CN105931460A - Variable speed limit control strategy optimization method for continuous bottleneck section of expressway - Google Patents

Abstract

The invention discloses a variable speed limit control strategy optimization method for a continuous bottleneck section of an expressway. The method comprises the steps: arranging a detector and a variable speed limit indication board at the continuous bottleneck section of the expressway; carrying out the mining of a variable speed limit control parameter value with the optimal control effect based on the genetic algorithm; judging the traffic flow operation data based on the actual measurement traffic flow data of the expressway; controlling the change frequency of the speed limit value through a variable speed limit control period; reducing the time-space fluctuation of the traffic flow speed of a trunk line through employing upstream and downstream speed smoothening factors; and carrying out the calculation of coordinative control among a plurality of speed limit signs based on the maximum speed limit difference between the adjacent sections. The method irons out a defect of randomness of core parameter values in the variable speed limit control, enables the speed limit value of the variable speed limit control to be continuously changed step by step in space and time, reduces the big fluctuation of the speed limit value in space and time, effectively reduces the traffic flow fluctuation and chaos caused by the abrupt change of the speed limit value, reduces the traffic accident risk, and reduces the severity of accidents.

Description

A kind of variable speed-limit control strategy optimization method of the continuous bottleneck road of through street

Technical field

The invention belongs to technical field of traffic control, particularly relate to a kind of continuous bottleneck road of through street based on genetic algorithm The variable speed-limit control strategy optimization method of section.

Background technology

Variable speed-limit controls as a kind of traffic control strategy being increasingly widely used in improving through street traffic safety, It controls effect and determines that the algorithm that process is used is closely related with variable speed-limit value.Genetic algorithm is as a kind of closed loop knot Structure, can more preferably control controlling parameter value and controlling anti-to control strategy of effect of effect by constantly excavating have Feedback regulation process, effectively promotes effect and the reasonability of variable speed-limit control speed limit that variable speed-limit controls.Therefore, base The variable speed-limit control strategy optimization method of the continuous bottleneck road of through street in genetic algorithm, continuous by genetic algorithm The optimum value excavating variable speed-limit control core parameter realizes the optimization of variable speed-limit control effect.

The key parameter value related in current variable speed-limit control strategy relies primarily on engineer experience's subjectivity and determines, and And the variable speed-limit value that diverse location is the most in the same time exists jumping characteristic, variable speed-limit value fluctuation the most frequently is easily caused The potential safety hazard of variable speed-limit control area.Research in the past is also not set up through street severity of injuries forecast model, Make to investigate variable speed-limit to control cannot severity of injuries be considered during effect.The present invention proposes based on genetic algorithm The variable speed-limit control strategy optimization method of the continuous bottleneck road of through street, control plan compared to conventional variable speed-limit Slightly, the strategy that the present invention proposes considers the traffic control impact on the vehicle accident order of severity, is effectively improved variable limit Speed controls effect, makes time adjacent segments and the change of variable speed-limit value spatially more continuous simultaneously.

Summary of the invention

The problem to be solved in the present invention is: adjacent segments speed limit was coordinated by variable speed-limit control strategy shortage in the past Control, and on through street, continuous bottleneck road traffic flow fluctuating margin is greatly the main inducing of vehicle accident, is simultaneous for The key parameter value of the variable speed-limit control strategy of bottleneck road has subjective random continuously.The present invention proposes a kind of base The variable speed-limit control strategy optimization method of the continuous bottleneck road of through street in genetic algorithm, uses genetic algorithm to can Become the value that in Control for Speed Limitation strategy, optimum core controls parameter to be optimized, reach variable speed-limit in current time and control week Utilize the rate smoothing factor progressively to be adjusted by speed limit to desired value during the phase, speed limit continuous change spatially is set simultaneously Change.During before overcoming, variable speed-limit controls, core controls randomness and the traffic flow big amplitude wave on space-time of parameter value Dynamic.

Technical solution of the present invention is:

The present invention proposes the variable speed-limit control strategy optimization of a kind of continuous bottleneck road of through street based on genetic algorithm Method, utilizes principle of genetic algorithm to obtain its optimum for the key control parameter related in variable speed-limit control strategy and takes Value, in actual control, current time reaches during the variable speed-limit control cycle to count each speed(-)limit sign position speed limit Calculating, be modified speed limit according to the upstream and downstream rate smoothing factor and adjacent segments maximum speed limit value difference, this method is to reality In border by variable speed-limit control strategy smooth through street main line traffic flow speed space-time fluctuation significant.Real Example shows, the variable speed-limit control strategy optimization method that the present invention proposes has good effect, and the strategy after optimization can be effective Reduce contingency occurrence probability and the degree of danger of the continuous bottleneck road of through street.

