CN108538065B - A Coordinated Control Method for Urban Arterial Roads Based on Adaptive Iterative Learning Control - Google Patents

Abstract

A coordinated control method for urban arterial roads based on adaptive iterative learning control includes: a. determining key intersections: for the controlled arterial roads, determining the intersection with the greatest traffic demand as the key intersection; b. initializing the public signal cycle , green-signal ratio, phase difference; c. optimize the green-signal ratio of key intersections; d. optimize the green-signal ratio of non-critical intersections; e. loop, repeat steps c and d every 3 to 5 signal periods. The invention takes the intersection as the control object, uses the flow correlation of the upstream and downstream intersections to coordinate the intersections, and calculates the green light time of each phase of the signal light through the iterative learning control method (ILC) according to the real-time road flow conditions. The closed-loop controller parameters are tuned in real time through a de-pseudo strategy. The invention reduces the real-time calculation amount of the main road control, improves the traffic efficiency of the main road, and the effect is better than the traditional timing control scheme, and provides an effective method for the coordinated control of the urban main road.

Description

A Coordinated Control Method for Urban Arterial Roads Based on Adaptive Iterative Learning Control

technical field

The invention relates to the technical field of traffic signal control, in particular to an urban arterial road coordinated control method based on adaptive iterative learning control.

Background technique

With the development of my country's social economy and the improvement of people's living standards, more and more cars have entered ordinary households, and problems such as traffic accidents, traffic congestion, environmental pollution and energy consumption are becoming more and more serious. Travel time, travel safety, environmental quality and Quality of life is constrained by traffic conditions.

Urban road traffic signal control is an extremely important aspect of modern urban traffic management. The quality of its management and control level will directly affect the effect of urban road traffic operation. In the urban road network, the main road bears a huge traffic load. Therefore, achieving good traffic signal control on the main road is the focus of urban traffic smoothing measures.

The research on modern urban traffic signal control theory shows that the realization of dynamic coordinated control of traffic signals on urban arterial roads, especially by realizing optimal conditions of signal timing, regulating traffic flow and making it evenly distributed in arterial roads, will greatly improve the road network. It is the best choice for traffic signal control in urban traffic peak hours, improving the traffic capacity of the main road itself and the traffic overflow phenomenon of the surrounding roads.

As an efficient urban traffic coordinated control method, the urban arterial road coordinated control method based on adaptive iterative learning control has the following characteristics: 1. Ensure the overall traffic balance of the arterial road, thereby reducing the overflow of the arterial road intersection; 2. .The green light at the intersection is more efficient, thereby improving the traffic capacity of the main road; 3. Adjusting the timing plan according to real-time data can quickly respond to changes in traffic demand and improve the stability of the road network; 4. It can make the main road Bear greater traffic demand, thereby reducing pressure on the rest of the road network and improving traffic conditions throughout the road network.

Foreign arterial road traffic signal coordination control methods have been researched, such as: J.D.C.Little first proposed the MAXBAND algorithm, which provides a set of optimized traffic signal phase differences for urban arterial roads including n intersections S1,...,Sn , so that as many motor vehicles as possible can pass through the main road without stopping within the set speed range. N.H.Gartner proposed MULTIBAND on the basis of the MAXBAND method, and improved many important features, such as the setting of clearing time, the control of left-turn vehicles, and the realization of different bandwidths for different sections of the trunk line. However, these research results do not effectively take advantage of the daily repetitive nature of traffic demand changes, and with the expansion of the road network, the amount of computation increases rapidly.

SUMMARY OF THE INVENTION

The present invention overcomes the above shortcomings of the prior art and provides a method for coordinated control of urban arterial roads based on adaptive iterative learning control, thereby reducing the probability of congestion on arterial roads and improving urban travel efficiency.

