CN110517510A - Based on the Intersections timing designing method for improving Webster function and genetic algorithm - Google Patents
Based on the Intersections timing designing method for improving Webster function and genetic algorithm Download PDFInfo
- Publication number
- CN110517510A CN110517510A CN201910795812.2A CN201910795812A CN110517510A CN 110517510 A CN110517510 A CN 110517510A CN 201910795812 A CN201910795812 A CN 201910795812A CN 110517510 A CN110517510 A CN 110517510A
- Authority
- CN
- China
- Prior art keywords
- function
- vehicle
- delay
- webster
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0137—Measuring and analyzing of parameters relative to traffic conditions for specific applications
- G08G1/0145—Measuring and analyzing of parameters relative to traffic conditions for specific applications for active traffic flow control
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/07—Controlling traffic signals
- G08G1/08—Controlling traffic signals according to detected number or speed of vehicles
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Traffic Control Systems (AREA)
Abstract
The present invention relates to a kind of based on the Intersections timing designing method for improving Webster function and genetic algorithm, include: that vehicle delay data are obtained according to conventional cross mouth vehicle delay function Webster function, actual vehicle is generated according to sample and is delayed data;The vehicle delay data that the vehicle delay data and point sample obtain to Webster function generate carry out data fitting, construct improved intersection vehicles delay function;Using improved intersection vehicles delay function as first object majorized function, Optimal Signals lamp timing scheme is obtained using the improved adaptive GA-IAGA for including penalty factor;Improved adaptive GA-IAGA is obtained into Optimal Signals lamp timing scheme as training set;The corresponding traffic signal timing scheme of practical wagon flow is obtained using XGBOOST regression model.The method of the present invention is practical, computational accuracy is high, fast response time, can greatly improve vehicle in the convenience of intersection, have broad application prospects.
Description
Technical field
The present invention relates to wisdom traffic field more particularly to a kind of friendships based on improvement Webster function and genetic algorithm
Prong traffic signal timing optimization method.
Background technique
Urban road get on the car quantity day from increase so that traffic jam issue is more serious, the whole nation resulting from
Economic loss reaches more than one hundred million members.Important component of the intersection as urban road, the alleviation of congestion problems is for city
The good development of city's traffic is got twice the result with half the effort.Existing research shows that the traffic signal timing of optimization intersection can be effectively relieved this and ask
Topic.When the magnitude of traffic flow reaches a certain level, by adjust optimize traffic signal timing, can be separated from the time be interweaved,
The traffic flow of conflict, make vehicle intersection run it is unimpeded, intersection congestion is reduced or avoided, by traffic accident occur probability
It is preferably minimized.
Currently, each each Intersections in big city generally use fixed timing, fixed timing has relative to real-time time-mixed
There is the characteristic convenient for management, but use fixed timing that cannot carry out real-time matching to wagon flow, is likely to result in vehicle flowrate delay
Larger problem.About the research of traffic signal timing optimization, information of vehicle flowrate usually becomes the effective information for differentiation.Signal
Lamp timing designing is generally divided into two classes: the first kind according to the difference of scheme, is calculating vehicle Delay Model, is confirming objective optimization
Function simultaneously solves in a kind of research: some scholars make summary for signal lamp intersection delay model and introduce, and delay is discussed
Application of the computation model in intersection is designed.Also there are some scholars using bilayer model, the above layer model target is passenger's warp
Help loss reduction, and lower layer is then that exhaust emissions is minimum, establishes Bi-level Programming Models, and count by optimization method of traditional genetic algorithm
Calculate model minimum value.Advantage is that model construction is perfect, the disadvantage is that genetic algorithm convergence rate is excessively slow, and lacks proof scheme just
The evaluation index of true property.Second class, some scholars are confirmed with traffic simulation optimizes traffic signal timing, with Vissim simulation software
Rule-based traffic signal timing optimization method is constructed, rule-based method is changed into static traffic timing, in actual number
On the basis of, the experimental results showed that journey time, delay, the evaluation methods such as queue length can all have good optimum results.
The advantages of Vissim simulation model is to be readily understood by, and disadvantage is the absence of rigorous mathematical formulae and derives.
Summary of the invention
The present invention is a kind of based on the Intersections timing designing method for improving Webster function and genetic algorithm, can
Solve that Webster traffic signal timing solving model efficiency existing in the prior art is lower, solving precision is lower, genetic algorithm is asked
The disadvantages of speed is slow is solved, while the present invention has scalability strong, the good feature of robustness.
Technical solution used by its technical problem of solution of the present invention is:
A kind of Intersections timing designing method based on improvement Webster function and genetic algorithm, comprising:
S1 obtains vehicle according to conventional cross mouth vehicle delay function Webster function and is delayed data, produced according to sample
Raw actual vehicle is delayed data;The vehicle that the vehicle delay data and point sample obtain to Webster function generate is delayed number
According to data fitting is carried out, improved intersection vehicles delay function is constructed;
S2, using improved intersection vehicles delay function as first object majorized function, using including penalty factor
Improved adaptive GA-IAGA obtains Optimal Signals lamp timing scheme;
Improved adaptive GA-IAGA is obtained Optimal Signals lamp timing scheme as training set by S3;Using XGBOOST regression model
Obtain the corresponding traffic signal timing scheme of practical wagon flow.