Accompanying drawing explanation

Fig. 1 is the variable speed-limit control strategy flow chart of the continuous bottleneck road of through street.

Fig. 2 is that Traffic flow detecting device and variable speed-limit direction board arrange schematic diagram in continuous bottleneck road.

Fig. 3 order logit model structure figure.

Fig. 4 is that rate smoothing factor pair traffic flow speed affects schematic diagram.

Fig. 5 is variable speed-limit control strategy optimization flow chart based on genetic algorithm.

Detailed description of the invention

The present invention is that the basic procedure of ultimate principle based on genetic algorithm and variable speed-limit control strategy proposes variable limit The method that core parameter in speed control strategy is optimized with control effect, by genetic algorithm and Traffic Flow Simulation software Between data exchange and iteration constantly excavate the optimization value of core parameter in variable speed-limit control strategy, based on free way The discriminating data traffic flow running rate that the Traffic flow detecting device in each section, road gathers, current time reaches variable speed-limit control During the integral multiple in cycle, each speed(-)limit sign position speed limit is updated, during renewal, need to consider that upstream and downstream speed is put down simultaneously Speed limit is modified by the sliding factor and adjacent segments maximum speed limit value difference, it is achieved coordinate between multiple speed(-)limit signs to control, base The flow chart of the variable speed-limit control strategy of the continuous bottleneck road of through street in genetic algorithm is as shown in Figure 1.

The first step determine that through street continuous bottleneck road scope and rational position supporting arrange Traffic flow detecting device and Variable speed-limit direction board.Variable speed-limit direction board is used for issuing speed limit, it should be noted that variable speed-limit value uniform requirement For the multiple of 5mph, therefore the updated value of speed limit approximate closest to the integer value of 5mph.With 30s as time Between the cycle obtain continuous bottleneck road position traffic flow data in real time by Traffic flow detecting device.Continuously in bottleneck road The Traffic flow detecting device of supporting setting and variable speed-limit direction board schematic diagram are as shown in Figure 2.

Second step is to build the real-time accident risk prediction model for the continuous bottleneck road of through street, in traffic flow Each position x is calculated during emulation_iThe vehicle accident probability of happening at place and death by accident/injured probability.Continuous for through street The real-time accident risk prediction model of bottleneck road is divided into two steps: (1) sets up street accidents risks forecast model, calculates thing Therefore probability of happening；(2) severity of injuries is divided into death/injured accident and only property loss accident two class, sets up accident tight Weight degree forecast model, calculates death/injured contingency occurrence probability.It should be noted that only when accident risk prediction mould When type shows have accident to occur, just can consider to utilize severity of injuries forecast model to calculate severity of injuries.The present invention Accident risk prediction model set up by employing order logit model, and structure chart is as it is shown on figure 3, the basic binomial logit of model Model form is as follows:

$P(Y=1)=\frac{1}{1+e^{-g\left(X\right)}}---\left(1\right)$

Wherein, P (Y=1) is vehicle accident probability of happening or injured/death by accident probability, and g (X) is utility function, i.e. from becoming The linear combination of amount X, expression formula is as follows:

G (X)=β₀+β₁x₁+…+β_kx_k (2)

Wherein, x_kFor the value of traffic flow variable k, β_kCoefficient for variable k.

Position x is calculated when using formula (1)_iDuring place's vehicle accident probability of happening, the utility function in formula (2) is as follows:

$\begin{array}{c}{g}_i\left(x\right)=-2.672+0.074x_{i1}+0.060x_{i2}+0.050x_{i3}+0.119x_{i4}+0.092x_{i5}\\ +0.026x_{i6}+1.057x_{i7}-0.049x_{i8}-0.856x_{i9}+0.508x_{i10}\end{array}---\left(3\right)$

Wherein, x_i1For upstream detector average occupancy, x_i2Poor for upstream detector velocity standard, x_i3For detected downstream Device velocity standard is poor, x_i4For the difference of adjacent lane occupation rate, x_i5For the difference of upstream and downstream detector flow, x_i6For upstream and downstream The difference of occupation rate, x_i7For upstream and downstream detector spacings, x_i8For width of roadway, x_i9For shoulder width, x_i10For curve section Indicator variable.

Position x is calculated when using formula (1)_iDuring place's vehicle accident death/injured probability, the effectiveness letter in formula (2) Number is as follows:

g_j(x)=2.129-0.033x_j1-0.056x_j2-0.335x_j3-0.036x_j4 (4)

Wherein, x_j1For upstream detector average occupancy, x_j2For downstream detector average discharge, x_j3Indicate for peak period Variable, x_j4For width of roadway.

3rd step is to judge that whether current time is the integral multiple that variable speed-limit controls the cycle, if then to each speed(-)limit sign Position speed limit calculates, and does not the most do any next one traffic flow data that is operated into and detects the cycle.

