CN114926997B - Heuristic on-line network signal optimization method based on performance weighting - Google Patents
Heuristic on-line network signal optimization method based on performance weighting Download PDFInfo
- Publication number
- CN114926997B CN114926997B CN202210475421.4A CN202210475421A CN114926997B CN 114926997 B CN114926997 B CN 114926997B CN 202210475421 A CN202210475421 A CN 202210475421A CN 114926997 B CN114926997 B CN 114926997B
- Authority
- CN
- China
- Prior art keywords
- intersection
- signal timing
- individual
- road network
- gene
- 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.)
- Active
Links
Classifications
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/07—Controlling traffic signals
- G08G1/081—Plural intersections under common control
- G08G1/083—Controlling the allocation of time between phases of a cycle
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/04—Constraint-based CAD
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/08—Probabilistic or stochastic CAD
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a heuristic on-line network signal optimization method based on performance weighting, which comprises the following steps: step 1: initializing road network signal configuration information aiming at a selected road network; step 2: acquiring traffic flow and signal timing data of a current optimization period of a road network, and generating an initial population at the same time; step 3: aiming at the individuals in the current population, calculating a road network weighting performance index to obtain individual fitness; step 4: generating a new population by genetic manipulation and individual adjustment; step 5: if the evolution is terminated, outputting the current optimal time, turning to the step 6, otherwise turning to the step 3; step 6: and (4) entering the next optimization period, and turning to the step (2). The invention can realize the self-adaptive adjustment of the signal timing of the road network intersection.
Description
Technical Field
The invention relates to the technical field of optimization of regional road network signal timing, in particular to a heuristic on-line network signal optimization algorithm based on performance weighting.
Background
The intersection is the throat of the urban road network, is a plurality of places of traffic jam and accidents, and can reduce traffic incidents and improve the running efficiency of the road network when optimizing the intersection signals. In order to prevent and reduce traffic jams at urban intersections and improve the running efficiency of urban traffic networks, intersection signal timing must be reasonably performed. At present, traffic signal control can be divided into three types of point control, line control and surface control according to the control range. In comparison, both point control and line control are difficult to meet the signal coordination requirement for regional road networks, so research on the surface control is more important. The face control is also called regional control, and is a control mode for uniformly coordinating signal timing of all intersections in a certain region of a city so as to reasonably use road resources and dredge traffic. .
The more mature surface control systems currently exist as SCATS (Sydney Coordinated Adaptive Traffic System) and SCOOT (Split Cycle Offset Optimizing Technique). SCATS is a real-time scheme selection control system, presets a plurality of signal timing schemes, and selects corresponding timing schemes according to actual traffic flow conditions, so that flexibility in coping with traffic flow changes is limited. SCOOT is a real-time scheme generation type control mode, traffic movement is measured and tracked in real time through a vehicle detector, and timing of a signal controller is optimized by utilizing an online traffic model and a corresponding control parameter optimization program.
Besides SCATS and SCOOT systems, the academic world often adopts a microscopic simulation method to optimize regional signal timing, namely, under a selected optimization algorithm framework, the microscopic simulation method is adopted to calculate and obtain the road network operation efficiency under different signal timing, so that the corresponding optimal timing is obtained through optimization search. The method can solve the traffic model adaptability problem in the signal optimization process, can directly and practically use the road network operation efficiency as an optimization target, has rationality in theory, but has large operation amount of microscopic simulation and low efficiency in the actual application process, is difficult to meet the real-time requirement of road network signal timing, and is free from considering the smoothness requirement of actual signal change in the optimization process, and lacks step change constraint.
Disclosure of Invention
The invention aims to solve the technical problem of overcoming the defects of the prior art and providing a heuristic on-line network signal optimization method based on performance weighting.
The invention adopts the following technical scheme for solving the technical problems:
a heuristic on-line network signal optimization method based on performance weighting comprises the following steps:
step 1, initializing road network signal configuration information: determining the total number m of intersections on the current road network, determining the weight of each intersection on the road network, determining the saturation flow rate s of each lane group of each intersection, and determining the short-time online rolling time window length v; initializing genetic algorithm parameters: population size L, selection probability P s Probability of crossover P c Probability of variation P v Maximum iteration number T;
step 2, obtaining road network traffic flow and signal timing data required by the current optimization period t, and generating an initial population; setting a loop parameter of the iteration times of the genetic algorithm as i, and enabling i=1;
step 3, calculating the fitness of all individuals of the current population;
step 4, genetic operation and individual adjustment are carried out, and a new population is generated;
step 5, if i is less than T, making i=i+1, and returning to the step 3; if i is more than or equal to T, taking the individual with the largest adaptability in the current population as a road network signal timing scheme of the current optimization period, and entering a step 6;
and 6, rolling a time window, entering the next optimization period, enabling t=t+1, and entering the step 2.
Further, in the step 2, traffic flow and signal timing data in the period t-1 are collected and used as road network traffic flow and signal timing data required by the current optimization period t, wherein the traffic flow data refers to total traffic flow Q of entrance roads of all intersections, and the signal timing data refers to green light duration of all phases of all intersections.
