Expressway ramp confluence optimization control method for mixed traffic flow

Abstract

The invention relates to the technical field of intelligent network connection, and discloses a mixed traffic flow-oriented highway ramp confluence optimization control method, which comprises the steps of selecting different control methods according to state information of vehicles entering a formation control area, indirectly guiding manual driving vehicles to form a mixed formation mode through intelligent network connection vehicles, and controlling the formation control area according to the mixed formation mode. The driving states of all vehicles in the motorcade are monitored in real time; and dynamically calculating a confluence sequence and confluence time according to the driving state of the vehicle team entering the confluence control area, and dynamically adjusting the driving path and speed of the vehicle according to the confluence sequence and the confluence time. According to the method, the passing efficiency in the confluence process is greatly improved, and particularly in a complex traffic flow environment, the problem that in a traditional method, complex dynamic interaction during vehicle team confluence cannot be processed is solved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent network connection technology, and in particular to a highway ramp merging optimization control method for mixed traffic flows. Background Art

[0002] With the rapid development of intelligent network technology, connected autonomous vehicles (CAVs) can obtain road information and the dynamic operating status of surrounding vehicles in real time through communication technologies such as vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I), and use perception and decision-making mechanisms to achieve precise control of vehicle operation. This technology has significantly improved traffic efficiency and driving safety. However, it will take a long time to achieve a traffic environment completely dominated by CAVs, which means that for a long time in the future, traditional human-driven vehicles (HDVs) and CAVs will coexist for a long time, forming a mixed traffic flow scenario. With the emergence of mixed traffic flows, technologies such as vehicle collaborative control and merging sequence optimization have become key issues that need to be solved urgently.

[0003] Although there have been some studies on vehicle cooperative control in mixed traffic flow ramp merging scenarios, the existing technology still has some shortcomings. First, many studies have weak robustness to sudden traffic conditions during vehicle formation. Especially in complex dynamic traffic environments, how to quickly adapt to traffic flow changes and adjust vehicle speed and trajectory through real-time control methods is still an important difficulty.

[0004] Secondly, most existing studies focus on the formation and local optimization of vehicle formations. However, in merging scenarios, the optimization of the merging sequence of the fleet is often overlooked. Traditional single-vehicle sequence optimization methods are difficult to effectively handle the complexity of merging fleets, especially in mixed traffic flows, and cannot well consider the coordination and interaction between CAVs and HDVs. There are large differences between CAVs and HDVs in terms of vehicle behavior, reaction speed, and adaptability to traffic flow, which leads to the failure of traditional methods to fully consider the dynamic interaction relationship between vehicles when optimizing the merging sequence, thereby affecting the traffic efficiency and safety during the merging process. Especially in high-density traffic flow environments, the optimization of merging order and time is of great significance to improving traffic efficiency.

[0005] Finally, existing research usually focuses on theoretically optimal solutions, but pays insufficient attention to real-time performance and computational complexity in actual traffic environments. Especially in a mixed traffic flow environment, CAVs need to quickly adapt to the uncertainties of HDVs, which leads to a large contradiction between the computational complexity and real-time performance of existing methods in high-dynamic complex scenarios. Existing technologies usually rely on computationally intensive optimization algorithms and are difficult to ensure real-time response. Especially in the case of heavy traffic and complex traffic conditions, the response speed and computational efficiency of the system are still urgent problems to be solved. In addition, many methods fail to effectively handle computational complexity, resulting in failure to meet the requirements in practical applications with high real-time requirements. Summary of the Invention

[0006] In view of the above deficiencies in the prior art, the present invention provides a freeway ramp merging optimization control method for mixed traffic flow.

[0007] In order to achieve the above object of the invention, the technical solution adopted by the present invention is as follows:

[0008] A freeway ramp merging optimization control method for mixed traffic flow, comprising the following steps:

[0009] Select a control method according to the state information of vehicles entering the formation control area, indirectly guide human-driven vehicles to form a mixed formation mode through connected and automated vehicles, and monitor the driving states of all vehicles in the fleet in real time;

[0010] Dynamically calculate the merging sequence and merging time according to the driving state of the fleet entering the merging control area, and dynamically adjust the driving paths and speeds of the vehicles according to the merging sequence and merging time.

[0011] Optionally, the control method selected according to the state information of vehicles entering the formation control area includes:

[0012] Obtain the state information of vehicles entering the formation control area;

[0013] Judge whether the current vehicle has reached the formation trigger point; if so, proceed to the next step and monitor the state information of the fleet in real time; otherwise, continue to monitor the state information of the vehicle;

[0014] Judge whether the type of the current vehicle is a connected and automated vehicle; if so, proceed to the next step; otherwise, follow the FVD car-following model to drive;

[0015] Judge whether the type of the vehicle following the current vehicle is a human-driven vehicle; if so, actively decelerate the current vehicle using a PID controller and follow the FVD car-following model for the following vehicle; otherwise, proceed to the next step;

[0016] Determine whether the leading vehicle type of the current vehicle is an intelligent connected vehicle; if so, the current vehicle follows the CACC car-following model; otherwise, the current vehicle follows the ACC car-following model.

[0017] Optionally, when the PID controller is used to actively decelerate the current vehicle, the PID controller is optimized by combining the particle swarm optimization algorithm and the simulated annealing method. Whether to accept the current solution is determined by comparing the fitness value of the current solution with the current global optimal solution, which is expressed as:

[0018]

[0019] where p represents the acceptance probability, G best represents the global optimal solution, J represents the fitness value, T represents the simulated annealing temperature, T = max(T·α, T min ), max represents the maximum value function, T min represents the minimum simulated annealing temperature, and α represents the temperature decay coefficient.

[0020] Optionally, determine whether the fleet size exceeds the maximum formation size according to the status information of the fleet; if so, trigger a new fleet for the excess vehicles and determine the length of the current fleet at the same time; otherwise, continue to monitor the status information of the fleet.

[0021] Optionally, dynamically calculate the merging time according to the driving state of the fleet entering the merging control area, including:

[0022] Taking the minimum merging time as the optimization objective and the allowable time range to reach the merging point and the safe merging distance of the fleet as the constraint conditions, solve to obtain the optimal merging time.