Preferably, the improved intersection vehicles delay function of construction, specifically includes:
S11 is prolonged by the vehicle that conventional cross mouth vehicle delay function Webster function calculates some specified intersection
Accidentally;Webster function is as follows:
Wherein, dnIndicate the delay of n-th position vehicle;C indicates signal lamp cycle duration;λnIndicate the green letter of the n-th phase
Than;qnIndicate n-th position vehicle flowrate;xnIndicate n-th position saturation degree;
S12 is handled wagon flow data using the method for sample, and the vehicle of calculating intersection all directions is practical to be prolonged
Accidentally;The vehicle flowrate data that point Sample Method uses include being parked in interior vehicle number of leading the way, stagnation of movement vehicle number and non-stagnation of movement vehicle number;
S13 carries out data fitting, constructs improved intersection vehicles delay function, as follows:
minfnew(C,λ1,λ2,λ3,λ4)=F (fold(C,λ1,λ2,λ3,λ4))
Wherein, function F (x) indicates to calculate the fitting function that real data generates, and wherein x is that Webster function calculates
The vehicles average delay come, i.e. fold(C,λ1,λ2,λ3,λ4), it is delayed by the calculated improved intersection vehicles of fitting function
Function is fnew(C,λ1,λ2,λ3,λ4).Signal lamp cycle is codetermined by the green time and the yellow temporal summation that dodges of each phase,
Middle yellow sudden strain of a muscle time ty2 seconds are fixed, complete red time thFix 3 seconds;It on the other hand is that conclusion is required to facilitate signal lamp hardware layout just
It is prompt, it is desirable that each phase green time is the positive integer in [15,60] range.
Preferably, using improved intersection vehicles delay function as first object majorized function, using include punishment because
The improved adaptive GA-IAGA of son obtains Optimal Signals lamp timing scheme, specifically includes:
S21, using improved intersection vehicles delay function as first object majorized function, argument of function includes each
Phase split and signal lamp cycle, dependent variable are intersection vehicles total delay, and vehicle flowrate is set as constant value in function;
The input of initialization population quantity, the number of iterations, boundary definition and first object majorized function, each population the inside include
Different signal lamp cycles and each phase split;
S22 sets the crossover probability and mutation probability of population;Operation is carried out in such a way that crossing formula is made a variation;
The fitness for not meeting the population of constraint condition in improved intersection vehicles delay function is set as negative by S23,
And this negative is referred to as penalty factor;Meet calculating using improved intersection vehicles delay function for constraint condition population
Fitness;The population quantity N that all fitness in the former generation are negative is counted, N number of new population is regenerated later and is added to and work as
Preceding filial generation guarantees to remain unchanged when the population quantity of former generation, repeats step S21 and S22 always later;
S24 iterates to defined the number of iterations until convergence, obtains Optimal Signals lamp timing scheme.
Preferably, improved adaptive GA-IAGA is obtained into Optimal Signals lamp timing scheme as training set;It is returned using XGBOOST
Model obtains the corresponding traffic signal timing scheme of practical wagon flow, specifically includes:
S31 constructs the Intersections timing based on improved intersection vehicles delay function and improved adaptive GA-IAGA
Optimized model, and a plurality of leggy vehicle flowrate in preset time is generated at random, calculate the signal lamp of corresponding time vehicle flowrate
Timing, in this, as model training collection;
S32 calculates the internal relation of vehicle flowrate and the timing of corresponding signal lamp using extreme gradient lift scheme XGBOOST,
Second objective optimization function is as follows:
Wherein,Indicate the model prediction result that extreme gradient lift scheme XGBOOST takes turns in t-1With
True output yiBetween difference, referred to as loss function;Loss function is existedAccording to Taylor expansion, wherein giIndicate one
Rank derived function;hiIndicate second order derived function;Ω(ft) indicating regularization term, n indicates the number of iterations of XGBOOST;ft(Xi) indicate
Extreme gradient lift scheme XGBOOST takes turns in t to sample XiPrediction result;XiIndicate sample Xi;
S33 calculates the corresponding signal of practical wagon flow according to the extreme gradient lift scheme XGBOOST that wagon flow data calculate
Lamp timing scheme, evaluation index are subject to relative error, as follows:
Beneficial effects of the present invention are as follows:
(1) a kind of Intersections timing designing method based on improvement Webster function and genetic algorithm of the present invention,
Delay time at stop for reducing public vehicles in intersection can obtain so that the effect that intersection total delay reduces;
(2) since there are also certain differences for traditional Webster Delay Model and actual intersection Delay Model, originally
Invention is delayed the one improved intersection vehicles with degree of precision of method acquisition being fitted to actual delay using theory and prolonged
Accidentally function;
(3) population that improved adaptive GA-IAGA of the invention intersects population, generation is unsatisfactory for constraint condition after variation is taken
Deletion and again recruit population are until convergence in population;By using improved adaptive GA-IAGA to improved intersection vehicles delay function
Global minimum is sought, to reach the purpose for solving traffic signal timing;
(4) present invention proposes that a kind of signal lamp genetic algorithm based on XGBOOST regression model solves acceleration model;It will lose
Model partition is online and offline two parts, by calculating training set pair under line by thinking of the propagation algorithm in conjunction with regression model
The corresponding regression model of the traffic signal timing answered accelerates the data calculating speed of line upper returning model.
Detailed description of the invention
Fig. 1 is that the present invention is based on the Intersections timing designing methods for improving Webster function and genetic algorithm
Functional block diagram;
Fig. 2 is the flow chart of the intersection vehicles delay function construction method in the present invention based on data fitting;
Fig. 3 is the flow chart of the genetic algorithm based on penalty factor in the present invention;
Fig. 4 is the flow chart of the signal lamp genetic algorithm solution acceleration model in the present invention based on XGBOOST regression model;
Fig. 5 is the flow chart of the Intersections optimization verifying model in the present invention based on Vissim simulation model.
Specific embodiment
The present invention is further described in detail with reference to the accompanying drawings and embodiments.
