摘要
基于模型预测控制(MPC)理论的智能车纵向速度控制问题可以转换为二次规划问题。针对该二次规划(QP)问题,利用一种改进的有效集(IASM)二次规划算法减少MPC计算成本。该方法包含两步:首先对等式约束引进一种降维算法;然后利用梯度投影方向对有效集算法的搜索方向进行改进。改进的QP算法减少了迭代次数,降低了MPC纵向控制的计算成本。仿真结果证明了该方法的有效性。
The problem of longitudinal speed control of intelligent vehicle based on MPC( model predictive control) theory can be converted to QP( quadratic programming) problem. Considering this QP problem,the paper proposes a method which can reduce MPC costs with IASM( improved active set method). Firstly,it introduces a dimension reduction algorithm for equality constraints. Then,it improves the search direction of active set algorithm with gradient-projection direction. The improved QP algorithm reduces the number of iterations and the computational cost of MPC longitudinal control.The simulation result verified the effectiveness of the method.
出处
《军事交通学院学报》
2016年第10期49-53,共5页
Journal of Military Transportation University
基金
国家自然科学基金项目(91120306
91220301)