[0023] Optionally, taking the minimum merging time as the optimization objective and the allowable time range to reach the merging point and the safe merging distance of the fleet as the constraint conditions, specifically:

[0024]

[0025] where min represents the minimum value function, max represents the maximum value function, represents the merging time allocated to fleet i, represents the shortest time required for the first vehicle of fleet i at the current vehicle speed and position to reach the merging point through the maximum acceleration a max ; represents the longest time required for the first vehicle of fleet i at the current vehicle speed and position to reach the merging point through the minimum acceleration a min ; v max represents the maximum speed, v i (t) is the speed of the first vehicle of fleet i at the current moment t, x i$(t)$ is the distance of the first vehicle in platoon $i$ from the merging point at the current time $t$, $x(t)$ represents the vehicle position at the current time $t$, $t$ i represents the merging time of the first vehicle in platoon $i$, $t$ i-1 represents the merging time of the second vehicle in platoon $i$, $\Delta t$ 1 represents the minimum allowable safety gap for consecutive platoons in the same lane to avoid rear-end collisions, $h$ 1 represents the minimum safety time interval for consecutive vehicles in the same lane to pass through the merging point, $n$ represents the number of vehicles in the platoon, $h$ HDV represents the safe following time interval of the HDV, $t$ j represents the merging time of the first vehicle in platoon $j$, $\Delta t$ 2 represents the minimum allowable safety gap for consecutive platoons in different lanes to avoid rear-end collisions, $h$ 2 represents the minimum safety time interval for consecutive vehicles in different lanes to pass through the merging point.

[0026] Optionally, calculate the merging sequence dynamically according to the driving state of the platoon entering the merging control area, including:

[0027] Construct the current stage state triple according to the number of platoons with allocated right-of-way on the main road and ramp and the allocation result of the right-of-way in the current stage;

[0028] Construct the state transition equation and state space diagram according to the transfer to the next state after making a decision based on the current stage state;

[0029] According to the arrival time constraint relationship between two platoons, take the minimum value of the maximum arrival time in the path from the initial state to the terminal state as the discriminant function, and gradually calculate the local optimal solution of each stage through the recurrence formula to finally obtain the global optimal path.

[0030] Optionally, gradually calculate the local optimal solution of each stage through the recurrence formula specifically as:

[0031]

[0032] Among them, $f$ k represents the discriminant function, $S$ k represents the current stage state, $m$ k represents the number of platoons with allocated right-of-way on the main road, $n$ k represents the number of platoons with allocated right-of-way on the ramp, $r$ k represents the allocation result of the right-of-way in the current stage, $\min$ represents the minimum value function, $MAT$ k represents the maximum arrival time from the initial state $S$ 0 to the terminal state $S$ k , $P$ k-1 represents from the previous state $S$ k-1 to the current state $S$k The path, where max represents the maximum value function, represents the shortest time Δt required for the first vehicle in platoon i at the current vehicle speed and position to reach the merging point with a maximum acceleration a max The shortest time Δt required to reach the merging point, 1 represents the minimum allowable safety gap between consecutive platoons in the same lane to avoid rear-end collisions, represents the longest time Δt required for the first vehicle in platoon i at the current vehicle speed and position to reach the merging point with a minimum acceleration a min The longest time Δt required to reach the merging point, 2 is the minimum allowable safety gap between consecutive platoons in different lanes to avoid rear-end collisions.

[0033] Optionally, dynamically adjust the driving path and speed of the vehicle according to the merging sequence and merging time, including:

[0034] Construct a Hamiltonian function based on the objective function and the vehicle dynamics state equation with the goal of minimizing energy consumption;

[0035] Determine the necessary optimality conditions for each vehicle according to the Pontryagin maximum principle;

[0036] Obtain the optimal control input based on the Hamiltonian function and the necessary optimality conditions for each vehicle;

[0037] Solve for the speed and position of the vehicle according to the optimal control input and the vehicle dynamics state equation.

[0038] Optionally, solving for the speed and position of the vehicle according to the optimal control input and the vehicle dynamics state equation specifically includes:

[0039]

[0040] where, represents the speed of platoon i at the current time t, b i , c i , d i , e i represent state constants, represents the position of platoon i at the current time t.

[0041] The present invention has the following beneficial effects:

[0042] (1) The present invention dynamically adjusts the control strategy of vehicle platoons through a PID controller that combines the Particle Swarm Optimization (PSO) algorithm and the Simulated Annealing (SA) algorithm. This technical solution can obtain the dynamic changes of traffic flow in real time, quickly adjust the speed and trajectory of vehicles, and ensure that the system can respond quickly to sudden traffic conditions. It avoids the low efficiency or traffic accident risks caused by untimely parameter adjustment in traditional methods. Through this adaptive control strategy, the system can efficiently cope with different traffic flow conditions, ensuring the stability and safety of traffic flow, thereby enhancing the real-time adaptability.

[0043] (2) The present invention combines the Dynamic Programming (DP) algorithm and the Pontryagin Maximum Principle (PMP) to optimize the collaborative work of vehicle merging sequences and trajectory planning. This technical solution accurately calculates the merging timing and optimal trajectory of each vehicle fleet, enabling vehicles to smoothly connect within the merging area and minimizing vehicle conflicts and passing delays to the greatest extent. For example, the interaction relationship between CAVs and HDVs is fully considered, and CAVs can adjust their speed and trajectory according to the behavior of surrounding HDVs, thus achieving a more efficient and safe merge. This optimization of dynamic interaction greatly improves the passing efficiency during the merging process, especially in complex traffic flow environments, avoiding the complex dynamic interaction problems that cannot be handled by traditional methods when vehicle fleets merge.

