CN114973660B

CN114973660B - Traffic decision method of model linearization iterative updating method

Info

Publication number: CN114973660B
Application number: CN202210522052.XA
Authority: CN
Inventors: 朱煜钰; 吴青娥; 肖娜
Original assignee: Huanghe Science and Technology College
Current assignee: Huanghe Science and Technology College
Priority date: 2022-05-13
Filing date: 2022-05-13
Publication date: 2023-10-24
Anticipated expiration: 2042-05-13
Also published as: CN114973660A

Abstract

A traffic decision method of a model linearization iterative update method relates to the technical field of traffic informatization, and specifically comprises the following steps: step 1, calculating traffic state parameters; step 2, establishing a state prediction model; step 3, model iteration updating; step 4, intelligent decision: and providing an area road game algorithm, implementing traffic optimization, area scheduling, dredging traffic jams and ensuring smooth roads. The invention has better decision effect, faster processing speed and stronger anti-noise capability, and the simulation result shows that the decision accuracy is higher than 90% in a dense target environment; under the medium density environment, the correct decision rate of the reasoning decision method is 91.28 percent, the processing speed is 0.0936s, and the processing speed is less than 1s, so that theoretical ideas and decision support are provided for government construction, city planning, road long-width design and the like, and a target detection and tracking method is also provided for places and departments using video streams.

Description

Traffic decision method of model linearization iterative updating method

Technical Field

The invention relates to the technical field of traffic informatization, in particular to a traffic decision method of a model linearization iterative updating method.

Background

With the rapid development of economy and society, urban traffic congestion is increasingly serious, especially in the case of frequent traffic jams, jams and accidents of some extra large cities, and the state estimation of a highway traffic flow model is also the basis of the work of highway event detection, traffic analysis, traffic prediction and the like.

In order to solve the problems of traffic congestion and blockage, since the 60 s of the 20 th century, developed countries have begun to conduct traffic control studies, the main method of which is to control and optimize the whole traffic system by establishing a mathematical model of traffic flow using operational laws and optimal control theory. Although these studies have been widely used, current traffic prediction decision methods ignore the randomness and complexity of the traffic system and the limitations of the traffic flow model, resulting in poor decision-making effect, slow processing speed and poor noise immunity.

Disclosure of Invention

Aiming at the situation, in order to overcome the defects of the prior art, the invention provides the traffic decision method of the model linearization iteration updating method, which has better decision effect, faster processing speed and stronger noise resistance compared with the prior art.

The invention provides the following technical scheme: the method comprises the following steps:

step 1, calculating traffic state parameters: digging out useful data to analyze actual traffic conditions, and calculating traffic flow, traffic flow density and traffic flow speed;

step 2, building a state prediction model: according to the parameters calculated in the step 1, a traffic flow prediction model is established, wherein the traffic flow prediction model comprises congestion degree and influence state estimation modeling of surrounding road branches in a future period;

step 3, iterative updating of the model: in traffic flow analysis application, traffic flow data under various conditions are learned, and prediction model parameters are modified until the requirements are met;

step 4, intelligent decision: and providing an area road game algorithm, implementing traffic optimization, area scheduling, dredging traffic jams and ensuring smooth roads.

Preferably, the iterative update of the model in the step 3 adopts an iterative update method to perform information interaction fusion on different segment state models, and correspondingly update the state, the state error, the measurement error, the process gain value K (k+1) and the covariance P (k+ 1|k) thereof.