The present invention is a method for coordinated control of urban arterial roads based on adaptive iterative learning control, which is used in an urban road traffic area including several consecutive adjacent intersections, including the following steps:

a. Identify key intersections: For the accused main road, identify the intersection with the greatest traffic demand as the key intersection.

b. Initialize the public signal period, green signal ratio, and phase difference: for key intersections, obtain the intersection signal period according to the Webster method, and use it as the intersection public signal period; each intersection adopts a four-phase timing scheme, respectively Calculate the initial green-signal ratio according to its phase-to-flow ratio; at the same time, the phase difference between adjacent intersections of the main road is calculated by dividing the length of the road segment by the average speed of the road segment.

c. Optimize the green-signal ratio of key intersections: According to the real-time traffic data, the closed-loop control rate of the iterative learning control is determined by the de-false control method, and then the open-closed-loop iterative learning control method is used by using the error of the previous iteration and the error of this iteration. Optimize the green-to-signal ratio.

d. Optimize the green-signal ratio of non-critical intersections: start from the adjacent intersections of key intersections, coordinate in pairs, take the upstream intersection as the main intersection and the downstream intersection as the slave intersection, carry out the master-slave coordinated control design, and complete in turn Green-signal ratio optimization for non-critical intersections.

e. Cycle, repeat steps c and d every 3 to 5 signal cycles.

趋向于理想占有率o_d，并使非关键交叉口主干道方向上车流的道路占有率趋向于主路口对应流向的道路占有率，即

The present invention is used as a coordinated control method for urban arterial roads. The optimization goal of steps c and d is to make the road occupancy rate of the key traffic flow in each phase of the key intersection within the limited control time [0, K].

tends to the ideal occupancy rate o _d , and makes the road occupancy rate of the traffic flow in the direction of the main road of the non-critical intersection tends to the road occupancy rate of the corresponding flow direction of the main intersection, namely

In step c, the algorithm steps for optimizing the green-to-signal ratio are as follows:

1) Determine the closed-loop controller of iterative learning control using the pseudo-removal strategy

使得其中各参数对应的控制器均确保相应的迭代学习控制收敛。然后确定去伪策略中对于每一个候选参数对应控制器的虚拟参考的计算方法：First determine a set of candidate controller parameters

The controller corresponding to each parameter ensures the corresponding iterative learning control convergence. Then determine the calculation method of the virtual reference of the controller corresponding to each candidate parameter in the de-pseudo strategy:

Finally determine the performance metrics of each candidate controller:

计算它对应控制器的虚拟参考

和性能指标

若该候选控制器满足性能指标，则该控制器为非伪控制器，加入控制系统，若不满足，则在剩余的候选控制器参数中选择最大的一个进行计算，直至有候选参数对应的控制器满足性能指标。where α and β are setting vectors. Pick the largest one from the set of candidate controller parameters

Calculate the virtual reference of its corresponding controller

and performance indicators

If the candidate controller satisfies the performance index, the controller is a non-pseudo controller and is added to the control system. If not, the largest one of the remaining candidate controller parameters is selected for calculation until there is a control corresponding to the candidate parameter. The device meets the performance specifications.

2) Calculate the green light time of the key intersection according to the controller obtained from the pseudo-removal strategy, where the learning law of the iterative learning control is set as:

为开环迭代学习控制部分，

为闭环迭代学习控制部分，

为闭环学习控制率，k^o为开环学习控制率。where u _n (k) is the green light time of the k-th sampling period of the n-th iteration, and e _n (k) is the error of the k-th sampling time of the n-th iteration process,

For the open-loop iterative learning control part,

is the closed-loop iterative learning control part,

is the closed-loop learning control rate, and k ^o is the open-loop learning control rate.

In view of the increasingly prominent urban traffic problems, and the current situation that urban arterial roads are the main artery of urban traffic, the traffic load is constantly increasing. By coordinating the signal timing of urban arterial roads, the existing road infrastructure can be reasonably utilized to reduce the occurrence of congestion. probability. The invention takes the intersection as the control object, uses the flow correlation of the upstream and downstream intersections to coordinate the intersections, and calculates the green light time of each phase of the signal light through the iterative learning control method (ILC) according to the real-time road flow conditions. The closed-loop controller parameters are tuned in real time through a de-pseudo strategy.