[0044] (3) The present invention significantly reduces the computational complexity through the Dynamic Programming (DP) algorithm while ensuring the optimization quality. Traditional optimization methods usually require a large amount of computing resources and are difficult to respond quickly in high-density and complex traffic flow environments. The DP algorithm of the present invention quickly solves the merging sequence and merging time through efficient state space search and strategy planning, effectively reducing the computational burden. For example, under high traffic density conditions, the system can still quickly obtain the optimal merging decision, ensuring real-time performance in the case of large traffic volume and complex traffic conditions. This technical solution enables the system to meet real-time requirements, provide rapid decision support, and solve the problems of excessive computational complexity and slow response in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a schematic diagram of the highway ramp merging scenario;

[0046] Figure 2 It is the technical roadmap of the present invention;

[0047] Figure 3 It is a schematic diagram of the flow of a highway ramp merging optimization control method for mixed traffic flow according to the present invention;

[0048] Figure 4 It is a schematic diagram of the change in the sum of squared errors;

[0049] Figure 5 Schematic diagram of PID parameter change

[0050] Figure 6(a) is a schematic diagram of the speed curve simulation of the leading vehicle at a constant speed

[0051] Figure 6(b) is a schematic diagram of the spacing curve simulation of the leading vehicle at a constant speed

[0052] Figure 7(a) is a schematic diagram of the speed curve simulation of the formation control with PID combined with PSO and SA

[0053] Figure 7(b) is a schematic diagram of the spacing curve simulation of the formation control with PID combined with PSO and SA

[0054] Figure 8(a) is a schematic diagram of the simulation of the curve of recovery time varying with scale

[0055] Figure 8(b) is a schematic diagram of the simulation of the curve of the decay rate of speed fluctuation varying with scale

[0056] Figure 8(c) is a schematic diagram of the simulation of the curve of the decay rate of spacing fluctuation varying with scale

[0057] Figure 9 Schematic diagram of the comprehensive performance index of different fleet sizes

[0058] Figure 10 Schematic diagram of the construction of the state space of dynamic programming

[0059] Figure 11 Schematic diagram of the comparison of the average vehicle speed under the main line / ramp flow ratio of 65:35

[0060] Figure 12 Schematic diagram of the analysis of the average vehicle delay under the main line / ramp flow ratio of 65:35

[0061] Figure 13 Schematic diagram of the comparison of the average vehicle speed under the main line / ramp flow ratio of 80:20

[0062] Figure 14 Schematic diagram of the analysis of the average vehicle delay under the main line / ramp flow ratio of 80:20

[0063] Figure 15 Schematic diagram of the comparison of the calculation time between dynamic programming and Gurobi in ramp merge control Specific implementation method

[0064] The specific embodiments of the present invention will be described below to facilitate those skilled in the art of this technology to understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those of ordinary skill in the art of this technology, as long as various changes are within the spirit and scope of the present invention defined and determined by the appended claims, these changes are obvious, and all inventions made using the concept of the present invention are within the scope of protection.

[0065] Aiming at the shortcomings of the prior art, the present invention mainly solves the following technical problems:

[0066] (1) Poor real-time adaptability. The existing control methods cannot adapt to the rapid changes in traffic flow in mixed traffic flow in a timely manner. The response of the system is often lagged, resulting in a decrease in traffic efficiency.

[0067] (2) Insufficient dynamic interaction between the merging sequence and trajectory planning. The traditional methods fail to fully consider the dynamic interaction between the vehicle platoon merging sequence and trajectory planning during the merging process, resulting in low passing efficiency and conflicts between vehicles during merging.

[0068] (3) High computational complexity. The existing optimization-based control methods usually need to process complex computational models, resulting in a large amount of computation and difficulty in meeting the real-time requirements. Especially in the high-density traffic flow environment, the computational efficiency of the system becomes a bottleneck.

[0069] The present invention aims to solve the above deficiencies of the existing control methods under mixed traffic flow. The specific objectives include improving the real-time adaptability of the system, optimizing the dynamic interaction between the merging sequence and trajectory planning, and reducing the computational complexity. To achieve these objectives, the following technical solutions are proposed:

[0070] (1) Improve real-time adaptability. The present invention introduces a PID controller dynamic adjustment mechanism that combines the particle swarm optimization (PSO) algorithm and the simulated annealing (SA) algorithm, enabling the control system to automatically optimize the control parameters according to the real-time changes in traffic flow, quickly respond to the changes in traffic flow, and ensure the efficient and stable operation of the system in a dynamic environment.

[0071] (2) Optimize the dynamic interaction between the merging sequence and trajectory planning. Based on the existing vehicle platoon control, the present invention combines the dynamic programming (DP) algorithm and the Pontryagin maximum principle (PMP) to optimize the collaborative work of the vehicle platoon merging sequence and trajectory planning. By accurately determining the merging timing and the optimal trajectory of each platoon, the vehicles can achieve seamless connection during the merging process, improving the merging efficiency and traffic safety.

[0072] (3) Reduce the computational complexity and improve the real-time performance. The present invention efficiently solves the merging sequence and merging time through the dynamic programming (DP) algorithm, reduces the computational complexity, ensures that the system can quickly solve in a high-density and dynamically complex traffic flow environment, and achieves real-time response. This method significantly improves the real-time performance and ensures the ability to quickly adapt and make decisions in complex traffic environments.

[0073] The application scenario of the present invention is the highway ramp, which mainly includes a platoon control area and a merging control area, as Figure 1 shown. The technical route covers the specific strategies in the two control areas, as Figure 2 shown.

[0074] In the platoon control area, the vehicle type entering the platoon area determines the choice of control strategy. Different control methods are adopted according to whether the vehicle is a CAV (Connected and Autonomous Vehicle) or an HDV (Human-Driven Vehicle). The HDV is indirectly guided by the CAV to form a 1 + n hybrid platoon mode, effectively reducing the impact of the randomness of HDV behavior on traffic efficiency. At the same time, the PID control parameters of the leading vehicle CAV are optimized in real time by combining the Particle Swarm Optimization (PSO) algorithm and Simulated Annealing (SA). This mode can monitor the driving states of all vehicles in the platoon in real time, thus improving the overall stability of the platoon.

[0075] In the merging area control, first, the dynamic programming algorithm (DP) is used to minimize the delay as the goal, efficiently solve the merging sequence and merging time, and determine the best timing for each platoon to reach the merging point, thus ensuring the smoothness of the merging process. Then, combined with the Pontryagin Maximum Principle (PMP), the driving paths and speeds of the vehicles are dynamically adjusted according to the optimized merging timing, so that each platoon can merge smoothly and coordinately. Through this process, the system can effectively maintain stability and operation efficiency in high-density traffic flow, reduce the risk of traffic conflicts and accidents, and improve the overall traffic safety and fluency.