Actual measurement average speed of traffic flow v (x according to i upstream and downstream position, certain section_i+1, t) with v (x_i-1, t) calculate the mesh in this section Mark speed limit, proposes rate smoothing factor a for realizing the effect of smooth velocity perturbation, and the value of smoothing factor a is to smooth The impact of effect as shown in Figure 4, is worth the actual speed closer to downstream road section i-1 of the speed limit in the biggest then section i, Then the computing formula of the target speed limit of certain section i is as follows:

T_VSL(x_i, t)=α v (x_i+1,t)+(1-α)·v(x_i-1,t) (5)

Wherein,

T_VSL(x_i, t) it is the section i target speed limit at moment t；

A is the rate smoothing factor (0 < a < 1)；

v(x_i+1, t) with v (x_i-1, t) it is respectively the actual speed exported at section, upstream i-1 and downstream road section i+1 detector.

After trying to achieve section target speed limit, in conjunction with change step Δ V and target speed limit T_VSLTwo factors can determine that position x_iVariable speed-limit sign at the change range value of t, computing formula is as follows:

${\mathrm{\&Delta;V}}_{SL}(x_i,t)=\left\{\begin{array}{ccc}-\mathrm{\&Delta;}V,& if& T_{VSL}(x_i,t+\mathrm{\&Delta;}t)<V_{SL}(x_i,t)-\mathrm{\&Delta;}V\\ 0,& if& V_{SL}(x_i,t)-\mathrm{\&Delta;}V\&le;T_{VSL}(x_i,t+\mathrm{\&Delta;}t)\&le;V_{SL}(x_i,t)+\mathrm{\&Delta;}V\\ \mathrm{\&Delta;}V,& if& T_{VSL}(x_i,t+\mathrm{\&Delta;}t)>V_{SL}(x_i,t)+\mathrm{\&Delta;}V\end{array}\right.---\left(6\right)$

Wherein,

Δ V is that variable speed-limit value changes amplitude；

T_VSLFor target speed limit；

V_SL(x_i, t) it is position x in the i of section_iVariable speed-limit sign at the speed limit of t.

Consider coordination between multiple speed(-)limit sign after controlling to position x_iVariable speed-limit sign in the change step value of t Being modified updating, computing formula is as follows:

${\mathrm{\&Delta;V}}_{SL}(x_i,t)=\left\{\begin{array}{cc}-{\mathrm{\&Delta;V}}^{\&prime;},& \begin{array}{cc}if& T_{VSL}(x_i,t+\mathrm{\&Delta;}t)\&le;T_{VSL}(x_{i+1},t+\mathrm{\&Delta;}t)-{\mathrm{\&Delta;V}}^{\&prime;}\end{array}\\ 0,,& allothercases\end{array}\right.---\left(7\right)$

Wherein,

Δ V ' is adjacent segments maximum speed limit value difference.

It is calculated position x by formula (7)_iVariable speed-limit sign in the end value of the change step of t, by it Bring the final speed limit that can obtain in the i of section in the computing formula of speed limit in section i as follows into:

V_SL(x_i, t+ Δ t)=V_SL(x_i,t)+ΔV_SL(x_i,t) (8)

Wherein,

V_SL(x_i, t+ Δ t) is position x in the i of section_iVariable speed-limit sign at the speed limit of t+ Δ t；

ΔV_SL(x_i, t) it is position x_iVariable speed-limit sign in the amplitude of variation of t.

In the section i of calculating gained in formula (8), the computing formula of speed limit will can obtain the final speed limit in the i of section Speed limit scope is allowed to compare with section, if exceeding section to allow speed limit scope [V_min,V_max], the most only issue section corresponding edge Boundary's speed limit；If being not above section to allow speed limit scope, then final speed limit is approximated immediate 5mph integer Value again.Variable speed-limit direction board releasing position x finally by supporting setting_iThe speed limit at place.

4th step is to judge whether current time actual measurement traffic flow running rate reaches variable speed-limit and control stop condition, works as road When transport need does not decline in section, then enter the next traffic flow data detection cycle；When in section, transport need declines Time, then variable speed-limit value progressively returns to give tacit consent to speed limit, and the computing formula of speed limit is as follows:

V_SL(x_i, t)=V_SL(x_i,t)+ΔV,if v(x_i,t)<v(x_i+1,t)and v(x_i, t)=V_SL(x_i,t) (9)

5th step determines that the value of four key parameters in the variable speed-limit control strategy in the 3rd step and the 4th step Scope and change step, determine the optimization object function for genetic algorithm simultaneously.