The invention has the advantages that the real-time calculation amount of the main road control is reduced, the traffic efficiency of the main road is improved, the effect is better than the traditional timing control scheme, and an effective method is provided for the coordinated control of the urban main road.

Description of drawings

1 is a schematic diagram of a part of a certain main road in a city using the method of the present invention;

Detailed ways

The present invention will be further described below through the accompanying drawings and embodiments.

As shown in Fig. 1, an urban road traffic area including several consecutive adjacent intersections using the method of the present invention has a total of 3 intersections, which are represented as {1, 2, 3} by a sequence of natural numbers, in which the east-west road is the main road , the north-south roads are secondary roads or branch roads, and the east-west traffic is generally significantly larger than the north-south direction. Define the upward direction of the main road from west to east, and the downward direction from east to west. The phase division of traffic flow at each intersection is as follows: Phase 1 is east-west straight and right turn; phase 2 is east-west left and right turn; phase 3 is north-south straight and right turn; phase 4 is north-south left and right turn.

A method for coordinated control of urban arterial roads based on adaptive iterative learning control of the present invention includes the following steps:

a. Identify key intersections: For the accused main road, identify the intersection with the greatest traffic demand as the key intersection.

b. Initialize the public signal period, green signal ratio, and phase difference: For key intersections, obtain the intersection signal period according to the Webster method, and use it as the intersection public signal period; each intersection adopts a four-phase timing scheme, respectively. Calculate the initial green-signal ratio according to its phase-to-flow ratio; at the same time, the phase difference between adjacent intersections of the main road is calculated by dividing the length of the road segment by the average speed of the road segment.

c. Optimize the green-signal ratio of key intersections: According to the real-time traffic data, the closed-loop control rate of the iterative learning control is determined by the de-false control method, and then the open-closed-loop iterative learning control method is used by using the error of the previous iteration and the error of this iteration. Optimize the green-to-signal ratio.

d. Optimize the green-signal ratio of non-critical intersections: start from the adjacent intersections of key intersections, coordinate in pairs, take the upstream intersection as the main intersection and the downstream intersection as the slave intersection, carry out the master-slave coordinated control design, and complete in turn Green-signal ratio optimization for non-critical intersections.

e. Cycle, repeat steps c and d every 3 to 5 signal cycles.

The selection steps for key intersections in step a are:

Obtain the traffic demand data {Q1, Q2, Q3} of each intersection on the controlled road section, and select the largest one as the key intersection. As shown in Figure 1, the key intersection in the control area is intersection 1.

趋向于理想占有率o_d，并使非关键交叉口主干道方向上车流的道路占有率趋向于主路口对应流向的道路占有率，即

In steps c and d, the goal of optimizing the green signal ratio is to make the road occupancy rate of the key traffic flow at each phase of the key intersection within the limited control time [0, K].

tends to the ideal occupancy rate o _d , and makes the road occupancy rate of the traffic flow in the direction of the main road of the non-critical intersection tends to the road occupancy rate of the corresponding flow direction of the main intersection, namely

The process of optimizing the calculation of the green-signal ratio of key intersections in step c is as follows:

1) Determine the closed-loop controller of iterative learning control using the pseudo-removal strategy

使得其中各参数对应的控制器均确保相应的迭代学习控制收敛。然后确定去伪策略中对于每一个候选参数对应控制器的虚拟参考的计算方法：First determine a set of candidate controller parameters

The controller corresponding to each parameter ensures the corresponding iterative learning control convergence. Then determine the calculation method of the virtual reference of the controller corresponding to each candidate parameter in the de-pseudo strategy:

Finally determine the performance metrics of each candidate controller:

计算它对应控制器的虚拟参考

和性能指标

若该候选控制器满足性能指标，则该控制器为非伪控制器，加入控制系统，若不满足，则在剩余的候选控制器参数中选择最大的一个进行计算，直至有候选参数对应的控制器满足性能指标。where α and β are setting vectors. Pick the largest one from the set of candidate controller parameters

Calculate the virtual reference of its corresponding controller

and performance indicators

If the candidate controller satisfies the performance index, the controller is a non-pseudo controller and is added to the control system. If not, the largest one of the remaining candidate controller parameters is selected for calculation until there is a control corresponding to the candidate parameter. The device meets the performance specifications.