[0076] As Figure 3 shown, an optimization control method for highway ramp merging for mixed traffic flow provided by an embodiment of the present invention includes the following steps S1 and S2:

[0077] S1. Select different control methods according to the status information of the vehicle entering the platoon control area, indirectly guide the human-driven vehicle by the connected and autonomous vehicle to form a hybrid platoon mode, and monitor the driving states of all vehicles in the platoon in real time;

[0078] In an optional embodiment of the present invention, step S1 selects different control methods according to the status information of the vehicle entering the platoon control area, including:

[0079] Obtain the status information of the vehicle entering the platoon control area;

[0080] Determine whether the current vehicle has reached the formation trigger point; if so, proceed to the next step and monitor the status information of the vehicle fleet in real time; otherwise, continue to monitor the status information of the vehicle.

[0081] Determine whether the current vehicle type is a connected and automated vehicle (CAV); if so, proceed to the next step; otherwise, follow the FVD car-following model to drive.

[0082] Determine whether the type of the vehicle following the current vehicle is a human-driven vehicle; if so, actively decelerate the current vehicle using a PID controller and follow the FVD car-following model for the following vehicle; otherwise, proceed to the next step.

[0083] Determine whether the type of the vehicle in front of the current vehicle is a connected and automated vehicle (CAV); if so, follow the CACC car-following model for the current vehicle; otherwise, follow the ACC car-following model for the current vehicle.

[0084] Step S1 also includes determining whether the size of the vehicle fleet exceeds the maximum formation size based on the status information of the vehicle fleet; if so, trigger a new vehicle fleet for the excess vehicles and determine the length of the current vehicle fleet; otherwise, continue to monitor the status information of the vehicle fleet.

[0085] The vehicle formation control strategy adopted in this embodiment is as follows:

[0086] Step1 When the vehicle enters the formation control area, the roadside unit sensing device obtains the basic information of the vehicle, including vehicle ID, position, speed, and acceleration.

[0087] Step2 The system determines whether the vehicle has reached the formation trigger point. If the vehicle has not reached the trigger point, continue to monitor its status; if it has reached the trigger point, the vehicle enters the next stage for type judgment.

[0088] Step3 Determine whether the vehicle is a CAV. If the vehicle is a CAV, further determine the type of the following vehicle. If the following vehicle is an HDV, the CAV uses a PID controller to actively decelerate, and the following HDV follows the FVD car-following model. If the following vehicle is a CAV, select an appropriate control strategy based on the type of the vehicle in front. If the vehicle in front is a CAV, use the CACC car-following model; if the vehicle in front is an HDV, use the ACC car-following model.

[0089] Step4 After the formation starts, the system monitors the status of the vehicle queue in real time. If the queue exceeds the maximum formation size, the excess vehicles will trigger a new vehicle fleet and determine the length of the current queue; if a new CAV vehicle joins, the CAV will trigger a new vehicle fleet again to ensure that the vehicle formation meets the control requirements.

[0090] Step 5. Confirm whether the queue is stable, that is, parameters such as the spacing and speed of all vehicles meet the control objectives.

[0091] Step 6. After the queue is stable, the entire formation control process ends, and the vehicles will run at the target speed and with a stable vehicle distance.

[0092] In an alternative embodiment of the present invention, when the current vehicle is actively decelerated using a PID controller in step S1, the PID controller is optimized by combining the particle swarm algorithm and the simulated annealing method.

[0093] PID control is a classic feedback control method, and its name comes from the three control actions in the controller, namely, proportional (P), integral (I), and derivative (D). The general form of the PID control algorithm is:

[0094] e(t) = r(t) - c(t)

[0095]

[0096] When optimizing the PID controller parameters using the particle swarm optimization (PSO), the particle swarm is first initialized, and the particles are randomly distributed in the solution space and given an initial velocity. The state of a particle is determined by its position and velocity. The position represents the current parameter combination, and the velocity determines the search direction and step size. The particle adjusts its velocity and position according to the historical best solution P best and the global best solution G best to gradually approach the optimal value. The fitness function usually measures the accuracy and stability of the PID control system in vehicle formation through system performance indicators such as the integral square error (ISE), integral absolute error (IAE), or peak error. The fitness function is:

[0097]

[0098] Assume there are K particles in the particle swarm, and the position of each particle in the search space is represented by the vector x i and the search velocity is represented by the vector v i . The velocity and position update formulas for the particles are as follows:

[0099] v i (t + 1) = ω·v i (t) + c 1 ·r 1 ·(P best,i - x i (t)) + c 2 ·r 2 ·(G best - x i (t))

[0100] x i (t + 1) = xi (t) + v i (t + 1)

[0101] Among them, ω is the inertia weight, which determines the balance between the global search and the local search of the particle; c 1 and c 2 are learning factors, which respectively control the learning ability of the particle to its own optimal position and the global optimal position; r 1 and r 2 are random numbers within the range of [0, 1].

[0102] Since the formation system faces a time-varying, uncertain and dynamically complex environment, there are multiple local minima in the solution space of parameter optimization, and PSO may cause the algorithm to fall into the local optimum. By embedding the probability acceptance mechanism of the simulated annealing method (SA) into PSO, if the fitness value J of the new solution is not better than the current global optimal solution G best , then it is judged whether to accept this solution through probability:

[0103]

[0104] T = max(T·α, T min )

[0105] Among them, p represents the acceptance probability, G best represents the global optimal solution, J represents the fitness value, T represents the simulated annealing temperature, T = max(T·α, T min ), max represents the maximum value function, T min represents the given minimum simulated annealing temperature, and α represents the temperature decay coefficient.

[0106] Figure 4 and Figure 5 illustrate that combining the PSO and SA algorithms can effectively balance the rapidity, stability and accuracy of the controller, and minimize the error and improve the control performance of the system by gradually optimizing the PID parameters.

[0107] To verify the effectiveness of the proposed formation method of the present invention, numerical simulation experiments are carried out using Python in this section.

