CN1053696A - Self-learning intelligent co-ordinative controlling of urban traffic - Google Patents
Self-learning intelligent co-ordinative controlling of urban traffic Download PDFInfo
- Publication number
- CN1053696A CN1053696A CN 91100825 CN91100825A CN1053696A CN 1053696 A CN1053696 A CN 1053696A CN 91100825 CN91100825 CN 91100825 CN 91100825 A CN91100825 A CN 91100825A CN 1053696 A CN1053696 A CN 1053696A
- Authority
- CN
- China
- Prior art keywords
- traffic
- timing parameter
- vehicle
- control
- subsystem
- 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.)
- Withdrawn
Links
Images
Landscapes
- Traffic Control Systems (AREA)
Abstract
The present invention relates to a kind of urban traffic control method, it is applicable to the grade crossing that crosses in length and breadth in urban road network, red, green, yellow chrominance signal lamp is set manages the control vehicle '.It is by the automatic detection subsystem of vehicle, information to controlled traffic process actual measurement, carrying out traffic controls in real time, and extract optimal control parameter the actual measurement controlled variable with from the timing parameter knowledge base and compare, carry out the self study process, determine to be fit to the timing parameter value (T, g, ) of new transport environmental condition, this method can reach timing parameter optimum, real-time good (can change timing parameter in per 30～120 seconds), and advantages such as bicycle detection are arranged.
Description
The present invention relates to a kind of urban traffic control method, it is applicable to the grade crossing that urban road network crosses in length and breadth, red, green, yellow chrominance signal lamp is set manages the control vehicle '.
At present, in the city various vehicles in travels down just like fluid in pipeline, flow, so people are referred to as traffic flow.In the crisscross road net in city, to build the viaduct except that minority intersection with good conditionsi, most of roads intersection in length and breadth all can only be grade crossings, so that travelling of red, green, yellow chrominance signal fluorescent tube reason and control vehicle to be set.The urban traffic road is the lifeblood in a city, no matter traffic is smooth economically or all can bring a series of serious consequences in the social influence.For example: a car is stopping once by the intersection, can increase the wearing and tearing that bring vehicle brake because of stop, oil consumption, exhaust and noise pollution, and the time of occupant's wait, intersection for an intermediate flow, the vehicle number that per hour passes through is about 1500/hour, if reduce parking rate 1%, only can save 10,000 yuan with regard to oil consumption intersection in an every year, about 500-1000 intersection, a big city, its economic worth is appreciable, therefore, how to adopt more advanced, more reliable, more efficient methods is coordinated the control urban traffic flow, is vital problem in the urban modernization.
As everyone knows, realize that in the world the method that regional coordination is controlled in real time has two kinds, a kind of is with Australian SCATS(Sydney Co-Ordinated Adaptive Traffic System) be the real-time system of selection of timing parameter of representative, another kind is the SCOOT(Split Cycle Offset Optimiza-tion Technique with Britain) be the real-time traffic situation analogy method of representative.
For the purpose of narrating conveniently, earlier the important parameter in traffic control method field is described as follows:
1. traffic lights generally are circulation change " red-green-Huang ", and red greenish-yellow signal shows all required times, and we are called period T, the tg-green time, ty-yellow time (general 1-2 second) tr-red time.
Obvious T=tg+ty+tr
The general signal lamp cycle shortest time is 30 seconds, and long period is that (too short vehicle had still just started, and bicycle, pedestrian pass the crossing needs the minimum time in 120 seconds; Oversize, people are misinterpreted as opertaing device easily and have broken and dislike and even disobey signal lamp commander).
2. split g, the ratio in green time and cycle, i.e. g=tg/T, it has described the allocation proportion of cycle on all directions.
Concerning single crossing, only need T and g just to describe the action of signal lamp fully.
For two above road nets in crossing, also has a controlled variable phase differential ψ.
3. phase differential ψ: two adjacent intersections, if the cycle is identical, the difference in the initial moment of red time on the same direction (or green time) is called phase differential ψ, obviously, if phase differential is provided with well, the fleet that runs into green light at crossing I1 still runs into green light to crossing I2, so just can avoid stopping and wait for.Concerning a crossing, it and adjacent intersection from all directions all have the phase differential problem.
Period T, split g and phase differential ψ are three controlled variable (also being timing parameter) in the urban traffic control, city has a lot of intersections (being provided with n) road net very complicated, and many highway sections link to each other (supposing to have m bar highway section) are arranged between crossing and the crossing.Obviously at a distance of near more crossing, the relation between the traffic flow is close more, then irrelevant more (correlativity is little) far away more.For the ease of control and management, often a urban road traffic network is divided into several subareas (supposing to be divided into e subarea) and controls.Cycle in subarea is consistent, 1 periodic quantity is so just arranged, n split value, m phase difference value.So T, g, ψ are vectors, T is 1 dimension, and g is the n dimension, and ψ is the m dimension.In one-period (30 seconds～120 seconds) scope, the controlled variable that will determine concerning the traffic control in a city has n * m * 1.Say that from the number of controlled variable this is a very complicated problems.And the mutual combination of these parameters then is the bigger permutation and combination problem of number.