Four the cores control parameters comprised for the variable speed-limit control strategy of continuous bottleneck road are respectively as follows: variable speed-limit Control cycle T, upstream and downstream rate smoothing index a, speed limit change amplitude, ao V and adjacent segments maximum speed limit value difference Δ V '. The variable speed-limit control cycle is 30 seconds to 5 minutes, and upstream and downstream rate smoothing index span is 0.1 to 0.9, speed limit Value change amplitude from 5mph to 30mph, adjacent segments maximum limit speed difference from 5mph to 30mph, above-mentioned core parameter Span is as shown in table 1.

Table 1 variable speed-limit controls parameter value

Traffic control generally believes can at utmost reduce street accidents risks, reduction accident occur after the order of severity and The variable speed-limit control strategy not dramatically increasing transit time is optimal strategy, thus be accordingly used in variable speed-limit control strategy excellent The object function of change method is shown below:

$Fitness=-\&lsqb;\mathrm{\&gamma;}\&CenterDot;\left(\frac{R_{VSL}-R_{No}}{R_{No}}\right)+\mathrm{\&mu;}\&CenterDot;\left(\frac{I_{VSL}-I_{No}}{I_{No}}\right)+\mathrm{\&eta;}\&CenterDot;\left(\frac{T_{VSL}-T_{No}}{T_{No}}\right)\&rsqb;---\left(10\right)$

Wherein,

Fitness is fitness；

γ, μ, η (γ+μ+η=1) are weight coefficient, in accident risk, severity of injuries, obtain between the travel time Balance；

R_VSLAnd R_NOIt is respectively under variable speed-limit controls and without controlling lower accident risk, Wherein, P_i,t(crash=1) it is the section i accident probability in t, can be according to the real-time accident risk set up in second step Forecast model is calculated, and K is total simulation time, and N is section number；

T_VSLAnd T_NOIt is respectively under variable speed-limit controls and without total transit time under controlling,Its In, d_iT () is the section i vehicle number in t, Δ t is simulation time step-length, and K is total simulation time, and N is section Number；

I_VSLAnd I_NOIt is respectively under variable speed-limit controls and without controlling lower severity of injuries, it should be noted that accident is serious Degree is predicted under only in vehicle accident, a situation arises the most meaningful, therefore only general when the vehicle accident of prediction during calculating

When rate exceedes accident generation threshold value, being just predicted the order of severity, the computing formula of severity of injuries is as follows:

$I=\frac{\underset{t=1}{\overset{K}{\&Sigma;}}\underset{i}{\overset{N}{\&Sigma;}}{I}_{i,t}(FI=1)}{M}---\left(11\right)$

$I_{i,t}(FI=1)=\left\{\begin{array}{cc}{P}_{i,t}(FI=1),& \begin{array}{cc}if& P_{i,t}(crash=1)\&GreaterEqual;threshold\end{array}\\ 0,& else\end{array}\right.---\left(12\right)$

$M=\left\{\begin{array}{cc}M+1,& \begin{array}{cc}if& P_{i,t}(crash=1)\&GreaterEqual;threshold\end{array}\\ M,& otherwise\end{array}\right.---\left(13\right)$

Wherein,

K is total simulation time；

N is section number；

Threshold is the threshold value that judgement accident occurs；

M is the time span that accident risk exceedes threshold value；

I_i,t(FI=1) it is death/injured accident probability, can calculate according to the real-time accident risk prediction model set up in second step Obtain.

6th step is that the variable speed-limit control flow for the first step to the 4th step builds Traffic Flow Simulation model, by emulation Mode input output relevant traffic flow data, based on to determining in the 5th step four of genetic algorithm and Traffic Flow Simulation model The value of key parameter is optimized, and variable speed-limit control strategy optimization flow process based on genetic algorithm is as shown in Figure 5.? Genetic algorithm part need to determine the value of four key parameters related in genetic algorithm: Population Size M is 20, maximum Algebraically T is 100, crossover probability Pc is 0.9, mutation probability Pm is 0.1.Based on genetic algorithm basic procedure to variable Control for Speed Limitation strategy key parameter value is optimized, and idiographic flow is as follows:

1, initialize: arranging evolutionary generation enumerator is t=0, maximum evolutionary generation T is set, arranges in genetic algorithm and hand over Fork probability P c and mutation probability Pm.

2, initial population: stochastic generation M is individual as initial population P (0), comprises each variable limit in m-th individuality Speed controls parameter value information coding m (x₁,x₂,x₃,x₄)。

3, individual evaluation: parameter value individual in colony P (t) is input in Traffic Flow Simulation model, to variable speed-limit control The lower traffic flow of system emulates, and is estimated controlling effect, calculates each individuality according to fitness function Fitness Fitness.

4, Selecting operation: selecting winning individuality, individual selected probability from colony is that ideal adaptation degree is whole Colony's fitness comprehensively middle proportion.After individuality is chosen, random group conclusion of the business is matched.