Preferably, the method for reinitializing the state condition of the iterative updating method is as follows:

is provided withIs the probability of the traffic flow state at time k-1,/->Are known; setting pi again _ji ＝Pr{m _k ＝m ⁽ⁱ⁾ |m _k-1 ＝m ^(j) Consists of m at time k-1 _k-1 M from traffic state to time k _k The probability of a transition of the traffic state and it is m dependent _k Traffic conditions as known conditions; the predicted probability of the traffic state at the i-th segment k moment is defined as:

the above equation can be predicted by measurements from time 1 to time k-1;

based on measurements at all times k-1, at the next time k, if the state is estimated to be state m _k ＝m ⁽ⁱ⁾ Then at the current k-1 time, state m _k-1 ＝m ^(j) The probability of (2) isDefinitions->The following are provided:

the hybrid estimation of the states is:

the mixed state error covariance is:

preferably, the specific method of state interaction of the iterative updating method is as follows:

prediction state error covariance:

corresponding measurement error covariance:

interaction gain:

the update state is:

the update covariance is:

preferably, the specific method for updating the probability of the state transition of the iterative updating method is as follows:

state interaction update, state probability update calculation is defined as:

wherein ,the definition is as follows: /> Is the measurement error of the traffic flow at the moment k of the ith section,/>Is corresponding to->Error covariance of (2);

the final update of the state prediction is:

updated final gain:

the corresponding state error hybrid total covariance is updated as:

preferably, the specific method for updating the application of the model iteration is as follows: discussing an interval section with an intersection, and only discussing roads with three equal-length road sections, namely an entrance, a driving road section and an exit, for each road in the intersection;

only the first road section inlet section and the third road section outlet section provide detection data of vehicle flow and average speed as input and output measurement values of a state estimation model, and the head end flow q of the first road section is taken ₀ Average velocity v ₀ Taking the actual measurement value of the third path segment end as an input quantity and taking the density rho ₃ Average velocity v ₃ As an output quantity;

according to the model equation, the state traffic expression of each segment is as follows:

wherein ,ρ₄ (k)、v ₄ (k) Density ρ of the previous time of the previous road segment ₃ (k-1) and velocity v ₃ (k-1) approximation.

The definition is as follows:

and (3) making:

then, there are:

and (3) making:

and the output is as follows:

the system state equation can be expressed as:

x(k+1)＝f[x(k)]+Γ[x(k)]w(k)，

y(k+1)＝h[x(k+1)]+v(k+1),

wherein Γx (k) is the noise driving matrix;

the formula is as follows:

calculating a partial guide matrix:

wherein ：

i.e. a recursive equation set can be utilized

Calculating;

the system noise covariance matrix Q (k) and the measurement noise covariance matrix R (k) are all obtained by real-time calculation of measurement data generated by simulation.

Preferably, the intelligent decision in step 4 is performed by the video of the current moment, and the decision system gives a prediction of the road condition of the driver and gives a clear road near the driving road, and further indicates an exit of the driving road and an entrance of the clear road.

Preferably, the decision rule of the intelligent decision is as follows:

if "condition z is a, then decision e is B",

then "if z is a', decision e is what should B? ";

the method can be defined as follows:i.e., conclusion B 'can be obtained by synthesizing A' with the reasoning relation from A to B;

where a is a condition set of states, B is a decision set based on condition a, and the above rule can be described simply as follows:

it is known that when A and B, the output is C, i.e., there is an inference rule:

IF A′AND B′，THEN C′

find what should the decision output C be when a 'and B'? The following steps can be used:

step (1): let d=a×b, let D _xy ＝μ _A (x)∧μ _B (y) obtaining the D matrix as

Step (2): writing D as a column vector DT, i.e., dt= [ D ₁₁ ,d ₁₂ ,…,d _1n ,d ₂₁ ,…d _mn ] ^T ；

Step (3): solving a relation matrix R, wherein R=DT×C;

step (4): from a ', B ', D ' =a ' ×b ';

step (5): converting D 'into a row vector DT' according to the method of step (2);

step (6): solving fuzzy reasoning output, i.e

Preferably, the rule established by the intelligent decision model is as follows:

Q _i (t) vector representing the number of vehicles waiting at the i-th intersection at time t, Q _i (t)＝[Q _i,E (t),Q _i,S (t),Q _i,W (t),Q _i,N (t)]Here Q _i,E (t),Q _i,S (t),Q _i,W (t),Q _i,N (t) each represents the number of vehicles waiting at time t in the 4 directions of east, south, west and north of the i-th intersection;