Further, the specific steps of generating the initial population P (0) in the step 2 are as follows:
step 21, acquiring signal timing data in a t-1 period;
step 22, encoding the chromosome by adopting a real number encoding mode: one chromosome corresponds to one individual in the population, namely corresponds to a road network signal timing scheme; one chromosome has theta gene sites, the circulation parameter of the gene sites is set as sigma, and one gene site corresponds to one signal timing parameter, namely the green light time length of a certain intersection phase;
step 23, setting the constraint condition of signal timing parameters as [ G ] σ ′-ε,G σ ′+ε]，G σ ' is the value of the signal timing parameter corresponding to the gene position sigma in the t-1 period, namely the green light duration of the phase corresponding to the gene position sigma in the t-1 period; epsilon is the maximum adjustment step length of the preset signal timing parameters;
step 24, generating a first individual P (0) of the initial population according to the signal timing scheme of the t-1 period and the real number coding mode 1 The individual has theta gene sites, each gene site corresponds to a signal timing parameter G of t-1 time period σ ' i.e. the green light duration of a certain intersection phase in the t-1 period;
step 25, consider signal timing parameter constraint [ G ] σ ′-ε,G σ ′+ε]Randomly generating signal timing parameter values for gene locus sigma using
Wherein:
a signal timing parameter value corresponding to the gene position sigma;
G σ ' represents the value of the signal timing parameter corresponding to the gene position sigma in the t-1 period;
epsilon is the maximum adjustment step length of the preset signal timing parameters;
int () is a rounding function, rounding real numbers to 0;
rand () is a random number generator that generates random numbers in the range of 0-1;
after signal timing parameter values meeting the constraint conditions of the signal timing parameters are randomly generated for all gene positions, a new individual is formed, the above operation is repeated until L-1 individuals are generated, and the generation of the initial population is completed.
Further, in the step 3, signal timing parameters corresponding to different gene positions in the individual are analyzed, and the saturation B of each intersection is calculated respectively and is used as the performance index of the current intersection; and calculating a road network weighting performance index W according to the weight of each intersection on the road network, and taking the reciprocal of W as the fitness function F of the current individual.
Further, the specific step of calculating the fitness of the individual l in the step 3 is:
step 31, setting the number of intersections as j, and enabling j=1;
step 32, obtaining the saturation flow rate s of each lane group of the intersection, and analyzing the current signal timing scheme of the intersection j based on the current individual gene value;
step 33, calculating the traffic capacity C of the intersection j under the current signal timing scheme j ：
Setting the total number of phases of the intersection j when the current signal is distributed as o, setting the phase circulation parameter of the intersection j as k, and calculating the traffic capacity c of the lane group corresponding to the phase k k The calculation formula is as follows:
wherein:
c k the traffic capacity of the lane group which can pass through the intersection j at the phase k is represented;
s k a saturation flow rate representing the set of lanes that can be traversed at phase k at intersection j;
g k representing the effective green time of phase k;
d j indicating that intersection j has a long signal period at the current signal timing.
Calculating the traffic capacity C of the intersection j j The calculation formula is as follows:
wherein:
C j indicating the traffic capacity of intersection j;
o is the total number of phases of intersection j when the current signal is distributed;
c k the lane group traffic capacity of intersection j at phase k is represented;
step 34, obtaining the total traffic flow Q of each entrance road of the intersection j j ；
Step 35, calculating saturation B of intersection j j The calculation formula is as follows:
wherein:
B j the saturation of the intersection j when the current road network signal is distributed is represented;
Q j representing the total traffic flow of each entrance road of the intersection j;
C j indicating the traffic capacity of intersection j;
step 36, if j is less than m, let j=j+1, return to step 32, calculate the saturation of the next intersection; if j is greater than or equal to m, entering a step 37;
step 37, calculating fitness F of individual l l ：
Calculating road network weighted performance index W of individual I l The calculation formula is as follows:
wherein:
W l the road network weighting performance index of the individual I under the current road network signal distribution is represented;
λ j the weight of the intersection j on the road network is represented;
B j the saturation of the intersection j when the current road network signal is distributed is represented;
m is the total number of intersections on the road network;
definition of fitness F of individual l l Is W l Inverse of (2), namely:
wherein:
F l the adaptation degree of the individual I in the current road network signal distribution is represented;
W l and the road network weighting performance index of the individual l under the current road network signal distribution is represented.
Further, the genetic operation in the step 4 includes a selection operation, a crossover operation and a mutation operation, and for the population after the genetic operation, it is determined whether the signal timing parameters corresponding to all the genetic loci of all the individuals satisfy the signal timing parameter constraint condition [ G ] σ ′-ε，G σ ′+ε]Wherein G is σ ' is the value of the signal timing parameter corresponding to the gene locus sigma in the t-1 period, epsilon is the maximum adjustment step length of the preset signal timing parameter, if yes, the step 5 is entered, otherwise, the individual which does not meet the constraint condition of the signal timing parameter is adjusted, and a new population is obtained.
Further, the specific steps of the genetic operation in the step 4 are as follows:
step 41, selecting operation:
firstly, the fitness of L individuals is ordered according to the size, and the front with the largest fitness is reservedIndividual.
