Multi-type vehicle formation robust control method based on event-driven mechanism

Abstract

The invention discloses a multi-type vehicle formation robust control method based on an event-driven mechanism. The method comprises the following steps: defining a vehicle longitudinal dynamics model and a state transition equation; a multi-step state prediction model is constructed, and combined prediction of state evolution and a control input sequence in a future time domain of the formation is realized; defining physical constraint and security constraint conditions of formation control; constructing a robust optimization control model, introducing disturbance boundary and prediction error modeling, and solving a robust control problem through a minimum-maximum form; designing an event triggering mechanism and a control mode switching rule, and dynamically switching a feedback control mode, a robust control mode and a tracking control mode according to an event type and a formation state; and executing the multi-mode control strategy. According to the method, a feedback control mode with relatively low calculation cost can be adopted during stable operation, and a robust optimization control mode is switched when disturbance is intense or the state is suddenly changed, so that unnecessary calculation resource overhead is effectively reduced, and the overall control efficiency is improved.

Description

Technical Field

[0001] This invention belongs to the field of intelligent traffic control technology, specifically relating to a robust control method for multi-type vehicle platooning based on an event-driven mechanism. Background Technology

[0002] With the accelerated implementation of intelligent connected vehicle (V2X) and autonomous driving technologies, the mixed operation of connected and automated vehicles (CAVs) and traditional human-driven vehicles (HVs) in the same transportation system has become an inevitable trend. Against this backdrop, vehicle platooning systems, as a crucial component of intelligent traffic control, are becoming an important research direction in the field of traffic control due to their potential to improve road efficiency, reduce energy consumption, decrease emissions, and enhance driving safety.

[0003] The core objective of vehicle platooning control is to achieve speed synchronization and spacing coordination among multiple vehicles, enabling it to adapt to changing traffic environments while maintaining platoon stability. Traditional vehicle platooning strategies are mostly based on stable communication between vehicles, a closed platoon structure, and consistent vehicle types, and are primarily suitable for highly collaborative autonomous vehicle environments. However, in real-world traffic systems, CAVs (Consumer-Agent Vehicles) and HVs (Hardware-Vehicle Vehicles) coexist. Issues such as random lane changes, acceleration / deceleration uncertainties, communication delays, and packet loss by non-connected vehicles make traditional control strategies difficult to directly adapt to current mixed traffic scenarios.

[0004] First, while existing vehicle platooning models based on Model Predictive Control (MPC) can theoretically handle various constraints and optimize control strategies, they are highly dependent on the prediction of the preceding vehicle's state and struggle to address prediction errors caused by HV (High Vehicle) state uncertainty, easily leading to system performance degradation or even instability. More importantly, as the platoon size increases and the prediction time domain expands, the computational complexity of the MPC model grows exponentially, making it difficult to complete high-frequency solutions in a short time, resulting in control delays and limiting its real-time application capabilities. Second, during platooning operation, vehicles not only face external traffic disturbances such as non-platooning vehicles entering the platoon and changes in road conditions, but also need to maintain a stable state for most of the time. However, existing control strategies often employ fixed optimization frequencies, failing to rationally switch control models according to actual changes in traffic conditions. This approach results in redundant computation and wasted resources when traffic is stable, and fails to respond promptly during severe disturbances, affecting safety and control robustness. Furthermore, communication systems still face the risk of frequent interruptions in complex urban environments. If information exchange between vehicles is hindered, it will affect the coordinated execution of control commands, thereby weakening the platoon's coordination capabilities and even causing the formation structure to disintegrate. Therefore, control methods relying solely on fixed topologies and continuous optimization calculations are insufficient to meet the comprehensive requirements of safety, adaptability, and efficiency in complex scenarios.

[0005] To address the aforementioned issues, there is an urgent need to develop a robust and flexible control strategy capable of achieving efficient formation control under uncertain disturbance conditions. On one hand, this strategy should incorporate robust optimization methods, modeling prediction errors as disturbance variables and introducing them into the control model to enhance the system's stability and safety under external disturbances. On the other hand, it should possess event-awareness capabilities, dynamically switching control modes based on the current formation state. When traffic conditions are stable, a low-complexity feedback control mode should be used to save computational resources, while automatically switching to a robust optimization mode to ensure control quality when disturbances intensify or structural changes occur, truly realizing a multi-mode control mechanism of "state-driven, on-demand control." Summary of the Invention

[0006] To address the shortcomings of existing vehicle platooning control systems in mixed traffic environments in terms of control accuracy, real-time performance, and system robustness, this invention proposes a robust control method for multi-type vehicle platooning based on an event-driven mechanism. This method aims to improve the operational stability and traffic efficiency of mixed vehicle platooning under disturbance conditions, while reducing the control computation burden and achieving a dynamic balance between control performance and resource consumption.

[0007] To solve the above-mentioned technical problems, the technical solution provided by the present invention is as follows: A robust control method for multi-type vehicle platooning based on an event-driven mechanism, the method comprising: Step 1: Define the vehicle's longitudinal dynamics model and state transition equations; Step 2: Construct a multi-step state prediction model to achieve joint prediction of the future state evolution of the formation and the control input sequence in the future time domain; Step 3: Define the physical and safety constraints for formation control; Step 4: Construct a robust optimization control model, introduce disturbance boundary and prediction error modeling, and solve the robust control problem through minimax form; Step 5: Design the event triggering mechanism and control mode switching rules, and dynamically switch between feedback control, robust control and tracking control modes according to the event type and formation status; Step 6: Execute the multi-mode control strategy and calculate the control input by calling the corresponding control model based on the current state.

[0008] Furthermore, in step 1, the vehicle longitudinal dynamics model is constructed based on the vehicle spacing error and speed difference, and the state transition equation introduces the acceleration increment as an optimization objective term.

[0009] Furthermore, the vehicle longitudinal dynamics model is specifically defined as follows: the formation state prediction equation is defined as follows: ; in, and for The state vector at any given time; It is the identity matrix; The sampling interval; matrix For control matrix; To control the input vector; ; For the lead vehicle at all times Control input; The number of vehicles following (controlled vehicles) in a vehicle platoon; In the state vector and The definition of is: ; ; in, For the first The distance error between the vehicle and the vehicle in front; For the first The speed difference between the vehicle and the vehicle in front; ; ; in, The desired headway; The state transition equation is specifically as follows: Let the system state be... The state transition equation is defined as follows: ; in, The acceleration of the manually driven vehicle in front of the convoy's lead vehicle; Considering the complexity of the traffic environment and the dynamic characteristics of vehicles, the prediction error is defined as: ; in, To predict acceleration; Characterize the prediction error boundary; Acceleration increment Add this to the optimization objective function to reconstruct the state equations: ; in, For state variables; Definitions of each matrix: .

[0010] Furthermore, step 2 specifically includes: defining a multi-step state prediction model: For a given prediction time domain It is necessary to deduce the future. The state of step 1, for the 2nd step Step prediction ( ): ; Future The states of each step are stacked into a vector. The control inputs and disturbance terms are stacked as follows: and : ; Considering that the state equation uses acceleration increments As a control variable, subsequent constraints will use this control variable. ,Will Converted to acceleration in the prediction time domain: ; in, for Extended vector of vehicle formation control output at any time; It can be represented as: ; Wherein, the state transition matrix Describe the effect of the initial state on future states: ; This describes the cumulative effect of control inputs on future states, and its structure is a lower triangular block matrix with elements... express right Impact: ; perturbation matrix Describe the cumulative effect of the perturbation term on future states, structure and similar: ; Furthermore, through rigorous derivation of the definition Step position: ; This yields the predicted vehicle positions in the time domain in matrix form: ; in, ; for Vehicle platoon position vector at any given moment; .

[0011] Furthermore, step 3 specifically includes: defining velocity boundary constraints, acceleration boundary constraints, and safety distance constraints. Velocity boundary constraints: ; in, and These are the lower and upper bounds of the velocity, respectively. Defined as ; It is a unit vector; Its linear constraint form is: ; The constraint matrix is ​​in the following specific form: ; Acceleration boundary constraints: ; The constraint matrix is ​​as follows: ; Safety distance constraints: To prevent collisions between vehicles, use As a vehicle exist The basic safety distance at any given time is constrained as follows: ; in, For the safe distance of the vehicle; For minimum safe distance, Extending the above formula to the entire formation, we get: Single-step safety constraints for formation at any given moment: ; in, ; for Predicted location of the navigation vehicle at all times; The definition of is: ; Furthermore, the formation safety constraints are defined throughout the entire prediction time domain: ; Wherein, the constraint matrix and The definition of is: .

[0012] Furthermore, step 4 specifically includes: defining an objective function. The target discrete form is expressed as: ; in, ; It is a symmetric penalty matrix; and These are the penalty submatrices for the state and control variables, respectively; The terminal penalty matrix is ​​used to ensure the stability of the control system; The robust control problem can be constructed into the following minimax form: ; in, and Let these represent the feasible sets of control input and disturbance, respectively; due to the uncertainty set... The existence of transforms the problem into the following equivalent form: ; make ,in Therefore, we get: ; Define state constraints: Considering safety constraints Includes predicted lead vehicle position When the disturbance satisfies At that time, the upper bound is used to replace the uncertainty term, thus obtaining the information of the navigating vehicle. Location at any given time: ; Replacing the above equation in matrix form yields: ; in, It is a unit vector. It is a lower triangular identity matrix; substituting... We can obtain: ; Therefore, the constraint is defined as: ; in, For the expansion matrix: ; in, .

[0013] Furthermore, step 5 specifically includes: Event triggering mechanisms include: General events: triggered by changes in formation length, the entry of the preceding vehicle, or a violation of physical constraints by control inputs; State-based events: triggered by the transition between a stable state, a robust state, or a risky state of the formation; The control mode switching rules are as follows: When a normal event occurs, switch to the robust control model; In a steady state, LQR pure feedback control is used; In robust mode, a tracking control combining robust MPC and LQR feedback is employed; In a risky state, the robust control model is continuously invoked until the state reverts to a robust state. The criteria for determining the stable state are as follows: Within the time step, the vehicle spacing and speed difference remain within the steady-state threshold range, i.e. ; in, and These represent the lower and upper bounds of the steady-state threshold, respectively. For the front The historical state vector of a time step; The robust state satisfies ; in, for Robust control of the feasible region at each moment; The risk state is when the formation leaves the robust control domain.

[0014] Furthermore, step 5 also includes feedback control and state deviation correction, specifically: when the vehicle formation is in a stable state, an LQR is used as the feedback controller to achieve efficient real-time control of the formation. The expression for the LQR controller is: ; in, Represents the feedback gain matrix. To expand control inputs; The optimal control law can be obtained by solving the equations. ; When the formation is in a robust state, a scheme combining robust control MPC and LQR feedback is adopted. After the robust control MPC model outputs the planned trajectory, the system uses feedback control to track the trajectory, calculates the error between the current state and the planned state at each time step, and corrects it using the LQR controller. Time step The state error is: ; in, and They represent Tracking deviation at each time step, actual state versus planned state; This leads to the expression for the tracking LQR controller: ; in, This indicates the tracking and correction control quantity.

[0015] The final control input is a combination of the planning input and the feedback input: .

[0016] Furthermore, step 6 specifically includes: multi-mode control including a planning module, a state tracking module, and a feedback control module for robust model predictive control; At each time step, update the state of the vehicle formation itself, calculate the state vector, and detect whether an event has occurred; If a general event occurs, the system switches to a robust control model to optimize the control output; if no general event occurs, it further determines whether the formation is stable; if stable, it switches to a more stable model. If the formation state does not meet the stability condition, it is determined whether the robust state is met, i.e., the formation state is within the robust control feasible region. If the robust state is met, a tracking control mode combining robust control MPC and LQR feedback is adopted. If the formation does not meet the above conditions, it means that the current state has deviated from the robust control region, and the robust control model is called for rolling optimization control.

[0017] Furthermore, the method also includes single-group performance evaluation analysis of the control model to verify the effectiveness of the robust control model.

[0018] The present invention has the following technical effects: 1. An event-driven mechanism is introduced, which enables dynamic switching of control strategies by real-time detection of vehicle platooning status and external traffic disturbances. This allows the system to adopt a low-computational-cost feedback control method when running smoothly, and switch to a robust optimization control mode when there are strong disturbances or sudden changes in state, thereby effectively reducing unnecessary computational resource overhead and improving overall control efficiency.

[0019] 2. By constructing a robust control model based on disturbance boundary modeling, the system can maintain good stability and reliability in complex traffic scenarios with prediction errors and environmental uncertainties, effectively improving its adaptability to unpredictable external disturbances. Attached Figure Description

[0020] Figure 1 This is a diagram showing the changes in the formation state of different vehicle models under continuous acceleration and deceleration scenarios according to the present invention. Figure 2 This is a diagram showing the changes in the formation state of different vehicle models under rapid acceleration and deceleration scenarios according to the present invention. Figure 3 This is a comparison diagram of the safe distance between vehicle formations and the actual distance under continuous acceleration / deceleration and rapid acceleration / deceleration scenarios according to the present invention; Figure 4 This is a comparison diagram of the state response and acceleration response of the lead vehicle in a formation under continuous acceleration / deceleration and rapid acceleration / deceleration scenarios according to the present invention; Figure 5 This is a spatiotemporal evolution diagram of traffic flow for the robust control strategy under different CAV penetration rates according to the present invention; Figure 6 This diagram illustrates the event detection and triggering process of the robust control strategy under different penetration rates according to the present invention. Figure 7 This is a flowchart of the method of the present invention. Detailed Implementation

[0021] To make the objectives, advantages and features of the present invention more apparent, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] This invention proposes a multi-mode robust control method based on an event-driven mechanism, such as... Figure 7 As shown, it includes the following steps: Step 1: Define the vehicle's longitudinal dynamics model and state transition equations Step 1.1: Define the vehicle's longitudinal dynamics model The formation state prediction equation is defined as follows: ; in, and for The state vector at any given time; It is the identity matrix; The sampling interval; matrix For control matrix; To control the input vector; ; For the lead vehicle at all times Control input; The number of vehicles following (controlled vehicles) in a vehicle platoon; In the state vector and The definition of is: ; ; in, For the first The distance error between the vehicle and the vehicle in front; For the first The speed difference between the vehicle and the vehicle in front; ; in, This represents the desired headway.

[0023] Step 1.2: Define the state transition equation Record system status The state transition equation is defined as follows: ; in, This refers to the acceleration of the manually driven vehicle in front of the convoy's lead vehicle.

[0024] Considering the complexity of the traffic environment and the dynamic characteristics of vehicles, the prediction error is defined as: ; in, To predict acceleration; Characterize the prediction error boundary.

[0025] In order to ensure the smoothness of vehicle acceleration in formation control, this invention will increment the acceleration. Add this to the optimization objective function to reconstruct the state equations: ; in, For state variables; Definitions of each matrix: ; Step 2: Define the multi-step state prediction model For a given prediction time domain It is necessary to deduce the future. The state of step 1, for the 2nd step Step prediction ( ): ; Future The states of each step are stacked into a vector. The control inputs and disturbance terms are stacked as follows: and : ; Considering that the state equation uses acceleration increments As a control variable, subsequent constraints will use this control variable. The present invention will Converted to acceleration in the prediction time domain: ; in, for Extended vector of vehicle formation control output at any time. It can be represented as: ; Wherein, the state transition matrix Describe the effect of the initial state on future states: ; This describes the cumulative effect of control inputs on future states. Its structure is a lower triangular block matrix, with elements... express right Impact: ; perturbation matrix Describe the cumulative effect of the perturbation term on future states, structure and similar: ; Furthermore, through rigorous derivation of the definition Step position: ; This yields the predicted vehicle positions in the time domain in matrix form: ; in, ; for Vehicle platoon position vector at any given moment; .

[0026] Step 3: Define vehicle formation constraints: To comprehensively consider the limitations imposed by vehicle hardware and road conditions on vehicle operating status and control output, as well as safety during operation, this invention defines speed boundary constraints, acceleration boundary constraints, and safety distance constraints.

[0027] (1) Velocity boundary constraints: ; in, and These are the lower and upper bounds of the velocity, respectively. Defined as ; It is a unit vector.

[0028] Its linear constraint form is: ; The constraint matrix is ​​in the following specific form: ; (2) Acceleration boundary constraints: ; The constraint matrix is ​​as follows: ; (3) Safety distance constraints: To prevent collisions between vehicles, this invention uses... As a vehicle exist The basic safety distance at any given time is constrained as follows: ; in, For the safe distance of the vehicle; For minimum safe distance, Extending the above formula to the entire formation, we get: Single-step safety constraints for formation at any given moment: ; in, ; for Predicted location of the navigation vehicle at all times; The definition of is: ; Furthermore, the formation safety constraints are defined throughout the entire prediction time domain: ; Wherein, the constraint matrix and The definition of is: ; Step 4: Define the robust control model: Step 4.1: Define the objective function The discrete form of the optimization objective can be expressed as: ; in, ; It is a symmetric penalty matrix; and These are the penalty submatrices for the state and control variables, respectively; The terminal penalty matrix is ​​used to ensure the stability of the control system.

[0029] The robust control problem can be constructed into the following minimax form: ; in, and Let these represent the feasible sets of control input and disturbance, respectively. Due to the uncertainty set... Given the existence of [the problem], this invention transforms the problem into the following equivalent form: ; Considering that the objective function in the above formula contains uncertainties, it cannot be solved directly. Let... ,in Therefore, we get: ; Step 4.2: Define state constraints: Considering safety constraints Includes predicted lead vehicle position When the disturbance satisfies When this happens, its upper bound can be used to replace the uncertainty term, thus obtaining the information about the navigating vehicle. Location at any given time: ; Replacing the above equation in matrix form yields: ; in, It is a unit vector. It is a lower triangular identity matrix. Substitute... We can obtain: ; Furthermore, the constraint can be defined as: ; in, For the expansion matrix: ; in, ; Step 5: Define the event triggering mechanism and control mode switching rules; Step 5.1: Define the event ① General events: These refer to events that do not depend on the formation state (position, speed, acceleration), mainly caused by changes in driving conditions or vehicle physical constraints. Changes in driving conditions include two types: changes in formation length and the entry of a vehicle from the front. First, changes in formation length arise from vehicles joining or leaving the formation, causing changes in the control model structure, thus rendering previously planned trajectories invalid. Furthermore, since the state of newly joined vehicles differs significantly from that of other vehicles in the formation, their states need to be re-processed to stabilize. Second, the entry of an external vehicle from the front of the formation significantly disrupts formation operation, causing considerable disturbance to the state of the lead vehicle. Continuing to execute the previous control strategy would increase safety risks, thus requiring a replanning of the trajectory to ensure safety. In contrast to changes in driving conditions, physical constraints refer to constraints arising from the vehicle's physical properties, including speed limits, acceleration limits, and acceleration increment limits. The system checks at each time step whether the control input violates physical constraints, i.e. ; in, and They represent the first The control input at each time step and the feasible control region based on physical constraints. Such violations mean that the calculated control input is infeasible, and the system will trigger a reprogramming process under any control mode.

[0030] ② State-based model switching events: Unlike general events, state-based model switching events focus on the platooning state under various control modes. They are mainly triggered by three types of state changes: stable state, robust state, and risky state. A stable state indicates that the vehicle's driving state has reached a stable state, and the system can switch to a more efficient pure feedback control model. A robust state indicates that the vehicle's driving is not stable, but the state always remains within the robust control domain, possessing redundancy to combat external disturbances; therefore, a control mode combining robust control and feedback control can be adopted. A risky state indicates that the vehicle platoon, under the influence of strong external disturbances, no longer meets robust safety constraints; therefore, a robust control model needs to intervene to guide it back to the robust control domain to ensure safety.

[0031] Step 5.2: Define the running status ① Steady State: During platooning operations, if the vehicles are in a steady state, meaning the distance between vehicles and the speed difference are within a preset steady-state range, the criteria for determination are as follows: ; in, and These represent the lower and upper bounds of the steady-state threshold, respectively. For the front The historical state vector at each time step.

[0032] In history If the vehicle spacing and speed difference remain within a preset steady-state range within a given time step, the current state is considered to be stable. In this case, the system can employ a pure feedback control model to improve control efficiency and reduce computational resource consumption.

[0033] ② Robust State: If the vehicle formation state does not meet the steady-state conditions, it will exit pure feedback control and instead determine whether the robust state is met. If the robust constraint conditions are met, i.e., the slack variables are... The definition of a robust state is: ; in, for The robust control feasible region at time step. In this state, the system will adopt a control mode combining robust control MPC and LQR. After the robust control model provides the planned trajectory, feedback control is used to track the trajectory and reduce process error.

[0034] ③ Risk state: If the vehicle formation state does not meet the robust state conditions, it will exit robust control and enter a risk state. At this time, it is necessary to continuously call the robust control model to guide the vehicle formation state back to the robust state to ensure safety.

[0035] Step 5.3: Define feedback control and state deviation correction When the vehicle platoon is in a stable state, an LQR (Low-Quickness Ratio) controller is used as the feedback controller to achieve efficient real-time control of the platoon. The expression for the LQR controller is: ; in, Represents the feedback gain matrix. To expand control inputs.

[0036] The optimal control law can be obtained by solving the Riccati equation. Furthermore, the system can generate optimal control inputs based on the feedback matrix and the current state of the formation. When the formation is in a robust state, a combination of robust control MPC and LQR feedback is used. After the robust control MPC model outputs the planned trajectory, the system will use feedback control to track the trajectory, calculate the error between the current state and the planned state at each time step, and correct it using the LQR controller.

[0037] Time step The state error is: ; in, and They represent Tracking deviation at time steps, actual status versus planned status.

[0038] This leads to the expression for the tracking LQR controller: ; in, This indicates the tracking and correction control quantity.

[0039] The final control input is a combination of the planning input and the feedback input: ; Step 6: Define the multi-mode control strategy execution algorithm, specifically including: The algorithm comprises three core modules: a planning module for robust model predictive control, a state tracking module, and a feedback control module. First, the current state of the formation is determined based on the aforementioned criteria, thus identifying the control model to be used subsequently. After determining the current state, the control input for the next time step is calculated based on the control model corresponding to each state. Specifically, this includes: (1) At each time step, update the state of the vehicle formation itself, calculate the state vector, and detect whether an event has occurred.

[0040] (2) If a general event occurs, such as a vehicle cutting in or a change in platoon length (vehicles joining or leaving, platoon merging), the system switches to robust control model optimization to solve for the control output. If no general event occurs, the system further determines whether the platoon is stable. If it is stable, the system switches to feedback control. If the platoon state does not meet the stability condition, the system determines whether it meets the robust state condition, i.e., the platoon state is within the robust control feasible region. If the robust state condition is met, the system adopts a tracking control mode that combines robust control MPC and LQR feedback. If the platoon does not meet the above conditions, it means that the current state has deviated from the robust control region, and therefore the robust control model needs to be called for rolling optimization control.

[0041] Step 7: Performance Evaluation and Analysis of Single Formation Control Model To verify the effectiveness of the robust control model proposed in this paper, different disturbance scenarios are constructed to analyze the control performance of a single formation.

[0042] (1) Changes in the state of vehicle formation under different disturbance scenarios ① Performance of a vehicle platoon under scenarios where its lead vehicle continuously accelerates and decelerates: In this scenario, the lead vehicle in the formation will accelerate and decelerate in a periodic oscillation manner, with its speed curve being a sine curve and its acceleration range being within... . Figure 1 Figures (a) and (b) illustrate the acceleration increment and acceleration variation of the robust control model in this scenario. As can be seen from the figures, at the start of the simulation, due to the initial unsteady state, the vehicle formation experiences a brief state change before stabilizing. Even with continuous and significant changes in the lead vehicle's state, the vehicle formation does not respond drastically but smoothly tracks the lead vehicle's state changes. Furthermore, the figures also show that the acceleration increment is limited to [a certain value]. Within the range, the vehicle acceleration is limited to Within the specified range, this demonstrates that the robust control model proposed in this paper can effectively handle constraints while tracking the state changes of the lead vehicle. Figure 1Figures (c), (d), (e), and (f) respectively illustrate the changes in acceleration increment and acceleration of the nominal control model and the LQR model under continuous acceleration and deceleration scenarios. From the figures, it can be seen that because essentially the same control model was used, the nominal control model and... Figure 1 The robust control models in (a) and (b) exhibit similar state responses, differing only in their response to the initial state at the start of the simulation. This characteristic is expected, as this scenario does not violate robust safety boundaries and relies solely on basic physical constraints and an optimization objective function for formation control. Compared to constrained control models, Figure 1 The LQR control models in (e) and (f) output large acceleration increments at the beginning of the simulation experiment, exceeding the range of physical constraints. However, during the subsequent tracking process, they behaved similarly to the MPC-based control models described above. This characteristic indicates that LQR is suitable for unconstrained control problems that do not touch the constraint boundaries.

[0043] ② The performance of a vehicle platoon under scenarios where its lead vehicle accelerates or decelerates rapidly: In this scenario, the lead vehicle initially travels smoothly, then decelerates at maximum speed. Rapid deceleration, then maximum acceleration Rapid acceleration. Furthermore, its preset acceleration also includes... The uniform random perturbation. Figure 2 The figures illustrate the acceleration increment and acceleration variations of the robust control model, nominal control model, and LQR model under rapid acceleration and deceleration scenarios. These figures show that the state responses of the different control models are essentially the same, and they can generally track the disturbances of the lead vehicle in a smooth manner, with each state remaining within the physical constraints. It should be noted that in unconstrained scenarios, the MPC and LQR models are equivalent, and since LQR is the optimal control model in the infinite time domain, its output control quantity is the optimal solution. Therefore, this experiment demonstrates that the MPC model proposed in this paper has good stability.

[0044] (2) Changes in the safety redundancy space of the robust control model in the face of disturbances Figure 3Figures (a) and (b) illustrate the changes in the safe and actual distances between the robust control model and the two baseline models under continuous acceleration and deceleration scenarios. As can be seen from the figures, the actual distances of all control models are greater than the safe distances in this scenario, while the robust control model reserves a larger safety redundancy space compared to the two baseline models. It should be noted that in this scenario, none of the models have reached the safety boundary, therefore the safety constraints are not effective. However, since the objective function of the robust control model also includes an error term, sufficient safety redundancy space can still be reserved in this scenario. Unlike continuous acceleration and deceleration scenarios, in rapid acceleration and deceleration scenarios, the spacing of vehicle formations changes drastically, requiring safety constraints to ensure their safety. Figure 3 Figures (c) and (d) illustrate the changes in safe and actual distances between the robust control model and the two baseline models under rapid acceleration and deceleration scenarios. The figures show that the safety constraints of vehicle formation may be violated in this scenario. The actual vehicle distances of the nominal control model and the LQR model exceed the safety boundaries, while the actual distance of the robust control model remains within the safe distance range and even has redundancy. This comparison clearly demonstrates that the robust control model can effectively handle constraints when there are errors in state prediction and, while ensuring safety boundaries, tracks the state changes of the lead vehicle. Furthermore, Figure 4 The figures illustrate the changes in acceleration increment and acceleration response of the lead vehicle in a convoy under continuous acceleration / deceleration and rapid acceleration / deceleration scenarios, respectively. The figures reveal that the main difference between the MPC-based control model and the LQR model in these scenarios lies in their ability to handle constraints. While the nominal control model and the robust control model exhibit similar performance across different time periods, their responses differ only in the initial state.

[0045] Step 8: Impact Analysis of Control Strategies on Traffic Flow This invention designs a traffic flow simulation experiment to evaluate the performance of the proposed control strategy in complex traffic scenarios. In the experiment, a single-lane road is designed to conduct the traffic flow simulation, focusing on the longitudinal dynamics of vehicles. Figure 5 The spatiotemporal diagrams of traffic flow under robust control strategies at different penetration rates are presented. The diagrams show that as the penetration rate of intelligent connected vehicles increases, more vehicles form platoons, significantly alleviating traffic congestion. This demonstrates that the robust control model based on multi-step MPC proposed in this chapter can effectively suppress traffic flow disturbances. Figure 6The diagram illustrates the mode-switching process of the vehicle formation during operation. It can be observed that near the ramp, continuous vehicle entry significantly disrupts traffic flow, leading to drastic changes in the vehicle formation's state. Therefore, the vehicle formation in the diagram concentrates on state replanning near the ramp. Furthermore, this phenomenon becomes more pronounced with the increasing penetration rate of intelligent connected vehicles. These results validate the effectiveness of the multi-mode robust control strategy in switching control modes based on the state of vehicles within the formation.

[0046] The contents not described in detail in this specification are existing technologies known to those skilled in the art.

[0047] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A robust control method for multi-type vehicle platoon based on event-driven mechanism, characterized in that, The method comprises: Step 1: defining a vehicle longitudinal dynamics model and a state transition equation; Step 2: constructing a multi-step state prediction model to realize joint prediction of state evolution and control input sequence in the future time domain of the platoon; Step 3: defining physical constraints and safety constraint conditions of the platoon control; Step 4: constructing a robust optimization control model, introducing disturbance boundary and prediction error modeling, and solving the robust control problem through minimax form; Step 5: designing an event triggering mechanism and control mode switching rule to dynamically switch feedback control, robust control and tracking control modes according to event types and platoon states; Step 6: executing a multi-mode control strategy to calculate control inputs according to the current state.

2. The method of claim 1, wherein, In step 1, the vehicle longitudinal dynamics model is constructed based on vehicle spacing error and speed difference, and the state transition equation introduces acceleration increment as the optimization objective item.

3. The method of claim 2, wherein, The vehicle longitudinal dynamics model is specifically defined as follows: ; wherein and is state vector at time instant is the identity matrix; is the sampling interval; matrix to control the matrix; to control the input vector; ； a control input for the lead vehicle at time t; a number of following vehicles (controlled vehicles) in the vehicle platoon; in the state vector with is defined as: ; ; wherein, is the distance error between the vehicle and the preceding vehicle; is the distance error between the vehicle and the preceding vehicle; For the first Vehicle and the speed difference of the preceding vehicle; ; ; wherein, T is the desired headway time; The state transition equation is specifically as follows: Let the system state be... The state transition equation is defined as follows: ; wherein, is the acceleration of the manually driven vehicle in front of the platooning lead vehicle; Considering the complexity of the traffic environment and the vehicle dynamics characteristics, the prediction error is defined as: ; wherein, is the predicted acceleration; Characterize the prediction error boundary; The acceleration increment is added to the optimization objective function, and the state equation is reconstructed: ; wherein is a state variable; The definition of each matrix is as follows: 。 4. The method of claim 3, wherein, Step 2 specifically includes defining a multi-step state prediction model: For a given prediction horizon , the state of the future step needs to be derived for the first step prediction ): ; Stacking future steps as vectors Stacking states of steps as vectors , control inputs and disturbance terms as and , respectively: ; Taking into account the state equation with acceleration increment As control variable, the subsequent constraint will use the control variable , the is converted into acceleration in the prediction horizon: ; wherein, is the momentary vehicle platoon control output; may be expressed as: ; where the state transition matrix Describes the influence of the initial state on the future state: ; The cumulative effect of control inputs on future states is described by a lower triangular block matrix with elements representing the effect of on ; Disturbance matrix Describes the cumulative effect of disturbance terms on future states, structure and Similar: ; Furthermore, by strict derivation definition Step's position: ; Then the matrix form of the vehicle position in the prediction time domain is obtained: ; wherein ; for the instant vehicle platoon position vector; 。 5. The method of claim 4, wherein, Step 3 specifically includes defining speed boundary constraints, acceleration boundary constraints and safety distance constraints: The linear constraint form of the speed boundary constraint is: ; wherein and are lower and upper bounds for the velocity, respectively; defined as ; is a unit vector; The specific form of the constraint matrix is: ; The acceleration boundary constraint is: ; Where the constraint matrix is: ; The safety distance constraint is: ; Then, the platoon safety constraint in the entire prediction time domain is defined as: To guarantee collision avoidance between vehicles, the use of As vehicles In The basic safety distance at a moment of time is formulated as: ; wherein, is the safety time interval for the vehicle; for minimum safety clearance, Generalizing the above equation to the entire platoon, we have Platoon single-step safety constraint at time ; wherein ; To keep track of the predicted position of the vehicle; The definition of "C1-C6alkyl" is: ; Step 4 specifically includes defining the objective function. ; where the constraint matrix and is defined as: 。 6. The method of claim 5, wherein, The discrete form of the objective function is expressed as: The robust control problem can be constructed in the following minimax form: ; wherein ; is a symmetric penalty matrix; and are the penalty sub-matrices of states and controls, respectively; The terminal penalty matrix is used to ensure stability of the control system; Define the state constraint: ; where and denote the feasible sets of control inputs and disturbances, respectively; due to the set of uncertainties the problem is transformed into the following equivalent form: ; Let where and further obtain: ; Change the above formula to matrix form to get: Considering safety constraints Includes predicted lead vehicle position When the disturbance satisfies At that time, the upper bound is used to replace the uncertainty term, thus obtaining the information of the navigating vehicle. Location at any given time: ; Then the constraint is defined as: ; where is a unit vector, is a lower triangular unit matrix; substituting we obtain: ; Step 5 specifically includes: ; wherein is the spreading matrix: ; wherein .

7. The method of claim 6, wherein, The event triggering mechanism includes: General event: triggered by changes in platoon length, front vehicle cutting in or control input violating physical constraints; State-based event: triggered by switching from a steady state, a robust state or a risk state of the platoon; The control mode switching rule is: When a general event occurs, switch to the robust control model; When in a steady state, use LQR pure feedback control; When in a robust state, use tracking control combining robust MPC and LQR feedback; When in a risk state, continuously call the robust control model until it returns to the robust state; The risk state is when the platoon deviates from the robust control domain. The criterion for determining the steady state is that the The inter-vehicle distance and speed difference within the time step are within the steady state threshold range, i.e. ; wherein and respectively represent the lower and upper bounds of the steady-state threshold; for the previous history state vector of the time step; The robust state satisfies ; wherein, is robust control feasible region at the moment Step 5 also includes feedback control and state deviation correction, which is specifically:

8. The method of claim 7, wherein, When the vehicle platoon state is in a steady state, use LQR as a feedback controller to achieve efficient real-time control of the platoon. The expression of the LQR controller is: When the platoon is in a robust state, use a scheme combining robust control MPC and LQR feedback. After the robust control MPC model outputs the planned trajectory, the system uses feedback control to track the trajectory, calculates the error between the current state and the planned state at each time step, and uses the LQR controller for correction; ; wherein denotes a feedback gain matrix, is the extended control input; By solving the equation, the optimal control law can be obtained ; Then the expression of the tracking LQR controller is obtained: Time step The state error is: ; wherein, and respectively represent tracking bias, actual state and planned state at time step The final control input is the combination of the planned input and the feedback input: ; wherein, represents a tracking correction control amount; ​ 。 9. The method of claim 8, wherein, The step 6 specifically comprises: a planning module, a state tracking module and a feedback control module for multi-mode control including robust model predictive control; In each time step, the vehicle platoon self-state is updated, a state vector is calculated, and whether an event occurs is detected; If a general event occurs, the system switches to a robust control model to optimize and solve a control output; if no general event occurs, it is further judged whether the platoon is stable, and if the platoon is stable, it is switched to feedback control; if the platoon state does not satisfy the stable condition, it is judged whether a robust state is satisfied, i.e., the platoon state is within a robust control feasible region; if the robust state is satisfied, a tracking control mode combining robust control MPC and LQR feedback is adopted; if the platoon does not satisfy the above conditions, it is indicated that the current state is out of the robust control region, and then a robust control model is called to perform rolling optimization control.

10. The method of claim 9, wherein, The method further comprises a control model single-platoon performance evaluation analysis for verifying the effectiveness of the robust control model.

Images