Vehicle team adaptive cruise control method

Abstract

The invention discloses a motorcade self-adaptive cruise control method, which comprises the following steps of: A, establishing a dynamic safety demand distance model used for describing the demand of an intelligent network connection driving queue for a safety distance in a network connection environment; in combination with speed and acceleration information of front vehicles in the queue and influences of different positions of the vehicles in the queue, an information transmission coefficient and a position influence coefficient are introduced, and a dynamic safety demand distance strategy is established; b, improving a cooperative adaptive cruise control (CACC) model based on the dynamic safety demand distance model, and updating a time headway strategy into a dynamic safety demand distance strategy; introducing speed and interval delay to simulate a time delay relation between upper-layer planning and bottom-layer control; c, establishing a cooperative lane changing control strategy based on a self-adaptive spacing strategy, setting a slack spacing, a recovery spacing and a balance spacing according to different stages of forced lane changing of the vehicle, and performing self-adaptive change according to a lane changing stage indication variable; and constructing a collaborative lane changing optimization model to realize queue centralized control.

Description

Technical Field

[0001] The present invention relates to the technical field of fleet cruise control, and specifically designs an adaptive cruise control and cooperative lane-changing control method for a fleet of autonomous driving vehicles based on a dynamic safety requirement distance. Background Art

[0002] With the advancement of intelligent interconnection and autonomous driving technologies, the emergence of connected vehicles (CAVs) is widely recognized as having significant potential in improving traffic safety, efficiency and fuel economy. Among them, CAVs form a platoon called a connected truck platoon (CATP), which is regarded as a multi-agent system with a four-element architecture (node ​​dynamics, information flow topology, platoon geometry, and distributed controller). It uses real-time information interaction to accurately perceive a large-scale dynamic traffic environment and realizes platoon operation through driving model drive and collaborative control. Control methods such as collaborative adaptive cruise control, speed control, feedback control, and heuristic control are usually applied to vehicle control. However, CATP inevitably faces various complex traffic conditions, including acceleration / deceleration and braking in basic sections, lane changing and overtaking in interweaving areas. Therefore, from the perspective of the control system, more appropriate control strategies are needed to improve the stability of the platoon under interference.

[0003] The disadvantages of the prior art are as follows:

[0004] 1. Insufficient flexibility and adaptability: Existing headway strategies and truck platoon longitudinal control algorithms perform poorly in complex and dynamic traffic environments. For example, although the constant spacing (CS) strategy can ensure the stability of the platoon, it does not perform well in dynamic traffic environments and lacks the flexibility to cope with complex driving conditions. Similarly, proportional differential control (PD) and model predictive control (MPC) algorithms also have limited performance in the face of external disturbances and system uncertainties.

[0005] 2. Insufficient use of information in the connected environment: Existing headway strategies such as CTH and VTH fail to fully utilize the information transmission characteristics in the intelligent connected environment and only consider the direct information of the preceding vehicle. This results in these strategies failing to fully utilize the full vehicle queue information perceived in the connected environment, thus limiting the rapid response and precise control of vehicle dynamic changes.

[0006] 3. Low real-time performance: In the study of longitudinal control of truck platoons, although some advanced control algorithms such as MPC and machine learning algorithms can theoretically provide higher accuracy and control effects, they usually perform poorly in real-time applications. The high computational complexity and resource requirements make it difficult for these algorithms to be effectively applied in traffic control with high real-time requirements.

[0007] 4. Insufficient dynamic adaptability: Existing studies have focused on the lane-changing timing and the process of creating lane-changing gaps, ignoring the impact of different stages of lane-changing in the diverging area on CATP. Existing lane-changing control strategies fail to fully consider the adaptive needs of different stages and often cannot effectively respond to dynamic changes in the lane-changing stage, resulting in deficiencies in adjusting the safe distance between vehicles and maintaining queue stability during the lane-changing process. Summary of the invention

[0008] The objectives of the present invention are achieved through the following technical solutions.

[0009] Specifically, the present invention provides a fleet adaptive cruise control method, comprising:

[0010] Step A: Establish a dynamic safety requirement distance model to describe the safety distance requirements of the intelligent connected driving queue in the connected environment; combine the speed and acceleration information of the front vehicle in the queue, and the influence of the different positions of the vehicle in the queue, introduce the information transfer coefficient and position influence coefficient, and establish a dynamic safety requirement distance strategy;

[0011] Step B: Based on the dynamic safety requirement distance model, the collaborative adaptive cruise control model CACC is improved, and the headway strategy is updated to the dynamic safety requirement distance strategy; speed and distance delay are introduced to simulate the time lag relationship between the upper-level planning and the bottom-level control;

[0012] Step C: Establish a cooperative lane-changing control strategy based on the adaptive spacing strategy, set the relaxation spacing, recovery spacing, and equilibrium spacing for different stages of forced lane changing of vehicles, and make adaptive changes according to the indicator variables of the lane-changing stage; based on the needs of different stages in the lane-changing process, analyze the relationship between the sub-goals, establish the weight distribution of the objective function of each stage, and build a cooperative lane-changing optimization model in combination with the constraints of the vehicles in the queue to achieve centralized queue control.

[0013] The advantages of the present invention are:

[0014] 1. The present invention proposes a dynamic safety requirement spacing strategy for CATP. The traditional headway strategy only considers the information of the immediate preceding vehicle and fails to make full use of the information obtained in the networked environment. Therefore, the present invention analyzes the correlation between the preceding vehicle and the speed of the target vehicle, proposes the information transmission coefficient and the position influence coefficient, and establishes a dynamic safety headway strategy, which can perceive the speed information of all vehicles in front of the queue and improve the response speed of driving behavior.

[0015] 2. The present invention establishes an improved CACC control method based on dynamic safety requirement distance, and considers model stability. MPC is used as the upper-level planner to dynamically adjust control parameters according to objectives such as trajectory tracking, fuel economy and stability. PD control is used as the underlying control model to achieve longitudinal control under vehicle dynamic constraints. Existing studies generally ignore the time lag relationship between the upper-level planning model and the underlying controller. However, feedback delay has an important impact on queue stability and may even cause system instability. Based on the frequency domain analysis method, the present invention introduces speed delay and spacing delay, constructs a system transfer function, derives queue stability constraints, incorporates queue stability into sub-goals, and achieves stable driving of queues under complex conditions by dynamically adjusting parameters.