Second, for the rest ofIndividual performs selection operation:
for each of the individuals, a random number x is generated using the following formula:
x＝rand()
wherein: rand () is a random number generator that generates random numbers in the range of 0-1; if x < P s ，P s For selecting probability, reserving the individual, otherwise, not reserving;
finally, traversing all individuals, and if the number L 'of the finally reserved individuals is smaller than L, randomly generating L-L' new individuals to finish the selection operation;
step 42, cross operation:
firstly, directly reserving two individuals with highest fitness, and carrying out pairwise random pairing on the rest L-2 individuals;
then, for each individual pair, a random number y is first generated using the following equation:
y＝rand()
wherein: if y < P c ，P c If the crossing probability is the crossing probability, the individual pair performs the crossing operation, the gene value of the individual pair after the crossing gene position is selected is exchanged, otherwise, the crossing operation is not performed;
finally, traversing all chromosome pairs to finish the crossing operation;
step 43, mutation operation:
first, for each individual, a random number ω is generated:
ω＝rand()
wherein: if ω < P v ，P v If the probability is variation probability, the individual is mutated, otherwise, no mutation occurs;
and finally, traversing all individuals to finish the mutation operation.
Further, in the crossover operation, the crossover gene locus z is generated by the following expression:
z＝int(rand()*θ+1)
wherein: θ is the total number of gene loci of an individual.
Further, in the mutation operation, a mutation gene position is generated by the following formulaThe method comprises the following steps:
the mutated gene locus values are:
τ＝int(10*rand())。
further, the individual adjustment method in the step 4 is as follows: adjusting the gene position values which do not meet the constraint of the signal timing parameters by using the following formula:
ρ＝G σ ′-ε+2ε*rand()
wherein:
ρ represents the adjusted gene locus value;
G σ ' is the value of the signal timing parameter corresponding to the gene locus sigma in the last period;
epsilon is the maximum adjustment step length of the preset signal timing parameters;
rand () is a random number generator that can generate random numbers in the range of 0-1.
Compared with the prior art, the technical scheme provided by the invention has the following technical effects:
the invention adopts intersection entrance lane detector data, carries out heuristic optimization on the timing of a line network signal based on a genetic algorithm framework, and generates an initial population based on the initialized line network signal configuration information, wherein an individual in the population corresponds to a group of signal timing parameters of the line network, and a gene position on the individual corresponds to one signal timing parameter, namely the green light duration of one phase. And aiming at the individuals, calculating the saturation of the intersections, and calculating to obtain road network weighted performance indexes according to the position importance weight of each intersection, wherein the reciprocal of the road network weighted performance indexes is used as the fitness of the individuals. Genetic manipulation is then performed to generate a new population, wherein the frame skip constraint on the signal timing parameters is "parameter value of last period ± maximum adjustment step", and the individual not satisfying the constraint is adjusted. Judging whether to terminate the evolution according to the iteration times, if so, outputting an optimal value of the road network signal timing at the current time period, and entering the next time period to realize self-adaptive real-time efficient adjustment of the road network intersection signal timing.
Drawings
FIG. 1 is an overall flowchart of an algorithm;
FIG. 2 is a flowchart of individual fitness calculation.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in detail with reference to the accompanying drawings and specific examples.
The heuristic on-line network signal optimization algorithm based on performance weighting overcomes the defects of the traditional method for adjusting the timing of the real-time scheme signal based on the running condition of the line network, and is specifically as follows:
(1) The road network weighting performance index is provided, the optimization efficiency is improved, and the collaborative optimization of the intersections is realized. In the traditional surface control method, road network performance indexes are adopted as optimization targets, the calculation process is complex, real-time running conditions of the road network are difficult to comprehensively consider effectively, influence of important intersections on the road network is ignored, and coordination among the intersections is difficult to achieve. The invention provides the road network weighting performance index, combines the unit performance into the road network performance, realizes coordination among the unit intersections, has simple and convenient performance index calculation, and can improve the optimization efficiency.
(2) The genetic algorithm is improved, and the global searching capability and efficiency of the optimization method are improved. The prior signal timing optimization method mostly adopts a microscopic simulation method to calculate the road network performance index, which is time-consuming and difficult to realize real-time signal adjustment, and does not meet the actual application requirements. The road network weighting performance index adopted by the invention is simple and convenient to calculate, the searching capability of an optimization algorithm is enhanced by increasing the population quantity and the iteration times in the optimization process, in addition, the elite strategy adopted in the optimization process ensures the gradual optimization of the optimization result, and the global searching capability of the optimization method is improved by enhancing the randomness factor in the genetic operation.
(3) And signal timing parameter constraint is provided, so that the practicability of the method is improved. The traditional signal timing optimization algorithm based on the road network is either a scheme selection type, an optimal control mode is selected from a plurality of timing schemes preset in advance according to real-time acquisition and statistical analysis of road network traffic information, or a scheme generation type is adopted, simulation calculation is carried out on a model through the real-time traffic information, and further the optimal road network signal timing scheme is determined, consideration in the actual signal switching process is insufficient, and the practicability of the method is limited. The invention takes the parameter value of the last time period plus or minus the maximum adjustment step length as the signal timing parameter constraint in the optimization model, and improves the continuity of the road network signal timing, thereby improving the practicability of the signal optimization algorithm in reality.