Q _i vector representing threshold number of vehicles at intersection i, Q _i ＝[Q _i,E ,Q _i,S ,Q _i,W ,Q _i,N ]，

Here Q _i,E ,Q _i,S ,Q _i,W ,Q _i,N The thresholds respectively representing the number of the waiting vehicles in different directions can be modified according to specific situations;

s represents all of the decisionsThe set of possible policies or actions, all possible policies of a decision are a finite set, where all policies are rule sets, set s= { S ₁ ,s ₂ ,…,s _n Each s _i Is the following rule:

s ₁ : if Q _i,E (t)＞Q _i,E And Q is _i,W (t)＞Q _i,W But Q is _i,S (t)＜Q _i,S And Q is _i,N (t)＜Q _i,N Then the things go straight, after 10 seconds, the things turn left to 30 seconds, and the red light of the double left turns is lighted;

s ₂ : if Q _i,S (t)＞Q _i,S And Q is _i,N (t)＞Q _i,N But Q is _i,E (t)＜Q _i,E And Q is _i,W (t)＜Q _i,W Then going straight north and south, turning on both the left and south after 10 seconds until both the left and south turns on the red light after 30 seconds;

s ₃ : if Q _i,E (t)＜Q _i,E And Q is _i,W (t)＜Q _i,W And Q is _i,N (t)＜Q _i,N But Q is _i,S (t)＞Q _i,S Then vehicles with northeast and northeast left turns are required to pass;

or vehicles in the south and north can select nearby intersections to bypass into other unblocked roads;

s ₅ : if Q _i,E (t)＞Q _i,E And Q is _i,W (t)＞Q _i,W And Q is _i,N (t)＞Q _i,N But Q is _i,S (t)＜Q _i,S At the same time |Q _i,E (t)-Q _i,E |＜|Q _i,N (t)-Q _i,N I and Q _i,W (t)-Q _i,W |＜|Q _i,N (t)-Q _i,N I and Q _i,S (t)-Q _i,S |＜|Q _i,N (t)-Q _i,N The vehicles turning to the south and the left or the vehicles turning to the north can select nearby intersections to be shunted into other unblocked roads;

……。

preferably, the decision coordination rule of the intelligent decision is: dividing the whole traffic coordination process in the area into three layers, wherein the lower layer is the coordination between the intersection and the adjacent intersection; the middle layer is the coordination between the regional road section and the intersection; the upper layer is the coordination between the regional segment and the adjacent regional segment;

p1, if the number of vehicles queued at the intersection 1 exceeds a threshold value, sending a request to the adjacent intersection 2;

p2. adjacent crossing responds to the request, and constructs game tree, letters on branches of game tree represent rule strategy, and expressions in block diagram represent advantages of rule comparison, by searching game tree, according to formulaSearching a balance point;

and P3, if the balance exists, the action strategy of the decision is the strategy for achieving the balance, each decision controls the road junction device according to the strategy, the coordination is finished, and if the balance does not exist, a request is sent to the decision of the road junction decision;

p4. the regional decision response request performs game coordination on the intersection decisions which are administrated by the regional decision response request, seeks an equalization point, and if the equalization point does not exist, the regional decision sends a request to an adjacent decision;

p5. adjacent area decisions respond to the requests, game coordination is performed, balance points are sought, and if the balance points do not exist and coordination fails, each decision keeps the original strategy unchanged.

Compared with the prior art, the invention has the beneficial effects that:

firstly, the application range is wide: the invention has more accurate prediction estimation value, has high prediction speed and good evacuation congestion effect, can fully solve the problem of road congestion, provides theoretical ideas and decision support for government construction, urban planning, road length, width design and the like, and also provides a target detection and estimation method for places and departments using video streams;

and (II) the accuracy is high: in simulation analysis, the invention has better decision effect, higher processing speed and stronger anti-noise capability, the decision accuracy rate is up to more than 90% in a dense target environment, and the correct decision rate is 91.28% in a medium density environment, which is 8% higher than the Agent method in the prior art; the treatment speed is 0.0936s, less than 1s;

and (III) saving resources: the invention can roughly estimate the maximum road section number between the two detectors on the road, namely, the furthest distance between the two detectors, which has direct guiding significance on saving cost in the installation process of traffic equipment and reduces resource waste.