5, crossing operation: set a cross point so that certain probability is random in individuality string, carry out when intersecting before this point or After two each and every one body portion structures be interchangeable, and generate two new individualities.

6, mutation operator: to the individual one or more locus of sequence random choose in colony, and with certain probability pair The genic value of these locus does and changes.

7, colony is produced: colony P (t), after selection, intersection, mutation operator, obtains colony P (t+1) of future generation, Iterative computation for a new round.

8, end condition: when iterations reaches maximum algebraically T, genetic algorithm terminates；Otherwise, 3 to 7 are repeated Step is until end condition meets.

9, parameter decoding: the individuality with maximum adaptation degree is exported as optimal solution, and by optimal solution (x₁,x₂,x₃,x₄) It is decoded into the value of four key parameters of the variable speed-limit control strategy of correspondence.

The core of new generation constantly using the generation of above-mentioned genetic algorithm to have more preferably control effect controls parameter value, will renewal After core control parameter value bring in the 3rd step and the 4th step, by suitable phantom output relevant traffic stream join Number calculates fitness function, constantly carries out data exchange and iteration finally gives variable speed-limit between genetic algorithm and phantom The optimal solution of control strategy core parameter.

7th step is that the variable speed-limit obtained in the 5th step is controlled cycle T, upstream and downstream rate smoothing index a, speed limit The optimization value changing amplitude, ao V and adjacent segments maximum speed limit value difference Δ V ' is brought in the 3rd step and the 4th step as each pass The recommendation value of bond parameter, then carries out variable limit according to the first step to the 4th step strategy bottleneck road continuous to through street Speed controls.

Below in conjunction with the accompanying drawings the control strategy optimization method of invention is carried out presented example:

Assume the continuous bottleneck road of a certain through street, Traffic flow detecting device is set with the spacing of average 0.5 mile.Section Interior traffic flow conditions is that free stream velocity is 65mph in section, and road section capacity is 1950veh/h/ln, the traffic capacity Fall is 8.5%, and kinematic wave spread speed is 9.2mph, and stop and go Model ParameterValue is 0.2, and φ takes Value is 0.1.It is that speed limit V is given tacit consent in section that variable speed-limit controls situation_SL(default) being 70mph, section allows speed limit Scope is [0mph, 70mph].

In simulation software, build the phantom in through street section, use genetic algorithm excellent to variable speed-limit control strategy Change can obtain the value of four core parameters and be respectively as follows: variable speed-limit control cycle T=30s, upstream and downstream rate smoothing index A=0.85, speed limit change amplitude, ao V=5mph, adjacent segments maximum speed limit value difference Δ V '=5mph.

The detection cycle of Traffic flow detecting device is 30s, 30 points 30 seconds when current time is 9, is that variable speed-limit controls week The integral multiple of phase 30s, so updating each position variable speed-limit value.The limit that on section, the current variable speed-limit of 3# detector controls Speed value V_SL(x₃, t) it being 70mph, the actual measurement traffic flow data of 2# and the 4# detector of upstream and downstream is respectively v (x₄, t)=65mph With v (x₂, t)=60mph, from formula (5), current time T_VSL(x₃, t)=α v (x₄,t)+(1-α)·v(x₂,t) =0.85 65+0.15 60=64.25mph, then T_VSL(x₃, t+ Δ t)=64.25 < (70-5)=V_SL(x₃, t)-Δ V, by formula (6) Understand variable speed-limit sign current time change step Δ V at 3# detector_SL(x₃, t) be-5mph.Current time T_VSL(x₃, t)=64.25≤70-5T_VSL(x₄, t)-Δ V ', formula (7) understand 3# detector variable speed-limit sign when t The change step value Δ V carved_SL(x₃, t)=-5mph, carry it into formula (8) and understand the final speed limit at 3# detector For 70-5=65mph, in section allows speed limit scope [0,70] and be the multiple of 5mph, therefore the corresponding variable limit of 3# It is 65mph that speed direction board issues current variable speed-limit value.

Judge that current time transport need does not reduce according to the real-time traffic flow data of 3# Traffic flow detecting device, therefore enter Enter next cycle continuation variable speed-limit to control.

