Regional boundary traffic signal control method and system based on model predictive control

Abstract

The invention provides a regional boundary traffic signal control method and system based on model predictive control, and belongs to the technical field of intelligent traffic. The method specifically comprises the following steps: partitioning a road network region; collecting traffic parameters of each region of the road network, and fitting an MFD curve of each region based on the collected data; establishing an optimization model by taking minimization of the number of vehicles in a road network as a target; an optimal boundary control value is obtained through optimization solution of an MPC controller with a built-in SLSQP algorithm, so that boundary traffic flow and boundary traffic signal lamp timing of each region of the road network are controlled; a traffic flow is simulated based on SUMO, and from a current time step to a next time step, an MPC controller with a built-in SLSQP algorithm is used for continuous iterative optimization solution to realize rolling optimization. And modeling the traffic state of each region by using MFD to obtain the flow dynamic state of the region. By establishing a network structure to simulate traffic dynamics, the MPC method is realized to control transfer of boundary traffic flow and dynamically control timing of traffic signal lamps, so that traffic congestion is relieved.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent transportation technology, and in particular relates to a method and system for controlling regional boundary traffic signals based on model predictive control. Background Art

[0002] With the acceleration of urbanization, traffic congestion has become a serious challenge for urban management and development. Frequent regional traffic congestion has become a major factor affecting urban economic development and social activities. The optimization and control methods of traffic flow based on regional congestion status have become a key issue in the field of traffic control. Factors such as the scale of the traffic network, the coordination of different areas of the network, the propagation of road congestion, and the interaction between drivers and traffic management systems constitute the complexity that arises when developing urban road network models and control schemes. Therefore, the macroscopic fundamental diagram (MFD) provides a unimodal low-scattering relationship between network vehicle density and network spatial average density for different network areas by simplifying the microscopic modeling task of the urban network. Alternatively, the road accumulation of MFD is defined as the number of vehicles in the area, and the trip completion flow is defined as the output flow of the area. After the trip completion flow increases to a critical point with the accumulation of traffic flow, the additional vehicles in the network will cause a sharp drop in the trip completion flow. In order to solve the problem of low efficiency in boundary control problem solving, the model predictive control (MPC) strategy is introduced. Model predictive control (MPC), also known as rolling horizon control, can be described as follows: at each sampling moment, based on the current measurement information obtained, an optimization algorithm is used to solve a finite time domain open-loop optimization problem online, and the first element of the obtained control sequence is applied to the controlled object. The above process is repeated at the next sampling moment, that is, the optimization problem is refreshed and solved again with the new measurement value. A model predictive boundary control framework is established to dynamically obtain the control parameters of each time period according to the flow conservation equation and control constraints, and then the control action is fed back to the traffic flow system to guide the signal control of traffic lights. Through the SUMO (Simulation of Urban Mobility) micro-simulation software and the Pycharm software through the TraCI (Traffic Control Interface) traffic control development interface, the SLSQP (Sequential Least Squares Programming) algorithm is used to solve the control parameters and synchronously update the boundary intersection signal light configuration scheme. The MPC strategy is applied to traffic flow simulation, and the boundary traffic lights are controlled with the help of efficient control performance, thereby alleviating road network congestion.

[0003] However, the MPC strategy has certain limitations in traffic flow, including: 1) MPC needs to solve an optimization problem in each control cycle, and its computational complexity increases significantly with the increase of the dimension of the control variable and the length of the prediction time domain. In large-scale traffic networks, the scale of the optimization problem may be too large to be solved within the actual time constraints. 2) MPC strategies are usually able to achieve optimal control locally, but in complex urban traffic networks, the optimization of a single traffic light may not be consistent with the global optimization goal. 3) The parameters in MPC (such as the length of the prediction time domain, the control cycle, the constraints, etc.) have a significant impact on the performance of the system, but these parameters usually rely on manual experience adjustment and have limited applicability. Summary of the invention

[0004] The purpose of the present invention is to overcome the problem that the optimization scale in existing large-scale traffic networks is too large and cannot be solved within the constraint time, and the existing technology only performs a single optimization of traffic lights, resulting in inaccurate optimization. A method and system for regional boundary traffic signal control based on model predictive control is provided.