Traffic control the complex nature of the problem also is its randomness.When vehicle sends, and walks what route, by which intersection, with what speed travels on road, what fortuitous event can occur on the road, and driver's psychological condition how, the vehicle of various kinds (motor vehicle, bicycle; Lorry, common electrical automobile, minibus, motorcycle ...) how to constitute or the like series of factors and do not determine, but randomness, and this random law be difficult to describe, even find the statistical law that is suitable for for the moment soon also can inapplicable (promptly aging).
Traffic control the complex nature of the problem also is its dynamic.Vehicle is moving on road, and the different moment are in different positions.From urban road network, the traffic flow distribution is inequality on the different road nets constantly; Road net and facility thereof neither be unalterable, repair the roads, change its course, increase the performance that platform etc. all can change road net; The people's activities rule neither be static, among one day, among the week, in January, the first quarter, all can change among 1 year; The formation of the vehicles is also slowly changing.All these is relevant with traffic control.Traffic control must adapt to the dynamic of traffic activity.
The urban traffic control of China is more complicated, more is difficult to control, and its reason is: one) road net imperfection, path area not enough (by for each person, considering from the planned land use angle); Function imperfection (lacking modern facility, sign etc.) category of roads is not high; Network planning unreasonable (from the distribute traffic amount, the road overload of power that has, the road that has is too light) etc.Two) vehicle constitutes complicated: the many trades mark of motor vehicle type are many, and the newness degree difference is big, and the rideability difference is big.Three) motor vehicle, bicycle even pedestrian mix row on same road, and bicycle accounts for significant proportion, and the road that uses Correct Lane accounts for total road ratio little (Beijing is a shade better, all is on duty mutually in other city).Four) people's traffic law idea is poor, and the pedestrian runs red light, and jaywalk etc. happen occasionally.Five) a lot of old urban renewals in city are carried out simultaneously with new enlarging, and the road change is big, also has other reason in addition, will study the traffic control method that is fit to China's national situation in a word, could solve the urban traffic control problem of China.
Controlled target and index:
People always wish that the road is clear, do not block, and arrive the destination as soon as possible, and few as far as possible the parking waited less when running into the crossing.In traffic control, generally get following two amounts as evaluation index:
1. mean delay t
d: all assign on all vehicles that arrive the crossing in preset time the loss of time when vehicle passes through the crossing, be mean delay.
2. parking ratio P
S: stop of several accounts for the number percent of whole arrival crossing vehicle fleet, and the vehicle of green lights such as queuing is many more, and the parking ratio is big more.
The few more parking of stop frequency waited for that the time of letting pass is short more when on behalf of people, these two indexs wish by the intersection, and it is many more just in time to meet the vehicle that green light need not stop, and it is good more then to control effect.No matter on control method, whether these two desired values are measured,, are representing the quality of controlling effect always objectively can form this two indexs.
For the optimal control parameter, always to construct a target and go out several I with above performance index, objective function I has finally reflected people to control effect total evaluation viewpoint, and objective function I and controlled variable have certain relation, is write as so it is the function of controlled variable: I(T, g, φ).General target function type is write as: (for an intersection)
I（T、g、φ）＝k
dt
d+k
sp
s
k
d-delay weighting coefficient, k
s-stop frequency weighting coefficient has been represented the attention degree of supvr to stop frequency and delay respectively.
If institute's road network of controlling always has N intersection, then total objective function I(T, g, φ)=I
Weigh the quality of whole road network control effect, so that the I minimum is a standard, controlled variable T, g, φ select well, arrange in pairs or groups well, then I(T, g, φ) just less, otherwise just bigger.
How to select the controlled variable (T, g, φ) of (determining) each each intersection of cycle, and make controlled variable can adapt to the continuous variation of traffic flow, remain target function value minimum (the control effect the is best) problem of the present invention's research just.
People have done the research of decades to the urban traffic control problem, develop into and generally acknowledged that now urban regional coordination control in real time is best control mode, so-called regional coordination is meant that controlled is a zone and not merely be a crossing, neither one several crossings (line) on the road, a but zone of forming by many crossings, the size in zone, shape is decided according to this city concrete condition, coordination is meant between each intersection controlled variable will be considered to influence each other, to mate mutually, can not consider the quality of indivedual intersections by light, and will consider the effect quality in whole Be Controlled zone.So-called control in real time is meant and will determines the parameter timing at the traffic behavior that occurs at that time, neither at the definite controlled variable of the traffic behavior of estimating in advance, more not that controlled variable has appearred just determining later in traffic behavior, the good variation that could really adapt to traffic behavior of real-time, the control effect that obtains.