Claims (7)

1. a variable speed-limit control strategy optimization method for the continuous bottleneck road of through street, is characterized in that comprising the following steps:

1) determine the reasonable distance that continuous bottleneck road scope and Traffic flow detecting device are arranged, close in through street section Reason position is the most supporting arranges Traffic flow detecting device and variable speed-limit control direction board.By the periodically inspection of Traffic flow detecting device Survey the traffic flow data in Nei Ge section, continuous neck region；

2) build the real-time accident risk prediction model for the continuous bottleneck road of through street, can calculating accident respectively send out Raw probability and death by accident/injured probability, concrete steps include:

201) accident risk prediction model set up by employing order logit model, and model is divided into two steps: first, sets up traffic Accident risk prediction model, Y=1 represents vehicle accident and occurs, and Y=0 represents without vehicle accident；Secondly, accident is set up tight Weight degree forecast model, Y=1 represents injured/death by accident, and Y=0 represents only property loss accident；Above-mentioned two step In, all following basic binomial logit model forms of employing:

$P(Y=1)=\frac{1}{1+e^{-g\left(X\right)}}$

Wherein, P (Y=1) is vehicle accident probability of happening or injured/death by accident probability, and g (X) is utility function, i.e. from becoming The linear combination of amount X, expression formula is as follows:

G (X)=β₀+β₁x₁+…+β_kx_k

Wherein, x_kFor the value of traffic flow variable k, β_kCoefficient for variable k；

Gather the traffic flow data of 5 minutes before and after vehicle accident in research section occurs and the traffic fluxion not having an accident According to, variable parameter in above-mentioned model is fitted；

202) step 201) in constructed accident risk prediction model, for predicted position x_iThe vehicle accident at place is sent out The model of raw probability comprises 10 notable variablees, and its utility function is as follows:

g_i(x)=-2.672+0.074x_i1+0.060x_i2+0.050x_i3+0.119x_i4+0.092x_i5

+0.026x_i6+1.057x_i7-0.049x_i8-0.856x_i9+0.508x_i10

Wherein, x_i1For upstream detector average occupancy, x_i2Poor for upstream detector velocity standard, x_i3For detected downstream Device velocity standard is poor, x_i4For the difference of adjacent lane occupation rate, x_i5For the difference of upstream and downstream detector flow, x_i6For upstream and downstream The difference of occupation rate, x_i7For upstream and downstream detector spacings, x_i8For width of roadway, x_i9For shoulder width, x_i10For curve section Indicator variable；

203) step 201) in constructed accident risk prediction model, for predicted position x_iThe vehicle accident at place is tight The model of weight degree probability comprises 4 notable variablees, and its utility function is as follows:

g_j(x)=2.129-0.033x_j1-0.056x_j2-0.335x_j3-0.036x_j4

Wherein, x_j1For upstream detector average occupancy, x_j2For downstream detector average discharge, x_j3Indicate for peak period Variable, x_j4For width of roadway；

3) judge that whether current time is the integral multiple that variable speed-limit controls the cycle, if then to each speed(-)limit sign position Speed limit calculates, and does not the most do any next one traffic flow data that is operated into and detects the cycle, and concrete steps include:

301) according to step 1) in the actual measurement traffic flow of i upstream and downstream position, certain section that detects of Traffic flow detecting device average Speed v (x_i+1, t) with v (x_i-1, t), the target speed limit according on equation below calculating section i:

T_VSL(x_i, t)=α v (x_i+1,t)+(1-α)·v(x_i-1,t)

Wherein,

T_VSL(x_i, t) it is the section i target speed limit at moment t；

A is the rate smoothing factor (0 < a < 1)；

v(x_i+1, t) with v (x_i-1, t) it is respectively the actual speed exported at section, upstream i-1 and downstream road section i+1 detector；

302) based on step 301) in the current time position x that tries to achieve_iThe target speed limit T of variable speed-limit direction board_VSL(x_i, T), can determine that position x in conjunction with change step Δ V_iVariable speed-limit sign at the change range value of t, computing formula is such as Under:

${\mathrm{\&Delta;V}}_{SL}(x_i,t)=\left\{\begin{array}{cc}-\mathrm{\&Delta;}V,& if{T}_{VSL}(x_i,t+\mathrm{\&Delta;}t)<V_{SL}(x_i,t)-\mathrm{\&Delta;}V\\ 0,& if{V}_{SL}(x_i,t)-\mathrm{\&Delta;}V\&le;T_{VSL}(x_i,t+\mathrm{\&Delta;}t)\&le;V_{SL}(x_i,t)+\mathrm{\&Delta;}V\\ \mathrm{\&Delta;}V,& if{T}_{VSL}(x_i,t+\mathrm{\&Delta;}t)>V_{SL}(x_i,t)+\mathrm{\&Delta;}V\end{array}\right.$

Wherein,

Δ V is that variable speed-limit value changes amplitude；

T_VSLFor target speed limit；

V_SL(x_i, t) it is position x in the i of section_iVariable speed-limit sign at the speed limit of t；

303) after considering that coordination between multiple speed(-)limit sign controls, to step 302) in calculated position x_iVariable Speed(-)limit sign is modified updating in the change step value of t, and computing formula is as follows:

${\mathrm{\&Delta;V}}_{SL}(x_i,t)=\left\{\begin{array}{cc}-{\mathrm{\&Delta;V}}^{\&prime;},& if{T}_{VSL}(x_i,t+\mathrm{\&Delta;}t)\&le;T_{VSL}(x_{i+1},t+\mathrm{\&Delta;}t)-{\mathrm{\&Delta;V}}^{\&prime;}\\ 0,,& allothercases\end{array}\right.$