[0005] In order to achieve the above object, the present invention adopts the following technical solution: In a first aspect, the present invention provides A further improvement of the present invention is that Compared with the prior art, the present invention has the following beneficial effects: The present invention provides a method and system for controlling regional boundary traffic signals based on model predictive control, comprising the following steps: partitioning a road network area; collecting traffic parameters of each area of ​​the road network, and fitting an MFD curve of each area based on the collected data; establishing an optimization model based on the MFD curve of each area with the goal of minimizing the number of vehicles in the road network; based on the established optimization model, optimizing and solving an MPC controller with a built-in SLSQP algorithm to obtain an optimal boundary control value, and controlling the boundary vehicle flow of each area of ​​the road network and the timing of boundary traffic lights according to the optimal boundary control value; simulating traffic flow based on SUMO, from the current time step to the next time step, continuously iterating and optimizing and solving the MPC controller with a built-in SLSQP algorithm to achieve rolling optimization. By using MFD to model the traffic status of each area, considering the impact of the current number of road vehicles on the results, and setting the boundary traffic flow as a control variable, the flow dynamics of the area can be obtained within the actual time constraint, solving the problem of untimely calculation due to the large scale of the existing optimization problem, which causes traffic congestion. By simulating traffic dynamics, the MPC method is used to control the transfer of boundary traffic flow and dynamically control the timing of traffic lights to achieve global optimization, thereby alleviating traffic congestion.

[0006] Furthermore, by fitting the MFD curve by partition, the traffic characteristics of each area can be reflected more accurately, providing strong support for traffic management and planning, and providing data support for traffic signal control, thus realizing a more intelligent and efficient signal control strategy.

[0007] Furthermore, the MPC controller with built-in SLSQP algorithm continuously iterates and optimizes the solution to achieve rolling optimization without relying on manual experience adjustment. The system can predict the traffic status at the next moment based on the traffic status at the current moment, so as to adjust the timing of traffic lights in time, control the traffic flow at the boundaries of each area, and avoid congestion.

[0008] Furthermore, simulation experiments of the present invention show that the MPC method proposed in the present invention has better control effect than the traditional method in terms of overall performance index and simulation. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 It is a flow chart of the present invention; Figure 2 A schematic diagram of a two-region road network based on SUMO simulation in the present invention; Figure 3 It is a schematic diagram of the fitted MFD of two regions in the present invention; Figure 4 It is a schematic diagram of the prediction framework of the MPC model in the present invention; Figure 5 Schematic diagram of three different experimental traffic demands in the present invention; Figure 6 It is a schematic diagram of the cumulative number of vehicles in two areas under three different experimental conditions in the present invention; Figure 7 It is a boundary control value curve diagram under three different experimental conditions in the present invention; Figure 8 Schematic diagram of the green light timing of the boundary signal light under three different experimental conditions in the present invention.

[0010] Table 1 shows the property indicators under the first experimental condition of the present invention; Table 2 shows the property indicators under the second experimental condition of the present invention; Table 3 shows the property indicators under the third experimental condition in the present invention. DETAILED DESCRIPTION

[0011] In order to further understand the content of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the embodiments are only for explaining the present invention and are not intended to limit it.

[0012] Embodiment 1: like Figure 1As shown, a method for controlling regional boundary traffic signals based on model predictive control comprises the following steps: S1: Divide the road network area; S2: Collect traffic parameters of each area of ​​the road network and fit the MFD curve of each area based on the collected data; S3: Based on the MFD curves of each area, an optimization model is established with the goal of minimizing the number of vehicles on the road network; S4: Based on the established optimization model, the MPC controller with built-in SLSQP algorithm is used to optimize and solve the optimal boundary control value, and the boundary traffic flow and boundary traffic light timing of each area of ​​the road network are controlled according to the optimal boundary control value; S5: Based on SUMO simulation of traffic flow, from the current time step to the next time step, the MPC controller with built-in SLSQP algorithm continuously iterates and optimizes the solution to achieve rolling optimization.

[0013] Specifically, in S1, Figure 2 As shown in the figure, the basic information of the road network in the study area, such as the length of each section and the number of lanes, is set according to the actual situation. The heterogeneous traffic network that divides the road network structure into two homogeneous areas is considered. , internal area (referred to as Area 1), External Area (Referred to as Area 2). The road network is divided into two sub-areas with a containment relationship. The traffic flow between the two sub-areas is mainly exchanged through 12 crossroads. The traffic flow inside the area is large, so only the boundary control value of the external area entering the internal area is set.