[0016] 3. The present invention proposes an adaptive collaborative control strategy suitable for CATP for lane changing of vehicles in the diversion area. Existing studies have focused on the lane changing time and the process of creating lane changing gaps, ignoring the impact of different stages of lane changing in the diversion area on CATP. Therefore, the present invention proposes an adaptive spacing strategy, including relaxation spacing, recovery spacing and equilibrium spacing, which can create spacing in time when sensing a lane-changing vehicle and quickly restore the stable operation of the queue after the vehicle leaves the main line, thereby reducing the impact of forced lane changing in the diversion area on the queue's own benefits and the overall benefits of traffic flow. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0018] Figure 1 A schematic diagram of the driving condition of a CATP pilot vehicle according to an embodiment of the present invention is shown.

[0019] Figure 2 A schematic diagram comparing driving behaviors with different headway strategies according to an embodiment of the present invention is shown.

[0020] Figure 3 A comparative schematic diagram showing the improvement trend of the DSRTH strategy according to an embodiment of the present invention is shown.

[0021] Figure 4 A schematic diagram comparing driving characteristics of different car-following models according to an embodiment of the present invention is shown.

[0022] Figure 5 A schematic diagram showing a comparison of improvement effects of the improved CACC model according to an embodiment of the present invention is shown.

[0023] Figure 6A schematic diagram of traffic flow rate results under different vehicle type ratios according to an embodiment of the present invention is shown.

[0024] Figure 7 A schematic diagram of analyzing the impact of queue size on traffic efficiency according to an embodiment of the present invention is shown.

[0025] Figure 8 A schematic diagram of vehicle trajectories with different queue sizes according to an embodiment of the present invention is shown.

[0026] Fig. 9 A schematic diagram of a diversion area simulation environment according to an embodiment of the present invention is shown.

[0027] Fig.10 A schematic diagram showing the degree of improvement of average travel time according to an embodiment of the present invention is shown.

[0028] Fig.11 A schematic diagram showing the degree of improvement of the average driving speed according to an embodiment of the present invention is shown.

[0029] Fig.12 A schematic diagram of average travel time distribution for different queue sizes according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0030] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0031] A cooperative autonomous driving truck platoon refers to a group of trucks traveling at a small distance from each other, which can improve the fuel economy and safety of the platoon, while also improving traffic efficiency. However, factors such as speed changes within the truck platoon and external traffic flow speed disturbances can easily lead to platoon instability, resulting in reduced traffic efficiency. In order to optimize the speed control of a cooperative autonomous driving truck platoon, and better improve the stability, safety, and traffic efficiency of basic sections and diversion areas of highways, the present invention proposes an improved adaptive cruise control model for a cooperative autonomous driving truck platoon, and proposes a cooperative lane change control strategy for vehicles forced to change lanes in the diversion area. The model aims to ensure safety while improving the stability, fuel economy, and traffic efficiency of the truck platoon, while also improving the overall traffic flow status.

[0032] In order to solve the problems mentioned in the prior art, the present invention proposes a dynamic safety requirement distance, and based on this, proposes an improved adaptive cruise control model, and proposes a cooperative lane change control strategy for vehicles that are forced to change lanes in the diverging area. A series of technical solutions of this method are as follows:

[0033] (1) Establish a dynamic safety requirement distance model to describe the safety distance requirements of intelligent connected driving queues in a connected environment. The target distance of the following vehicle is an important manifestation of ensuring safety and pursuing efficiency. Combined with the speed, acceleration and other information of the front vehicle in the queue, as well as the influence of different positions of the vehicle in the queue, the information transmission coefficient and position influence coefficient are introduced to establish a dynamic safety requirement distance strategy. The specific steps are as follows:

[0034] Step 1: Remember that the VTH spacing strategy is as follows:

[0035] X des =(t0-c r ·υ r )·υ k

[0036] Among them, X des is the desired safety distance; v k is the speed of the following vehicle k; t0 is the headway time; c r is the coefficient to be calibrated; v r is the speed difference between the front and rear vehicles. This strategy is suitable for homogeneous manually driven vehicles. In a connected environment, the use of existing headway strategies will lead to excessive headway, resulting in a waste of road resources. It will also increase the frequency of frequent merging of vehicles in adjacent lanes, resulting in the splitting of the queue, thereby reducing the efficiency of information transmission. Therefore, considering the information of other vehicles within the vehicle communication range is of great significance for the spacing strategy.

[0037] Step 2: Introduce information perception coefficient and position influence coefficient:

[0038] Since CAV can obtain real-time acceleration information of surrounding vehicles through communication and perception technology, the headway strategy can be improved by introducing information perception coefficient and position influence coefficient to adapt to the vehicle characteristics of CATP.

[0039] Due to the restrictions of information transmission rules, some vehicles may be beyond the communication range or do not have the characteristics of network communication. In order to make better use of this information, the information perception coefficient σ is introduced k , reflecting the topological structure of information transmission between vehicles. k The value of is 0 or 1, where σ k =1 means that the status information of vehicle k can be obtained, σ k =0 means that the vehicle k status information cannot be obtained.

[0040] A study shows that the farther the front vehicle is from the target vehicle, the less influence it has on the target vehicle. In order to analyze the specific influence of the front vehicle on the target vehicle, the correlation analysis of the status information of vehicles at different positions in the queue and the target vehicle is performed based on the intelligent driving data set ACC Data. The results show that the farther the vehicle is from the vehicle, the less influence it has on the vehicle speed, and the results are significant. Therefore, considering the vehicle position, the position influence coefficient is introduced, expressed as θ k , the value range of this coefficient is (0, 1), which is the degree of influence of vehicle k on the target vehicle.

[0041] The Dynamic Safety Requirement Time Headway (DSRTH) strategy is established as follows:

[0042]

[0043] In the formula, v r，k represents the speed difference between the target vehicle and the preceding vehicle; σ j is the information perception coefficient of vehicle j; θ j is the position influence coefficient of vehicle j; a j is the acceleration of vehicle j; t0, c v and c a is the parameter to be calibrated, c v represents the influence of the speed difference with the preceding vehicle, c a Represents the influence of the acceleration of the vehicle in front; m is the vehicle number at the end of the queue.