Therefore, the method adopts a rolling optimization time period strategy on the basis of initializing road network signal configuration information, dynamically adopts detectors to obtain intersection traffic flow data, calculates to obtain saturation indexes of all units, calculates road network weighting performance indexes through weights of all intersections on a road network, and then optimizes the road network signal timing in real time by adopting a genetic algorithm framework by taking a parameter value + -maximum adjustment step length of a last time period as a signal timing parameter constraint, thereby increasing the adaptability of the algorithm and improving timing optimization efficiency.
The method for adjusting the timing of the road network signal of the invention is shown in figure 1, and mainly comprises the following steps:
and step 1, initializing road network signal configuration information. For a selected road network, determining the total number m of intersections on the current road network, determining the weight lambda of each intersection on the road network according to the position characteristics of the intersection, and simultaneously determining the saturation flow rate s of each lane group of each intersection. Determining short time according to signal timing requirementThe line scroll time window length v is typically set to 5-10 minutes. Initializing genetic algorithm parameters: population size L (L is even), selection probability P s The crossover probability is P c Probability of variation is P v The maximum iteration number is T. And setting an optimization period corresponding to the current moment as t.
And step 2, obtaining road network traffic flow and signal timing data required by the current optimization period t, and generating an initial population. The optimization of the road network signal timing is completed at the starting point of the current optimization period, so that traffic flow and signal timing data in the t-1 period are required to be collected as the optimization input data of the t period. The traffic flow data refers to the total traffic flow Q of the entrance channels of all intersections, and the signal timing data refers to the green light duration of all phases of all intersections. Generating an initial population P (0), setting a loop parameter of the iteration number of the genetic algorithm as i, and enabling i=1.
And step 3, calculating the fitness of all individuals in the current population. For a single individual, analyzing signal timing parameters corresponding to different gene positions in the individual, respectively calculating the saturation B of each intersection as a performance index of the current intersection, calculating a road network weighting performance index W according to the weight of each intersection on the road network, and taking the reciprocal of the road network weighting performance index W as an adaptability function F of the current individual.
And 4, performing genetic operation and individual adjustment to generate a new population. Genetic operations include selection operations, crossover operations, and mutation operations, after which a new population is obtained. Judging whether signal timing parameters corresponding to all gene positions of all individuals are in [ G ] according to the new population σ ′-ε，G σ ′+ε]Within the constraint, where G σ ' is the signal timing value of the current gene position in the t-1 period, epsilon is the preset maximum adjustment step length, and is generally selected to be 5-10 seconds. And adjusting individuals which do not meet the constraint to obtain an adjusted new population.
Step 5, judging the termination of the evolution iteration according to the iteration times, if i is less than T, enabling i=i+1, returning to the step 3, and calculating the fitness of the new population individuals; if i is not less than T, taking the individual with the largest adaptability in the current population as a road network signal timing scheme of the current optimization period, and entering the step 6.
And 6, rolling a time window, entering the next optimization period, enabling t=t+1, and entering the step 2.
In one embodiment, the specific steps of generating the initial population P (0) in step 2 are:
step 21, acquiring signal timing data in a t-1 period;
and step 22, encoding the chromosome by adopting a real number encoding mode. One chromosome corresponds to one individual in the population, namely corresponds to a road network signal timing scheme; one chromosome has theta gene sites, the circulation parameter of the gene sites is set as sigma, and one gene site corresponds to one signal timing parameter, namely the green light time length of a certain intersection phase;
step 23, setting the constraint condition of signal timing parameters as [ G ] σ ′-ε，G σ ′+ε]，G σ ' is the value of the signal timing parameter corresponding to the gene locus sigma in the t-1 period, namely the green light duration of the corresponding phase of the gene locus sigma in the t-1 period, epsilon is the maximum adjustment step length of the preset signal timing parameter, and is generally set to be 5-10 seconds;
step 24, generating individual P (0) 1 . Generating a first individual P (0) of the initial population according to a real number coding mode according to a signal timing scheme of a t-1 period 1 The individual has theta gene sites, each gene site corresponds to a signal timing parameter G of t-1 time period σ ' i.e. the green light duration of a certain intersection phase in the t-1 period;
step 25, at P (0) 1 On the basis of which L-1 individuals are generated. Consider constraint [ G ] σ ′-ε，G σ ′+ε]Generating a signal timing parameter value for the locus σ using the following formula
Wherein:
a signal timing parameter value corresponding to the gene position sigma;
G σ ' represents the value of the signal timing parameter corresponding to the gene position sigma in the t-1 period;
epsilon is the maximum adjustment step length of the preset signal timing parameters;
int () is a rounding function, rounding real numbers to 0;
rand () is a random number generator that can generate random numbers in the range of 0-1.
Through the above method, the signal timing parameter value corresponding to the gene position sigma can be randomly obtainedAfter values meeting constraint conditions are randomly generated for all gene positions, a new individual is formed, the above operation is repeated until L-1 individuals are generated, and the generation of the initial population is completed.