Shown in Figure 1, the present invention is a kind of based on the Intersections timing for improving Webster function and genetic algorithm
Optimization method, the delay time at stop for reducing public vehicles in intersection can obtain the effect so that the reduction of intersection total delay,
Include:
S1 obtains vehicle according to conventional cross mouth vehicle delay function Webster function and is delayed data, produced according to sample
Raw actual vehicle is delayed data;The vehicle that the vehicle delay data and point sample obtain to Webster function generate is delayed number
According to data fitting is carried out, improved intersection vehicles delay function is constructed;
S2, using improved intersection vehicles delay function as first object majorized function, using including penalty factor
Improved adaptive GA-IAGA obtains Optimal Signals lamp timing scheme;
Improved adaptive GA-IAGA is obtained Optimal Signals lamp timing scheme as training set by S3;Using XGBOOST regression model
Obtain the corresponding traffic signal timing scheme of practical wagon flow;
S4, the Intersections based on Vissim simulation model optimize verifying.
Means using fitting be because tradition Webster Delay Model and actual intersection Delay Model also
Certain difference, so being delayed the method being fitted to actual delay using theory obtains the improved friendship with degree of precision
Prong vehicle delay function is used as objective optimization function.Shown in Figure 2, the detailed process of the step S1 is successively are as follows:
(11), prolonged by the vehicle that conventional cross mouth vehicle delay function Webster function calculates some specified intersection
Accidentally, after calculating the delay of tradition Webster function vehicle, (12) are entered step.Shown in Webster function such as formula (1):
Parameter declaration is shown in Table 1:
Parameter needed for table 1webster function
(12), known formula 1 is tradition Webster Delay Model, but this theoretical Delay Model and actual road are handed over
There are also certain differences for prong Delay Model, are delayed data to calculate actual intersection, the present invention determines to use point sample
This method handles wagon flow data, calculates the vehicle delay of intersection all directions.(13) are entered step later.
(13), after the actual delay for obtaining intersection, traditional Webster delay function result is adopted to actual delay number
Obtaining the new Webster delay function with degree of precision according to the method for fitting is used as specifying the vehicle of intersection flat
Equal delay function includes some constraint conditions inside function shown in the function such as formula (2) after being fitted, specific as follows shown:
minfnew(C,λ1,λ2,λ3,λ4)=F (fold(C,λ1,λ2,λ3,λ4))
Function F (x) is represented as calculating the fitting function that real data generates, and wherein X is that tradition Webster delay estimation goes out
The vehicles average delay come, i.e. fold(C,λ1,λ2,λ3,λ4), it is f by the calculated vehicle delay function of fitting functionnew(C,
λ1,λ2,λ3,λ4);Signal lamp cycle is codetermined by the green time and the yellow temporal summation that dodges of each phase, wherein yellow dodge time th
2 seconds are fixed, complete red time thFix 3 seconds;It on the other hand is that conclusion is required to facilitate signal lamp hardware layout convenient, it is desirable that each phase
Green time is the positive integer in [15,60] range.
Shown in Figure 3, the population that the S2 intersects population, generation is unsatisfactory for constraint condition after variation takes deletion simultaneously
Again recruit population is until convergence in population.The intersection vehicles total delay that first part is acquired by using improved adaptive GA-IAGA
Function seeks global minimum, to reach the purpose for solving traffic signal timing, detailed process is successively are as follows:
(21), firstly, it is consistent with traditional genetic algorithm, it is specified that population quantity, the number of iterations, boundary definition and model
Input.Mode input is using n traffic signal timing being randomly generated as initialization population, after completing initialization of population
, enter step (22);
(22), secondly, with reference to traditional genetic algorithm, according to fixed intersection α, variation β probability, parent population is selected
The process select, intersect, to make a variation generates new progeny population, enters step (23);
(23), then, due to the constraint condition that signal lamp cycle includes green time, yellow time and complete red time,
So necessarily there is progeny population not meet constraint, the population's fitness is set as negative at this time, this negative is referred to as penalty factor,
Show that this population is that performance is poor, negative is because the population that genetic algorithm always selects fitness high, allows this kind of population can
It is on the contrary then population is allowed to wither away to survive down, meet fitness calculating process data with reference to described in S1 of constraint condition population
The intersection vehicles delay function of fitting.The population quantity N when fitness all in former generation are negative is counted later, later again
It generates N number of new population and is added to current filial generation, guarantee to remain unchanged when the population quantity of former generation, repeat step always later
(22)-(23);
(24), finally, the number of iterations as defined in iterating to is until convergence, obtains Optimal Signals lamp timing scheme.
To solve the slower characteristic of genetic algorithm convergence rate, by thinking of the genetic algorithm in conjunction with regression model, by mould
Type is divided on line, two parts under line are accelerated by calculating the corresponding regression model of the corresponding traffic signal timing of training set under line
The data calculating speed of line upper returning model.Shown in Figure 4, the detailed process of the step S3 is successively are as follows:
(31), it firstly, according to S1 and S2, constructs based on the intersection for improving Webster function and improved adaptive GA-IAGA
Message signal lamp signal timing optimization model, and a plurality of leggy vehicle flowrate in preset time is generated at random, calculate corresponding time vehicle
The traffic signal timing of flow, in this, as model training collection.
(32), consider that genetic algorithm convergence rate is slower, model uses regression effect in the small situation of data volume still preferable
Extreme gradient lift scheme (XGBOOST), calculate vehicle flowrate and the timing of corresponding signal lamp internal relation, objective optimization function
As shown in formula (3):
To sample (x, y), traditional regression models are typically directly based on model outputWith true output yiBetween difference
NotIt is lost to calculate, only when identical, loss is just zero, but unlike XGBOOST, it is proposed
The thought of stochastical sampling feature and stochastical sampling sample avoids data over-fitting, and only makes relative to traditional regression models
With single order derived function giCalculating parameter convergence process, XGBoost use second order derived function hiAccelerate convergence rate;In addition model is also
Introduce regularization term Ω (ft), avoid model over-fitting.Such methods usually can be very good prediction data.