[0044] Step 3: Parameter calibration:

[0045] DSRTH strategy involves t0, c v and c a There are three parameters to be calibrated, among which t0 has the same meaning as the headway time, c v represents the influence of the speed difference with the preceding vehicle, c a Indicates the influence of the acceleration of the vehicle in front. The model parameters are calibrated based on ACCData, and the experiment lasts for 289.7 seconds. The dataset consists of 2897 frames, including the speed and spacing information of the vehicles. All frames in the dataset are used for calibration. The particle swarm optimization algorithm (PSO) is used as the calibration method, and the root mean square error (RMSE) is selected to measure the error between the actual spacing and DSRTH. The calibration results are as follows:

[0046]

[0047] (2) Improve the Cooperative Adapted Cruise Control (CACC) model based on the dynamic safety requirement distance. Based on the existing CACC model, the headway strategy is updated to the dynamic safety requirement distance strategy. Considering the impact of time lag on system stability, speed and distance delays are introduced to simulate the time lag relationship between upper-level planning and underlying control. In order to improve fuel economy, tracking accuracy and stability, model predictive control (MPC) is used to dynamically adjust the underlying controller parameters. The specific steps are as follows:

[0048] Step 1: Improved CACC model considering time delay and dynamic safety demand distance:

[0049]

[0050] In the formula, e k is the gap error; x k-1 and x k are the longitudinal positions of vehicles k-1 and k respectively; v k is the speed of the target vehicle; T is the headway; Δt is the update step size; k p and k d is the control coefficient; L k is the vehicle length; t1 and t2 represent the spacing delay and speed delay, respectively.

[0051] Step 2: Dynamic parameter adjustment based on MPC:

[0052] Combining the constraints and the objective function, an optimization model is established under the MPC framework to dynamically adjust the PD control parameters. This method can predict the behavior of the queue within a certain time range in the future and optimize the driving behavior of the target vehicle based on the prediction results. The iteration step size of this model is 0.1s, and the prediction step size is set to N p =10, the control step size is set to N c = 5. Use linear prediction to predict the vehicle's future N p The vehicle status within the step is predicted. The model is established as follows:

[0053]

[0054] Where J is the objective function, ρ v is the speed tracking weight coefficient, ρ x is the trajectory tracking weight coefficient, ρ c is the fuel economy weight coefficient, ρ s is the queue stability weight coefficient, v k is the speed of the target vehicle, Xdes is the desired safety distance, x k-1 and x k represents the longitudinal position of vehicles k-1 and k, L k is the vehicle length, σ k is the information perception coefficient of vehicle k.

[0055]

[0056] In the formula, x k-1 and x k represents the longitudinal position of vehicles k-1 and k, d safety is the safety distance; v k is the speed of the target vehicle, a k represents the acceleration of vehicle k; a max is the maximum acceleration of the vehicle, b max is the maximum deceleration of the vehicle, and the parameter a in the model max and b max All are set to 3m / s 2 ; Δt is the update step size; k p and k d is the control coefficient; ω is the angular frequency.

[0057] Taking into account the dynamic changes of the vehicle and the environment, the state variables are defined as follows:

[0058] z=[xυa]

[0059] The state update is defined as follows:

[0060]

[0061] (3) Establish a collaborative lane-changing control method based on an adaptive spacing strategy. Considering that some vehicles in the highway diversion area have the need to leave the main line, a collaborative lane-changing strategy is established to actively create gaps for vehicles to make forced lane changes, while optimizing speed control to reduce the impact on traffic flow. For different stages of forced lane changes, relaxation spacing, recovery spacing, and equilibrium spacing are proposed, and adaptive changes are made according to the indicator variables of the lane-changing stage. Based on the needs of different stages in the lane-changing process, the relationship between the sub-goals is analyzed, the weight distribution of the objective function of each stage is established, and the constraints of the vehicles in the queue are combined to construct a collaborative lane-changing optimization model to achieve centralized queue control. The specific steps are as follows:

[0062] Step 1: Adaptive Spacing Strategy:

[0063] The lane-changing process is divided into three stages: relaxation, equilibrium, and recovery, and the corresponding spacing is constructed.

[0064] Relaxation spacing:

[0065]

[0066] In the formula, is the remaining distance for vehicle k to change lanes; is the number of lane changes; x lc is the difference between the lane change sign position and the lane change point position; Defines the desire of lane-changing vehicle k to leave the current lane, The lane-changing pressure of vehicle a at vehicle k is , where vehicle a is the adjacent rear vehicle in the target lane of vehicle k; X des，a is the desired safety distance of vehicle a, X relax，k Create a safe gap for CATP to change lanes.

[0067] Balanced Spacing:

[0068]

[0069] In the formula is the expected return to equilibrium distance of vehicle k, where sk is the distance between vehicle k and the preceding vehicle, X ave is the queue equilibrium spacing, n is the queue leader vehicle number, and m is the queue end vehicle number.

[0070] Restore spacing:

[0071]

[0072] In the formula is the expected recovery distance that vehicle k expects to achieve; s k+1 is the distance between the kth vehicle and the following vehicle k+1, X ave is the queue equilibrium distance, X des，k is the desired safety distance of vehicle k.

[0073] In order to determine the stage of the CAT vehicle in the lane change coordination process, an indicator variable is introduced to describe the progress of the vehicle in the coordinated lane change. The specific meanings are as follows:

[0074]

[0075] The relationship between the current spacing and the required spacing is used as the basis for judgment to establish an adaptive adjustment mechanism. The algorithm flow is as follows:

[0076]

[0077] Step 2: Collaborative lane-changing optimization model:

[0078]

[0079] In the formula, s kis the distance between the target vehicle and the preceding vehicle; v eff is the expected road speed; β safe , β fuel , β eff , β fol , β eq is the sub-objective weight of the objective function; N p is the prediction step length of model predictive control; S ave is the equilibrium distance between the target vehicle and the preceding vehicle.