[0014] Specifically, in S2, if Figure 3 As shown in the figure, the Macro Fundamental Diagram (MFD) is a powerful and effective modeling tool for monitoring and managing large-scale urban road networks. It not only describes the general relationship between network traffic flow and network traffic operation level on homogeneous road networks with uniform vehicle density or congestion distribution, but also reflects the functional relationship between the number of vehicles in the network and the vehicle flow that can complete the journey, that is, the completed flow. For a complete MFD, the horizontal axis represents the cumulative number of vehicles in the road network, and the vertical axis represents the completed flow of vehicles in the road network. As the cumulative number of vehicles in the road network continues to increase, the state of the road network continues to deteriorate, and the road network state develops from the free flow stage to the critical congestion stage, then to the congestion stage, and finally reaches the blocking stage. When the cumulative number of vehicles in the area is small, the average speed of regional vehicles is high, the internal traffic conditions are good, but the road section utilization rate is not high. This is the free flow stage; in the critical congestion stage, as the cumulative number of vehicles reaches the critical cumulative value When the area is near, the regional completion flow is close to the optimal state, and the road section utilization rate is further improved. In the control strategy, the cumulative number of vehicles in the control area is generally maintained at As the cumulative number of vehicles in the area exceeds And gradually increase to the maximum cumulative , the area begins to be congested, the average speed of vehicles drops sharply and eventually reaches 0, the area is completely locked, and no vehicles can complete their journeys, the area goes from the congestion stage to the blocking stage. The E2 lane detector is used to detect the traffic flow in each section, and the cumulative number of vehicles and the trip completion flow at each time step are accumulated to obtain the total number of vehicles and the trip completion flow at each time step. The cumulative number of vehicles and the trip completion flow For the two divided areas, each has its corresponding MFD curve and its corresponding functional relationship, and their respective MFD fitting curves are obtained by polynomial function fitting.

[0015] Specifically, in S3, the optimization control objectives of the urban traffic road network have various forms, such as optimizing the delay time of the entire road network (the total travel time of vehicles), the overall output within the road network, etc. For the two-region urban road network, minimizing the number of vehicles in the road network is to allow vehicles to complete travel faster and more in a certain period of time. The present invention proposes to establish an optimization model with the goal of minimizing the cumulative number of vehicles in the two-region urban road network:

[0016] Subject to:

[0017]

[0018]

[0019]

[0020]

[0021] First, time is discretized into equally spaced intervals, that is, ,and is the simulation time interval. It is area 1 The total number of vehicles at the time, It is area 2 The total number of vehicles at the time.

[0022] is the ratio of boundary roads to total roads in region 1, is the ratio of boundary roads to the total roads in region 2. Boundary roads are roads from region 2 to region 1. is the boundary control value for region 2 to enter region 1, and They are the trip completion flows for area 1 and area 2, and They represent The total flow entering and leaving area 1 at time t, therefore, formula (2) is the flow conservation equation for area 1.

[0023] In formula (3) It is area 2 in Vehicles entering zone 1 at any time, is the ratio of the external area roads leaving area 2 to the total roads in area 2, Represents The total flow entering area 2 at any time, and Represents The total flow leaving area 2 at time t, so formula (3) is the flow conservation equation for area 2.

[0024] Formulas (4)-(6) describe the control state of the boundary controller. for The boundary control value when area 2 enters area 1 at the moment, To control the time domain, For time domain prediction, the boundary control values ​​of adjacent time steps should be kept within a small difference. The range of the boundary control value should be within [0,1]. Reasonable control based on the boundary control value can improve the stability and controllability of the system.

[0025] The essence of the boundary control strategy is to use traffic signals to control the number of vehicles in the congested network area to avoid regional network saturation and improve network performance. By combining the influence of the conditions at the regional boundary, an MPC-based regional boundary control strategy is proposed to improve the network performance of the internal congested area and keep the vehicles in an orderly outer area. In order to prevent the queue from spreading from the inner area to the outer area, the boundary intersection is used to dynamically control the traffic signal timing. When the cumulative number of vehicles at the intersection exceeds the optimal threshold, the boundary control value is used to dynamically adjust the boundary traffic signal timing.