(33), it is corresponding to calculate practical wagon flow for the XGBOOST regression model calculated according to step (32) according to wagon flow data
Traffic signal timing scheme, evaluation index are subject to relative error, shown in calculation formula such as formula (4).
It is shown in Figure 5, propose a kind of Intersections optimization verification method based on Vissim simulation model, verifying
Scheme correctness, specifically includes: based on the intersection for improving Webster function and improved adaptive GA-IAGA according to S1 and S2
Traffic signal timing optimization needs to carry out evaluation verifying before and after optimizing Intersections.The detailed process of the S4 is successively are as follows:
(41), firstly, it is necessary to be investigated for the physical factor of specified intersection, for example, area, quantity of leading the way etc., structure
The traffic simulation road network that part tallies with the actual situation, concurrently sets emulation data collection point, facilitates investigation vehicle delay data.Later
Enter step (42).
(42), secondly, according to practical vehicle flowrate and original traffic signal timing collectively as the input parameter of simulation model,
Computer sim- ulation mean delay is compared with the intersection vehicles delay function data being fitted based on data, verifies the quasi- of function
Conjunction degree quality.Evaluation index is subject to relative error, shown in calculation formula such as formula (4).(43) are entered step after the completion.
△ indicates absolute error in formula 3, is the error of true value L and measured value.
(43), based on the Intersections for improving Webster function and improved adaptive GA-IAGA according to S1 and S2
As a result, being put into the traffic simulation road network of step building, observation is calculated the traffic signal timing that timing calculates based on emulation road network
Optimization post-simulation average vehicle delay be based on step 1, the relative error of theoretical average vehicle delay after 2 optimizations calculated,
The correctness of evaluation procedure S1 and S2.Evaluation index is subject to relative error, shown in calculation formula such as formula (4).
Embodiment 1
Step 1, the intersection vehicles delay function construction process based on data fitting are as follows:
This example uses 22 Feng Zelu in the city A April in 2018-neoasozin South Road Intersections morning 10:00-11:
00 video data is analyzed, the part of step 1 be in order to obtain the vehicle delay function for meeting specified intersection, so:
1) firstly, generating vehicle flowrate text data according to intersection video data, data file is carried out using excel file
Storage, data file constitute (such as table 4) by 13 fields;Intersection wagon flow video data is provided by Department of Communications, the city A, is the city A
April in 2018 22 Feng Zelu-neoasozin South Road intersection video data, file size 8.50GB (9,131,431,375 words
Section), storage format is MP4 file.Method by manually counting vehicle converts video data to vehicle flowrate text data, textual data
According to format it is as shown in table 2.
2 wagon flow data of table
2) secondly, according to the city A 22 Feng Zelu April in 2018-neoasozin South Road intersection in 10:00-11:00
Traffic signal timing calculates the vehicles average delay of all directions, and data format is as shown in table 3.
3 delay estimation data requirements table of table
It is known: signal lamp of the city A 22 Feng Zelu April in 2018-neoasozin South Road intersection in 10:00-11:00
Period 141 seconds, the Huang sudden strain of a muscle time 2 seconds, complete red time 3 seconds, the green time length of all directions are as follows:
First phase: it turns right, turn right to the east of west 32 seconds to straight trip+right-hand rotation, south to the north in north to south;
Second phase: it turns right, turn right west to east 32 seconds to straight trip+right-hand rotation, north to south in south to the north;
Third phase: to the east of west to straight trip, west to east to straight trip 32 seconds;
4th phase: south turns right to the north, turns left+turn right to the east of west, north to south is turned right, west to east
It turns left+turns right 25 seconds;
The split of each phase of signal lamp: λ can be calculated by green time and cycle time1=0.228571;λ2
=0.228571;λ3=0.228571;λ4=0.178571;, the road essential information according to provided by Department of Communications, the city A obtains later
The maximum traffic volume 1800 of intersection surrounding road is taken, thirdly, all directions traffic flow text flowmeter obtained according to part 1
Road saturation degree is calculated, the road saturation degree of all directions is respectively as follows:
x1=0.17;x2=0.228889;x3=0.09;x4=0.072222;x5=0.342222;x6=0.104444;
x7=0.114444;x8=0.22;x9=0.104444;x10=0.091111;x11=0.323333;x12=
0.118889
Later, according to known split, saturation degree, the data of vehicle flowrate calculate tradition Webster delay function jointly
As a result, being respectively as follows:
d1=43.65073;d2=44.27062;d3=25.72185;d4=48.19098;d5=45.51452;d6=
25.87984;
d7=43.08161;d8=44.17593;d9=25.87984;d10=48.35619;d11=45.30238;d12=
26.03978
3) it after calculating the result for completing tradition Webster delay function, according to the content of step 1), needs result
The real data generated to sample is drawn close, and is carried out data fitting, is reduced the error of function.The data such as table 4 that point sample needs
It is shown:
4 sample data demand schedules of table
The city A 22 Feng Zelu April in the 2018-neoasozin South Road intersection morning 10:00- provided according to Department of Communications, the city A
The video data of 11:00, the data in available intersection periphery section, as shown in table 5, calculating process is as follows: (to the east of
For westwards keeping straight on)
Table 5- point sample data demand:
Then the vehicles average delay of east-west direction can be calculated by table 2: d1=36.06271777, similarly, it is each to calculate remaining
The point sample vehicle in a direction is delayed, and as a result knows:
d1=36.06271777;d2=44.21052632;d3=35.45454545;d4=42.33471074;d5=
64.83050847;
d6=30.47619048;d7=42.06293706;d8=57.5625;d9=25.61320755;d10=
40.90909091;
d11=36.58878505;d12=38.15217391;
4) result that sample is calculated with tradition Webster delay function is analyzed, both discoveries are basic
Linear relationship is presented, so decision will be traditional by the way of linear regression in order to promote the precision of Webster delay function
On the truthful data that Webster Function Fitting is calculated to sample, functional form after being fitted: f (x)=0.68x+14.63,
The Webster delay function model with degree of precision is obtained with by above-mentioned function.