[0080] The goal is to minimize the joint cost function composed of the safety, fuel, efficiency, tracking, and balance costs of all vehicles. Since the objective function includes multiple goals such as safety, fuel, efficiency, tracking, and balance, the degree of expectation of vehicles for the goals varies at different stages of the lane change. Therefore, Multi-Criteria Decision Analysis (MCDA) taking into account prior knowledge is selected, which is suitable for multi-objective optimization solution in the collaborative lane change stage. Based on the theory of multi-criteria decision analysis, combined with the preference for the goals at different stages of the lane change, the weight coefficient is determined, and the coefficient value is as follows:

[0081]

[0082] In the following stage, since the distance between vehicles can be kept safe and the probability of danger in the following state is small, a smaller safety weight coefficient is used. At the same time, in the following state, the weight coefficient of fuel is increased to reduce the amplitude of vehicle speed changes, try to keep it stable, and improve the fuel economy of the queue. In addition, the overall traffic efficiency can be improved by adjusting the free flow speed of tracking. In this state, the queue itself maintains a relatively uniform small distance, and there is no need to pay attention to the balance of the queue.

[0083] During the gap creation phase, the queue needs to accelerate and decelerate to create a larger gap for vehicles changing lanes. The distance between vehicles will drop sharply at the moment of lane change, so it is necessary to increase the focus on safety and reduce the pursuit of efficiency targets (free flow speed) in order to create gaps by decelerating.

[0084] During the gap recovery phase, a large gap appears in the queue. Therefore, vehicles behind the large gap need to speed up as quickly as possible to shorten the gap and reach an equilibrium gap. At this time, increasing the equilibrium target weight can improve the recovery speed during optimization.

[0085] In the equilibrium stage, since the equilibrium interval is for the front vehicle with a large interval, tracking the expected interval is no longer the main goal, but the goal is to restore the vehicle interval to the equilibrium interval as soon as possible. In this stage, the value of the equilibrium weight is increased.

[0086] The lane-changing phase indicator variable switches the weight coefficients at different phases to achieve adaptive variable weights. This adaptive variable weight method can adjust the weights of different objectives according to different lane-changing phases to better reflect the current decision-making environment and needs, improve the flexibility and adaptability of decision-making, and enable CATP to better respond to different decision-making situations and changes.

[0087] The beneficial effects brought by the technical solution of the present invention are as follows:

[0088] 1) In order to describe the safety distance requirements of intelligent connected driving queues, a dynamic safety requirement distance strategy was established. Compared with the traditional constant headway strategy and variable headway strategy, the queue recovery stabilization time was shortened by 12.3% and 4.5%, respectively, verifying the advantages of the dynamic safety requirement distance strategy in response speed and queue recovery stability performance.

[0089] 2) An improved Cooperative Adapted Cruise Control (CACC) model is proposed. The headway strategy is updated to a dynamic safety requirement distance strategy. Considering the impact of time lag on system stability, speed and spacing delays are introduced to simulate the time lag relationship between upper-level planning and underlying control, and the queue stability judgment condition is derived based on the frequency domain analysis method. In order to improve fuel economy, tracking accuracy and stability, the Model Predictive Control (MPC) is used to dynamically adjust the underlying controller parameters. Compared with the Intelligent Driver Model (IDM) model and the CACC model before improvement, the time for the queue to recover stability is shortened by 19.4% and 3.2%.

[0090] 3) Establish an adaptive spacing strategy for forced lane-changing behavior, and propose a multi-objective optimized coordinated lane-changing control method. It can effectively advance the lane-changing time and quickly restore the stable operation of the queue after completing the lane change, avoiding queue segmentation caused by other vehicles changing lanes and entering the queue, which has a positive impact on traffic flow efficiency. The improvement effect is particularly significant when the traffic flow is large. When the traffic flow is 1800veh / h, the average driving time on the off-ramp is shortened by 19.6%, and the average driving speed is increased by 10.3%.

[0091] Example

[0092] 1) Evaluation of the effectiveness of dynamic safety demand distance strategy

[0093] In order to analyze the model effect of the dynamic safety demand distance strategy applied to the speed change scenario, the reaction speed, speed change duration and stability of CATP were used to evaluate the effect of this strategy. The experimental environment used the speed of the pilot car in ACC Data as the execution condition, and selected a period of 50 seconds to 100 seconds, during which the speed changed significantly. The speed, acceleration and displacement changes of the pilot car are as follows Figure 1 shown.

[0094] In order to verify the effectiveness of the DSRTH strategy, IDM is used as the vehicle following model, and the DSRTH strategy results are compared with the CTH and VTH strategies. The headway strategy parameters in the table below are the results after calibration with the ACC Data dataset.

[0095] Different headway strategy parameter values

[0096]

[0097] In order to quantitatively analyze the characteristics of different headway strategies in terms of reaction speed, speed change duration and stability, three characteristics are used for evaluation: the time when the vehicle speed valley occurs, the speed valley, and the time when the acceleration returns to 0. The speed and expected spacing of vehicles at different positions in the queue are as follows Figure 2 shown.

[0098] By quantitatively analyzing the time when the vehicle speed valley occurs, the speed valley, and the time when the acceleration returns to 0, the following table shows the degree of improvement of the DSRTH strategy. Comparative analysis of driving behaviors at different stages:

[0099] Speed ​​maintenance period: The period from 50 to 60 seconds represents the driving condition of constant speed. Since the acceleration of the leading vehicle is zero during this period, it is not affected by the acceleration and speed difference of the leading vehicle. The three strategies maintain a consistent speed and are affected by t0, resulting in the difference in expected headway distance. The results show that the DSRTH strategy has the shortest expected headway distance and can maintain a smaller gap during the speed stage.

[0100] Comparison of evaluation indicators of different headway strategies

[0101]

[0102]

[0103] Deceleration-Acceleration Period: The period of 60 to 80 seconds represents the scenario of acceleration after deceleration. This scenario is mainly used to compare the response and tracking capabilities of different strategies to the rapid speed change of the leading vehicle. For the leading vehicle V1, since the information obtained by different strategies is similar, relatively consistent results are produced, and the speed change trend is consistent. However, for the following vehicles V2-5, the DSRTH strategy responds faster than the VTH strategy and the CTH strategy. This difference is more obvious in the vehicles at the end of the queue.

[0104] In addition, compared with the VTH strategy, the DSRTH strategy leads by 1-3 seconds in speed change rate. Since the speed of the front vehicle starts from 22m / s, decelerates and then accelerates, the speed finally stabilizes at 20m / s. The V5 speed valley is 3.0% higher than the CTH strategy and 1.8% higher than the VTH strategy. Compared with the CTH strategy, the time to reach the speed valley is shortened by 6.1%. Analyzing the improvement effects of the two strategies, the rear vehicles (V3-V5) in the convoy show faster responses under the DSRTH strategy than the CTH and VTH strategies, accompanied by relatively small changes.

[0105] Speed ​​recovery period: The period of 80 to 100 seconds represents the stage of queue recovery stability. In this stage, the speed recovery time is mainly used as the evaluation indicator of queue stability recovery. The recovery time of the DSRTH strategy is 2% to 13% shorter than that of the CTH strategy, showing comparable performance to the VTH strategy. The advantage of the DSRTH strategy lies in the speed change stage. a The parameters respond actively to the state of the leading vehicle. This allows the vehicle to quickly return to a stable constant speed state after completing the speed change. The DSRTH strategy is particularly effective for the rear vehicles in the platoon, such as Figure 3 shown.

[0106] The results show that during the speed oscillation, the front vehicle in the DSRTH queue can respond quickly and expand the expected headway distance to ensure safety. At the same time, the rear vehicle in the queue responds earlier to the acceleration of the front vehicle, resulting in a smaller change in the expected headway distance and improved fuel economy. 2) Evaluation of the effect of the improved CACC model

[0107] a. Verification of the effectiveness of CATP:

[0108] In order to verify the improvement effect of the model on CATP, the IDM is compared with the unmodified PATH laboratory CACC model for evaluation. In order to avoid the system error caused by the headway strategy, the three models all use the DSRTH strategy. The three models are calibrated through ACC Data to obtain parameter values ​​to improve the accuracy of the model. The parameter values ​​are as follows.

[0109]

[0110] Using the driving scenario of the pilot car in ACC Data as the simulation condition, the numerical simulation method is used to analyze the changes in speed, acceleration and position of the five following cars during acceleration and deceleration, as shown below. Figure 4 shown.

[0111] Comparison of evaluation indicators of different vehicle following models

[0112]

[0113]

[0114] Comparative analysis of driving behaviors at different stages:

[0115] Speed ​​maintenance period: During the period of 50 to 60 seconds, the vehicle queue travels at a constant speed. Since the acceleration of the front vehicle is basically 0, the vehicle is not affected by the acceleration and speed difference of the front vehicle. Therefore, in the speed maintenance stage, the driving behaviors of the three strategies are basically consistent.

[0116] Deceleration-acceleration period: 60 seconds to 80 seconds is the stage of rapid speed change, which aims to verify the response ability of different following models to speed changes. Figure 4 (d)-(f), the time difference between V1 and V5 reaching the speed valley is used as an indicator of the transmission time of the external disturbance within the truck queue. The time interval of the IDM model is 13.5 seconds, the time interval of the PATH-CACC model is 11.2 seconds, and the time interval of the improved CACC model is 6 seconds, which has a significant improvement in response speed.

[0117] It is worth noting that in order to pursue the goal of reducing energy consumption and improving the stability of the queue, the queue expects to reduce the amplitude of speed changes. Therefore, the effect of the queue on suppressing the amplitude of disturbance propagation is reflected by comparing the time when the vehicle speed valley occurs and the speed valley. The results show that the minimum speed value of the improved CACC model is improved and the time of reaching the valley speed is advanced. Specifically, when V5 reaches the speed valley, compared with the IDM and CACC models, the minimum speed is 2.1% higher than the IDM model and 3.6% higher than the CACC model; the time is 12.2% earlier than the IDM model and 8.6% earlier than the CACC model. Since the sub-goals of control quantity and trajectory tracking are introduced into the objective function of the improved CACC model, it shows a more balanced advantage in speed response and stability.

[0118] Speed ​​recovery period: After 80 seconds, the queue enters the speed recovery phase. Figure 4(g)-(i), in the IDM model, the acceleration of V5 tends to zero at 109.5 seconds; in the CACC model, it tends to zero at 91.2 seconds; the improved CACC model tends to zero at 88.3 seconds, 19.4% earlier than the IDM model and 3.2% earlier than the CACC model. By analyzing this indicator, the results show that the improved CACC model has a significantly enhanced ability to restore stability when disturbed by the speed of the vehicle ahead.

[0119] The improvement effects of the three indicators of vehicles at different positions in the improved CACC model are as follows: Figure 5 shown.

[0120] The results show that as the vehicle moves to the rear of the queue, the improvement in the time when the vehicle speed valley occurs, the speed valley, and the acceleration returns to zero are more significant. This highlights the impact of information from the leading vehicle on the driving behavior of the following vehicle. The implementation of the improved CACC strategy has better rapid response and rapid recovery effects.

[0121] b. Fuel economy improvement effect for CATP:

[0122] When multiple trucks are driving in CACC mode, CATP is formed. At this time, the leading vehicle acts as a windbreaker, effectively reducing the air resistance of the rear vehicle and acting as a protective barrier. At the same time, the rear vehicle will change the wake pattern of the front vehicle and reduce the air resistance of the front vehicle. Therefore, after the formation of the queue, the air resistance coefficients of vehicles at different positions in the queue will change. It is necessary to design the air resistance correction coefficient based on the position in the queue. Hussein considered different models and different positions and proposed the following expression for the air resistance correction coefficient of the vehicle queue.

[0123]

[0124] Where, X k is the inter-vehicle distance. When car k is the pilot car, X k is the distance between the target vehicle and the following vehicle; when k is the following vehicle, X k is the distance between the target vehicle and the preceding vehicle.

[0125] The present invention uses five trucks arranged longitudinally, and the pilot vehicle travels under the ACC Data condition. A simulation experiment on the fuel consumption of the truck queue is carried out by TruckSim. TruckSim is a software for truck simulation, and the platform can be used to verify the effectiveness of the formulated vehicle control strategy. The present invention selects the TS-2A-LCF-VAN 5.5T model as the queue vehicle, combines it with TruckSim for joint simulation, uses the improved CACC model to calculate the vehicle speed and then passes it to the target vehicle to achieve control, and uses the engine model corresponding to TS-2A-LCF-VAN 5.5T to calculate the fuel consumption. In order to verify the improvement of fuel economy by different headway strategies and following models, the headway strategy uses the CTH strategy, VTH strategy, and DSRTH strategy, and the following model uses the IDM model, CACC model, and improved CACC model. Five experimental scenarios are obtained by combination, as shown in the following table.

[0126] Fuel consumption test scenario

[0127]

[0128]

[0129] The fuel consumption test results of different combination strategies are shown in the following table.

[0130] Fuel consumption test results

[0131]

[0132] In terms of the headway strategy, scenario 1 was used as the control group. In scenario 2, the use of the VTH strategy enabled timely response to the speed of the vehicle ahead, reducing vehicle speed changes, which reduced fuel consumption by 0.04%. In scenario 3, the use of the DSRTH strategy enabled driving at a smaller headway, reducing the air resistance coefficient, and reducing fuel consumption by 0.17% compared to scenario 1.

[0133] In terms of the car-following model, scenario 3 was used as the control group. In scenario 4, the CACC model controlled based on the difference between the expected and actual spacing, reducing the speed change range and reducing overall fuel consumption by 0.33%. In scenario 5, the improved CACC model used MPC to predict the speed change of the leading vehicle and dynamically adjusted the parameters, and used PD to control the spacing error, significantly reducing fuel consumption by 3.89%. Fuel consumption was effectively saved.

[0134] c. Improvement effect on traffic flow:

[0135] In order to evaluate the effect of CATP in traffic flow, SUMO was used to conduct simulation experiments to simulate the traffic flow scenario of the basic section of the highway. The simulation scenario is a basic section of the highway with a two-lane length of 5 km. The speed limit for cars is 120 km / h, and the speed limit for trucks is 80 km / h. The simulation time is set to 500 seconds, and the vehicle arrival probability conforms to the negative exponential distribution with a parameter λ of 0.65. There are three types of vehicles on the road: manually driven cars, manually driven trucks, and trucks equipped with CACC systems. Among them, the manually driven vehicles use the IDM model, and the CAT uses the improved CACC model. In order to avoid the influence of systematic errors and random factors, each experiment is repeated 5 times and the average value is taken. When calculating the indicators, the results observed between 200 seconds and 500 seconds during the simulation are used to ensure that the traffic flow is in a stable state and the consistency of the evaluation process is ensured.

[0136] Model ratio analysis:

[0137] It is known that trucks usually drive in the outermost lane on highways and the proportion of trucks is usually smaller than that of cars, but the proportion of trucks in vehicles varies at different times of the day, which leads to differences in the proportion of vehicle models. At the same time, with the continuous updating of intelligent network technology, the penetration rate of CAT will continue to increase, and there will be differences in penetration rates during the development process. Therefore, it is necessary to analyze the traffic flow under different vehicle model proportions and penetration rates. In the experiment, the proportion of cars and the proportion of CATP were changed to 0%, 20%, 40%, 60%, 80%, and 100%. Taking traffic flow rate as the evaluation indicator, the results are as follows.

[0138] Comparison results of traffic flow rate under different vehicle ratios

[0139]

[0140] In order to better illustrate the impact of different vehicle type ratios on traffic flow, the changing trend of traffic flow rate when the vehicle type ratio changes is visualized as follows: Figure 6 shown.

[0141] In terms of CATP penetration, increasing penetration can increase traffic flow, except when the proportion of cars is 100%, there are no trucks on the road, resulting in no CAT. Figure 6The traffic flow rate in the middle shows an upward trend from right to left. This is because CATP can improve the utilization rate of road resources by driving at a small distance. Therefore, when the proportion of vehicle models remains unchanged, the higher the CATP penetration rate, the more vehicles equipped with the CACC system, and the more significant the positive impact on traffic efficiency. By observing the trend of traffic flow rate changes, when the penetration rate ranges from 0% to 60%, the effect of improving traffic flow rate is more significant. When the penetration rate exceeds 60%, the marginal benefit of improving traffic flow rate decreases. Therefore, in the early stage of CATP development, as the penetration rate increases, there is a more obvious improvement in traffic efficiency.

[0142] In terms of the proportion of cars, when the CATP ratio is between 0% and 60%, an increase in the proportion of cars will increase traffic flow rate, but when the CATP ratio is between 60% and 100%, increasing the proportion of cars will instead cause a decrease in traffic flow rate. This means that when the CATP ratio is higher, the traffic efficiency of trucks may be slightly higher than that of cars.

[0143] Cohort size analysis:

[0144] In the CATP configuration, the maximum queue size is a key parameter. However, the impact of fleet size on traffic flow has not been fully studied. Therefore, in order to study the impact of different queue sizes on traffic flow, the experimental scenario is set as follows: maintaining a constant proportion of vehicle types (CATP penetration rate is 100%, the proportion of cars is 60%, which is basically consistent with the daily environment), controlling the queue size from 1 to 10, and considering the extreme case of CATP penetration rate of 0. Traffic flow rate, vehicle average speed and speed variance are used as evaluation indicators. Traffic flow rate is used to evaluate traffic efficiency, average speed is used to evaluate the vehicle's own benefits, and speed variance is used to evaluate the stability of traffic flow. The experimental results are as follows Figure 7 shown.

[0145] The results show that the traffic flow rate increases with the increase of the queue size, especially when the queue size is from 1 to 5. This is because the CACC model is used when the following car in the queue follows the front car, and when the leading car follows the front car, it degenerates to the ACC model because the front car is a manually driven vehicle. Therefore, although the number of CATs remains unchanged in the simulation experiment, the increase in the queue size leads to more CACC vehicles in the CAT, thus improving the traffic flow rate.

[0146] However, when the queue size is 6 to 10, the improvement in traffic efficiency is not significant. Figure 7 As shown in (b), except for the increase in CATP penetration from 0% to 100%, the effect of increasing the cohort size on the average speed is negligible. Figure 7In (c), as the size of the queue increases, the speed variance gradually increases, which indicates that the speed fluctuations within the traffic flow increase and the stability decreases. It can be deduced that as the size of the queue increases, the stability of the queue decreases, offsetting the improvement brought by the reduction of the headway. Therefore, when the queue size is 6 to 10, the improvement of traffic flow rate is not significant, and the instability increases rapidly. CATP does not recommend the use of a formation that is too large. For a queue size of 3 to 5, the speed change is not obvious, and the traffic flow rate is significantly improved.

[0147] In order to analyze the relationship between CATP and the internal mobility bottleneck of traffic flow, the vehicle trajectory diagrams under different queue sizes are analyzed as follows Figure 8 The colors in the figure correspond to the speed values, and the dotted line represents the trajectory of CATP.

[0148] Can be from above Figure 8 It is observed that the congestion caused by the increase in queue size propagates downstream. In the simulation scenario of a two-lane road, since vehicles travel at different speed limits, manually driven vehicles can obtain speed benefits by changing lanes. When the queue size is small, there is no significant negative impact on the lane changing process, the flexibility of lane changing is relatively high, and the number of lane changes is relatively high. However, when the truck queue is in the form of a small spacing and the queue size is long, the vehicles in the inner lane face certain difficulties when they hope to change lanes to the lane where the CATP is located. They can only overtake or wait for the CATP to pass. Therefore, when the vehicle density is high, lane changing becomes more difficult, resulting in the formation of a bottleneck at the CATP. From a theoretical analysis, the formation of a CATP queue can improve traffic efficiency by maintaining a small following distance, but as the queue size increases, the possibility of traffic bottlenecks increases, which will lead to the propagation of disturbances and ultimately reduce the stability of the overall traffic flow.

[0149] 3) Evaluation of cooperative lane-changing control effect

[0150] In order to evaluate the effect of CATP coordinated lane change control in traffic flow, Fig. 9 Traffic flow is added to the road shown. Trucks account for 40% of the traffic flow and cars account for 60%, which is more in line with the proportion of vehicles in daily life. Among cars, 40% of vehicles need to get off the ramp, so they need to leave the main line by forced lane change in the diversion area. The effectiveness is verified under different traffic flow and different queue size scenarios.

[0151] a. Traffic density analysis:

[0152] In order to verify the effectiveness of the cooperative control strategy under different traffic densities, the single-lane flow rate is set to [600, 1000, 1400, 1800], and the double-lane conversion coefficient is 0.8. The effectiveness of the formation control and adaptive formation strategy is verified in the diversion area. When the control model is the improved CACC model, the following model uses the CACC model improved based on the dynamic safety requirement distance, and does not use the cooperative lane change control; when the control model is the cooperative lane change control, the following model uses the CACC model improved based on the dynamic safety requirement distance, and uses the cooperative lane change control. The average driving time and average driving speed are calculated as evaluation indicators. The main purpose is to analyze the impact of different control methods on vehicle driving efficiency under different traffic densities. The calculation results are as follows.

[0153] Average travel time of traffic flow

[0154]

[0155]

[0156] By analyzing the average travel time of traffic flow, the conclusion shows that the improved CACC model has a shorter average travel time than the uncontrolled formation. Fig.10 It shows that, taking the uncontrolled experimental results as the control group, the improvement degree of the other two strategies is calculated. When the traffic flow is low (600veh / h / lane and 1000veh / h / lane), the improved CACC model and the cooperative lane-changing control strategy have a small improvement on the average driving time. This is because when the traffic density is low, the distance between vehicles is larger and they travel more in a free-flow mode. Therefore, improving the vehicle spacing strategy has little effect on traffic flow.

[0157] As the traffic flow increases (1400 veh / h / lane and 1800 veh / h / lane), the improvement of the average driving time by the improved CACC model gradually decreases. This is because the improved CACC following model does not consider the impact of forced lane changes. Therefore, CATPs travel at small intervals, and vehicles on the off-ramp that need to change lanes can only complete the lane change by waiting for the CATP to pass, which has a certain negative impact on traffic efficiency.

[0158] At the same time, after the introduction of the collaborative lane-changing control strategy, the average driving time has been greatly improved. When the off-ramp vehicles have the need to change lanes, CATP can actively create gaps and restore the queue, and it has greatly improved the traffic efficiency of the off-ramp vehicles and the main line. When the traffic flow is 1800veh / h, the average driving time of the off-ramp is shortened by 19.6%, and the average driving time of the main line is shortened by 1.5%.

[0159] The average driving speed reflects the vehicle's own benefits to a certain extent. With the increase of background traffic volume, the average driving speed of vehicles on the main line and on the off-ramp has decreased. However, the improvement of the coordinated lane-changing control strategy compared with the uncontrolled scenario has continued to increase, and the greater the traffic flow, the more significant the improvement effect.

[0160] Average speed of traffic flow

[0161]

[0162]

[0163] By analyzing the average speed of traffic flow, the results show that traffic flow has a significant impact on the average speed. With the increase of traffic flow, the average speed shows a gradual downward trend, whether on the main line or the off-ramp. This is because the increase in traffic flow leads to a smaller distance between vehicles, and the speed of vehicles is more affected by other vehicles, resulting in a decrease in speed.

[0164] Under the same traffic flow, the improved CACC model and the coordinated lane-changing control strategy can both increase the average speed of the main line and the off-ramp compared to the uncontrolled case. This shows that these two control strategies can alleviate traffic congestion and improve traffic efficiency to a certain extent.

[0165] When the traffic flow is low (600veh / h / lane and 1000veh / h / lane), the improvement of the average speed of the off-ramp by the improved CACC model and the coordinated lane-changing control strategy is relatively small. When the traffic flow is high (1400veh / h / lane and 1800veh / h / lane), the coordinated lane-changing control strategy improves the average speed more significantly. When the traffic flow is 1800veh / h, the average speed of the off-ramp increases by 10.3%, and the average speed of the main line increases by 2.3%, which is consistent with the improvement trend of the average travel time, such as Fig.11 b. Queue size analysis:

[0166] In order to verify the effectiveness of the cooperative control strategy under different queue sizes, the control queue size was set from 1 to 10 while keeping the CACC penetration rate consistent, and the experiment was carried out under the traffic flow state (1800veh / h / lane) where the cooperative lane change control had the most significant improvement effect. Since the queue size was 0 in the uncontrolled state and there was no difference under different queue sizes, only the comparison of whether the cooperative lane change control strategy was applied was used for comparison, and the results are as follows.