1.SCATS the real-time system of selection of timing parameter
The SCATS method is in data acquisition and processing, there is the technique and skill of many details aspects such as intersection annunciator and CPU (central processing unit) information processing, but its basic principle is: the contrast relationship of drafting a cover timing parameter and traffic rank before control system puts into operation, promptly, select the timing parameter combination of corresponding optimum at the volume of traffic of different brackets.This timing parameter drafted in advance of cover and the syntagmatic of the volume of traffic are stored in the central control computer.Central authorities' controller then provides wagon flow throughput data by the wagon detector that is located at each crossing, selects suitable timing parameter automatically.And according to the real-time control of selected timing parameter combination implementation to the road network traffic signals.
Illustrate that with reference to Fig. 1 the SCATS method is as follows:
A. Be Controlled traffic process: refer to be managed the flow process of (comprising motor vehicle and bicycle) of traffic flow on the road net in the control area, it is the Be Controlled object.
B. the automatic detection subsystem of vehicle: it is formed by being distributed in the wagon detector of burying underground each intersection each track certain position (SCATS is embedded in the stop line place) (SCATS has only the motor vehicle wagon detector).It measures the information that each intersection vehicle arrives in time.
C. wagon flow data acquisition and processing subsystem: it gathers at that time wagon flow data from each detecting device according to the moment of design, handle through data smoothing then, add up, processing such as verification provide the traffic value that reflects each crossing all directions lane occupancy size at last.
D. real-time control subsystem: form by central computer and tens of crossing controllers.The traffic value of each intersection and direction lane is selected the foundation of timing parameter combination as central computer, certain traffic value syntagmatic correspondence the combination of certain timing controlled variable, and this corresponding relation is to calculate according to people's subjective experience or according to certain statistics in advance to be set on the central control computer.
K. traffic lights driver sub-system: be made up of tens of (intersection one) traffic signals lamp drivers and changing of traffic lights, the driver drives traffic lights change light color, to command travelling of each intersection vehicle.
Mainly there are following three shortcomings in this method:
(1) detecting device is embedded in the stop line place, the wagon flow data that record each crossing are to record under the timing parameter of having implemented at local crossing, it can only be used to provide the foundation of each cycle selection timing parameter of back, and this just causes the imperfect reason of real-time.
(2) the controlled variable combination is to be provided with in advance with the corresponding relation of volume of traffic combination, so just brings two problems:
A) be provided with the untested shortcoming of subjective mind in advance.Controlled variable should be fit to the Changing Pattern of the volume of traffic, and people can ask naturally: whether the corresponding relation of the volume of traffic that is provided with and controlled variable combination has in advance reflected this rule, and the SCATS method fails to address this problem.
B) the corresponding relation reflection of both just having admitted to be provided with is provided with rule constantly, but the rule when whether reflecting actual motion, traffic environment, condition have changed, and this rule also can change thereupon, and SCATS does not consider this problem.Though it is to measure traffic value in real time, the corresponding relation of traffic value and controlled variable may be out-of-date, and this has also caused its bad Another reason of real-time.
(3) SCATS does not consider that bicycle does not have the bicycle detecting device, and this does not become problem to developed country, and is a very important problem to the situation that China's bicycle accounts for significant proportion.
2.SCOOT real-time traffic situation analogy method:
Central control computer had both been undertaken the real time traffic data Treatment Analysis, carried out again the network diagram intersection traffic signal control.Compare with the real-time system of selection of SCATS timing parameter, its fundamental difference is: this method does not need to store any set timing scheme in advance, does not need to determine in advance the corresponding selection relation of a cover timing controlled variable and the volume of traffic yet.The real time modelling method is to rely on certain traffic mathematical model that is stored in central computer, the real time traffic data that records is analyzed, and timing parameter is optimized and revised.The optimization of timing parameter is that prediction value with integrated objective function (delay time at stop, stop frequency etc.) is a foundation.
Illustrate that with reference to Fig. 2 the SCOOT method is as follows:
A, B, C, K and SCATS methodological function are similar.Different is: vehicle detecting sensor is embedded in the exit at crossing, upstream among the B.The data that C deals are not traffic values, but the wagon flow-time of crossing, upstream section (detecting sensor position) is graphic.The D traffic model: traffic model is the core.It comprises two parts: 1) vehicle queue length prediction model changes prediction according to upstream section car-time and goes out the queue length of being obstructed of this highway section downstream road junction vehicle and change.2) congested in traffic prediction model: the wagon flow crowding of the wagon flow crowding prediction downstream road junction inflow point in the exit, crossing, upstream that obtains from C.As one of foundation of adjusting signal time distributing conception.