The specific steps that step 2, the genetic algorithm based on penalty factor are realized:
Step 2 content is to propose a kind of genetic algorithm based on penalty factor, is unsatisfactory for generating after population intersection, variation
The population of constraint condition take deletion and again recruit population up to convergence in population.By using improved adaptive GA-IAGA to first
The intersection vehicles total delay function acquired is divided to seek global minimum, to reach the purpose for solving traffic signal timing, step is such as
Shown in lower:
1) firstly, using the improvement Webster delay function of generation as objective optimization function, argument of function includes each
Phase split and signal lamp cycle C, λ1,λ2,λ3,λ4, dependent variable is intersection vehicles total delay, by vehicle flowrate in function
Be set as constant value, shown in vehicle flowrate table reference table 6: on the south northwards and for the western all directions wagon flow of east orientation:
Table 6- vehicle flowrate tables of data
Genetic algorithm first the step of be initialization population, so initialization population of the present invention 60, iteration 80 times, each kind
Group the inside all includes different signal lamp cycle and each phase split, such as:
2) secondly, setting the intersection of population, mutation probability 0.1,0.9, population of the invention is the decimal system due to number
Floating type, so operation is carried out by the way of directly being made a variation with crossing formula, with population q1,q2For, it is assumed that it is random
The number of generation represents two populations and is intersected within crossover probability range, is selected inside each population first to which
Independent variable carries out crossover operation, and number is assumed to be 0.65, carries out crossover operation to first independent variable, then updating as follows:
100.30860831055195 updated value are as follows: 0.65*100.30860831055195+ (1-0.65) *
130.52731634028126=110.8851561
130.52731634028126 updated value are as follows: 0.65*130.52731634028126+ (1-0.65) *
100.30860831055195=119.9507685
The update of remaining independent variable is similar to illustrated above;
The reason that makes a variation is similar to its, selects the independent variable of variation first, and assumes the number generated at random in mutation probability
Within range, representing some population and make a variation, number is assumed to be 0.09, mutation operation is carried out to first independent variable, and
Assuming that the valued space of first independent variable is [100,180], then updating as follows:
100.30860831055195 updated value are as follows: 100.30860831055195+0.5* (180-
100.30860831055195) * 0.09=103.89472093
The mutation operation of remaining population independent variable is similar as shown above.
3) since signal lamp has a various constraint conditions, for example, signal lamp green time must [15,
30] within, the sum of split is necessary for 1 etc., then carry out step 1) and 2) operation necessarily lead to some be not inconsistent contract later
Beam condition population, for this kind of populations, the present invention determines the fitness of this kind of populations being set as negative, and this negative
Referred to as penalty factor, because genetic algorithm always selects the population by high fitness to survive, remaining population is then slow
Slowly be eliminated, and the population of negative then never may survival go down, so the genetic algorithm based on penalty factor always can be with
Preferably obtain the globally optimal solution of objective optimization function.
4) similarly with traditional genetic algorithm, step 4) will iteration step 2)~3) until objective optimization function
Convergence, shown in convergency value reference table 7:
Table 7- genetic algorithm convergence process
Step 3, the signal lamp genetic algorithm based on XGBOOST regression model solve acceleration model, and specific step is as follows:
The present embodiment is to solve the slower characteristic of genetic algorithm convergence rate, is proposed genetic algorithm in conjunction with regression model
Thinking, by model partition be line on, two parts under line, by under line calculate training set it is traffic signal timing corresponding time corresponding
Return model, accelerates the data calculating speed of line upper returning model.
According to step 1 traffic signal timing scheme corresponding with the vehicle flowrate that step 2 exports as training set, specific steps are such as
Shown in lower:
1) firstly, generating n vehicle flowrate record at random, vehicle flowrate distribution meets Poisson distribution, and by step 2
The corresponding traffic signal timing scheme of the different car flow informations of method calculating, since n item record is for doing regression model, so to mention
High accuracy, it is proposed that n value is 1000 or so, and in this way while meeting data volume demand, computer asks the time of optimal solution
Will not be excessively slow, specific example is shown in Table 8:
The corresponding traffic signal timing scheme table of the random wagon flow of table 8-
15 datas are only listed in table 5, and prepare to carry out using 1000 row data as training set in the present invention
Regression training.
2) then, consider genetic algorithm convergence rate it is slower, model use the small situation of data volume under regression effect still compared with
Good extreme gradient lift scheme (XGBOOST) calculates the internal relation of vehicle flowrate and the timing of corresponding signal lamp, objective optimization letter
Number is as shown in formula 3:
XGBOOST is carried out using the from xgboost.sklearn import XGBRegressor of python exploitation to return
Return training, training set is divided into 80% training set, and 20% as test set verifying model accuracy.
The results are shown in Table 9 for model prediction (for predicting first phase split):
Table 9- model prediction Comparative result table
The experimental results showed that the regression model relative error based on XGBoost Ensemble Learning Algorithms is maintained at 12.25%,
Reference value with higher, while by can greatly accelerate genetic algorithm for regression model and genetic algorithm simultaneous
Convergence rate.
3) finally, according to the XGBOOST regression model that step 2 is calculated according to wagon flow data, it is corresponding to calculate practical wagon flow
Traffic signal timing scheme.Practical vehicle flowrate data are as shown in table 10, it is contemplated that table length, only with each east-west direction wagon flow in north and south
Data citing.