In one embodiment, as shown in fig. 2, the specific calculation steps of the individual fitness in step 3 are as follows:
step 31, determining the total number m of intersections on a road network, setting the cycle parameters of the number of the intersections as j, and enabling j=1;
step 32, obtaining the saturation flow rate s of each lane group of the intersection, and analyzing the current signal timing scheme of the intersection j based on the current individual gene value;
and step 33, calculating the traffic capacity Cj of the intersection j under the current signal timing scheme. Setting the total number of phases of the intersection j when the current signal is distributed as o, and setting the phase circulation parameter of the intersection j as k. Firstly, calculating the traffic capacity c of a lane group corresponding to the phase k k The calculation formula is as follows:
wherein:
c k the traffic capacity (vehicle/h) of the lane group that can pass at phase k at intersection j;
s k a saturation flow rate (vehicle/h) representing the set of lanes that can be travelled by intersection j at phase k;
g k representing the effective green time(s) of phase k;
d j indicating that intersection j has a long signal period at the current signal timing.
Then, the traffic capacity C of the intersection j is calculated j The calculation formula is as follows:
wherein:
C j indicating the traffic capacity of intersection j;
o is the total number of phases of intersection j when the current signal is distributed;
c k the lane group traffic capacity (vehicle/h) at phase k at intersection j is indicated.
Step 34, obtaining the total traffic flow Q of each entrance road of the intersection j according to the step 2 j ；
Step 35, calculating saturation B of intersection j j The calculation formula is as follows:
wherein:
B j the saturation of the intersection j when the current road network signal is distributed is represented;
Q j representing the total traffic flow of each entrance road of the intersection j;
C j indicating the traffic capacity of intersection j.
Step 36, if j is less than m, let j=j+1, return to step 32, calculate the saturation of the next intersection; if j is greater than or equal to m, entering a step 37;
step 37, calculating fitness F of individual l l . First, a road network weighting performance index W of an individual l is calculated according to the intersection weight lambda determined in the step 1 l The calculation formula is as follows:
wherein:
W l the road network weighting performance index of the individual I under the current road network signal distribution is represented;
λ j the weight of the intersection j on the road network is represented;
B j the saturation of the intersection j when the current road network signal is distributed is represented;
m is the total number of intersections on the road network.
Then, the fitness F of the individual l is defined l Is the performance index W l Inverse of (2), namely:
wherein:
F l the adaptation degree of the individual I in the current road network signal distribution is represented;
W l and the road network weighting performance index of the individual l under the current road network signal distribution is represented.
In one embodiment, the specific steps of genetic manipulation and individual adjustment in step 4 are:
step 41, selecting operation. The goal of the selection is to randomly select some individuals of the current population while retaining a fixed proportion of the optimal individuals and generating new individuals to form a new population.
Firstly, the fitness of L individuals is ordered according to the size, and the front with the largest fitness is reservedIndividual.
Second, for the remainderEach of the individualsThe random number x is first generated using the following equation:
x＝rand()
wherein: rand () is a random number generator that generates random numbers in the range of 0-1. If x < P s ，P s To select a probability, the individual is retained, otherwise not retained.
And finally traversing all the individuals, and if the number of the finally reserved individuals is smaller than L, randomly generating new individuals to complement the number of individuals in the preset population, so as to complete the selection operation.
Step 42, cross operation. The two parent individuals are crossed to generate two new individuals through gene exchange so as to enhance the searching capability of the genetic algorithm.
First, the two individuals with the highest fitness are directly kept, and the remaining L-2 individuals are paired pairwise randomly.
Then, for each chromosome pair, a random number y is first generated using the following equation:
y＝rand()
wherein: rand () is a random number generator that can generate random numbers in the range of 0-1. If y < P c ，P c If the crossing probability is the crossing probability, the chromosome pair performs crossing operation, and the gene value of the chromosome pair after the crossing gene position is selected is exchanged, otherwise, the chromosome pair does not cross. In the crossover operation, the crossover gene position z is generated by the following formula:
z＝int(rand()*θ+1)
wherein: int () is a rounding function, rounding real numbers to 0; the rand () is a random number generator that can generate random numbers in the range of 0-1; θ is the total number of loci per chromosome/individual.
And finally, traversing all chromosome pairs to finish the crossing operation.
Step 43, mutation operation. And (3) carrying out mutation operation on certain gene positions of certain individuals in the population, enhancing the random searching capability of a genetic algorithm, maintaining the diversity of the population and preventing local convergence.
First, for each individual, a random number ω is generated:
ω＝rand()
wherein: r is (r)and () is a random number generator that generates a random number in the range of 0-1. If ω < P v ，P v If the probability is variation, the individual is mutated, otherwise, no mutation occurs. Wherein the variant gene positionThe method comprises the following steps:
wherein: the rand () is a random number generator, which can generate random numbers in the range of 0-1, and the int () is a rounding function, rounding real numbers to 0; the mutated gene locus values are:
τ＝int(10*rand())
wherein: int () is a rounding function, rounding real numbers to 0; rand () is a random number generator that generates random numbers in the range of 0-1.
