CN103035124B - Based on the traffic congestion monitoring forecasting procedure of band dissipative term macroscopic traffic flow - Google Patents

Abstract

Be difficult to directly to the technological deficiency of traffic congestion monitoring forecast to overcome existing traffic flow model, the invention provides a kind of traffic congestion monitoring forecasting procedure based on band dissipative term macroscopic traffic flow, the method obtains car speed, density and flow information by the video image of CCTV camera, forecast by the traffic congestion of generation according to newly-established traffic congestion model, solve the technical matters that traffic congestion can not be forecast in time.

Description

Based on the traffic congestion monitoring forecasting procedure of band dissipative term macroscopic traffic flow

Technical field

The present invention relates to a kind of modeling method, particularly a kind of traffic congestion monitoring forecasting procedure based on band dissipative term macroscopic traffic flow.

Background technology

Communications and transportation is closely-related significant problem with national economy.The transportation network setting up unimpeded prosperity is the set objective of national development, the up-to-dateness of traffic and transportation system and the advanced degree of traffic administration, is the important symbol of a measurement modernization of the country development; Therefore, communications and transportation cause, particularly highway transportation, be subject to the great attention of every country government, obtains in recent years and develop rapidly; Communications and transportation whether unimpeded, to the development of urban economy, the quality of life of people, the international fame of regional and even whole country has very important impact, and in order to alleviate traffic jam issue, domestic and international academia all drops into larger energy research model problem.Traffic flow model research is the basis of traffic system research, simultaneously as one of the fundamental research content of intelligent transportation system (abbreviation ITS), also significant to the development of ITS.

The research of traffic flow theory starts from nineteen thirties, and what traffic flow theory the earliest adopted substantially is probability theory method.After the fifties, along with developing rapidly of auto industry, road traffic flow increases severely, in traffic flow, vehicle independence is more and more less, probability theory method is no longer applicable, so people are from different conceptual frameworks, proposes various new traffic flow theory model.

Existing traffic flow theory model is divided three classes substantially: microvisual model, mesoscopic model and macromodel.Microvisual model is that traffic flow is considered as far from equilibrium state, a large amount of dispersion, interactional self-driven particle.Mesoscopic model is then the gas movable model based on probability description, also has scholar to incorporate this class model into microscopic approach.Macromodel is then compressible continuous fluid medium traffic flow being considered as being made up of a large amount of vehicle, the average behavior of research vehicle body, the not explicit appearance of individual character of single unit vehicle.In macroscopic traffic flow, traffic flow is regarded as the compressible continuous fluid medium be made up of a large amount of vehicle, and the average behavior of research vehicle collective, the individual character of single unit vehicle does not highlight.Macroscopic traffic flow portrays traffic flow with the average density ρ of vehicle, average velocity v and flow q, study they the equation that meets.

Compared with microvisual model, macromodel can portray the collective behavior of traffic flow better, thus for designing effective traffic control strategy, simulation and estimating that the traffic engineering problem such as effect of road geometry modification provides foundation.Numerical evaluation aspect, simulation Macro-traffic Flow required time is studied number of vehicles in traffic system with institute and is had nothing to do, with studied road, numerical method choose and middle space x, time t discrete steps Δ x relevant with Δ t.So macroscopic traffic flow is comparatively suitable for the traffic flow problem of the traffic system processing a large amount of vehicle composition.In macro traffic model, the continuous traffic flow model that Lighthill and Whitham proposes has illustrative, Richards also independently proposes similar model simultaneously, these two models are collectively referred to as LWR model, LWR model is with ρ (x, t) and v(x, t) represent that t is positioned at average density and the average velocity of the traffic flow of x place, they meet following continuity equation

$\frac{\∂\mathrm{\ρ}}{\∂t}+\frac{\∂\left(\mathrm{\ρv}\right)}{\∂x}=0$

Above formula represents number of vehicles conservation.For making equation close, assume that the speed-density relation of balance simultaneously

v(x,t)=V _e[ρ(x,t)]