Wherein,

Δ V ' is adjacent segments maximum speed limit value difference；

304) by step 303) in result of calculation bring following formula into, can obtain the final speed limit in the i of current time section:

V_SL(x_i, t+ Δ t)=V_SL(x_i,t)+ΔV_SL(x_i,t)

Wherein,

V_SL(x_i, t+ Δ t) is position x in the i of section_iVariable speed-limit sign at the speed limit of t+ Δ t；

ΔV_SL(x_i, t) it is position x_iVariable speed-limit sign in the amplitude of variation of t；

305) by step 304) the middle section permission speed limit scope [V calculating gained current time and preset_min,V_max] carry out Relatively.If exceeding section to allow speed limit scope, the most only issue section corresponding edge circle speed limit；If being not above section to allow Speed limit scope, then issuing steps 304) in final speed limit in the section i that determines；

306) by the variable speed-limit direction board issuing steps 305 of supporting setting) in defined location x_iThe speed limit at place；

4) transport need situation of change is surveyed based on each bottle-neck zone, it is judged that current time actual measurement traffic flow running rate whether Reach variable speed-limit and control stop condition；When transport need does not decline in section, then maintain variable speed-limit controlled state Enter the next traffic flow data detection cycle；When in section, transport need declines, then variable speed-limit value progressively returns to Acquiescence speed limit；

5) span and the change step of four key parameters in continuous bottleneck road variable speed-limit control strategy are determined, Determining the optimization object function for genetic algorithm, concrete steps include simultaneously:

501) four cores control parameters that the variable speed-limit control strategy for continuous bottleneck road comprises are respectively as follows: variable Control for Speed Limitation cycle T, upstream and downstream rate smoothing index a, speed limit change amplitude, ao V and adjacent segments maximum speed limit Difference Δ V '.The variable speed-limit control cycle is 30 seconds to 5 minutes, and upstream and downstream rate smoothing index span is 0.1 to 0.9, Speed limit change amplitude is from 5mph to 30mph, and adjacent segments maximum limit speed difference is from 5mph to 30mph；

502) consider that variable speed-limit controls the impact on accident risk, severity of injuries and travel time, builds such as simultaneously The object function of the variable speed-limit control strategy optimization method shown in following formula:

$Fitness=-\&lsqb;\mathrm{\&gamma;}\&CenterDot;\left(\frac{R_{VSL}-R_{No}}{R_{No}}\right)+\mathrm{\&mu;}\&CenterDot;\left(\frac{I_{VSL}-I_{No}}{I_{No}}\right)+\mathrm{\&eta;}\&CenterDot;\left(\frac{T_{VSL}-T_{No}}{T_{No}}\right)\&rsqb;$

Wherein,

Fitness is fitness；

γ, μ, η (γ+μ+η=1) are weight coefficient, in accident risk, severity of injuries, obtain between the travel time Balance；

R_VSLAnd R_NOIt is respectively under variable speed-limit controls and without controlling lower accident risk；

T_VSLAnd T_NOIt is respectively under variable speed-limit controls and without total transit time under controlling；

I_VSLAnd I_NOIt is respectively under variable speed-limit controls and without controlling lower severity of injuries, it should be noted that accident is serious Degree is predicted under only in vehicle accident, a situation arises the most meaningful, therefore only general when the vehicle accident of prediction during calculating When rate exceedes accident generation threshold value, being just predicted the order of severity, the computing formula of severity of injuries is as follows:

$I=\frac{\underset{t=1}{\overset{K}{\&Sigma;}}\underset{i}{\overset{N}{\&Sigma;}}{I}_{i,t}(FI=1)}{M}$

$I_{i,t}(FI=1)=\left\{\begin{array}{cc}{P}_{i,t}(FI=1),& if{P}_{i,t}(crash=1)\&GreaterEqual;threshold\\ 0,& else\end{array}\right.$

Wherein,

K is total simulation time；

N is section number；

Threshold is the threshold value that judgement accident occurs；

M is the time span that accident risk exceedes threshold value；

I_i,t(FI=1) being death/injured accident probability, the real-time accident risk prediction model built according to second step is calculated；

6) based on step 2) to step 4) build Traffic Flow Simulation model, based on genetic algorithm and Traffic Flow Simulation model Between data transmission and iteration to step 5) in four key parameters in its span, its value is optimized.