And finally, traversing all individuals to finish the mutation operation.
Step 44, individual adjustment. Judging whether the signal timing parameters corresponding to all genetic loci of all individuals in the new population are in [ G ] σ ′-ε，G σ ′+ε]And (3) in the constraint, namely within the parameter value +/-maximum adjustment step length of the t-1 period, if yes, entering a step 5, and if not, adjusting the gene position value which does not meet the signal timing parameter constraint by using the following formula:
ρ＝G σ ′-ε+2ε*rand()
wherein:
ρ represents the adjusted gene locus value;
G σ ' is the value of the signal timing parameter corresponding to the gene locus sigma in the last period;
epsilon is the maximum adjustment step length of the preset signal timing parameters;
rand () is a random number generator that can generate random numbers in the range of 0-1.
The foregoing is merely illustrative of the present invention, and the present invention is not limited thereto, and any changes or substitutions easily contemplated by those skilled in the art within the scope of the present invention should be included in the scope of the present invention.
Claims (4)
1. A heuristic on-line network signal optimization method based on performance weighting is characterized by comprising the following steps:
step 1, initializing road network signal configuration information: determining the total number m of intersections on the current road network, determining the weight of each intersection on the road network, determining the saturation flow rate s of each lane group of each intersection, and determining the short-time online rolling time window length v; initializing genetic algorithm parameters: population size L, selection probability P s Probability of crossover P c Probability of variation P v Maximum iteration number T;
step 2, obtaining road network traffic flow and signal timing data required by the current optimization period t, and generating an initial population; setting a loop parameter of the iteration times of the genetic algorithm as i, and enabling i=1;
step 3, calculating the fitness of all individuals of the current population;
step 4, genetic operation and individual adjustment are carried out, and a new population is generated;
step 5, if i is less than T, making i=i+1, and returning to the step 3; if i is more than or equal to T, taking the individual with the largest adaptability in the current population as a road network signal timing scheme of the current optimization period, and entering a step 6;
step 6, rolling a time window, entering the next optimization period, enabling t=t+1, and entering the step 2;
in the step 2, collecting traffic flow and signal timing data in a period t-1 as road network traffic flow and signal timing data required by the current optimization period t, wherein the traffic flow data refers to total traffic flow Q of entrance roads of all intersections, and the signal timing data refers to green light duration of all phases of all intersections;
the specific steps for generating the initial population P (0) in the step 2 are as follows:
step 21, acquiring signal timing data in a t-1 period;
step 22, encoding the chromosome by adopting a real number encoding mode: one chromosome corresponds to one individual in the population, namely corresponds to a road network signal timing scheme; one chromosome has theta gene sites, the circulation parameter of the gene sites is set as sigma, and one gene site corresponds to one signal timing parameter, namely the green light time length of a certain intersection phase;
step 23, setting the constraint condition of signal timing parameters as [ G ] σ ′-ε,G σ ′+ε]，G σ ' is the value of the signal timing parameter corresponding to the gene position sigma in the t-1 period, namely the green light duration of the phase corresponding to the gene position sigma in the t-1 period; epsilon is the maximum adjustment step length of the preset signal timing parameters;
step 24, generating a first individual P (0) of the initial population according to the signal timing scheme of the t-1 period and the real number coding mode 1 The individual has theta gene sites, each gene site corresponds to a signal timing parameter G of t-1 time period σ ' i.e. the green light duration of a certain intersection phase in the t-1 period;
step 25, consider signal timing parameter constraint [ G ] σ ′-ε,G σ ′+ε]Randomly generating signal timing parameter values for gene locus sigma using
Wherein:
a signal timing parameter value corresponding to the gene position sigma;
G σ ' represents the value of the signal timing parameter corresponding to the gene position sigma in the t-1 period;
epsilon is the maximum adjustment step length of the preset signal timing parameters;
int () is a rounding function, rounding real numbers to 0;
rand () is a random number generator that generates random numbers in the range of 0-1;
after randomly generating signal timing parameter values meeting the constraint conditions of the signal timing parameters for all gene positions, forming a new individual, repeating the above operations until L-1 individuals are generated, and completing the generation of an initial population;
in the step 3, analyzing signal timing parameters corresponding to different gene positions in an individual, and respectively calculating the saturation B of each intersection as the performance index of the current intersection; calculating a road network weighting performance index W according to the weight of each intersection on the road network, and taking the reciprocal of W as a fitness function F of the current individual;