Objective function calculates:
k
Di, t
Di, p
SiDefine same front.Mi-traffic congestion degree weighting coefficient.The congested in traffic degree in this highway section of ci-.
By G timing parameter being done trace adjustment (small step is long to be adjusted), generally to change a step be 3 seconds to 5 seconds, be that cycle direction green time, each change of phase differential difference (increasing or minimizing) 5 seconds (3 seconds) are simulated by traffic model and obtained adjusted target function value, contrast with the current programme desired value, get target function value wherein few as preferred version, realize traffic control by the timing parameter drive signal lamp that optimizes.
This method exists following 3 shortcomings:
The quality of 1) working control effect depends primarily on the accuracy of traffic model.Vehicle drives to downstream road junction and relies on the model prediction to come out from the crossing, upstream, because the road holding difference is very big, car type is very complicated, vehicle mixes row under China's actual conditions, fortuitous event occurs often on the highway section, mix a driving phase mutual interference, these all have influence on the order of accuarcy of traffic model, the target function value (the control effect that calculates) that causes the timing parameter actual effect that obtains on this basis and analog computation to come out may be inconsistent, thereby influenced the optimization of timing parameter.
2) traffic model is artificial fixed in advance, the statistical law when the traffic flow statistics rule of foundation may not necessarily reflect actual motion when determining traffic model, and in this sense its real-time is also good inadequately.
3) set up traffic model many assumed conditions are arranged, just as, suppose the motor vehicle speed Normal Distribution, this and China's actual conditions (vehicle, car type, vehicle performance constituent ratio are external much complicated) fall far short, and China's motor vehicle speed distributes and very difficultly describes with some definite statistical distribution.Therefore also being not suitable for China uses.
In view of this, purpose of the present invention is considered the truth of China's traffic behavior again at the defective that existing urban transportation traffic control method exists, and proposes a kind of new self-learning intelligent co-ordinative controlling of urban traffic.This method is the application of self study Based Intelligent Control in urban traffic control, it is by the information of the automatic detection subsystem of vehicle to controlled traffic process actual measurement, carrying out traffic controls in real time, then relatively the optimal control parameter of surveying parameter and from the timing parameter knowledge base, extracting, carry out the self study process, determine to be fit to the timing parameter value (T, g, φ) of new transport environmental condition.This new method is compared with existing urban traffic control method: the timing parameter optimum, do not rely on mathematical model yet, real-time good (every 30-120 can change timing parameter second) has been considered the influence to motor vehicles of bicycle detection and bicycle, and can have been adapted to complicated traffic environment etc.
In order to realize purpose of the present invention, the invention is characterized in: by the information of the automatic detection subsystem B of vehicle controlled A actual measurement, offer traffic status identification subsystem C, timing parameter inference machine G is from timing parameter knowledge base F, extract optimum timing parameter (T, g, φ), send real-time control subsystem D then to, traffic lights driver sub-system K is according to the time sequencing and the timing parameter (T of control subsystem D arrangement in real time, g, φ), control controlled traffic process A again, the automatic detection subsystem B of vehicle is one group of new timing parameter (T, g, φ) and the control effect (t of corresponding actual measurement
d, p
s) offer and control effect discriminating subsystem E, calculate the target function value I of optimum timing parameter, and this functional value I compared with target function value under the optimum in history same traffic behavior, its result offers timing parameter learning machine H, determine whether to upgrade the timing parameter value under the corresponding traffic behavior (T, g, φ) among the F, by D, K, control A more then, the controlled variable that while D also carries out reality is passed to H and is noted.
Further feature of the present invention is: when the target function value that newly obtains among the timing parameter learning machine H when optimal objective function value is good in history, occur repeatedly so repeatedly, represent the incompatible new transport environmental condition of original timing parameter value (T, g, φ), H can change the timing parameter value automatically.When the target function value that newly obtains among the timing parameter learning machine H when the optimization objective function value is good in history, so repeated multiple times represents that new timing parameter value is more suitable for new transport environmental condition, above-mentioned two kinds of new timing parameter values just remain in F.
Further feature of the present invention is: controlled traffic process A comprises the traffic flow of motor vehicle and bicycle.
Further feature of the present invention is: the automatic detection subsystem B of vehicle is the data that detect motor vehicle and bicycle respectively, offers traffic status identification subsystem C.
Further feature of the present invention is: traffic status identification subsystem C is made up of traffic behavior library and data analysis recognizer, the information of wagon flow through the data analysis recognizer identify each intersection, all directions, each track vehicle arrive, by, queueing condition, traffic state information is provided for timing parameter inference machine G through the traffic behavior library.
Fig. 1 is the SCATS method frame principle figure of prior art.
Fig. 2 is the SCOOT method frame principle figure of prior art.