Wherein, V _e[ρ (x, t)] is equilibrium rate.Following equation can be obtained like this

$\frac{\∂\mathrm{\ρ}}{\∂t}+\left\{V_e[\mathrm{\ρ}(x,t)\right]+\mathrm{\ρ}\frac{\∂}{\∂\mathrm{\ρ}}{V}_e\left[\mathrm{\ρ}(x,t)\right]\}\frac{\∂\mathrm{\ρ}}{\∂x}=0$

As time goes on, the nonlinear motion ripple described by above formula is evolved into sharp wave structure gradually, and wavefront is more and more precipitous, until vertically, which results in the discontinuous of traffic flow density; But the traffic density distribution in reality is so not extreme, in order to make shock wave structure smoothing to a certain extent, adds dissipative term in above formula object is the diffusional effect of reflection vehicle, makes nonlinear motion ripple continuous distribution, so model conversation is

$\frac{\∂\mathrm{\ρ}}{\∂t}+\left\{V_e\right[\mathrm{\ρ}(x,t)]+\mathrm{\ρ}\frac{\∂}{\∂\mathrm{\ρ}}{V}_e[\mathrm{\ρ}(x,t)\left]\right\}\frac{\∂\mathrm{\ρ}}{\∂x}=D\frac{{\∂}^2\mathrm{\ρ}}{\∂x^2}$

Wherein D is dissipation factor.

Suppose there is the equilibrium rate-density relationship as lower linear

Model is just converted into famous Burgers equation (see document J.M.Burgers.Amathematicalmodelillustratingthetheoryoftur bulence [J] .AdvancesinAppliedMechanics1948,2:171-199)

$\frac{\∂R}{\∂t}+R\frac{\∂R}{\∂x}=D\frac{{\∂}^{2}R}{{\∂x}^2}$

Function in formula

$R=v_m(1-\frac{2\mathrm{\ρ}}{{\mathrm{\ρ}}_m})$

But, Burgers equation can not directly provide traffic congestion condition, particularly directly provide the whole description of traffic jam issue when various traffic parameter change, make traffic system research worker be not easy to direct use, there is the technical matters being difficult to forecast traffic congestion.

Summary of the invention

Be difficult to directly to the technological deficiency of traffic congestion monitoring forecast to overcome Burgers traffic flow model, the invention provides a kind of traffic congestion monitoring forecasting procedure based on band dissipative term macroscopic traffic flow, the method obtains car speed, density and flow information by the video image of CCTV camera, forecast by the traffic congestion of generation according to newly-established traffic congestion model, solve the technical matters that traffic congestion can not be forecast in time.

The technical solution adopted for the present invention to solve the technical problems is: a kind of traffic congestion monitoring forecasting procedure based on band dissipative term macroscopic traffic flow, is characterized in adopting following steps:

When 1, obtaining car speed, density and flow information by the video image of CCTV camera, consider that actual monitored video camera works at crossing throughout the year, the impossible artificial frequent correction image Processing Algorithm of mode, select image processing algorithm according to the composition error performance index of the overall process of following image procossing:

min(e _z)=min{e _para{e _seg[e _pre(e _samp)]}}

In formula, e _zfor the global error of image zooming-out traffic parameter, min (e _z) be the e by selecting the image processing method of various combination to obtain _zminimum value, e _sampfor image sampling error, e _prefor image and process errors, e _segfor vehicles segmentation error in image, e _parafor the error according to segmentation image zooming-out traffic parameter;

2, the macroscopic traffic flow in given section is set up

$\frac{\∂\mathrm{\η}}{\∂t}+v_m(\frac{2}{{\mathrm{\ρ}}_m\mathrm{\η}}-1)\frac{\∂\mathrm{\η}}{\∂x}=D[\frac{{\∂}^2\mathrm{\η}}{{\∂x}^2}-\frac{2}{\mathrm{\η}}{\left(\frac{\∂\mathrm{\η}}{\∂x}\right)}^2]$

In formula, for state variable, x is position, and t is the time, and D is dissipation factor, and ρ is the average density of vehicle, v is that vehicle is on average fast, degree v _mfor maximal rate, ρ _mfor saturation traffic density during traffic congestion;

3, when state variable η is tending towards infinite in time, this section will be tending towards obstruction density and produce traffic congestion, send traffic more to block up forecast to this section;

4, the restriction of discontinuity Induction Control is adopted to sail sending a car of this section into.

