CN101826258B - Method for predicting simple accidents on freeways - Google Patents

Abstract

The invention relates to a method for predicting simple accidents on freeways, which can be used for predicting the occurrence number of traffic accidents on some road sections of freeways and belongs to the field of traffic safety. The traffic accidents are not only concerned with speeds and speed differences, but also are closely associated with traffic volume and horizontal and longitudinal line shapes. The invention establishes relational models comprising intersection angles of the accidents and horizontal curves, gradients of longitudinal slopes and the like by utilizing the method of statistical regression. Practice proves that the accident number predicted with the method can be preferably coincident with the actually occurred accident number, and therefore, the method provides reliable theories and basis for reforming accident areas, preventing the occurrence of the traffic accidents, and reducing the severities of accidents.

Description

Method for predicting simple accidents on freeways

Technical field

The present invention is a kind of method for predicting simple accidents on freeways, can be used to predict the generation number of certain road section traffic volume accident of highway, belongs to the traffic safety field.

Background technology

Statistics shows that highway all is the highest highway of safe coefficient, and on the contrary in China in worldwide.According to Japan's statistics, the accident rate of its highway per 100 km only is 1/2 or 1/3 of a common road, and the magnitude of traffic flow is about 10 times of common road; The statistics of the U.S. is that the accident rate of its highway per 100 km only is 1/10 of a common road, also is far superior to Japan.And in China, the magnitude of traffic flow of highway is basic to similar abroad, even is lower than a lot of countries, but according to Public Security Department of China Ministry of Public Security statistics, the per 100 km accident rate but is more than 4 times of common road.If factors such as the volume of traffic are taken into account, the traffic safety status of China's highway will be more troubling again.Therefore, be necessary to set up the perfect traffic accidents Forecasting Methodology of a cover, for the prediction and the prevention of China's traffic accidents provides theoretical foundation.

According to the data of American National major accident research institute (NCSS), traffic mortality and travelling speed gradient (Δ V)) biquadratic be directly proportional.Approximate function is as follows:

$\mathrm{Death}={\left(\frac{\mathrm{\ΔV}}{114.24}\right)}^4---\left(1\right)$

In the formula: Death-traffic mortality, %;

Δ V-running velocity gradient, km/h;

People such as Pei Yulong of Harbin Institute of Technology carry out regretional analysis to vehicle speed standard difference and hundred million truck kilometer accident rates on seven highways of China, have obtained the relational model of hundred million truck kilometer accident rates and vehicle speed standard deviation.Model shows: the speed of a motor vehicle distributes discrete more, and accident rate is high more, and model is as follows:

AR＝9.583e ^0.0553σ(2)

In the formula: AR---hundred million truck kilometer accident rates;

The standard deviation of σ---the speed of a motor vehicle (km/h).

The Du Boying of Tongji University carries out on the basis of analysis-by-synthesis at accident rate and the operating speed to Foreign Expressway, set forth on the highway traffic accident prediction method based on operating speed.

The computing formula of highway mortality ratio is as follows:

$I_{\mathrm{Death}}=[V^{(2E-09)\×V^{3.667}}-(1-V/\stackrel{\‾}{V})]\·{\left[\frac{\mathrm{\ΔV}}{114.24}\right]}^4---\left(3\right)$

I in the formula _Death---traffic mortality (inferior/10 ⁶Veh*km);

V---travelling speed (km/h);

Δ V---velocity gradient, the i.e. difference of section operating speed and average operating speed (km/h);

---average operating speed (km/h).

As far back as 1964, Solomon studied the speed of a motor vehicle and the relation of safety with regard to beginning, and the relation between speed and the accident is discussed mostly, does not consider the influence of the volume of traffic.Make a general survey of both at home and abroad, in the influence research to speed and safety, major part all is in the relation of discussing between travelling speed, velocity contrast and the accident, not have to quantize the variation such as the volume of traffic, so conclusion and inequality have stronger cogency.Because the special national conditions and the highway of China can not directly be indiscriminately imitated external speed of a motor vehicle theoretical model, also need to carry out concrete theoretical research and engineering practice, constantly sum up the speed of a motor vehicle safety theory of China's highway, propose reasonably to solve countermeasure.

Summary of the invention

The present invention has taken all factors into consideration the influence to accident of horizontal curve factor, vertical section factor, the volume of traffic on the basis that the domestic communication operation characteristic is analyzed, and has obtained effect preferably through actual verification.