The practical vehicle flowrate data of table 10-
By the way that 6 data of table are brought into the trained XGBOOST model of step 2, obtaining current time vehicle flowrate should
Traffic signal timing mode, as a result as shown in table 11:
Table 11- traffic signal timing table
Step 4, the Intersections optimization verifying model based on Vissim simulation model
The groundwork of step 4 is according to step 1 and 2 based on improvement Webster function and improved adaptive GA-IAGA
Intersections timing designing, need to optimize front and back to Intersections and carry out evaluation verifying, proof scheme correctness,
Key step is as shown in table 12:
Table 12- emulates data requirements
1) firstly, it is necessary to be investigated for the physical factor of specified intersection, for example, area, quantity of leading the way etc., component
The traffic simulation road network to tally with the actual situation, concurrently sets emulation data collection point, facilitates investigation vehicle delay data.According to A
Feng Zelu-neoasozin South Road intersection CAD the plan view and road essential information that Department of Communications, city provides can be emulated in Vissim
Traffic network is constructed in software emulates base map.The intersection peripheral path width is 3 meters of width.Quantity of leading the way be respectively west to
East enters 6 lane of intersection, and 3 lane of intersection, east orientation west enter 6 lane of intersection out, and 3 lane of intersection, south orientation north enter to intersect out
4 lanes of mouth, 2 lane of intersection, north orientation south enter 4 lane of intersection out, out 2 lane of intersection;It can be with according to above-mentioned essential information
The emulation base map of this intersection is constructed, emulation job is completed.
About traffic signal timing problem, it is known that: the city A 22 Feng Zelu April in 2018-neoasozin South Road intersection is 10:
Signal lamp cycle when 00-11:00 141 seconds, the Huang sudden strain of a muscle time 2 seconds, complete red time 3 seconds, the green time length of all directions
Are as follows:
First phase: it turns right, turn right to the east of west 32 seconds to straight trip+right-hand rotation, south to the north in north to south;
Second phase: it turns right, turn right west to east 32 seconds to straight trip+right-hand rotation, north to south in south to the north;
Third phase: to the east of west to straight trip, west to east to straight trip 32 seconds;
4th phase: south turns right to the north, turns left+turn right to the east of west, north to south is turned right, west to east
It turns left+turns right 25 seconds;
The split of each phase of signal lamp: λ can be calculated by green time and cycle time1=0.228571;λ2
=0.228571;λ3=0.228571;λ4=0.178571;, the road essential information according to provided by Department of Communications, the city A obtains later
Take the maximum traffic volume 1800 of intersection surrounding road.
In addition it to be compared with the data in step 1 and step 2, needs to establish data collection point in simulations, be used for
Quantization investigation is carried out to vehicle delay, it is all 12 that data collection point quantity is consistent with the quantity in step 1, each direction
Wagon flow each 3, four direction wagon flow altogether.Data collection point setting information is as shown in table 13:
Table 13- data collection point setting information
2) secondly, being counted according to practical vehicle flowrate and original traffic signal timing collectively as the input parameter of simulation model
Emulation mean delay is calculated, emulation delay table is as shown in Table 14:
Table 14- truthful data and emulation data comparison table
Analytical table 7 it can be found that true average vehicle delay with emulate average vehicle delay be it is similar, the two data are only
Have 9.97% relative error, such error cause be because the vehicle arrival in simulation model is uniformly to reach, and it is practical
The arrival of situation vehicle is then a chance event, but still illustrates that emulating data has some reference value.
3) based on the intersection signal for improving Webster function and improved adaptive GA-IAGA according to step 1 and step 2
The traffic signal timing that lamp timing calculates is as a result, be put into the traffic simulation road network of step building, and observation is based on emulation road network
The phase of the optimization post-simulation average vehicle delay of calculation and theoretical average vehicle delay after the optimization based on step 1 and step 2 calculating
To error, the correctness of evaluation procedure 1 and step 2.Data comparison is referring to shown in table 15:
Theory and emulation delay comparison after table 15- optimization
Analytical table 8 by the calculated theoretical vehicle after genetic algorithm optimization it can be found that being delayed and passing through genetic algorithm
And Vissim simulates the relative error that the emulation vehicle come is delayed only 4.77%, such error causes to be because of emulation mould
Vehicle arrival in type is uniformly to reach, and the arrival of actual conditions vehicle is then a chance event, it was demonstrated that this scheme timing exists
Equally there are better effects in truth.
Above-described embodiment is intended merely to illustrate the present invention, and is not used as limitation of the invention.As long as according to this hair
Bright technical spirit is changed above-described embodiment, modification etc. will all be fallen in the scope of the claims of the invention.
Claims (4)
1. a kind of based on the Intersections timing designing method for improving Webster function and genetic algorithm, which is characterized in that
Include:
S1 obtains vehicle according to conventional cross mouth vehicle delay function Webster function and is delayed data, is generated according to sample real
The vehicle on border is delayed data;To Webster function obtain vehicle delay data and put sample generate vehicle be delayed data into
The fitting of row data, constructs improved intersection vehicles delay function;
S2 uses the improvement including penalty factor using improved intersection vehicles delay function as first object majorized function
Genetic algorithm obtains Optimal Signals lamp timing scheme;
Improved adaptive GA-IAGA is obtained Optimal Signals lamp timing scheme as training set by S3;It is obtained using XGBOOST regression model
The corresponding traffic signal timing scheme of practical wagon flow.