Drawings

Fig. 1: the invention provides a system construction flow.

Detailed Description

The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

As shown in fig. 1, the invention provides a traffic decision method of a model linearization iterative update method, which comprises the following steps:

Specifically, in step 1, calculating traffic flow condition parameters: in the present embodiment, a section with an intersection in an area is discussed, and is appropriately divided into L sections, each section having a length delta _i About several hundred meters. Traffic detectors are arranged at the head and tail ends of the whole road, and measured data is provided as input of a dynamic model, including traffic flow q _i (k) Average velocity v _i (k) A. The invention relates to a method for producing a fibre-reinforced plastic composite The detection period T is 60 seconds, and is usually 10 to 60 seconds. Density ρ of traffic flow _i (k) Is made up of traffic flow q _i (k) And average velocity v _i (k) The calculation of these several parameters is as follows:

q _i (k)＝αρ _i (k)v _i (k)+(1-α)ρ _i+1 (k)v _i+1 (k)，

wherein ：

v(ρ)＝v _f exp[-(1b)(ρρ _cr ) ^b ]，

the main variables in the formula have the meanings as follows: q _i (k) For the flow rate of the vehicle flow from the ith section to the (i+1) th section at the kT moment, ρ _i (k) For the flow density at the ith section kT, v _i (k) Is the spatial average speed of the traffic flow at the ith section kT. Delta _i For the spatial sampling length, i.e. the length of the section of the i-th segment, it is generally denoted delta for simplicity. r is (r) _i (k) Inlet for the ith segment kT timeTraffic flow rate, s _i (k) The outlet vehicle flow rate at the i-th segment kT.

The parameters in the equation are v _f The free running speed of the traffic flow is represented, namely the maximum speed of the current traffic flow running; ρ _cr The vehicle flow density when the vehicle flow reaches the maximum is represented as critical density; b, τ, γ, ζ, λ, α, etc. are adjustment coefficients of the equation.

V (ρ) =v _f exp[-(1b)(ρρ _cr ) ^b ]The system is a steady-state traffic flow model and describes the speed-density relationship of steady-state traffic flow.

Specifically, a state prediction model is established in the step 2: for the condition of the traffic flow at time k and its measured values, the following nonlinear system is established:

x(k+1)＝f[x(k)]+Γ[x(k)]w(k)，

y(k+1)＝h[x(k+1)]+v(k+1)，

where w (k), v (k) is a zero-mean noise vector, and

E[w(k)]＝0,E[w(k)w′(j)]＝Q(k)δ _kj ，

E[v(k)]＝0,E[v(k)v′(j)]＝S(k)δ _kj ，

q (k) is the system noise variance matrix, S (k) is the measurement noise matrix, Γ [ x (k) ] is the noise drive matrix.

The initial state estimation is implemented and the values are measured as follows:

the corresponding state covariance is

The primary measurement error is as follows:

the corresponding measurement error covariance is:

the gain of interaction is obtained as follows:

the primary state estimate is updated as:

the primary state covariance is updated as:

wherein

At the given initial filtered valueAnd after the initial filtering covariance matrix P (0), the calculation sequence of the recursive algorithm is as follows:

is->Is->And P (2|1) … …, whereby the system state can be estimated by stepwise calculation. Obviously, the recurrence equations are all to be implemented online.

In order to make the prediction model more accurate in application, the state of the traffic flow model is continuously updated by utilizing state interaction of each segment, a corresponding partial guide matrix in a recurrence equation needs to be calculated to obtain an expression conforming to the recurrence equation, and then online recurrence calculation can be performed to obtain a system state estimated value at each moment.