2) Calculate the green light time of the key intersection according to the controller obtained from the pseudo-removal strategy, where the learning law of the iterative learning control is set as:

The process of optimizing the green-signal ratio of non-critical intersections in step d is to coordinate the intersections on the main road from the adjacent intersections of key intersections, and take the upstream intersection as the ideal downstream intersection. model, perform master-slave control, and complete the optimization of the green-signal ratio in turn. As shown in Figure 1, the optimization steps in this controlled area should be {2, 3}.

The specific embodiments described herein are merely illustrative of the spirit of the invention. This does not limit the right scope of the present invention. In fact, for more complex on-site situations, such as the existence of "T"-shaped intersections, part of the lanes are one-way streets, etc., the method described in the present invention can also be applied, as long as the method of simply changing the flow calculation can be considered.

Claims (1)

1. A city main road coordination control method based on adaptive iterative learning control is suitable for an urban road traffic area comprising a plurality of continuous adjacent intersections, and comprises the following steps:

a. determining a key intersection: for the controlled main road, determining an intersection with the largest traffic demand as a key intersection;

b. initializing common signal period, split ratio, phase difference: for a key intersection, acquiring an intersection signal period according to a Webster method, and taking the intersection signal period as an intersection public signal period; each intersection adopts a four-phase timing scheme, and the initial green signal ratio is calculated according to the phase flow ratio of the intersection; meanwhile, the phase difference between adjacent intersections of the main road is calculated by dividing the length of the road section by the average speed of the road section;

c. optimizing the green signal ratio of the key intersection: determining the closed-loop control rate of iterative learning control by using a pseudo-removing control method according to real-time traffic data, and then optimizing the split green ratio by using the open-closed loop iterative learning control method by using the error of the previous iteration and the error of the current iteration, wherein the optimization aims at the limited control time [0, K ]]In the method, the road occupancy of each phase key traffic flow at the key intersection is ensured

Tending towards the ideal occupancy o_dWherein

The subscript indicates the ith key phase of the key intersection j;

the steps for carrying out optimization calculation on the split green ratio are as follows:

1) determining a closed-loop controller for iterative learning control by utilizing a de-counterfeiting strategy;

first, a candidate controller parameter set is determined

Ensuring the convergence of corresponding iterative learning control by the controller corresponding to each parameter, and then determining a calculation method of virtual reference of the controller corresponding to each candidate parameter in the pseudo-removing strategy:

finally, determining the performance index of each candidate controller:

wherein α and β are set vectors, and the largest is selected from the set of candidate controller parameters

Calculating its virtual reference to the controller

And performance index

If the candidate controller meets the performance index, the controller is a non-pseudo controller, a control system is added, if the candidate controller does not meet the performance index, the largest one of the remaining candidate controller parameters is selected for calculation until the controller corresponding to the candidate parameter meets the performance index;

2) and calculating the green time of the key intersection according to the controller obtained by the pseudo-removing strategy, wherein the learning law of iterative learning control is set as follows:

wherein u is_n(k) For the green light of the kth sampling period of the nth iterationM, e_n(k) For the error at the kth sampling instant of the nth iteration,

for the open-loop iterative learning control portion,

in order to perform the closed-loop iterative learning control part,

for closed loop learning control rate, k^oThe open loop learning control rate;

d. optimizing the green signal ratio of the non-critical intersection: starting from adjacent intersections of the key intersections, coordinating every two intersections, taking an upstream intersection as a main intersection and a downstream intersection as a slave intersection, carrying out master-slave coordination control design, and sequentially finishing the optimization of the green-to-green ratio of non-key intersections, wherein the optimization target is that the limited control time is [0, K ]]In the method, the road occupancy of traffic flow in the main road direction of the non-critical intersection tends to the road occupancy of the corresponding flow direction of the main road, namely

Wherein

Subscripts denote the ith phase of n, m at adjacent crossings;

e. and c, circulating, and repeating the steps c and d every 3-5 signal periods.

Images