In genetic algorithm, the value of four key parameters is as follows: Population Size M is 20, maximum algebraically T be 100, Crossover probability Pc is 0.9, mutation probability Pm is 0.1.Initialize installation evolutionary generation enumerator is t=0, arranges maximum Evolutionary generation T, arranges crossover probability Pc and mutation probability Pm in genetic algorithm；Stochastic generation M is individual as initial Colony P (0), comprises each variable speed-limit and controls parameter value information coding m (x in m-th individuality₁,x₂,x₃,x₄)；By group Individual parameter value input Traffic Flow Simulation model in body P (t), calculates the fitness of each individuality according to simulation data result； Selecting winning individuality, individual selected probability from colony is that ideal adaptation degree is in whole colony fitness comprehensively middle institute Accounting example, the random groups of individuals after being chosen strikes a bargain and matches；A cross point is set so that certain probability is random in individuality string, When carrying out intersection, two each and every one body portion structures before or after this point are interchangeable, and generate two new individualities；General according to variation Rate is chosen some individuals in colony and is changed its partial parameters value；After selection, intersection, mutation operator pass through in colony P (t), Obtain colony P (t+1) of future generation；When iterations reaches maximum algebraically T, genetic algorithm terminates；Otherwise, at the beginning of repetition Each step that beginningization is later is until end condition meets；To have the individual corresponding optimal solution (x of maximum adaptation degree₁,x₂,x₃, x₄) translate into the value of four core parameters that variable speed-limit controls；

7) by step 6) in the variable speed-limit that obtains controls cycle T, upstream and downstream rate smoothing index a, speed limit change The optimization value of amplitude, ao V and adjacent segments maximum speed limit value difference Δ V ' brings step 3 into) and step 4) in as each key The recommendation value of parameter, then according to step 1) to step 4) control strategy through street encouragement bottleneck road is entered Row variable speed-limit controls.

The variable speed-limit control strategy optimization method of the continuous bottleneck road of a kind of through street the most according to claim 1, It is characterized in that described step 201) in employing order logit model set up accident risk prediction model, model is divided into two steps: First, setting up street accidents risks forecast model, Y=1 represents vehicle accident and occurs, and Y=0 represents without vehicle accident；Its Secondary, set up severity of injuries forecast model, Y=1 represents injured/death by accident, and Y=0 represents only property loss accident. In above-mentioned two step, all following basic binomial logit model forms of employing:

$P(Y=1)=\frac{1}{1+e^{-g\left(X\right)}}$

Wherein, P (Y=1) is vehicle accident probability of happening or injured/death by accident probability, and g (X) is utility function, i.e. from becoming The linear combination of amount X, expression formula is as follows:

G (X)=β₀+β₁x₁+…+β_kx_k

Wherein, x_kFor the value of traffic flow variable k, β_kCoefficient for variable k.

The variable speed-limit control strategy optimization method of the continuous bottleneck road of a kind of through street the most according to claim 1, It is characterized in that described step 301) in rate smoothing factor a that proposes be the effect for realizing smooth velocity perturbation, a value Speed limit in the biggest then section i is (0,1) closer to the actual speed of downstream road section, the span of a.

The variable speed-limit control strategy optimization method of the continuous bottleneck road of a kind of through street the most according to claim 1, It is characterized in that described step 305) in, in the section finally determined, need to approximate immediate 5mph whole for variable speed-limit value Several times value.

The variable speed-limit control strategy optimization method of the continuous bottleneck road of a kind of through street the most according to claim 1, It is characterized in that described step 4) in variable speed-limit value progressively return to give tacit consent to speed limit process as shown by the equation: V_SL(x_i, T)=V_SL(x_i,t)+ΔV,if v(x_i,t)<v(x_i+1,t)and v(x_i, t)=V_SL(x_i,t)。

The variable speed-limit control strategy optimization method of the continuous bottleneck road of a kind of through street the most according to claim 1, It is characterized in that described step 502) in optimization object function computing formula in, the computing formula of accident risk is as follows:

$R=\frac{\underset{t=1}{\overset{K}{\&Sigma;}}\underset{i}{\overset{N}{\&Sigma;}}{P}_{i,t}(crash=1)}{N\&times;K}$

Wherein,

P_i,t(crash=1) be the section i accident probability in t, can be according to step 2) in the real-time accident risk set up pre- Survey model is calculated；

K is total simulation time；

N is section number.

The variable speed-limit control strategy optimization method of the continuous bottleneck road of a kind of through street the most according to claim 1, It is characterized in that described step 502) in optimization object function computing formula in, the computing formula of total transit time is as follows:

$T=\frac{\underset{t=1}{\overset{K}{\&Sigma;}}\underset{i=1}{\overset{N}{\&Sigma;}}{d}_i\left(t\right)\&CenterDot;\mathrm{\&Delta;}t}{N\&times;K}$

Wherein,

d_iT () is the section i vehicle number in t；

d_iT () is the section i vehicle number in t；

Δ t is simulation time step-length；

K is total simulation time；

N is section number.