[0108] (1) Formation efficiency analysis

[0109] In the simulation environment, a heterogeneous mixed vehicle fleet consisting of five vehicles is designed, where the first vehicle is a CAV and the rest are HDVs. The initial states of the vehicles are shown in Table 1. Through the PID control strategy combining PSO and SA, real-time vehicle formation control is carried out, and a comparison is made with the traditional car-following model method, verifying the advantages of the proposed method in terms of formation efficiency and dynamic performance.

[0110] Table 1 Initial states of vehicles

[0111]

[0112] Comparing and analyzing Figure 6(a), Figure 6(b), Figure 7(a), and Figure 7(b), the PID control method combining PSO and SA significantly improves the formation efficiency and coordination. The leading vehicle solves the problem of excessive headway by actively decelerating and gradually resumes the target speed when the following vehicle approaches the safe headway, thus balancing speed and distance. This control method completes the formation process within 40 seconds, which is about 33% faster than the 60 seconds of the traditional car-following model. At the same time, the speed and headway adjustment of the vehicle are smooth and stable, avoiding overshoot and oscillation. The global optimization of the algorithm enhances the adaptability to different initial conditions, enabling the vehicle to quickly converge and form a stable formation.

[0113] (2) Maximum formation scale analysis

[0114] To quantify the impact of the fleet size on the formation system, this experiment designs a fleet consisting of 1 leading vehicle (CAV) and several following vehicles (HDV), with the size ranging from 2 to 8 HDVs. The experiment analyzes the speed-time curves of fleets of different sizes and calculates three key indicators: formation recovery time (the time for all vehicles to recover to the target state), speed fluctuation attenuation rate (the ratio of the speed fluctuation amplitude of the last vehicle in the queue to the disturbance vehicle), and spacing fluctuation attenuation rate (the ratio of the spacing fluctuation amplitude of the last vehicle in the queue to the disturbance vehicle), as shown in Figure 8(a), Figure 8(b), and Figure 8(c).

[0115] Figure 9 It shows the impact of the fleet size on the system performance. When the fleet size is 5 vehicles, the comprehensive performance index reaches the highest value of 0.61, indicating that this size effectively balances cooperation and flexibility and shows the best stability and efficiency in terms of vehicle speed and spacing fluctuations. The queue under this size can quickly recover to a stable state, reduce fluctuations, and improve overall efficiency, fully demonstrating the cooperation advantage of CAV.

[0116] S2. Dynamically calculate the merging sequence and merging time according to the driving state of the fleet entering the merging control area, and dynamically adjust the driving path and speed of the vehicle according to the merging sequence and merging time.

[0117] In an alternative embodiment of the present invention, step S2 dynamically calculates the merging time according to the driving state of the fleet entering the merging control area, including:

[0118] Taking the minimum merging time as the optimization objective and the allowable time range to reach the merging point and the safe merging spacing of the fleet as the constraint conditions, the optimal merging time is obtained by solving.

[0119] The core goal of the merging sequence optimization in this embodiment is to reasonably allocate the merging time of the fleet and ensure that the fleet enters the merging area at the optimal time sequence, thereby improving traffic efficiency and ensuring the orderliness and safety of the merging process. The present invention adopts the following objective function:

[0120]

[0121] in, It represents the merging time assigned to fleet i, that is, the time when the fleet arrives at the merging point.

[0122] In the merging sequence optimization, the first vehicle in each convoy is the pilot vehicle, and its movement behavior directly affects the merging performance and traffic efficiency of the entire convoy. Therefore, it is necessary to set constraints for the vehicles:

[0123]

[0124] in, Indicates that at the current speed and position, the first vehicle in team i passes the maximum acceleration a max The minimum time required to reach the confluence point; Indicates that at the current speed and position, the first vehicle in team i passes the minimum acceleration a min The maximum time required to reach the confluence point; x i (t) is the distance between the first vehicle in fleet i and the merging point at the current moment; v i (t) is the speed of the first vehicle in fleet i at the current moment.

[0125] To avoid potential collision risks between vehicles, the time when the first vehicle in the convoy enters the merging point needs to meet the safety time interval requirements between the first vehicles in other convoys:

[0126]

[0127] Among them, Δt 1 is the minimum allowable safety gap between continuous convoys in the same lane to avoid rear-end collisions; n is the number of vehicles in the convoy; h ... HDV h is the safe following distance for HDV; 1 It is the minimum safe time interval for consecutive vehicles from the same lane to pass the merging point. 2 is the minimum allowable safety gap between continuous convoys in different lanes to avoid rear-end collisions; h 2 It is the minimum safe time interval for consecutive vehicles from different lanes to pass the merging point.

[0128] In an optional embodiment of the present invention, step S2 dynamically calculates the merging sequence according to the driving state of the convoy entering the merging control area, including:

[0129] Construct the current stage state triple according to the number of vehicle platoons with allocated right of way on the main road and ramp and the allocation result of the right of way in the current stage;

[0130] Construct the state transition equation and state space diagram according to the transfer of the current stage state to the next state after making a decision;

[0131] According to the arrival time constraint relationship between two vehicle platoons, take the minimum value of the maximum arrival time in the path from the initial state to the terminal state as the discriminant function, and gradually calculate the local optimal solution of each stage through the recurrence formula to finally obtain the global optimal path.

[0132] In this embodiment, to describe the merging sequence optimization problem, a triple S k (m k ,n k ,r k ) is constructed to represent the state of the current stage, where m k represents the number of vehicle platoons with allocated right of way on the main road, n k represents the number of vehicle platoons with allocated right of way on the ramp, and r k represents the allocation result of the right of way in the current stage, that is, whether to allocate to the main road vehicle platoon or the ramp vehicle platoon.

[0133] The state transition equation describes the transfer of the current state S k (m k ,n k ,r k ) to the next state after making a decision r k . The specific formula is as follows:

[0134] (1) When r k = 1 (the main road vehicle platoon obtains the right of way):

[0135] S k (m k ,n k ,1) = g(S k+1 (m k+1 + 1,n k+1 ,r k+1 ),1)

[0136] (2) When r k = 2 (the ramp vehicle platoon obtains the right of way):

[0137] S k (m k ,n k ,2) = g(S k+1 (m k+1 ,n k+1 + 1,r k+1 ),2)

[0138] To more intuitively display the evolution process of state variables and the state transition relationship in dynamic programming, taking three vehicle fleets on the main line and three vehicle fleets on the ramp as examples, combined with the above state definitions and transition equations, a state space diagram is constructed as shown in Figure 10 shown below.