Use the brief traffic hazard frequency of formula prediction highway step:

Road is a three-dimensional entity.General said route is meant the locus of center line of road.The projection of route on surface level is called the planimetric map of route.Vertically dissecing again along center line, the row expansion then is the skiagraph of route.The turning point of two slope sections is relaxed with one section curve for the ease of driving on the vertical section, is called vertical curve.

(1) determine prediction highway condition, collect road information, comprising:

The average corner of horizontal curve:

$\mathrm{Ave}\_\mathrm{angle}=\frac{\mathrm{\Σ}\left|{\mathrm{\α}}_i\right|}{n}---\left(4\right)$

The parametric representation meaning:

a _iThe corner of i bar horizontal curve in the expression highway section;

N represents to comprise in the highway section horizontal curve number;

Horizontal curve figure as shown in Figure 1

2. vertical curve factor:

A basic variable supposing every vertical curve is V (i), and its unit is the variation of every 100m vertical curve gradient.

$V\left(i\right)=\frac{|g_i-g_{i+1}|}{L_{\mathrm{xi}}}---\left(5\right)$

g _iThe expression gradient, g _i=tan θ;

The weight of vertical curve i:

VC: the slope change value after the weighting

VC＝∑ _iWV(i)×V(i) (6)

The weighting gradient of vertical section:

Ave_slope＝∑ _iWG(k)×|g _k|(7)

In the highway section on the k/weight in descending highway section

g _k: the gradient in k highway section, highway section

Vertical curve figure as shown in Figure 2

3. cart number percent is formed in traffic;

(2) calculate exposure variable EXPO

EXPO＝AADT*365*L*10 ^-6*Y

The parametric representation meaning:

AADT: annual average daily traffic

L: road section length

Y: prediction continues the time

(3) gather the cart ratio, utilize Stata9.0 software to carry out statistical study, adopt the regression analysis of rejecting backward: set up full model earlier, judgement according to index of correlation (| Z| value minimum) among the output result, one of each rejecting least meets the variable that enters model, no longer contains till the independent variable that does not meet criterion in regression equation.Therefore, remove incongruent independent variable successively and return again, obtain brief hazard model at last:

λ _i＝EXPO·EXP(-2.676614+0.0071095·Ave_angle+0.737331·VC+0.2539619·Ave?slope+6.14963·Truck)(8)

The parametric representation meaning:

λ _i: i section prediction accident number;

EXPO: expose variable;

Y: prediction continues the time;

The L:i road section length;

The average corner of horizontal curve in the Ave_angle:i highway section;

VC:: vertical curve index, the slope change value after the weighting;

Average_slope: vertical curve index, the weighting gradient of vertical section;

Truck: cart ratio.

The present invention has taken all factors into consideration the influence to accident of horizontal curve factor, vertical section factor, the volume of traffic on the basis that the domestic communication operation characteristic is analyzed, and has obtained effect preferably through actual verification.

Description of drawings:

Fig. 1 horizontal curve synoptic diagram; Fig. 2 vertical curve synoptic diagram; Fig. 3 horizontal curve example schematic; Fig. 4 vertical curve example schematic.

Embodiment:

As Fig. 3, Figure 4 shows that the flat vertical instance graph of a certain highway, this highway annual average daily traffic is 2500, to the annual investigation statistics of this highway, the cart proportion is 13%.

The first step: calculate the average corner of horizontal curve

$\mathrm{Ave}\_\mathrm{angle}=\frac{\mathrm{\Σ}\left|{\mathrm{\α}}_i\right|}{n}=\frac{\frac{5}{18}\mathrm{\π}+\frac{1}{3}\mathrm{\π}}{2}=\frac{11}{36}\mathrm{\π}$

Second step:

Vertical curve L ₁Length=10000*[2%-(3%)]=500m

Vertical curve L ₂Length=8000*4.5%=360m

This road section length=150+500+300+360+280=1590m

$V\left(1\right)=V\left(j\right)=\frac{|g_j-g_{j+1}|}{L_j}=\frac{2\%-(-3\%)}{5}=1\%$

$V\left(2\right)=V\left(i\right)=\frac{|g_i-g_{i+1}|}{L_i}=\frac{4.5\%}{3.6}=1.25\%$

VC: the slope change value after the weighting

VC＝∑ _iWV(i)×V(i)＝V(1)*WV(1)+V(2)*WV(2)＝0.54％

Longitudinal gradient i in the highway section ₁=-3% shared weight:

Longitudinal gradient i in the highway section ₂=2% shared weight:

Longitudinal gradient i in the highway section ₃=-2.5% shared weight:

The weighting gradient of vertical section:

Ave_slope＝∑ _jWG(k)×|g _k|＝g1*WG(1)+g2*WG(2)+g3*WG(3)＝3％*0.094+2％*0.189+2.5％*0.176＝0.011