[0026] Specifically, in S4, the MPC controller with built-in SLSQP algorithm is optimized and solved to control the vehicle flow at the boundary of the area and the timing of the boundary traffic lights according to the boundary control value. The details are as follows: MPC is a rolling horizon control strategy. In each time period, an optimization algorithm is used to solve a finite time domain optimal control problem to obtain the control action for each step in the time domain. The current state of the system is used as the initial state of the optimal control problem. At the same time, considering the disturbance and uncertainty of the prediction model, the optimal control sequence of the solution only applies the control action at the current time step to the control system. Compared with traditional control methods, MPC solves the optimal control problem in the current time domain in each time period, rather than optimizing the solution in the entire time domain. Therefore, model predictive control has more advantages in dealing with complex constraints.

[0027] The model predictive control (MPC) method mainly consists of the following three parts according to its working process: 1) Prediction Model The prediction model is the core foundation of MPC. Its main function is to predict the response of the system at a future moment through the historical information of the controlled object and the future control input, and use these prediction results as the input parameters of the optimization model. MPC does not have strict requirements on the form of the prediction model, but only needs to ensure that it has good prediction capabilities.

[0028] 2) Scrolling optimization The rolling time domain characteristics of MPC enable it to have the ability to adjust dynamically, so it is also called rolling time domain control. At each sampling moment, MPC predicts the behavior of the system in the future limited time domain based on the prediction model, and solves the control action with the optimal performance index as the goal through the optimization model. At the next sampling moment, the time window rolls forward, the optimization model is updated accordingly, and the optimization solution is recalculated to obtain the optimal control action at the current moment.

[0029] The optimization process of MPC is not completed offline at one time, but is carried out online at every moment. This rolling optimization mechanism enables MPC to dynamically adjust the target according to the predicted system behavior, ensuring that local optimal control is achieved at every moment. In this way, MPC can effectively deal with the uncertainty caused by the time-varying characteristics of the system and external disturbances, and improve the control effect.

[0030] 3) Feedback correction During the execution of MPC, the optimization algorithm is used to solve the optimal control sequence in a finite time domain at each sampling moment. However, due to system disturbances or model mismatch, usually only the first control action in the optimization sequence is applied to the actual system. At the next sampling moment, the actual output of the system will be fed back to the MPC controller and compared with the predicted output. Through error analysis and compensation correction, MPC further optimizes the control calculation at the next moment. Under this feedback correction mechanism, MPC continuously corrects the system error through closed-loop control, thereby improving the accuracy and robustness of control. Therefore, MPC is essentially a closed-loop optimization control strategy.

[0031] Furthermore, the MPC framework structure in this embodiment is as follows: Figure 4 As shown in the figure, the traffic state n(k) detected by the lane detector and the control input u(k) are input into the prediction model to calculate the traffic state of the next time step. The objective function is to minimize the cumulative number of vehicles in the two-area road network. The constrained optimization algorithm sequential quadratic programming SLSQP is used to solve the optimal control input sequence at the current moment.

[0032] The specific process of model predictive control is as follows: 1) Parameter initialization. Initialization control time domain is 10, prediction time domain is 15, the area boundary road r=12, the total number of roads in area 1 =36, total number of roads in area 2 =188, the number of roads leaving area 2 =28, is 1e-8.

[0033] 2) Get the initial state value, that is, the cumulative number of vehicles and the trip completion flow in area 1 and area 2.

[0034] 3) Initial solution setting. When the simulation time is 0, the initial solution is set to a fixed value. When the simulation starts, the initial solution of each cycle is set in a rolling manner, that is, the first value in the initial solution list is removed, and the first value of the solution obtained by optimizing the next time step is inserted at the end.

[0035] 4) Optimization solution. A mathematical model of the nonlinear optimization problem based on model predictive control is established, and the optimization algorithm SLSQP is used to solve it, and the control input sequence is obtained. The boundary control value is input into the next time step to achieve rolling optimization.

[0036] 5) Simulation loop: The control input sequence obtained by optimization is passed to the traffic simulation platform to obtain the traffic status of area 1 and area 2, thereby reflecting the traffic congestion status of the entire road network.

