汽车轨迹跟踪模型预测控制的加速求解方法

Accelerated Solution Method for Vehicle Trajectory Tracking Based on Model Predictive Control

针对汽车轨迹跟踪模型预测控制求解中存在的规模较大、求解效率较低的问题,提出一种基于时域分解的加速计算方法提高求解效率.首先引入全局一致性变量,将模型预测控制中邻接控制周期的时域耦合约束转化为全局一致性约束,实现时域解耦;随后在交叉方向乘子法框架下推导了时域分解后优化问题的分块更新方法,并设计了分块更新数值求解问题的停止准则,从而将大规模优化问题转化为小规模子问题;最后搭建了Simulink-CarSim平台进行了算法的仿真验证.仿真结果表明,在求解精度不变的情况下,求解耗时下降24.21%,从而实现模型预测控制问题的加速求解.

To solve the problems of large scale and low efficiency in the solution of intelligent vehicle trajectory tracking algorithm based on model predictive control,this paper proposes a time-domain splitting method to accelerate the calculation.First,global consistency variables are introduced to transform the time-domain coupling constraints of the adjacent control cycles into global consistency constraints,thus to achieve time-domain decoupling.Then,under the framework of Alternating Direction Method of Multipliers,the block updating method of optimization problem after time domain splitting is derived,and the stop criterion of block updating numerical solution is designed.Therefore,the large-scale optimization problem is converted into several small-scale sub-problems.Finally,the Simulink-CarSim platform is set up and the algorithm is verified by simulation.The simulation results show that,under the same accuracy of the solution,the efficiency of solution is improved by 24.21%on average with the proposed method,achieving accelerated solution for vehicle trajectory tracking based on model predictive control.