Fig. 3 is a control method frame principle figure of the present invention.
It is as follows to be described in detail control method of the present invention with reference to Fig. 3:
For sake of convenience, earlier code name among Fig. 3 is described as follows:
The controlled traffic process of A-; The automatic detection subsystem of B-vehicle; C-traffic status identification subsystem; The real-time control subsystem of D-; E-control effect is differentiated subsystem; F-timing parameter knowledge base; G-timing parameter inference machine; H-timing parameter learning machine; K-traffic lights driver sub-system.
A is identical with front SCATS and SCOOT with K, and remainder is all inequality.
The automatic detection subsystem of B(vehicle) with the front difference 3 points are arranged:
1) sensor burial place difference, this method design are embedded in the queuing system porch (promptly the distance from stop line is slightly larger than vehicle queue team leader place) and the stop line place in each each track, crossing.
2) because 1) so it can not only detect the vehicle that each crossing arrives, and can survey the situation of change of vehicle number between stop line and queuing system inlet at any time, can obtain the traffic behavior of crossing more accurately, can survey the performance index value (parking number percent, mean delay) of reflection control effect.This just provides data (information) source for self study intelligent controlling method.
3) two class vehicle automatic detectors are arranged is vehicle detector and bicycle detecting device for this method design, can obtain the data of bicycle at any time, can consider the influence of bicycle in the The whole control method, is fit to China's urban transportation characteristics.
C. traffic status identification subsystem, it is made up of traffic behavior library and data analysis recognizer.Carry out data analysis according to the information of the automatic detection subsystem B of vehicle and identify the traffic behavior pattern that each track vehicle of each intersection all directions arrives, passes through, lines up, in time provide traffic state information at that time to G.
D. real-time control subsystem: it with Fig. 1 in D among the SCATS distinguish and be:
1) it is not according to traffic value but passes through timing parameter inference machine G reasoning according to traffic behavior (refer to each track of each crossing direction vehicle arrives, comprehensive condition by, queuing) and select timing parameter (T, g, φ).
2) the various combinations of timing parameter combination＜(T, g, φ)〉not artificial in advance the setting, but this is constantly accumulated experience in actual motion and brings in constant renewal in that knowledge generates by system, these timing parameters are provided by timing parameter knowledge base F.The timing parameter of selecting must be corresponding with the actual traffic state that occurred at that time, and guarantees it is optimized under the same in history traffic behavior.
E. control effect and differentiate subsystem: objective function is determined to be set among the E to go by the supvr in advance, and objective function has been represented the overall consideration to control effect assessment index as previously mentioned.The performance index value of reflection control effect constantly is provided by B, E constantly calculates target function value from performance index value, constantly with similar state in history under the objective function optimal value compare, identify the quality of control effect, identification result is delivered to H(timing parameter learning machine) go, as the foundation of upgrading timing parameter knowledge.
F. timing parameter knowledge base: it is depositing a large amount of timing parameter combinations, these timing parameter combinations are made up corresponding with traffic behavior and are for optimum in history through diagnostic test, in case G selects traffic behavior through reasoning and makes up the corresponding relation that makes up with timing parameter, parameter (T, g, φ) when F just provides an assembly for D.
G. timing parameter inference machine, it is the traffic behavior that provides according to C, carry out reasoning according to the inference rule between traffic behavior and the timing parameter, determine that one group provides traffic behavior corresponding optimum timing parameter with C, from F, come out to give D this group parameter extraction and go to implement.
H. timing parameter learning machine, it is to differentiate the mechanism that whether upgrades timing parameter knowledge, its function is constantly to absorb the best timing parameter of control effect to upgrade best timing parameter under the same in history traffic behavior.Learning machine has cover learning rules, learning rules are the criterions how the regulation machine is learnt, for example: if the target function value that newly obtains is not as optimal objective function value is good in history, and this situation has occurred repeatedly, original " knowledge " of this explanation may wear out (be transport environmental condition that incompatibility is new), opportunity to study changes timing parameter value automatically when determining this situation, when occurring this traffic behavior again next time, go practical application with the timing parameter value after changing, as the timing parameter value after changing than original effective (it is excellent that target function value becomes), and the timing parameter that this situation then illustrates new change repeatedly occurs and be more suitable for this traffic behavior, the new knowledge that this corresponding relation just should become in the timing parameter knowledge base remains.Newly " knowledge " replacement old " knowledge " (renewal) learning rules are stipulated exactly: a) several less important considerations change timing parameters appear in situation.B) change which timing parameter.C) become still toward little change toward big? d) change what etc. so that machine is had regulations to abide by.
Provide the timing parameter value of actual execution by D to H, the result by E provides it the control effect is differentiated provides its traffic behavior by C.
This as can be seen from Figure 4 control method is made up of two main loops: a real-time control loop of being made up of A → B → C → G → F → D → K → A, the self study loop that another is made up of A → B → E → H → F → D → K → A.