[0139] During the dynamic programming process, assume that vehicle fleets i and j are respectively assigned the right of way in stage k - 1 and stage k. The arrival times of vehicle fleets j and i satisfy the following relationship:

[0140]

[0141] where r k and r k-1 represent the decision variables in the current stage and the previous stage.

[0142] For state S k , its discriminant function f k (S k ) represents the minimum value of the maximum arrival time in the path from the initial state S 0 to the terminal state S k . The formula is as follows:

[0143]

[0144] where MAT k represents the maximum arrival time from the initial state S 0 to the terminal state S k ; P k-1 represents the path from the previous state S k-1 to the current state S k .

[0145] Based on the Bellman optimality principle, only the local optimal solutions of each stage need to be calculated step by step through the recurrence formula; through the recurrence of the local optimal solutions at each step, the global optimal path can be finally obtained. Therefore, the recurrence formula for the discriminant function value f k (S k ) can be derived through the following process:

[0146]

[0147] where f k represents the discriminant function, S k represents the state of the current stage, m k represents the number of vehicle fleets with the right of way assigned on the main line, n k represents the number of vehicle fleets with the right of way assigned on the ramp, r k represents the allocation result of the right of way in the current stage, min represents the minimum value function, MAT k represents the maximum arrival time from the initial state S0 The maximum arrival time to the terminal state S k , P k-1 represents the path from the previous state S k-1 to the current state S k , where max represents the maximum value function represents the shortest time required for the first vehicle in platoon i at the current vehicle speed and position to reach the merging point with the maximum acceleration a max , Δt 1 represents the minimum allowable safety gap for consecutive platoons in the same lane to avoid rear-end collisions represents the longest time required for the first vehicle in platoon i at the current vehicle speed and position to reach the merging point with the minimum acceleration a min , Δt 2 is the minimum allowable safety gap for consecutive platoons in different lanes to avoid rear-end collisions

[0148] In an alternative embodiment of the present invention, step S2 dynamically adjusts the driving path and speed of the vehicle according to the merging sequence and merging time, including:

[0149] Taking minimizing energy consumption as the optimization objective, constructing a Hamiltonian function according to the objective function and the vehicle dynamics state equation;

[0150] Determining the necessary optimality conditions for each vehicle according to the Pontryagin maximum principle;

[0151] Obtaining the optimal control input according to the Hamiltonian function and the necessary optimality conditions for each vehicle;

[0152] Solving for the speed and position of the vehicle according to the optimal control input and the vehicle dynamics state equation

[0153] This embodiment only considers the influence of longitudinal motion and uses a linear model of vehicle longitudinal dynamics to describe the dynamic characteristics of the vehicle. The dynamic equation can be described by the following state-space expression:

[0154]

[0155] where the matrices A and B represent the state matrix and input matrix of the system respectively, and τ is the first-order inertial time constant of the vehicle. The specific forms are as follows:

[0156]

[0157] The optimization objective is to minimize the integral of the square of the vehicle's acceleration, thereby reducing the vehicle's energy consumption and improving the driving smoothness. The objective function can be expressed as:

[0158]

[0159] In the trajectory optimization problem, the present invention uses the Hamiltonian Function to derive an analytical solution for the objective function.

[0160] H i (x i (t), u i (t)) = L i (x i (t), u i (t)) + λ T f i (x i (t), u i (t))

[0161] Where x i (t) and u i (t) are the state variable and control input of vehicle i at time t respectively, and λ is the co-state variable. According to the objective function and the vehicle dynamics state equation, the Hamiltonian function expression can be sorted out as:

[0162]

[0163] Where λ 1 and λ 2 are the co-state variables respectively. According to Pontryagin's Maximum Principle, the necessary condition for optimality of each vehicle is:

[0164]

[0165] According to the above formula, it can be deduced that The optimal control input u i (t) of the function has been obtained, and its form is:

[0166]

[0167] Substituting this control input into the vehicle dynamics model, the analytical expressions of the vehicle speed v i (t) and position p i (t) over time can be further solved, which are respectively:

[0168]

[0169]

[0170] Where represents the speed of platoon i at the current time t, b i , c i , d i , e irepresents the state constant, Represents the position of team i at the current time t.

[0171] The state constants can be solved by the simultaneous equations of the initial state and terminal state of each vehicle.

[0172]

[0173] in, represents the target merging time of fleet i; and are respectively the target merging time of fleet i Corresponding speed and position; and is the initial time for team i Time speed and position.

[0174] In order to verify the method of the present invention, an implementation case is given below to analyze the model and algorithm. SUMO and Python are used for joint development to verify the ramp merging control scheme. In the simulation experiment, in order to evaluate the performance of the formation control combined with the dynamic programming method proposed in the present invention, three control strategies are designed for comparative analysis: 1) First-in-first-out strategy (FIFO). No control is performed, as a comparison benchmark; 2) Dynamic programming (DP) strategy. No formation is performed, only the CAV is controlled; 3) Collaborative optimization strategy. A collaborative optimization method that combines formation control and dynamic programming. The three methods are simulated and analyzed for key indicators such as average vehicle speed and average delay in the ramp merging area. In order to verify the computational efficiency, an optimization method based on the Gurobi solver is also introduced, and a comparative analysis is performed with the dynamic programming method.

[0175] In order to ensure the continuous traffic flow of the simulated road section, the present invention constructs a vehicle generation model, which is expressed as follows:

[0176]

[0177] Among them, C mix represents the theoretical capacity of mixed traffic (unit: veh / h / lane); ρ represents the penetration rate of CAV; t c ,t a and t h They represent the expected time interval of CAV, the expected time interval between CAV and HDV, and the expected time interval of HDV, and their values ​​are 0.6s, 1.1s, and 1.5s respectively; e Indicates the equilibrium speed, usually the maximum speed on the main road; s 0、L represent the vehicle stationary interval and vehicle length respectively, with values of 2m and 4.5m in the present invention. When the penetration rate ρ of CAVs is 40%, 60% and 80% respectively, the theoretical traffic capacity C of the main line mix is 2368, 2686 and 3214 veh / h respectively.