The invention has the beneficial effects as follows: by all image method integrally, image processing algorithm is selected according to the composition error performance index of the overall process of following image procossing, and forecast by the traffic congestion of generation according to newly-established traffic congestion model, solve the technical matters that traffic congestion can not be forecast in time.

Below in conjunction with embodiment, the present invention is elaborated.

Embodiment

When 1, obtaining car speed, density and flow information by the video image of CCTV camera, consider that actual monitored video camera works at crossing throughout the year, the impossible artificial frequent correction image Processing Algorithm of mode, select image processing algorithm according to the composition error performance index of the overall process of following image procossing:

min(e _z)=min{e _para{e _seg[e _pre(e _samp)]}}

=k _sampk _prek _sege _para

In formula, e _zfor the global error of image zooming-out traffic parameter, min (e _z) be the e by selecting the image processing method of various combination to obtain _zminimum value, e _sampfor image sampling error, e _prefor image and process errors, e _segfor vehicles segmentation error in image, e _parafor the error according to segmentation image zooming-out traffic parameter, k _samp>1 is image sampling error coefficient, k _pre>1 is image and process errors coefficient, k _seg>=1 is the error coefficient of vehicles segmentation error in image;

2, the macroscopic traffic flow in given section is set up

$\frac{\∂\mathrm{\η}}{\∂t}+v_m(\frac{2}{{\mathrm{\ρ}}_m\mathrm{\η}}-1)\frac{\∂\mathrm{\η}}{\∂x}=D[\frac{{\∂}^2\mathrm{\η}}{{\∂x}^2}-\frac{2}{\mathrm{\η}}{\left(\frac{\∂\mathrm{\η}}{\∂x}\right)}^2]$

In formula, for state variable, x is position, and t is the time, and D is dissipation factor, and ρ is the average density of vehicle, v is that vehicle is on average fast, v _mfor maximal rate, ρ _mfor saturation traffic density during traffic congestion;

3, when state variable η is tending towards infinite in time, this section will be tending towards obstruction density and produce traffic congestion, send traffic more to block up forecast to this section;

4, the restriction of discontinuity Induction Control is adopted to sail sending a car of this section into.

Claims (1)

1., based on a traffic congestion monitoring forecasting procedure for band dissipative term macroscopic traffic flow, be characterized in adopting following steps:

(1) when obtaining car speed, density and flow information by the video image of CCTV camera, consider that actual monitored video camera works at crossing throughout the year, the impossible artificial frequent correction image Processing Algorithm of mode, select image processing algorithm according to the composition error performance index of the overall process of following image procossing:

In formula, for the global error of image zooming-out traffic parameter, for what obtained by the image processing method of selection various combination minimum value, for image sampling error, for Image semantic classification error, for vehicles segmentation error in image, for the error according to segmentation image zooming-out traffic parameter, for image sampling error coefficient, for Image semantic classification error coefficient, for the error coefficient of vehicles segmentation error in image;

(2) macroscopic traffic flow in given section is set up

In formula, for state variable, for position, for the time, for dissipation factor, for maximal rate, for average density, for saturation traffic density during traffic congestion;

(3) state variable is worked as when being tending towards infinite in time, this section will be tending towards obstruction density and produce traffic congestion, send traffic more to block up forecast to this section;

(4) restriction of discontinuity Induction Control is adopted to sail sending a car of this section into.