The course of work of real-time control loop:
Vehicle travels on controlled road net, controlled traffic process A is constantly carrying out, B constantly detects vehicle and arrives, by, the information of queuing, offer C, identify the foundation that the residing traffic behavior in each intersection of road net carries out reasoning for G as the selection timing parameter, timing parameter is not careless the selection, but determine according to the optimum down rule of same traffic behavior in history, depositing the corresponding relation and the timing parameter value of a large amount of traffic behaviors and timing parameter among the F, result by the G reasoning determines that the timing parameter that extracts optimum from F sends D to, D becomes control signal corresponding to timing parameter and gives K, the conversion of signal when K finishes lamp, drive traffic lights and press timing parameter official hour replacing light color, staff vehicles travels, the controlled traffic process of control A().
The course of work in self study loop:
Performance index value＜td, the ps of parameter when an assembly (T, g, φ)〉offer E, E basis＜td, ps〉calculate target function value I, in E, the target function value I under the same traffic behavior optimum on I and the history is compared, comparative result is given H, H is according to control effect identification result, press learning rules, determine whether to upgrade the timing parameter value under the corresponding traffic behavior among the F, revise the timing parameter among the F when needing.Overlap with real-time control loop from the process of F → D → K → A.Controlled variable makes H note the actual controlled variable of carrying out by D → H simultaneously.Constantly diagnostic test control effect is constantly upgraded timing parameter, guaranteed to offer among the F that the timing parameter of D and the corresponding relation of traffic behavior (knowledge) always adapt with at that time transport environmental condition, and timing parameter is optimum in real time always.
With the real-time system of selection of SCATS(timing parameter) advantage of comparing this method is:
1. timing parameter is optimum through verifying as, and SCATS is subjective the setting.
2. real-time is good, and this comprises following two aspect factors:
1) this method is to determine that by this cycle traffic behavior each cycle of this cycle timing parameter (30～120 seconds) timing scheme is all variable, and a control of SCATS (timing) scheme continues to carry out the change in 5 minutes of timing scheme 5 minutes.
2) timing parameter of this method is that constantly check is brought in constant renewal in, always constantly adapt with the variation of traffic environment, timing parameter optimal value reflection objective circumstances at that time, the real-time of this corresponding relation is good, and SCATS artificially is provided with timing parameter, can only reflect the timing parameter of thinking optimum when being provided with, and can not accomplish along with traffic environment constantly changes and the value of setting of change timing parameter.
3. this method has been considered the bicycle detection and has been influenced problem, and SCATS does not consider.
4. this method adaptability is strong, timing parameter generates under the condition on the spot, no matter on the spot how complicated traffic environment (road, means of transportation, outside transport need, weather, vehicles formation etc.) and how to change, and timing parameter has always been taken into account with the influence factor of the form of " knowledge " complexity no matter how.The control effect that all can not obtain particularly at this two-way mixed traffic flow (the motor vehicle non-locomotive both direction mixed running influences each other) SCATS and the SCOOT of China.We's rule can adapt to China's urban traffic flow characteristics.
With SCOOT(real-time traffic situation analogy method) to compare, the advantage of this method is:
1. this method does not need to set up the traffic mathematical model, and the SCOOT rule mainly relies on mathematical model.And the complicacy of traffic process, the energy of traffic mathematical model absolutely not accurate description particularly under the situation of China.
2. real-time is good, and identical with the front also have 2 reasons:
1) controlling schemes of SCOOT (timing parameter) also is continue to carry out 5 minutes, and this detecting device with it is embedded in the exit at crossing, upstream, the traffic in continuous prediction arrival downstream to line up graphic relevant.This method can accomplish that this cycle traffic behavior is corresponding in real time with this cycle controlled variable, and each cycle can change timing parameter.
2) rule of traffic model (SCOOT method) representative is that transport environmental condition has changed in the finite time, and SCOOT can not revise traffic model at any time, and the real-time of this corresponding relation is also bad.
3. the adaptability of this method is strong, and its reason and front (comparing with the SCATS method) the 4th is identical.
Claims (5)
1, a kind of self-learning intelligent co-ordinative controlling of urban traffic, it is characterized in that: by the information of the automatic detection subsystem B of vehicle controlled A actual measurement, offer traffic status identification subsystem C, timing parameter inference machine G is from timing parameter knowledge base F, extract optimum timing parameter (T, g, Φ), send real-time control subsystem D then to, traffic lights driver sub-system K is according to the time sequencing and the timing parameter (T of control subsystem D arrangement in real time, g, Φ), control controlled traffic process A again, the automatic detection subsystem B of vehicle, one group of new timing parameter (T, g, Φ) and the control effect (td of corresponding actual measurement, ps) offer the control effect and differentiate subsystem E, calculate the target function value I of optimum timing parameter, and this functional value I compared with target function value under the optimum in history same traffic behavior, its result offers timing parameter learning machine H, determine whether to upgrade the timing parameter value (T under the corresponding traffic behavior among the F, g, Φ), pass through D then, K, control A again, the controlled variable that while D also carries out reality is passed to H and is noted.
