting time delay maximum L, the temporal and spatial correlations coefficient threshold T between each section₁With history correlation coefficient threshold T₂；

(7) temporal and spatial correlations coefficient matrix generation module, for the time correlation coefficient ρ according to each section_ij(τ) with space correlation coefficient ρ_ijW () builds each observation section S_iWith prediction section S_jTemporal and spatial correlations coefficient matrix ρ (τ) ' under different time postpones τ, and calculate the temporal and spatial correlations coefficient ρ in each section_ij(τ) ', wherein i ∈ [1, N] and τ ∈ [0, L], the span of L is [8,12], and the computing formula of temporal and spatial correlations coefficient matrix ρ (τ) ' is:

Temporal and spatial correlations coefficient ρ_ij(τ) ' computing formula be:

ρ_ij(τ) '=ρ_ij(τ)ρ_ij(w)；

(8) history correlation matrix generation module, is used for generating prediction section S_jHistory correlation matrix ρ (t):

Wherein, the same period all for nearly M and same type of historical traffic are chosen as traffic flow sequence X_jHistory correlated series, be designated asM=1,2 ... the span of M, M is [3,5], described history correlation coefficient ρ_jmT the computing formula of () is:

(9) predictor chooses module, for according to described temporal and spatial correlations coefficient threshold T₁With history correlation coefficient threshold T₂Choosing the predictor relevant to predicting impact point, and carry out matrix reconstruction according to locus j selected by it with time delay τ, selection principle is:

If ρ_ij(τ) ' ＞ T₁, then will observation section S_iTraffic flow sequence X_iIn meet the traffic flow of condition and form new sequence as the first predictor, be denoted as X', X'=(x₁',x₂',...,x_p'), wherein p is the described traffic flow number meeting condition, if L₁It is the maximum of time delay, L in the first predictor₁=max{ τ | τ ∈ [0, L] and ρ_ij(τ) ' ＞ T₁, then the first predictor X' can state following matrix form as:

If ρ_jm(t) ＞ T₂, then by all history correlated series X meeting condition_jmT (), as the second predictor, is denoted as Y', Y'={y₁',y₂',...,y_q', wherein q is the historical traffic number meeting condition, and the second predictor Y' can state following matrix form as:

(10) forecast model construction module, it is by constructing the measurable section forecast model in the traffic flow of subsequent time using the first predictor and the second predictor as training sample.

A kind of traffic flow monitoring and controlling forecast system the most according to claim 5, it is characterized in that, in described data preprocessing module, the rule of the data not meeting traffic practical situation described in rejecting is: within a data update cycle, set the threshold range of total traffic flow data in each section respectively, if total traffic flow data in certain section collected falls in corresponding threshold range, then show that these group data are reliable, retain this group data；If total traffic flow data in certain section collected falls not in corresponding threshold range, then show that these group data are unreliable, and rejected.

A kind of traffic flow monitoring and controlling forecast system the most according to claim 6, is characterized in that, described stationary test module includes following submodule:

(1) syndrome module, for being in same type of observation section S_iTraffic flow sequence X_iWith prediction section S_jTraffic flow sequence X_jCarry out stationary test respectively；

(2) continuity check submodule, it is connected with syndrome module, for to not carrying out continuity check by the traffic flow sequence to be tested of stationary test, if the seriality of not meeting, described continuity check submodule uses average interpolation method that data are carried out polishing；

(3) misarrangement submodule, is connected with continuity check submodule, for deleting the data of apparent error, uses average interpolation method that data are carried out polishing simultaneously；

(4) difference processing submodule, connects misarrangement submodule and syndrome module, for the data after polishing carry out difference processing, and the data after difference processing is sent to syndrome module.