The 3rd step: calculate cart number percent

Know by condition: Truck=13%

The 4th step: expose variable and calculate

EXPO＝AADT*365*L*10 ^-6*Y＝2500*365*3*10 ^-6*1＝2.74

The 5th step: accident number prediction

λ _i＝EXPO·EXP(-2.676614+0.0071095·Ave_angle+0.737331·VC+0.2539619·Ave_slope+6.14963·Truck％)

＝2.74*EXP(-2.676614+0.0071095*0.96+0.737331*0.0054+0.2539619*0.011+6.14963*13％

＝2.74*0.155

＝1

Use respectively the Chongqing Chengdu-Chongqing at a high speed, the upper bound at a high speed, long ten thousand at a high speed, Yunnan sieve (gateway of the village) rich (rather) at a high speed, the road traffic of many highways such as the Beijing-Tianjin pool is according to top, the simple accidents on freeways forecast model verified the result is as showing

Highway	Road section length (km)	The volume of traffic (/ day)	Cart ratio (%)	The prediction duration (year)	The prediction traffic hazard	The actual traffic accident
							Luo Fu (K40+791～K41+390)	1	4300	46.2％	1	5	5
Luo Fu (k38+318～K40+39)	1.720	4300	46.2％	1	8	7
							Chengdu-Chongqing (k8+500～K9+764)	1.264	12473	48％	2	7	6
Chengdu-Chongqing (k2+100～K3+456)	1.56	17506	48％	2	4	4
							Long by ten thousand (K152+327～K154+772)	2.444	2310	32％	2	8	9
Long by ten thousand (K182+108～K183+507)	1.398	1500	32％	2	10	11
							The upper bound (K26+870～K27+900)	1.03	10748	15.7％	2	17	15
The upper bound (K24+789～K26+930)	2.414	30380	25.7％	2	29	25
							The Beijing-Tianjin pool (K6～K8)	2	18169	27.44％	4	66	64
The Beijing-Tianjin pool (K49～K56)	7	19400	43.11％	4	25	20

As can be seen from the table, prediction accident number and actual accidents number can reasonablely coincide, thereby the trouble-saving theoretical foundation that provides is provided.

Claims (1)

1. method for predicting simple accidents on freeways is characterized in that carrying out according to the following steps:

Route is meant the locus of center line of road; The projection of route on surface level is called the planimetric map of route; Vertically dissecing again along center line, the row expansion then is the skiagraph of route; The turning point of two slope sections is relaxed with one section curve for the ease of driving on the vertical section, is called vertical curve;

(1) determine prediction highway condition, collect road information, comprising:

1.1 the average corner of horizontal curve:

$\mathrm{Ave}\_\mathrm{angle}=\frac{\mathrm{\Σ}\left|{\mathrm{\α}}_i\right|}{n}$

The parametric representation meaning:

α _iThe corner of i bar horizontal curve in the expression highway section;

N represents to comprise in the highway section horizontal curve number;

1.2. vertical curve factor:

A basic variable supposing every vertical curve is V (i), and its unit is the variation of every 100m vertical curve gradient;

$V\left(i\right)=\frac{|g_i-g_{i+1}|}{L_{\mathrm{xi}}}$

g _iThe expression gradient, g _i=tan θ;

The weight of vertical curve i:

VC: the slope change value after the weighting

VC＝∑ _iWV(i)×V(i)

The weighting gradient of vertical section:

Ave_slope＝∑ _iWG(k)×|g _k|

In the highway section on the k/weight in descending highway section

g _k: the gradient in k highway section, highway section

1.3. traffic is formed, cart number percent;

(2) calculate exposure variable EXPO

EXPO＝AADT*365*L*10 ^-6*Y

The parametric representation meaning:

AADT: annual average daily traffic

L: road section length

Y: prediction continues the time

(3) utilize Stata9.0 software to carry out statistical study, adopt the regression analysis of rejecting backward: obtain brief hazard model at last:

λ _i＝EXPO·EXP(-2.676614+0.0071095·Ave_angle+0.737331·VC

+0.2539619·Ave_slope+6.14963·Truck)

The parametric representation meaning:

λ _i: i section prediction accident number;

EXPO: expose variable;

Y: prediction continues the time;

The L:i road section length;

The average corner of horizontal curve in the Ave_angle:i highway section;

VC: vertical curve index, the slope change value after the weighting;

Average_slope: vertical curve index, the weighting gradient of vertical section;

Truck: the cart ratio is obtained by historical data.

Images