the specific step of calculating the fitness of the individual l in the step 3 is as follows:
step 31, setting the number of intersections as j, and enabling j=1;
step 32, obtaining the saturation flow rate s of each lane group of the intersection, and analyzing the current signal timing scheme of the intersection j based on the current individual gene value;
step 33, calculating the traffic capacity C of the intersection j under the current signal timing scheme j ：
Setting the total number of phases of the intersection j when the current signal is distributed as o, setting the phase circulation parameter of the intersection j as k, and calculating the traffic capacity c of the lane group corresponding to the phase k k The calculation formula is as follows:
wherein:
c k the traffic capacity of the lane group which can pass through the intersection j at the phase k is represented;
s k a saturation flow rate representing the set of lanes that can be traversed at phase k at intersection j;
g k representing the effective green time of phase k;
d j the signal period of the intersection j when the current signal is distributed is long;
calculating the traffic capacity C of the intersection j j The calculation formula is as follows:
wherein:
C j indicating the traffic capacity of intersection j;
o is the total number of phases of intersection j when the current signal is distributed;
c k the lane group traffic capacity of intersection j at phase k is represented;
step 34, obtaining the total traffic flow Q of each entrance road of the intersection j j ；
Step 35, calculating saturation B of intersection j j The calculation formula is as follows:
wherein:
B j the saturation of the intersection j when the current road network signal is distributed is represented;
Q j representing the total traffic flow of each entrance road of the intersection j;
C j indicating the traffic capacity of intersection j;
step 36, if j < m, let j=j+1, return to step 32, calculate the saturation of the next intersection; if j is greater than or equal to m, entering a step 37;
step 37, calculating fitness F of individual l l ：
Calculating road network weighted performance index W of individual I l The calculation formula is as follows:
wherein:
W l the road network weighting performance index of the individual I under the current road network signal distribution is represented;
λ j the weight of the intersection j on the road network is represented;
B j the saturation of the intersection j when the current road network signal is distributed is represented;
m is the total number of intersections on the road network;
definition of fitness F of individual l l Is W l Inverse of (2), namely:
wherein:
F l the adaptation degree of the individual I in the current road network signal distribution is represented;
W l the road network weighting performance index of the individual I under the current road network signal distribution is represented;
the genetic operation in the step 4 comprises selection operation, crossover operation and mutation operation, and for the population after genetic operation, whether the signal timing parameters corresponding to all the gene positions of all individuals meet the signal timing parameter constraint condition [ G ] is judged σ ′-ε,G σ ′+ε]Wherein G is σ ' is the value of the signal timing parameter corresponding to the gene locus sigma in the t-1 period, epsilon is the maximum adjustment step length of the preset signal timing parameter, if yes, the step 5 is entered, otherwise, the individual which does not meet the constraint condition of the signal timing parameter is adjusted, and a new population is obtained;
the specific steps of the genetic operation in the step 4 are as follows:
step 41, selecting operation:
firstly, the fitness of L individuals is ordered according to the size, and the front with the largest fitness is reservedA subject;
second, for the rest ofIndividual performs selection operation:
for each of the individuals, a random number x is generated using the following formula:
x＝rand()
wherein: rand () is a random number generator that generates random numbers in the range of 0-1; if x < P s ，P s For selecting probability, reserving the individual, otherwise, not reserving;
finally, traversing all individuals, if the number L of the individuals is finally reserved ′ Less than L, randomly generate L-L ′ New individuals finish the selection operation;
step 42, cross operation:
firstly, directly reserving two individuals with highest fitness, and carrying out pairwise random pairing on the rest L-2 individuals;
then, for each individual pair, a random number y is first generated using the following equation:
y＝rand()
wherein: if y < P c ，P c If the crossing probability is the crossing probability, the individual pair performs the crossing operation, the gene value of the individual pair after the crossing gene position is selected is exchanged, otherwise, the crossing operation is not performed;
finally, traversing all chromosome pairs to finish the crossing operation;
step 43, mutation operation:
first, for each individual, a random number ω is generated:
ω＝rand()
wherein: if ω < P v ，P v If the probability is variation probability, the individual is mutated, otherwise, no mutation occurs;
and finally, traversing all individuals to finish the mutation operation.
2. The heuristic on-line network signal optimization method based on performance weighting according to claim 1, wherein the crossover gene locus z is generated during crossover operation using the following formula:
z＝int(rand()*θ+1)
wherein: θ is the total number of gene loci of an individual.