Specifically, in the step 3, the iterative update of the model adopts an iterative update method to perform information interaction fusion on different segment state models, and correspondingly update the state, the state error, the measurement error, the process gain value K (k+1) and the covariance P (k+ 1|k) thereof.

First, linear expansion is performed in the neighborhood of the point of interest, in this embodiment, the point of interest refers to the state of the moving object at the time of the ith segment k, and the expressions x (k+1) =f [ x (k) ]+Γ [ x (k) ] w (k) and y (k+1) =h [ x (k+1) ]+v (k+1) are respectively linearized as:

x _i (k+1)＝F _i (k)x _i (k)+u _i (k)+G _i (k)w _i (k)，

y _i (k)＝H _i (k)x _i (k)+v _i (k)。

the state update is implemented in different road segments, and the update method is as follows:

(1) The state condition of the iterative updating method is reinitialized:

is provided withIs the probability of the traffic flow state at time k-1,/->Are known; setting pi again _ji ＝Pr{m _k ＝m ⁽ⁱ⁾ |m _k-1 ＝m ^(j) Consists of m at time k-1 _k-1 M from traffic state to time k _k The probability of a transition of the traffic state and it is m dependent _k Traffic status as alreadyKnowing the conditions; the predicted probability of the traffic state at the i-th segment k moment is defined as:

the above equation can be predicted by measurements from time 1 to time k-1;

the hybrid estimation of the states is:

the mixed state error covariance is:

(2) The specific method for the state interaction of the iterative updating method comprises the following steps:

prediction state error covariance:

corresponding measurement error covariance:

interaction gain:

the update state is:

the update covariance is:

(3) The specific method for updating the probability of the state transition of the iterative updating method comprises the following steps:

state interaction update, state probability update calculation is defined as:

/>

the final update of the state prediction is:

updated final gain:

the corresponding state error hybrid total covariance is updated as:

the specific method for the model iterative updating application comprises the following steps: discussing an interval section with an intersection, and only discussing roads with three equal-length road sections, namely an entrance, a driving road section and an exit, for each road in the intersection;

/>

The definition is as follows:

and (3) making:

then, there are:

/>

and (3) making:

and the output is as follows:

the system state equation can be expressed as:

x(k+1)＝f[x(k)]+Γ[x(k)]w(k)，

y(k+1)＝h[x(k+1)]+v(k+1),

wherein Γx (k) is the noise driving matrix;

the formula is as follows:

calculating a partial guide matrix:

wherein ：

can be utilizedRecurrence equation set

Calculating;

Specifically, the intelligent decision in step 4 is performed by the video at the current moment, and the decision system provided by the embodiment gives a prediction of the road condition to the driver, and gives a clear road near the driving road, and further indicates a driving road exit and a clear road entrance. In the design of the road traffic decision system, the determination of speed limit, entrance adjustment rate, information release and other control amounts must take the traffic state of each road section in the road as feedback information.

The decision rules for intelligent decision are as follows:

if "condition z is a, then decision e is B",

then "if z is a', decision e is what should B? ";

IF A′AND B′，THEN C′

step (1): let d=a×b, let D _xy ＝μ _A (x)^μ _B (y) obtaining the D matrix as

Step (3): solving a relation matrix R, wherein R=DT×C;

step (4): from a ', B ', D ' =a ' ×b ';

step (6): solving fuzzy reasoning output, i.e

The rule established by the intelligent decision model is as follows:

s represents the set of all possible policies or actions in a decision, all possible policies of a decision are a finite set, all policies are rule sets, and S= { S is set ₁ ,s ₂ ,…,s _n Each s _i Is the following rule:

……。

these rules are common to the above: since there are 2 cases of traffic flow in each direction compared with the threshold value, four directions are considered simultaneously, 2 in total ⁴ Species, analog s ₁ 、s ₂ 、s ₃ 6 cases in total, then 2 ⁴ -6 = 10 cases requiring consideration of the difference between the flow in each direction and the threshold; however, there are 4 differences, each of which is 3 compared with 3 differences in the other direction, then there are 3×10=30, and together there are 30+6=36, so 36 rules are defined here, i.e. the set S has 36 elements S _i ，i＝1,2,…,36。

At a certain moment, if a certain rule s _i Compared with any rule, the rule s is optimal, so that the road is smoother, namely the crossing dredging time is short _i Game winning, noted asThe coordination model is the following game algorithm.