2, by the described self-learning intelligent co-ordinative controlling of urban traffic of claim 1, it is characterized in that: when the target function value that newly obtains among the timing parameter learning machine H when optimal objective function value is good in history, occur repeatedly so repeatedly, represent original timing parameter value (T, g, ψ) the new transport environmental condition of incompatibility, H can change the timing parameter value automatically, when the target function value that newly obtains among the timing parameter learning machine H during than optimization objective function value in history, repeated multiple times like this, represent that new timing parameter value is more suitable for new transport environmental condition, above-mentioned two kinds of new timing parameter values just remain in F.
3, by the described self-learning intelligent co-ordinative controlling of urban traffic of claim 1, it is characterized in that: controlled traffic process A comprises the traffic flow of motor vehicle and bicycle.
4, by the described self-learning intelligent co-ordinative controlling of urban traffic of claim 1, it is characterized in that: the automatic detection subsystem B of vehicle is the data that detect motor vehicle and bicycle respectively, offers traffic status identification subsystem C.
5, by the described self-learning intelligent co-ordinative controlling of urban traffic of claim 1, it is characterized in that: traffic status identification subsystem C is made up of traffic behavior library and data analysis recognizer, the information of wagon flow through the data analysis recognizer identify each intersection, all directions, each track vehicle arrive, by, queueing condition, traffic state information is provided for timing parameter inference machine G through the traffic behavior library.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 91100825 CN1053696A (en) | 1991-02-12 | 1991-02-12 | Self-learning intelligent co-ordinative controlling of urban traffic |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 91100825 CN1053696A (en) | 1991-02-12 | 1991-02-12 | Self-learning intelligent co-ordinative controlling of urban traffic |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN1053696A true CN1053696A (en) | 1991-08-07 |
Family
ID=4904844
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 91100825 Withdrawn CN1053696A (en) | 1991-02-12 | 1991-02-12 | Self-learning intelligent co-ordinative controlling of urban traffic |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1053696A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102034360A (en) * | 2010-12-10 | 2011-04-27 | 中兴通讯股份有限公司 | Method and device for realizing self-adaption of traffic light |
| CN101226692B (en) * | 2007-12-21 | 2011-08-31 | 青岛海信网络科技股份有限公司 | Coordination type influence traffic control method |
| CN102360531A (en) * | 2011-09-30 | 2012-02-22 | 哈尔滨工业大学 | Intelligent traffic light control method and system based on wireless sensor network |
| CN103761883A (en) * | 2014-01-29 | 2014-04-30 | 中国科学技术大学 | Self-learning method and system for traffic signal control |
| CN104408944A (en) * | 2014-11-10 | 2015-03-11 | 天津市市政工程设计研究院 | Lamp group based mixed traffic flow signal timing optimization method |
| CN104751627A (en) * | 2013-12-31 | 2015-07-01 | 西门子公司 | Traffic condition parameter determining method and device |
| CN105702054A (en) * | 2015-12-01 | 2016-06-22 | 中华电信股份有限公司 | Method for maximizing multipath main road signal continuous bandwidth |
| CN106327886A (en) * | 2016-08-29 | 2017-01-11 | 安徽科力信息产业有限责任公司 | Method and device for reducing impact on control efficiency of plane sensing signal from motor vehicle |
| CN107464012A (en) * | 2017-07-10 | 2017-12-12 | 中国电子科技集团公司第二十八研究所 | A kind of Urban Transportation based on parallel simulation supports system |
| CN111696342A (en) * | 2019-03-11 | 2020-09-22 | 阿里巴巴集团控股有限公司 | Traffic signal timing optimization method and device, electronic equipment and readable storage medium |
| CN112353549A (en) * | 2020-10-14 | 2021-02-12 | 深圳中科大唐科技有限公司 | Device and method for improving sleep breathing |
-
1991
- 1991-02-12 CN CN 91100825 patent/CN1053696A/en not_active Withdrawn
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101226692B (en) * | 2007-12-21 | 2011-08-31 | 青岛海信网络科技股份有限公司 | Coordination type influence traffic control method |
| CN102034360A (en) * | 2010-12-10 | 2011-04-27 | 中兴通讯股份有限公司 | Method and device for realizing self-adaption of traffic light |
| CN102034360B (en) * | 2010-12-10 | 2013-12-11 | 中兴通讯股份有限公司 | Method and device for realizing self-adaption of traffic light |