2. according to claim 1 a kind of based on the Intersections timing for improving Webster function and genetic algorithm
Optimization method, which is characterized in that the improved intersection vehicles delay function of construction specifically includes:
S11 is delayed by the vehicle that conventional cross mouth vehicle delay function Webster function calculates some specified intersection;
Webster function is as follows:
Wherein, dnIndicate the delay of n-th position vehicle;C indicates signal lamp cycle duration;λnIndicate the n-th phase split;qn
Indicate n-th position vehicle flowrate;xnIndicate n-th position saturation degree;
S12 is handled wagon flow data using the method for sample, calculates the vehicle actual delay of intersection all directions;Point
The vehicle flowrate data that Sample Method uses include being parked in interior vehicle number of leading the way, stagnation of movement vehicle number and non-stagnation of movement vehicle number;
S13 carries out data fitting, constructs improved intersection vehicles delay function, as follows:
minfnew(C,λ1,λ2,λ3,λ4)=F (fold(C,λ1,λ2,λ3,λ4))
Wherein, function F (x) indicates to calculate the fitting function that real data generates, and wherein x is what Webster function calculated
Vehicles average delay, i.e. fold(C,λ1,λ2,λ3,λ4), pass through the calculated improved intersection vehicles delay function of fitting function
For fnew(C,λ1,λ2,λ3,λ4).Signal lamp cycle is codetermined by the green time and the yellow temporal summation that dodges of each phase, wherein yellow
Dodge time ty2 seconds are fixed, complete red time thFix 3 seconds;On the other hand it to require conclusion to facilitate signal lamp hardware layout convenient,
Seeking each phase green time is the positive integer in [15,60] range.
3. according to claim 2 a kind of based on the Intersections timing for improving Webster function and genetic algorithm
Optimization method, which is characterized in that using improved intersection vehicles delay function as first object majorized function, using including punishing
The improved adaptive GA-IAGA of penalty factor obtains Optimal Signals lamp timing scheme, specifically includes:
S21, using improved intersection vehicles delay function as first object majorized function, argument of function includes each phase
Split and signal lamp cycle, dependent variable are intersection vehicles total delay, and vehicle flowrate is set as constant value in function;Initially
Change the input of population quantity, the number of iterations, boundary definition and first object majorized function, includes not inside each population
Same signal lamp cycle and each phase split;
S22 sets the crossover probability and mutation probability of population;Operation is carried out in such a way that crossing formula is made a variation;
The fitness for not meeting the population of constraint condition in improved intersection vehicles delay function is set as negative, and handle by S23
This negative is referred to as penalty factor;Meet calculating using improved intersection vehicles delay function for constraint condition population to adapt to
Degree;It counts as the population quantity N that all fitness are negative in former generation, regenerates N number of new population later and be added to current son
In generation, guarantees to remain unchanged when the population quantity of former generation, repeats step S21 and S22 always later;
S24 iterates to defined the number of iterations until convergence, obtains Optimal Signals lamp timing scheme.
4. according to claim 3 a kind of based on the Intersections timing for improving Webster function and genetic algorithm
Optimization method, which is characterized in that improved adaptive GA-IAGA is obtained into Optimal Signals lamp timing scheme as training set;Using
XGBOOST regression model obtains the corresponding traffic signal timing scheme of practical wagon flow, specifically includes:
S31 constructs the Intersections timing designing based on improved intersection vehicles delay function and improved adaptive GA-IAGA
Model, and a plurality of leggy vehicle flowrate in preset time is generated at random, the traffic signal timing of corresponding time vehicle flowrate is calculated,
In this, as model training collection;
S32 calculates the internal relation of vehicle flowrate and the timing of corresponding signal lamp using extreme gradient lift scheme XGBOOST, and second
Objective optimization function is as follows:
Wherein,Indicate the model prediction result that extreme gradient lift scheme XGBOOST takes turns in t-1With it is true defeated
Y outiBetween difference, referred to as loss function;Loss function is existedAccording to Taylor expansion, wherein giIndicate that single order leads letter
Number;hiIndicate second order derived function;Ω(ft) indicating regularization term, n indicates the number of iterations of XGBOOST;ft(Xi) indicate extreme ladder
Lift scheme XGBOOST is spent to take turns in t to sample XiPrediction result;XiIndicate sample Xi;
S33 calculates the corresponding signal lamp of practical wagon flow and matches according to the extreme gradient lift scheme XGBOOST that wagon flow data calculate
When scheme, evaluation index is subject to relative error, as follows:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910795812.2A CN110517510B (en) | 2019-08-27 | 2019-08-27 | Intersection signal lamp timing optimization method based on improved Webster function and genetic algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910795812.2A CN110517510B (en) | 2019-08-27 | 2019-08-27 | Intersection signal lamp timing optimization method based on improved Webster function and genetic algorithm |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110517510A true CN110517510A (en) | 2019-11-29 |
CN110517510B CN110517510B (en) | 2020-12-25 |
Family
ID=68626211
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910795812.2A Active CN110517510B (en) | 2019-08-27 | 2019-08-27 | Intersection signal lamp timing optimization method based on improved Webster function and genetic algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110517510B (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112905912A (en) * | 2021-03-30 | 2021-06-04 | 第四范式(北京)技术有限公司 | Timing scheme determination method and device |
CN113706860A (en) * | 2021-07-14 | 2021-11-26 | 石家庄铁道大学 | Intelligent timing traffic light control method based on raspberry group |
CN114093178A (en) * | 2022-01-21 | 2022-02-25 | 杭州海康威视数字技术股份有限公司 | Traffic signal control method and device, electronic equipment and storage medium |