[0178] The specific simulation parameter settings are shown in Table 2.

[0179] Table 2 Simulation experiment parameters

[0180]

[0181] (1) Analysis of traffic efficiency

[0182] According to Figure 11 、 Figure 12 、 Figure 13 and Figure 14 results, the cooperative optimization control performs optimally in all scenarios. Compared with the FIFO strategy, the speed improvement range is 10.4% - 61.5%; compared with the dynamic programming strategy, the improvement range is 8.2% - 37.5%; the average delay reduction ratios are 60% - 91.3% (compared with FIFO) and 61.9% - 86.6% (compared with dynamic programming) respectively, showing excellent efficiency and stability. Generally speaking, the cooperative optimization control realizes comprehensive optimization in two key indicators of speed and delay. Especially under high traffic volume and high penetration rate conditions, it shows significant efficiency improvement and delay reduction effects. The excellent performance of this control method in complex traffic scenarios proves its practical application value, becoming the optimal solution to improve traffic efficiency and reduce delays, and has great development potential.

[0183] (2) Analysis of algorithm efficiency

[0184] In ramp merging control, the real-time performance of the algorithm is a key factor determining its applicability and system performance. The experiment takes the flow distribution ratio (65 - 35 and 80 - 20) and penetration rate (40%, 60%, 80%) as variables, and selects 90% traffic volume as the analysis condition. In order to more intuitively compare the computational performance of the two methods, the figure uses logarithmic coordinates to show the trend of calculation time changing with the penetration rate, so as to evaluate the applicability and real-time performance of dynamic programming and Gurobi solver in different scenarios.

[0185] From Figure 15It can be seen that although both dynamic programming and Gurobi can achieve the global optimum in terms of the solution results, in terms of computational efficiency, dynamic programming shows obvious real-time advantages under high penetration conditions. Especially when the penetration rate reaches 80%, the computational time of dynamic programming is shortened by about 20 times compared with Gurobi, significantly improving the system response speed and making it more suitable for the ramp merging control scenario with high real-time requirements. In contrast, Gurobi has a long computational time, especially in a high-penetration environment, making it difficult to meet the real-time control requirements and restricting its application in dynamic traffic management systems. Therefore, from the perspective of real-time performance, the dynamic programming method has stronger practical feasibility in traffic management scenarios that require rapid response and efficient processing.

[0186] Compared with the prior art, the present invention significantly improves the real-time adaptability of the system, the dynamic interaction ability of the merging sequence and trajectory planning, and at the same time reduces the computational complexity, resulting in obvious technical effects. The following is a specific description in combination with the key technical solutions:

[0187] (1) Improve real-time adaptability: The present invention dynamically adjusts the control strategy of vehicle formations through a PID controller that combines the particle swarm optimization (PSO) algorithm and the simulated annealing (SA) algorithm. This technical solution can obtain the dynamic changes of traffic flow in real time, quickly adjust the speed and trajectory of vehicles, and ensure that the system can respond quickly to sudden traffic conditions. Through this adaptive control strategy, the system can efficiently cope with different traffic flow conditions, ensuring the stability and safety of traffic flow, thereby improving real-time adaptability.

[0188] (2) Optimize the dynamic interaction of the merging sequence and trajectory planning: In the merging scenario, the present invention combines the dynamic programming (DP) algorithm and the Pontryagin maximum principle (PMP) to optimize the collaborative work of vehicle merging sequences and trajectory planning. This technical solution accurately calculates the merging time and the optimal trajectory of each vehicle fleet, enabling vehicles to smoothly connect within the merging area, minimizing conflicts and passing delays between vehicles. For example, the interaction relationship between CAVs and HDVs is fully considered, and CAVs can adjust their speed and trajectory according to the behavior of surrounding HDVs, thereby achieving more efficient and safe merging. This optimization of dynamic interaction greatly improves the passing efficiency during the merging process, especially in a complex traffic flow environment, avoiding the complex dynamic interaction problems that cannot be handled by traditional methods when vehicle fleets merge.

[0189] (3) Reducing computational complexity and enhancing real-time performance: Through the dynamic programming (DP) algorithm, the present invention significantly reduces the computational complexity while ensuring the optimization quality. Traditional optimization methods usually require a large amount of computational resources and are difficult to respond quickly in high-density and complex traffic flow environments. The DP algorithm of the present invention quickly solves the merging sequence and merging time through efficient state space search and strategy planning, effectively reducing the computational burden of the system. For example, under high traffic density conditions, the system can still quickly obtain the optimal merging decision, ensuring real-time performance in the case of large traffic flow and complex traffic conditions. This technical solution enables the system to meet the real-time requirements, provide rapid decision-making support, and solve the problems of excessively high computational complexity and slow response in the prior art.

[0190] Specific embodiments are applied in the present invention to elaborate on the principles and implementation manners of the present invention. The description of the above embodiments is only used to help understand the method and its core idea of the present invention; at the same time, for those of ordinary skill in the art, based on the idea of the present invention, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present invention.

[0191] Those of ordinary skill in the art will realize that the embodiments described herein are for helping the reader understand the principles of the present invention, and it should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those of ordinary skill in the art can make various other specific deformations and combinations that do not depart from the essence of the present invention based on the technical revelations disclosed in the present invention, and these deformations and combinations are still within the protection scope of the present invention.

Claims

1. A highway ramp merging optimization control method for mixed traffic flow, characterized in that: The following steps are involved: Different control methods are selected based on the status information of vehicles entering the formation control area, and the intelligent networked vehicles are used to indirectly guide the manually driven vehicles to form a mixed formation mode, and the driving status of all vehicles in the fleet is monitored in real time; The merging sequence and merging time are dynamically calculated according to the driving status of the convoy entering the merging control area, and the driving path and speed of the vehicles are dynamically adjusted according to the merging sequence and merging time.