[0037] So far, traffic lights in most cities at home and abroad still use an offline signal control strategy with fixed phases and fixed timing schemes. From the time dimension, the fixed signal time control strategy of a single traffic light will cause waste of green time due to the uneven arrival of vehicles. In the spatial dimension, the timing schemes of traffic lights with similar topological structures at different locations within a certain range are exactly the same, but the scale of vehicles passing through these intersections is different, which will also cause overall waste of green time. Therefore, compared with the offline control strategy of fixed signal time, the control strategy of flexibly adjusting the signal light timing scheme according to the regional vehicle distribution and the scale of vehicle arrival at a single intersection can cope with more complex road traffic environments.

[0038] There are multiple signalized intersections at the border of the highly congested inner area 1, and each intersection has different traffic conditions. If there are limited border roads, it is possible that the border roads will be severely congested. We assume that no vehicles enter or leave the border roads to eliminate the interference of vehicles entering and leaving the border roads. This paper installs E2 lane detectors on each road to measure traffic flow in real time. The dynamic flow conservation model in S3 can grasp the number of vehicles accumulated in the area and predict the overall operation status of the inner area 1, and calculate the traffic flow from the outer area 2 to the inner area 1, as shown below: (7) in, is the time step External area Entering the inner area The transmission traffic flow, For external area The accumulated traffic flow in area 2, i.e., the trip completion flow in area 2, is the ratio of boundary roads to the total roads in area 2, that is, the ratio of the number of lanes from area 2 to area 1 to all lanes in area 2. As the first control input of the control optimization sequence, the traffic flow that can pass through the boundary road is calculated through the area boundary control value and the road proportion.

[0039] In time step The duration of the green light at each intersection is a function of the transfer traffic flow from area 2 to area 1 and the number of boundary roads. Multiply by the saturation flow rate The minimum duration of each green light is =30, maximum =52: (8) (9) The total number of vehicles accumulated in the critical phase of the boundary intersection is highly correlated with the green light time assigned to the phase. If a longer green light time is assigned to the critical phase of the boundary intersection with a large traffic volume, the number of vehicles in the overall road network will be relatively reduced. In the SUMO simulation software, the driving status of any vehicle at any time at each boundary intersection can be obtained in real time. Therefore, in the simulation, the judgment of the vehicle status is completely achievable, and the corresponding parameters can also be obtained in real time through the SUMO traffic control interface TraCI.

[0040] Specifically, in S5, SUMO simulates the traffic flow, repeats S4 to S5 from the current time step to the next time step, realizes rolling optimization, and obtains the experimental results and their analysis.

[0041] Embodiment 2: First, two different control strategies are set up, and the performance of the boundary traffic signal control strategy based on the MPC method is verified by comparing the simulation results. The specific simulation experiment design is as follows: Strategy 1: No control (NC). It assumes that the system control parameters are , set a fixed green light timing =41, used to simulate the changes in the traffic conditions in the area where no control action is applied to the road network. This strategy is usually used as a control group experiment to verify the performance of other control strategies.

[0042] Strategy 2: Boundary traffic signal control strategy based on model predictive control method. The boundary control value is solved by the MPC method embedded with the SLSQP algorithm, and the prediction time domain is set to 15 steps. In this way, the green light timing of the boundary traffic light is solved, and the minimum green light time is set to =30, maximum =52, saturation flow rate =1.

[0043] The boundary control optimization problem is solved by SUMO simulation through the Traci interface, the optimal control parameters of the system for each simulation time step are calculated, the boundary traffic light timing for each time step is calculated, and then three overall traffic demand performance indicators are obtained, based on which the control effects of the above two strategies are evaluated.

[0044] In addition, in order to simulate the process of a road network from the initial loading stage to the generation of congestion and then the dissipation of congestion, the traffic demand in the simulation process is given as follows: Figure 5 As shown, the simulation runs for 80 steps, the sampling interval is 90 seconds, and the total simulation time is 7200 seconds.

[0045] Three traffic demand simulation experiments were selected for two different control scenarios, and four overall performance indicators were used: cumulative vehicle flow, trip completion flow, total waiting time, and total travel time. As shown in Table 1, Table 2, and Table 3: Table 1: Property index-demand1

[0046] Table 2: Property index-demand2

[0047] Table 3: Property index-demand3

[0048] The three traffic demand simulation experiments of two control scenarios adopted four performance indicators: total cumulative vehicles, trip completion flow, total waiting time and total travel time as evaluation indicators for different scenarios. In traffic demand 1, compared with the NC strategy, the MPC strategy reduced the total cumulative vehicles by 40.91%, the trip completion flow by 22.02%, the total waiting time by 54.72%, and the total travel time by 39.54%. In traffic demand 2, compared with the NC strategy, the MPC strategy reduced the total cumulative vehicles by 16.59%, the trip completion flow by 6.64%, the total waiting time by 28.01%, and the total travel time by 16.24%. In traffic demand 3, compared with the NC strategy, the MPC strategy reduced the total cumulative vehicles by 34.49%, the trip completion flow by 18.41%, the total waiting time by 51.15%, and the total travel time by 32.92%. From the comparison of overall property indicators, under the MPC control strategy, they all decreased to varying degrees, proving the effectiveness of the MPC strategy.