Is provided with nRegular gaming is described as g= { S, U }, where S is a rule set and U is an advantage of comparing two rules, for example, the advantage is that its road is smoother, or the time for crossing clear is short, or congestion is cleared, etc. If there are β advantages, let U= { U ₁ ,u ₂ ,…,u _β }. If the strategy in the questionIs a Nash equilibrium, then must be satisfied

in the formula ：a strategy representing the ith rule selection; />A policy representing all rules except i; u (u) _j Representing the j-th advantage. Or reordered according to the advantages, and a balance under Nash equilibrium can be obtained.

The decision coordination rule of the intelligent decision is as follows: dividing the whole traffic coordination process in the area into three layers, wherein the lower layer is the coordination between the intersection and the adjacent intersection; the middle layer is the coordination between the regional road section and the intersection; the upper layer is the coordination between the regional segment and the adjacent regional segment;

In simulation analysis, the decision method provided by the invention is better than the existing Agent method, the invention not only has better decision effect and quicker processing speed, but also has stronger noise immunity, the decision accuracy is higher than 90% in a dense target environment, and the correct decision rate of the reasoning decision method is 91.28% in a medium density environment and is 8% higher than that of the existing Agent method; the processing speed is 0.0936s, less than 1s.

The invention is applied to each three equal-length road sections of an intersection, one entrance road, one driving road and one exit road of each road, the prediction model is used for obtaining the prediction of the traffic density and the traffic speed of the three road sections, a quantitative and qualitative combination method is adopted for comprehensive evaluation, and the number of state quantities which can be accurately estimated can be determined by the maximum time required by calculating the state data once under a certain condition and the estimation precision according to the setting of actual control needs and sampling time, so that the maximum road section number between two detectors on a road can be roughly estimated, namely the maximum distance between the two detectors is estimated, and the invention has direct guiding significance for saving expenses in the installation process of traffic equipment and reduces the resource waste.

In summary, the method has the advantages of accurate prediction and estimation value, high prediction speed and good evacuation congestion effect, solves the problem of road congestion, provides theoretical ideas and decision support for government construction, city planning, road length-width design and the like, and also provides a target detection and tracking method for places and departments using video streams.

The foregoing is only a preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art should be able to apply equivalents and modifications according to the technical scheme and the inventive concept thereof within the scope of the present invention.

Claims

1. A traffic decision method of a model linearization iterative update method is characterized by comprising the following steps:

the iterative updating of the model in the step 3 adopts an iterative updating method to perform information interaction fusion on different segment state models, and correspondingly updates states, state errors, measurement errors, a process gain value K (k+1) and covariance P (k+ 1|k) thereof;

the method for reinitializing the state condition of the iterative updating method comprises the following steps:

the above equation can be predicted by measurements from time 1 to time k-1;

the hybrid estimation of the states is:

the mixed state error covariance is:

step 4, intelligent decision: providing an area road game algorithm, implementing traffic optimization, area scheduling and dredging traffic jam, and ensuring the smoothness of the road;

wherein, the decision rule of the intelligent decision is as follows:

if "condition z is a, then decision e is B",

then "if z is a', decision e is what should B? ";

IF A′AND B′，THEN C′

Step (3): solving a relation matrix R, wherein R=DT×C;

step (4): from a ', B ', D ' =a ' ×b ';

step (6): solving fuzzy reasoning output, i.e

The intelligent decision in the step 4 is that a video at the current moment is used for giving a road condition prediction to a driver through a decision system, giving a clear road near a driving road, and further indicating a driving road outlet and a clear road inlet;

the rule established by the intelligent decision is as follows:

Q _i (t) vector representing the number of vehicles waiting at the i-th intersection at time t, Q _i (t)＝[Q _i,E (t),Q _i,S (t),Q _i,W (t),Q _i,N (t)]Here Q _i,E (t),Q _i ^, _S (t),Q _i,W (t),Q _i,N (t) each represents the number of vehicles waiting at time t in the 4 directions of east, south, west and north of the i-th intersection;

s ₂ : if Q _i,S (t)＞Q _i,S And Q is _i ^, _N (t)＞Q _i,N But Q is _i,E (t)＜Q _i,E And Q is _i,W (t)＜Q _i,W Then going straight north and south, turning on both the left and south after 10 seconds until both the left and south turns on the red light after 30 seconds;

s ₃ : if Q _i,E (t)＜Q _i,E And Q is _i,W (t)＜Q _i,W And Q is _i,N (t)＜Q _i,N But is provided withQ _i,S (t)＞Q _i,S Then vehicles with northeast and northeast left turns are required to pass;

s ₄ : if Q _i,E (t)＞Q _i,E And Q is _i,W (t)＞Q _i,W But Q is _i,S (t)＞Q _i,S And Q is _i,N (t)＞Q _i,N At the same time |Q _i,E (t)-Q _i,E |＞|Q _i,S (t)-Q _i,S I and Q _i,E (t)-Q _i,E |＞|Q _i,N (t)-Q _i,N I and Q _i,W (t)-Q _i,W |＞|Q _i,S (t)-Q _i,S I and Q _i,W (t)-Q _i,W |＞|Q _i,N (t)-Q _i,N I, then the things go straight, or vehicles in the south or north can choose nearby intersections to split into other clear roads;

s ₅ : if Q _i,E (t)＞Q _i,E And Q is _i,W (t)＞Q _i,W And Q is _i,N (t)＞Q _i,N But Q is _i,S (t)＜Q _i,S At the same time |Q _i,E (t)-Q _i,E |＜|Q _i,N (t)-Q _i,N I and Q _i,W (t)-Q _i,W |＜|Q _i,N (t)-Q _i,N I and Q _i,S (t)-Q _i,S |＜|Q _i,N (t)-Q _i,N And the vehicles going south, southwest and left turn or the vehicles going east and west can select nearby intersections to split into other unblocked roads.

2. The traffic decision method of the model linearization iterative updating method according to claim 1, wherein the specific method of the state interaction of the iterative updating method is as follows:

prediction state error covariance:

corresponding measurement error covariance:

interaction gain:

the update state is:

the update covariance is:

3. the traffic decision method of a model linearization iterative update method according to claim 1, wherein the specific method of updating the probability of state transition of the iterative update method is as follows:

state interaction update, state probability update calculation is defined as:

the final update of the state prediction is:

updated final gain:

the corresponding state error hybrid total covariance is updated as:

4. the traffic decision method of a model linearization iterative update method according to claim 1, wherein the specific method of the model iterative update application is: discussing an interval section with an intersection, and only discussing roads with three equal-length road sections, namely an entrance, a driving road section and an exit, for each road in the intersection;

The definition is as follows: and (3) making:

then, there are:

and (3) making:

and the output is as follows:

the system state equation can be expressed as:

x(k+1)＝f[x(k)]+Γ[x(k)]w(k)，y(k+1)＝h[x(k+1)]+v(k+1),

wherein Γx (k) is the noise driving matrix;

the formula is as follows:

calculating a partial guide matrix:

wherein ：

i.e. a recursive equation set can be utilized

Calculating;

5. The traffic decision method of a model linearization iterative update method according to claim 1, wherein the decision coordination rule of the intelligent decision is: dividing the whole traffic coordination process in the area into three layers, wherein the lower layer is the coordination between the intersection and the adjacent intersection; the middle layer is the coordination between the regional road section and the intersection; the upper layer is the coordination between the regional segment and the adjacent regional segment;