| CN102360531A (en) * | 2011-09-30 | 2012-02-22 | 哈尔滨工业大学 | Intelligent traffic light control method and system based on wireless sensor network |
| CN104751627A (en) * | 2013-12-31 | 2015-07-01 | 西门子公司 | Traffic condition parameter determining method and device |
| CN103761883A (en) * | 2014-01-29 | 2014-04-30 | 中国科学技术大学 | Self-learning method and system for traffic signal control |
| CN103761883B (en) * | 2014-01-29 | 2016-03-02 | 中国科学技术大学 | A kind of self-learning method of traffic signalization and system |
| CN104408944A (en) * | 2014-11-10 | 2015-03-11 | 天津市市政工程设计研究院 | Lamp group based mixed traffic flow signal timing optimization method |
| CN105702054A (en) * | 2015-12-01 | 2016-06-22 | 中华电信股份有限公司 | Method for maximizing multipath main road signal continuous bandwidth |
| CN106327886A (en) * | 2016-08-29 | 2017-01-11 | 安徽科力信息产业有限责任公司 | Method and device for reducing impact on control efficiency of plane sensing signal from motor vehicle |
| CN107464012A (en) * | 2017-07-10 | 2017-12-12 | 中国电子科技集团公司第二十八研究所 | A kind of Urban Transportation based on parallel simulation supports system |
| CN111696342A (en) * | 2019-03-11 | 2020-09-22 | 阿里巴巴集团控股有限公司 | Traffic signal timing optimization method and device, electronic equipment and readable storage medium |
| CN111696342B (en) * | 2019-03-11 | 2022-05-27 | 阿里巴巴集团控股有限公司 | Traffic signal timing optimization method and device, electronic equipment and readable storage medium |
| CN112353549A (en) * | 2020-10-14 | 2021-02-12 | 深圳中科大唐科技有限公司 | Device and method for improving sleep breathing |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108364467B (en) | Road condition information prediction method based on improved decision tree algorithm | |
| CN104778834B (en) | Urban road traffic jam judging method based on vehicle GPS data | |
| Huang et al. | Evaluation of lane reduction “road diet” measures on crashes and injuries | |
| Abdulhai et al. | Simulation of ITS on the Irvine FOT area using" PARAMICS 1.5" scalable microscopic traffic simulator: Phase I: Model calibration and validation | |
| Owais et al. | When to decide to convert a roundabout to a signalized intersection: Simulation approach for case studies in Jeddah and Al-Madinah | |
| CN1053696A (en) | Self-learning intelligent co-ordinative controlling of urban traffic | |
| CN106935044A (en) | A kind of site location optimization method for preferentially coordinating control based on bus signals | |
| Pilko et al. | Study of vehicle speed in the design of roundabouts | |
| Fernandes et al. | Assessment of corridors with different types of intersections: Environmental and traffic performance analysis | |
| CN106530710A (en) | Manager-oriented highway traffic index prediction method and system | |
| CN110070720B (en) | Calculation method for improving fitting degree of traffic capacity model of intersection road occupation construction area | |
| Pokhrel et al. | Performance Assessment of a Signalized Intersection: A Case Study of Jay Nepal Intersection | |
| Dong et al. | Multiobjective evaluation of left-turn waiting areas at signalized intersections in China | |
| Bakhsh | Traffic simulation modeling for major intersection | |
| Madhu et al. | Estimation of roadway capacity of eight-lane divided urban expressways under heterogeneous traffic through microscopic simulation models | |
| Fitzpatrick et al. | Comparison of above-sign and below-sign placement of rectangular rapid-flashing beacons | |
| Vajeeran et al. | Identification of Effective Intersection Control Strategies During Peak Hours | |
| Tong et al. | Real time dynamic regulation method of main road signal lights based on data clustering in the environment of internet of vehicles. | |
| Wasson et al. | Reconciled platoon accommodations at traffic signals | |
| Desta et al. | Impacts of autonomous vehicle driving logics on heterogenous traffic and evaluating transport interventions with microsimulation experiments | |
| Liu et al. | Baseline microscopic and macroscopic models: Deliverable D4. 1 of the CoEXist project | |
| HASAN et al. | Traffic assessment and optimization at signalized intersections: A review study | |
| Zeb et al. | HetroTraffSim: A Novel Traffic Simulation Software for Heterogeneous Traffic Flow | |
| สมิทธิ ภัทร คำ ประพันธ์ | A Study of Sam Yan Intersection Traffic Signal Management Optimization | |
| Zhang et al. | Developing calibration tools for microscopic traffic simulation final report part 1: Overview methods and guidelines on project scoping and data collection |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C06 | Publication | ||
| PB01 | Publication | ||
| C03 | Withdrawal of patent application (patent law 1993) | ||
| WW01 | Invention patent application withdrawn after publication |