[0049] like Figure 6 As shown in the figure, the changes in the cumulative number of vehicles in the two areas during the simulation process under the two control scenarios and three different traffic demands are shown. The road network traffic conditions are the same under the two control scenarios. In traffic demand 1 and traffic demand 3, between the simulation time of 1500s and 2000s, the cumulative number of vehicles under the MPC control strategy gradually began to decline until the end of the simulation, and the downward trend became more obvious. In traffic demand 2, the cumulative number of vehicles in the outer area under the MPC strategy began to decline at 4120s, and the cumulative number of vehicles in the inner area began to decline at 3060s. It can be seen that the control effect of the MPC strategy on traffic demand 1 and traffic demand 3 is more obvious, which proves that the model predictive control strategy is an efficient and ideal control strategy.

[0050] Since the boundary control value and green light timing are fixed in the NC strategy, Figure 7 The figure shows the changes of the boundary control values ​​of the road network under three different traffic demands under the MPC strategy. As the simulation time changes, the model predictive controller continuously controls the boundary control value of the region. In traffic demand 1, the boundary control value is stable from 0 to 3000s, indicating that the traffic flow is stable. Then it decreases due to some external factors (adjustment of traffic demand or traffic signal). After the decrease, the control value enters a lower and fluctuating stage, indicating that the boundary flow continues to fluctuate at a lower level. Traffic demand 2 also maintains a stable state at the beginning. The large fluctuations in the middle part may be the frequent adjustments made by the system to adapt to complex traffic conditions. In the last part, it remains at a low and stable level. This may indicate that in the subsequent time steps, the traffic flow is relatively sparse or the adjustment of signal timing no longer needs to change frequently. Traffic demand 3 also remains stable at the beginning. After the subsequent decrease, it stabilizes for a period of time and then rises again, which may be due to the recovery of traffic flow. Through different traffic demand MPC strategies, the boundary control value can be dynamically controlled to respond to changes in vehicle flow, reduce traffic pressure in a certain area, and thus optimize the overall traffic flow distribution.

[0051] This paper time-matches the 12 traffic lights at the boundaries of the area. Figure 8 The green light timing of a traffic light at the boundary signal light (2_1) under the MPC strategy for different traffic demands is used. The vehicle demand at each intersection on the boundary is solved by the boundary control value solved by MPC. The traffic signal green light timing is allocated according to the vehicle demand. The minimum traffic light timing is set to 30s and the maximum is 52s. Under the model predictive control strategy, the traffic light timing changes continuously as the simulation time progresses. Therefore, the model predictive control achieves the goal of alleviating traffic congestion by dynamically adjusting the boundary traffic light timing.

[0052] Embodiment 3: A regional boundary traffic signal control system based on model predictive control, comprising: Partitioning module, used to partition the road network area; The acquisition module and the MFD curve fitting module are used to collect traffic parameters of various areas of the road network and fit the MFD curve of each area based on the collected data; The optimization model building module is used to build an optimization model based on the MFD curves of each area with the goal of minimizing the number of vehicles on the road network; The optimization solution module is used to optimize and solve the established optimization model through the MPC controller with built-in SLSQP algorithm to obtain the optimal boundary control value, and control the boundary traffic flow and boundary traffic light timing of each area of ​​the road network according to the optimal boundary control value; The rolling optimization module is used to simulate traffic flow based on SUMO. From the current time step to the next time step, the MPC controller with built-in SLSQP algorithm continuously iterates the optimization solution to achieve rolling optimization.