CN114299713A (en) * | 2021-12-01 | 2022-04-08 | 重庆邮电大学 | Single-point intersection vehicle scheduling system and method based on self-adaptive signal lamp |
CN116189459A (en) * | 2023-04-26 | 2023-05-30 | 西南民族大学 | Intersection traffic signal lamp timing method |
CN117523823A (en) * | 2023-10-11 | 2024-02-06 | 吉林师范大学 | Regional traffic signal control optimization method based on quantum genetic algorithm |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103680127A (en) * | 2013-08-29 | 2014-03-26 | 中国科学院地理科学与资源研究所 | A method for calculating signal lamp control road intersection delays through the utilization of low sampling rate floating vehicle data |
CN104952261A (en) * | 2015-06-25 | 2015-09-30 | 广东工业大学 | Single intersection traffic signal optimizing control method |
CN106558226A (en) * | 2016-11-23 | 2017-04-05 | 中兴软创科技股份有限公司 | Traffic signal timing is assessed and method of real-time adjustment |
CN107331159A (en) * | 2017-08-17 | 2017-11-07 | 上海交通大学 | A kind of traffic major trunk roads velocity estimation apparatus based on coil checker data |
CN109785619A (en) * | 2019-01-21 | 2019-05-21 | 南京邮电大学 | Regional traffic signal coordination and optimization control system and its control method |
CN110164148A (en) * | 2019-05-28 | 2019-08-23 | 成都信息工程大学 | A kind of urban road crossing traffic lights intelligently matches period control method and control system |
-
2019
- 2019-08-27 CN CN201910795812.2A patent/CN110517510B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103680127A (en) * | 2013-08-29 | 2014-03-26 | 中国科学院地理科学与资源研究所 | A method for calculating signal lamp control road intersection delays through the utilization of low sampling rate floating vehicle data |
CN104952261A (en) * | 2015-06-25 | 2015-09-30 | 广东工业大学 | Single intersection traffic signal optimizing control method |
CN106558226A (en) * | 2016-11-23 | 2017-04-05 | 中兴软创科技股份有限公司 | Traffic signal timing is assessed and method of real-time adjustment |
CN107331159A (en) * | 2017-08-17 | 2017-11-07 | 上海交通大学 | A kind of traffic major trunk roads velocity estimation apparatus based on coil checker data |
CN109785619A (en) * | 2019-01-21 | 2019-05-21 | 南京邮电大学 | Regional traffic signal coordination and optimization control system and its control method |
CN110164148A (en) * | 2019-05-28 | 2019-08-23 | 成都信息工程大学 | A kind of urban road crossing traffic lights intelligently matches period control method and control system |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112905912A (en) * | 2021-03-30 | 2021-06-04 | 第四范式(北京)技术有限公司 | Timing scheme determination method and device |
CN112905912B (en) * | 2021-03-30 | 2024-02-02 | 第四范式(北京)技术有限公司 | Timing scheme determining method and device |
CN113706860A (en) * | 2021-07-14 | 2021-11-26 | 石家庄铁道大学 | Intelligent timing traffic light control method based on raspberry group |
CN114299713A (en) * | 2021-12-01 | 2022-04-08 | 重庆邮电大学 | Single-point intersection vehicle scheduling system and method based on self-adaptive signal lamp |
CN114299713B (en) * | 2021-12-01 | 2023-09-29 | 重庆邮电大学 | Single-point intersection vehicle scheduling system and method based on self-adaptive signal lamp |
CN114093178A (en) * | 2022-01-21 | 2022-02-25 | 杭州海康威视数字技术股份有限公司 | Traffic signal control method and device, electronic equipment and storage medium |
CN114093178B (en) * | 2022-01-21 | 2022-05-06 | 杭州海康威视数字技术股份有限公司 | Traffic signal control method and device, electronic equipment and storage medium |
CN116189459A (en) * | 2023-04-26 | 2023-05-30 | 西南民族大学 | Intersection traffic signal lamp timing method |
CN117523823A (en) * | 2023-10-11 | 2024-02-06 | 吉林师范大学 | Regional traffic signal control optimization method based on quantum genetic algorithm |
Also Published As
Publication number | Publication date |
---|---|
CN110517510B (en) | 2020-12-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110517510A (en) | Based on the Intersections timing designing method for improving Webster function and genetic algorithm | |
CN110164128A (en) | A kind of City-level intelligent transportation analogue system | |
CN108197739B (en) | Urban rail transit passenger flow prediction method | |
CN112216108B (en) | Traffic prediction method based on attribute-enhanced space-time graph convolution model | |
Celtek et al. | Real-time traffic signal control with swarm optimization methods | |
CN110021165A (en) | A kind of traffic flow forecasting method based on Autoencoder-LSTM Fusion Model | |
Jin et al. | A stochastic optimization framework for road traffic controls based on evolutionary algorithms and traffic simulation | |
CN113902011A (en) | Urban rail transit short-time passenger flow prediction method based on cyclic neural network | |
CN112365724A (en) | Continuous intersection signal cooperative control method based on deep reinforcement learning | |
CN103593535A (en) | Urban traffic complex self-adaptive network parallel simulation system and method based on multi-scale integration | |
CN103761138A (en) | Parameter correction method for traffic simulation software | |
CN111582750B (en) | Method and system for identifying reverse passengers and determining riding class of rail transit | |
CN113223305B (en) | Multi-intersection traffic light control method and system based on reinforcement learning and storage medium | |
CN104464310A (en) | Signal collaborative optimization control method and system of multiple intersections of urban region | |
CN106781489A (en) | A kind of road network trend prediction method based on recurrent neural network | |
CN112489426A (en) | Urban traffic flow space-time prediction scheme based on graph convolution neural network | |
Mao et al. | A comparison of deep reinforcement learning models for isolated traffic signal control | |
CN106228265A (en) | Based on Modified particle swarm optimization always drag phase transport project dispatching algorithm | |
CN113205698A (en) | Navigation reminding method based on IGWO-LSTM short-time traffic flow prediction | |
CN108985518A (en) | A kind of traffic congestion charging policy optimization method based on VISSIM emulation | |
Zhang et al. | Traffic light optimization based on modified Webster function | |
CN109522584B (en) | Tramcar departure timetable design method | |
Armas et al. | Evolutionary design optimization of traffic signals applied to Quito city | |
CN113724507A (en) | Traffic control and vehicle induction cooperation method and system based on deep reinforcement learning | |
Chen et al. | Traffic signal optimization control method based on adaptive weighted averaged double deep Q network |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |