Distributed Model Predictive Control Of Truck Platoons Under Intelligent Networked Environment

Cooperative control of truck platoons can effectively enhance the driving safety of trucks,increase traffic capacity,decrease traffic congestion,and reduce fuel consumption,which has a major social and economic value,thus it has become an important research direction and academic research hotspot of intelligent transportation system.With the rapid advancement of intelligent networked technology,real-time communication and accuracy between vehicles provide new conditions for cooperative control of truck platoons.With the rapid advancement of advanced automation technology,the computational speed of onboard processors presents new opportunities for cooperative control of truck platoons.However,the current control technologies for truck platooning still face the following challenges:(1)The study of string stability in truck platooning has primarily concentrated on frequency domain analysis.A pivotal challenge is how to transition string stability from frequency domain analysis to time domain analysis via distributed model predictive control,to ensure the string stability of the truck platoon.(2)For truck platoons navigating complex road environments,a significant challenge is how to integrate longitudinal and lateral control to simultaneously ensure the longitudinal tracking performance and lateral path-following performance of the truck platoon,as well as both longitudinal and lateral string stability.(3)Addressing the strong coupling characteristics of longitudinal and lateral motions,as well as the nonlinear characteristics of tires in truck platoons under cornering conditions,presents a significant challenge.It necessitates the development of an integrated longitudinal-lateral coupled truck platoon model,the design of a cooperative longitudinal-lateral distributed model predictive controller,the achievement of integrated control objectives for both longitudinal and lateral directions of the truck platoon,and the assurance of the algorithm’s realtime performance.This paper will address the aforementioned challenging issues through detailed research.The specific contents include:For truck platoons in longitudinal motion,this study explores how to translate string stability from frequency domain analysis to time domain analysis to ensure the platoon’s string stability.It proposes a distributed model predictive control algorithm that guarantees string stability.By incorporating time-domain inequality constraints in local optimization problems,the string stability of truck platoons is ensured.Additionally,by introducing terminal control laws,terminal penalty functions,and terminal constraint sets in local optimization problems,the asymptotic consensus of truck platoons are guaranteed.Considering practical scenarios,a combined feedforward and feedback lower-level controller is designed.The feedforward controller utilizes the inverse model of vehicle dynamics to convert control inputs determined by the upper-level controller into desired throttle opening and desired brake pressure needed for truck operation.Meanwhile,the feedback controller employs a PID controller to eliminate unmodeled dynamic and static errors.The joint simulation experiments conducted with Matlab and TruckSim demonstrate that the proposed control strategy ensures longitudinal tracking performance and longitudinal string stability of truck platoons under the leader-predecessor following communication topology.For truck platoons operating in complex road environments,to simultaneously ensure the longitudinal and lateral tracking performance of the platoon,and to mitigate the impact of modeling errors and external disturbances on platoon performance,a combined longitudinal and lateral distributed robust Tube model predictive control strategy is proposed.In the longitudinal direction,a sequential distributed robust Tube model predictive controller is employed.The Tube invariant set is computed based on the robust online reachable set.By designing time-domain inequality constraints within the local optimization problem,the longitudinal string stability of the truck platoon is ensured.Additionally,terminal control laws,terminal penalty functions,and terminal constraint sets are added to local optimization problems to ensure the longitudinal robust stability of the truck platoon.In the lateral direction,a sequential distributed robust Tube model predictive controller based on the preceding vehicle’s path generation is employed.Given that the computational burden of calculating robust online reachable sets increases with the system’s dimensionality and uncertainty,the minimal robust positive invariant set is used to compute the Tube invariant set.By adding time-domain inequality constraints and terminal constraints in local optimization problems,the lateral string stability and lateral robust stability of the truck platoon are ensured.Joint simulation experiments conducted using Matlab and TruckSim demonstrate that,during low and medium-speed operations of a fully loaded truck platoon,the proposed control strategy is capable of simultaneously ensuring both the longitudinal and lateral tracking performance,as well as the longitudinal and lateral string stability of the truck platoon.In response to the strong coupling characteristics of longitudinal and lateral movements of truck platoons under cornering conditions,a cooperative distributed model predictive control strategy for platoon longitudinal and lateral motions with guaranteed consistency is proposed.To avoid solving nonlinear optimization problems and to obtain terminal elements that ensure the asymptotic convergence of the predictive control system,a dynamic mode decomposition algorithm based on Koopman operator theory is utilized.This algorithm approximates nonlinear vehicle dynamics as a “global” linear model.Combined with a lane-keeping model,a linear parameter-varying model for the truck platoon is established.A cooperative distributed model predictive controller for both longitudinal and lateral directions is developed,incorporating terminal control laws,terminal penalty functions,and terminal constraint sets into the local optimization problem of the distributed model predictive controller.This approach ensures the asymptotic consensus of the truck platoon.The local optimization problems of distributed model predictive control are solved using the alternating direction method of multipliers to reduce the computational burden.Joint simulation experiments conducted using Matlab and TruckSim demonstrate that the proposed control strategy can ensure both the longitudinal tracking and lateral lane-keeping performance of the fully loaded truck platoon.Additionally,it meets the realtime requirements of the algorithm.For truck platoons driving on wet and slippery roads,in addition to considering the coupling characteristics of longitudinal and lateral motion,the impact of nonlinear tire characteristics on the safety of truck platoon operation should also be taken into account.To address this,a longitudinal-lateral coordinated distributed model predictive control strategy considering tire nonlinearity is proposed.To avoid solving nonlinear non-convex optimization problems,based on the Koopman operator theory,the vehicle dynamics model considering tire nonlinearity is approximated as a“global” linear model.By computing the maximum robust positively invariant set of the "global" linearized system as the terminal constraint set,the nonlinear non-convex optimization problem is transformed into a quadratic programming problem,reducing computational burden while ensuring the asymptotic consensus of the truck platoon.Additionally,by designing time-domain inequality constraints in the local optimization problems,the longitudinal string stability of the truck platoon is ensured.Joint simulation using Matlab and TruckSim demonstrates that the proposed control strategy ensures longitudinal tracking performance,longitudinal string stability,and lateral lane-keeping performance of the fully loaded truck platoon on low-adhesion road surfaces.Additionally,it meets the real-time requirements of the algorithm.In conclusion,this thesis explores the distributed model predictive control method for intelligent networked truck platoons,providing a viable control scheme for the safe and stable formation driving of truck platoons.