[0053] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for controlling regional boundary traffic signals based on model predictive control, characterized in that: The steps include: Zoning of road network areas; Collect traffic parameters of each area of ​​the road network and fit the MFD curve of each area based on the collected data; Based on the MFD curves of each area, an optimization model is established with the goal of minimizing the number of vehicles on the road network; Based on the established optimization model, the MPC controller with built-in SLSQP algorithm is used to optimize and solve the optimal boundary control value, and the boundary traffic flow and boundary traffic light timing of each area of ​​the road network are controlled according to the optimal boundary control value; Based on SUMO simulation of traffic flow, from the current time step to the next time step, the MPC controller with built-in SLSQP algorithm continuously iterates and optimizes the solution to achieve rolling optimization.

2. The method for controlling regional boundary traffic signals based on model predictive control according to claim 1, characterized in that: Partitioning the road network area specifically includes: dividing the road network area into two sub-areas containing each other, including the inner area and external areas , the inner area and external areas Traffic flow is exchanged at intersections.

3. The method for controlling regional boundary traffic signals based on model predictive control according to claim 1, characterized in that: Traffic parameters of various areas of the road network are collected through the E2 lane detector.

4. The method for controlling regional boundary traffic signals based on model predictive control according to claim 1, characterized in that: The optimization model formula established based on the MFD curves of each area with the goal of minimizing the number of vehicles on the road network is as follows: in, It is the internal area No. The total number of vehicles at the time, Is an external area No. Total number of vehicles at the time; Internal area occupied by boundary roads The proportion of total roads, Boundary roads occupy external areas The proportion of total roads, For external area Entering the inner area The boundary control value of and The internal area and external areas The trip is completed flow, and They represent Always enter the inner area and leaving the inner area Total flow rate; In formula (3) Is an external area exist Always enter the inner area of vehicles, Leaving the external area The external area roads account for the external area The proportion of total roads, Represents Always access external areas The total flow rate, and Represents Stay out of the outside area at all times Total flow; Formulas (4)-(6) describe the control state of the boundary controller. for External area at the moment Entering the inner area The boundary control value of To control the time domain, To predict the time domain, the boundary control values ​​of adjacent time steps are controlled within the difference between.

5. The method for controlling regional boundary traffic signals based on model predictive control according to claim 4, characterized in that: Based on the established flow conservation model, the MPC controller with built-in SLSQP algorithm is optimized and solved to obtain the optimal boundary control value. The boundary traffic flow of each area of ​​the road network and the timing of the boundary traffic lights are controlled according to the optimal boundary control value. The specific method is as follows: Real-time monitoring of internal areas and external areas The traffic state n(k), where k represents the time step; Solve using the optimization algorithm SLSQP to obtain the boundary control input u(k); Based on the traffic state n(k) and boundary control input u(k), the traffic state at the next time step is predicted.

6. The method for controlling regional boundary traffic signals based on model predictive control according to claim 5, characterized in that: The calculation formula for controlling the traffic flow at the boundaries of each area of ​​the road network according to the optimal boundary control value is as follows: (7) in, is the time step External area Entering the inner area The transmission traffic flow, For external area The accumulated traffic volume, The first control input for the boundary control optimization sequence.

7. The method for controlling regional boundary traffic signals based on model predictive control according to claim 5, characterized in that: The calculation formula for controlling the timing of traffic lights at the boundaries of each area of ​​the road network according to the optimal boundary control value is as follows: (8) in, For the time step External area Entering the inner area The green light duration, r represents the external area Entering the inner area is the number of lanes, and S represents the saturation flow rate.

8. The method for controlling regional boundary traffic signals based on model predictive control according to claim 5, characterized in that: The parameters of the traffic status are obtained in real time through the traffic control interface TraCI.

9. A regional boundary traffic signal control system based on model predictive control, characterized in that: include: Partitioning module, used to partition the road network area; The acquisition module and the MFD curve fitting module are used to collect traffic parameters of various areas of the road network and fit the MFD curve of each area based on the collected data; The optimization model building module is used to build an optimization model based on the MFD curves of each area with the goal of minimizing the number of vehicles on the road network; The optimization solution module is used to optimize and solve the established optimization model through the MPC controller with built-in SLSQP algorithm to obtain the optimal boundary control value, and control the boundary traffic flow and boundary traffic light timing of each area of ​​the road network according to the optimal boundary control value; The rolling optimization module is used to simulate traffic flow based on SUMO. From the current time step to the next time step, the MPC controller with built-in SLSQP algorithm continuously iterates the optimization solution to achieve rolling optimization.

Images