2. The highway ramp merging optimization control method for mixed traffic flow according to claim 1 is characterized in that: Different control methods are selected according to the status information of the vehicle entering the formation control area, including: Obtain status information of vehicles entering the formation control area; Determine whether the current vehicle has reached the formation trigger point; if so, proceed to the next step and monitor the status information of the fleet in real time; otherwise, continue to monitor the status information of the vehicle; Determine whether the current vehicle type is an intelligent connected vehicle; if so, proceed to the next step; otherwise, adopt the FVD following model to follow the vehicle; Determine whether the vehicle behind the current vehicle is a manually driven vehicle; if so, use a PID controller to actively decelerate the current vehicle, and use the FVD following model to follow the vehicle; otherwise, proceed to the next step; Determine whether the type of the preceding vehicle of the current vehicle is an intelligent connected vehicle; if so, adopt the CACC following model for the current vehicle to follow the vehicle; otherwise, adopt the ACC following model for the current vehicle to follow the vehicle.

3. The highway ramp merging optimization control method for mixed traffic flow according to claim 2 is characterized in that: When the current vehicle uses a PID controller for active deceleration, the PID controller is optimized by combining the particle swarm algorithm and the simulated annealing method. By comparing the fitness value of the current solution with the current global optimal solution, it is decided whether to accept the current solution, which is expressed as: Among them, p represents the acceptance probability, G best represents the global optimal solution, J represents the fitness value, T represents the simulated annealing temperature, T = max(T·α,T min ), max represents the maximum value function, T min represents the minimum simulated annealing temperature, and α represents the temperature attenuation coefficient.

4. The highway ramp merging optimization control method for mixed traffic flow according to claim 2 is characterized in that: Determine whether the fleet size exceeds the maximum fleet size based on the fleet status information; if so, trigger a new fleet for the exceeding vehicles and determine the length of the current fleet; otherwise, continue to monitor the fleet status information.

5. The highway ramp merging optimization control method for mixed traffic flow according to claim 1, characterized in that: The merging time is dynamically calculated based on the driving status of the convoy entering the merging control area, including: Taking minimizing the merging time as the optimization goal, and taking the allowed time range for reaching the merging point and the safe merging distance of the fleet as constraints, the optimal merging time is solved.

6. The highway ramp merging optimization control method for mixed traffic flow according to claim 5, characterized in that: The optimization goal is to minimize the merging time, and the constraints are the allowed time range to reach the merging point and the safe merging distance of the fleet, specifically: Among them, min represents the minimum value function, and max represents the maximum value function. represents the merging time assigned to fleet i, Indicates that the first vehicle in team i passes the maximum acceleration a at the current speed and position max The minimum time required to reach the confluence point; Indicates that the first vehicle in fleet i passes the minimum acceleration a at the current speed and position min The maximum time required to reach the confluence point, v max Indicates the maximum speed, v i (t) is the speed of the first vehicle in fleet i at the current time t, x i (t) is the distance from the first vehicle in platoon i to the merging point at the current time t, x(t) represents the vehicle position at the current time t, t i represents the merging time of the first vehicle in fleet i, t i-1 represents the merging time of the second vehicle in platoon i, Δt1 represents the minimum allowable safe gap between consecutive platoons in the same lane to avoid rear-end collision, h1 represents the minimum safe time interval for consecutive vehicles from the same lane to pass the merging point, n represents the number of vehicles in the platoon, and h HDV represents the safe following distance of HDV, t j represents the merging time of the first vehicle in platoon j, Δt2 represents the minimum allowable safe gap between continuous platoons in different lanes to avoid rear-end collision, and h2 represents the minimum safe time interval between continuous vehicles in different lanes passing the merging point.

7. The highway ramp merging optimization control method for mixed traffic flow according to claim 1, characterized in that: The merging sequence is dynamically calculated based on the driving status of the convoy entering the merging control area, including: The current stage state triplet is constructed according to the number of fleets that have been allocated the right of way on the main road and ramp and the allocation result of the right of way in the current stage; Construct state transition equations and state space diagrams based on the current stage state moving to the next state after making a decision; According to the arrival time constraint relationship between the two fleets, the minimum value of the maximum arrival time in the path from the initial state to the terminal state is used as the discriminant function. The local optimal solution of each stage is gradually calculated through the recursive formula, and finally the global optimal path is obtained.

8. The highway ramp merging optimization control method for mixed traffic flow according to claim 7, characterized in that: The local optimal solution of each stage is calculated step by step through the recursive formula: Among them, f k represents the discriminant function, S k Indicates the current stage status, m k Indicates the number of convoys that have been assigned the right of way on the main road, n k represents the number of fleets that have been assigned the right of way on the ramp, r k Indicates the allocation result of the right of way in the current stage, min indicates the minimum value function, MAT k Represents the transition from the initial state S0 to the terminal state S k The maximum arrival time, P k-1 Indicates that from the previous state S k-1 To the current state S k The path of max represents the maximum value function. Indicates that the first vehicle in team i passes the maximum acceleration a at the current speed and position max The shortest time required to reach the merging point, Δt1 represents the minimum allowable safety gap between consecutive convoys in the same lane to avoid rear-end collisions, Indicates that the first vehicle in fleet i passes the minimum acceleration a at the current speed and position min The maximum time required to reach the merging point, Δt2 is the minimum allowable safety gap between consecutive platoons in different lanes to avoid rear-end collisions.

9. The highway ramp merging optimization control method for mixed traffic flow according to claim 1, characterized in that: Dynamically adjust the vehicle's path and speed based on the merging sequence and merging time, including: Taking minimizing energy consumption as the optimization goal, the Hamiltonian function is constructed according to the objective function and the vehicle dynamics state equation; Determine the necessary conditions for optimality for each vehicle based on the Pontryagin maximum principle; The optimal control input is obtained according to the Hamiltonian function and the necessary optimality conditions for each vehicle; The vehicle's velocity and position are solved based on the optimal control input and the vehicle dynamics state equations.

10. The highway ramp merging optimization control method for mixed traffic flow according to claim 9, characterized in that: The speed and position of the vehicle are solved according to the optimal control input and the vehicle dynamics state equation as follows: in, represents the speed of team i at the current time t, b i , c i , d i , e i represents the state constant, Represents the position of team i at the current time t.

Images