[0167] Average travel time for different queue sizes

[0168]

[0169] By analyzing the average travel time under different queue sizes, the results show that in the highway diversion area, the average travel time first decreases and then increases with the increase of queue size, as shown below Fig.12 This is different from the downward trend of the basic road section. In the diverging area, due to the presence of lane-changing vehicles, when the queue size is too large, the positive impact of small-spacing driving on the road and the negative impact of lane-changing barriers caused by the queue offset each other, and even when the queue length exceeds 7, the traffic efficiency develops in a worse direction.

[0170] By comparing the average travel time of the improved CACC model and the cooperative lane-changing control strategy under different queue sizes, the results show that when the queue size is small, the improvement effect of adding the cooperative lane-changing control strategy on the average travel time is not significant, and the lane-changing vehicle has a shorter delay time waiting for the CATP to pass. As the queue size gradually increases (when the queue size is 1-7), the improvement effect on the main line and the off-ramp becomes more significant, but as the queue size further increases (when the queue size is 8-10), due to the long queue, 2-3 gaps are created inside the queue when processing the cooperative lane change, resulting in poor stability inside the queue. The process of creating gaps and restoring the stable operation of the queue has a great impact on the traffic flow. Therefore, in the highway diversion area, CATP is not suitable for using a longer formation size. It can be separated in advance into multiple queues with smaller queue sizes, which will improve the traffic efficiency of the diversion area to a certain extent.

[0171] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A fleet adaptive cruise control method, characterized in that: include: Step A: Establish a dynamic safety requirement distance model to describe the safety distance requirements of the intelligent connected driving queue in the connected environment; combine the speed and acceleration information of the front vehicle in the queue, and the influence of the different positions of the vehicle in the queue, introduce the information transfer coefficient and position influence coefficient, and establish a dynamic safety requirement distance strategy; Step B: Based on the dynamic safety requirement distance model, the collaborative adaptive cruise control model CACC is improved, and the headway strategy is updated to the dynamic safety requirement distance strategy; speed and distance delay are introduced to simulate the time lag relationship between the upper-level planning and the bottom-level control; Step C: Establish a cooperative lane-changing control strategy based on an adaptive spacing strategy, set a relaxation spacing, a recovery spacing, and an equilibrium spacing for different stages of forced lane changing of vehicles, and make adaptive changes according to the indicator variables of the lane-changing stage; Based on the requirements at different stages of the lane-changing process, the relationship between the sub-goals is analyzed, and the weight distribution of the objective function at each stage is determined. Combined with the constraints of the vehicles in the queue, a collaborative lane-changing optimization model is constructed to achieve centralized queue control.

2. The method according to claim 1, characterized in that The step A comprises: Step A1: Establish the VTH spacing strategy as follows: X des =(t0-c r ·v r )·v k Among them, X des is the desired safety distance; v k is the speed of the following vehicle k; 0 is the headway; c r is the coefficient to be calibrated; v r is the speed difference between the front and rear vehicles; Step A2: Introduce information perception coefficient and position influence coefficient: The dynamic safety requirement distance strategy is established as follows: In the formula, v r,k represents the speed difference between the target vehicle and the preceding vehicle; σ j is the information perception coefficient of vehicle j; θ j is the position influence coefficient of vehicle j; a j is the acceleration of vehicle j; t0, c v and c a is the parameter to be calibrated, c v represents the influence of the speed difference with the preceding vehicle, c a represents the influence of the acceleration of the vehicle in front; m is the number of the vehicle at the end of the queue; Step A3: Parameter calibration: The model parameters of the dynamic safety requirement distance strategy are calibrated, and the particle swarm optimization algorithm is used as the calibration method. The root mean square error is selected to measure the error between the actual spacing and the dynamic safety requirement distance strategy.

3. The method according to claim 2, characterized in that The step B comprises: Step B1: The improved CACC model considering time delay and dynamic safety demand distance is: In the formula, e k is the gap error; x k-1 and x k are the longitudinal positions of vehicles k-1 and k respectively; v k is the speed of the target vehicle; T is the headway; Δt is the update step size; k p and k d is the control coefficient; L k is the vehicle length; t1 and t2 represent the spacing delay and speed delay respectively; Step B2: Dynamic parameter adjustment based on MPC: Combining the constraints and objective function, an optimization model is established under the MPC framework to dynamically adjust the PD control parameters.

4. The method according to claim 3, characterized in that The step C comprises: Step C1: Adaptive spacing strategy: The lane-changing process is divided into three stages: relaxation, equilibrium, and recovery, and the corresponding spacing is constructed; Indicator variables are introduced to describe the progress of vehicles in cooperative lane changing; An adaptive adjustment mechanism is established by using the relationship between the current spacing and the required spacing as a basis for judgment; Step C2: Establish a collaborative lane-changing optimization model.

5. The method according to claim 4, characterized in that The step C further comprises: Multi-criteria decision analysis considering prior knowledge is selected for multi-objective optimization solution in the cooperative lane changing stage.

6. The method according to claim 5, characterized in that The multi-criteria decision analysis includes: During the following phase, a smaller safety weight coefficient is used, the fuel weight coefficient is increased, and the tracking free flow speed is adjusted to improve overall traffic efficiency.

7. The method according to claim 5, characterized in that The multi-criteria decision analysis includes: During the gap creation phase, the platoon accelerates and decelerates to create a larger gap for lane-changing vehicles to merge in, while reducing the pursuit of efficiency targets.

8. The method according to claim 5, characterized in that The multi-criteria decision analysis includes: In the gap recovery phase, a larger gap appears in the queue, and vehicles speed up to shorten the gap and reach an equilibrium gap, thereby increasing the equilibrium target weight.

9. The method according to claim 5, characterized in that The multi-criteria decision analysis includes: In the balancing stage, the value of the balancing weight is increased.

10. The method according to claim 5, characterized in that The multi-criteria decision analysis includes: Through the lane-changing stage indicator variable, the weight coefficients in different stages are switched to achieve adaptive variable weights.

Images