3. A heuristic on-line network signal optimization method based on performance weighting as recited in claim 1,wherein, in the mutation operation, the mutation gene position is generated by the following formulaThe method comprises the following steps:
the mutated gene locus values are:
τ＝int(10*rand())。
4. the heuristic on-line network signal optimization method based on performance weighting according to claim 1, wherein the individual adjustment method in step 4 is as follows: adjusting the gene position values which do not meet the constraint of the signal timing parameters by using the following formula:
ρ＝G σ ′-ε+2ε*rand()
wherein:
ρ represents the adjusted gene locus value;
G σ ' is the value of the signal timing parameter corresponding to the gene locus omega in the last period;
epsilon is the maximum adjustment step length of the preset signal timing parameters;
rand () is a random number generator that can generate random numbers in the range of 0-1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210475421.4A CN114926997B (en) | 2022-04-29 | 2022-04-29 | Heuristic on-line network signal optimization method based on performance weighting |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210475421.4A CN114926997B (en) | 2022-04-29 | 2022-04-29 | Heuristic on-line network signal optimization method based on performance weighting |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114926997A CN114926997A (en) | 2022-08-19 |
| CN114926997B true CN114926997B (en) | 2023-09-19 |
Family
ID=82806339
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210475421.4A Active CN114926997B (en) | 2022-04-29 | 2022-04-29 | Heuristic on-line network signal optimization method based on performance weighting |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114926997B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1889114A (en) * | 2006-07-17 | 2007-01-03 | 中国科学院地理科学与资源研究所 | Traffic signal off-line time distribution optimizing method basedon particle group operation method |
| JP2008242927A (en) * | 2007-03-28 | 2008-10-09 | Shinka System Sogo Kenkyusho:Kk | Genetic algorithm execution device and genetic algorithm execution method |
| CN106228819A (en) * | 2016-08-18 | 2016-12-14 | 广东工业大学 | The traffic signal optimization control method of a kind of multi-intersection and device |
| CN107610487A (en) * | 2017-08-23 | 2018-01-19 | 南京邮电大学 | Area Traffic Control System and method based on the dynamic random wagon flow phase difference coordination system |
| CN107705589A (en) * | 2017-11-06 | 2018-02-16 | 西南交通大学 | Bicyclic signal timing optimization method based on self-adapted genetic algorithm |
| CN113378486A (en) * | 2021-08-13 | 2021-09-10 | 深圳市城市交通规划设计研究中心股份有限公司 | Regional traffic signal optimization method and device, computing equipment and storage medium |
-
2022
- 2022-04-29 CN CN202210475421.4A patent/CN114926997B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1889114A (en) * | 2006-07-17 | 2007-01-03 | 中国科学院地理科学与资源研究所 | Traffic signal off-line time distribution optimizing method basedon particle group operation method |
| JP2008242927A (en) * | 2007-03-28 | 2008-10-09 | Shinka System Sogo Kenkyusho:Kk | Genetic algorithm execution device and genetic algorithm execution method |
| CN106228819A (en) * | 2016-08-18 | 2016-12-14 | 广东工业大学 | The traffic signal optimization control method of a kind of multi-intersection and device |
| CN107610487A (en) * | 2017-08-23 | 2018-01-19 | 南京邮电大学 | Area Traffic Control System and method based on the dynamic random wagon flow phase difference coordination system |
| CN107705589A (en) * | 2017-11-06 | 2018-02-16 | 西南交通大学 | Bicyclic signal timing optimization method based on self-adapted genetic algorithm |
| CN113378486A (en) * | 2021-08-13 | 2021-09-10 | 深圳市城市交通规划设计研究中心股份有限公司 | Regional traffic signal optimization method and device, computing equipment and storage medium |
Non-Patent Citations (3)
| Title |
|---|
| 凌墨 ; 吴桢 ; 郭建华 ; .考虑交通流不确定性的单交叉口信号配时方法.南通大学学报(自然科学版).2020,(第01期),127-132. * |
| 基于改进遗传算法的交通信号配时优化模型;李;刘瑶;周健;刘薇;洪良;;工业仪表与自动化装置(04);127-132 * |
| 胡海涛 ; 罗杰 ; .基于相位差的子区交通信号协调优化控制.计算机技术与发展.2018,(第06期),157-161. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114926997A (en) | 2022-08-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108197739B (en) | Urban rail transit passenger flow prediction method | |
| CN110032782B (en) | City-level intelligent traffic signal control system and method | |
| CN112365724B (en) | Continuous intersection signal cooperative control method based on deep reinforcement learning | |
| CN106971566B (en) | Self-adaptive control method based on genetic algorithm | |
| CN109215355A (en) | A kind of single-point intersection signal timing optimization method based on deeply study | |
| CN110136455A (en) | A kind of traffic lights timing method | |
| CN112216124A (en) | Traffic signal control method based on deep reinforcement learning | |
| CN106779198A (en) | A kind of congestion in road situation analysis method | |
| CN109269516B (en) | Dynamic path induction method based on multi-target Sarsa learning | |
| CN107331166B (en) | A kind of dynamic restricted driving method based on path analysis | |
| CN111951549A (en) | Self-adaptive traffic signal lamp control method and system in networked vehicle environment | |
| CN110766956B (en) | Fuzzy logic based traffic signal control method | |
| CN113516277B (en) | Internet intelligent traffic path planning method based on road network dynamic pricing | |
| CN110570672A (en) | regional traffic signal lamp control method based on graph neural network | |
| CN106781465A (en) | A kind of road traffic Forecasting Methodology | |
| CN113112823B (en) | Urban road network traffic signal control method based on MPC | |
| CN106781464A (en) | A kind of congestion in road situation method of testing | |
| CN110517510A (en) | Based on the Intersections timing designing method for improving Webster function and genetic algorithm | |
| CN112884229B (en) | Large-scale public place people flow guiding path planning method based on differential evolution algorithm | |
| CN113724507B (en) | Traffic control and vehicle guidance cooperative method and system based on deep reinforcement learning | |
| CN113392577B (en) | Regional boundary main intersection signal control method based on deep reinforcement learning | |
| CN112562363B (en) | Intersection traffic signal optimization method based on V2I | |
| CN114926997B (en) | Heuristic on-line network signal optimization method based on performance weighting | |
| CN115472023B (en) | Intelligent traffic light control method and device based on deep reinforcement learning | |
| CN116976045A (en) | Multi-target constraint simulation method for control subarea under non-congestion state |
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 |