中文摘要：

针对高超声速飞行器非线性、快时变、硬约束、大扰动和对象模型不确定性大等特点,研究离线鲁棒预测控制器的设计方法。在线鲁棒预测控制方法可以处理对象模型的不确定性,但是计算量大;常规离线鲁棒预测控制方法减小了在线计算量,但是牺牲了系统的调节时间等性能。提出基于单因素椭圆集序列的鲁棒预测控制策略,针对系统状态的m组一维子空间,运用常规离线鲁棒预测控制生成m组椭圆不变集序列,将在线优化转化成为在m组椭圆不变集中分别搜索当前状态所处最大椭圆不变集,最后通过简单的融合计算得到对应的控制律。该方法保证系统鲁棒性的同时,大幅度减少了在线计算时间,提高了系统响应速度。仿真算例验证了该方法的有效性。

英文摘要：

The offline robust predictive controller design for the nonlinear, time-varying and uncertain hypersonic vehicle with various constraints and disturbances is investigated in this paper. It is known that the online robust predictive control method can handle object model uncertainty, but a large quantity of computation is required. The conventional offiine robust predictive control method can greatly reduce the online computation at the expense of system setting time. This paper presents a new off-line robust model predictive control strategy based upon the univariate elliptic set sequence. Firstly, m-elliptic invariant set sequences are independently generated with the conventional offline robust predictive control method. Then the maximum elliptic invariant set which contains the actual state is obtained though some search scheme, and final control action is set up with simple fusion. The proposed method can greatly reduce the online computation with good robustness and satisfactory response speed. Simulation examples demonstrate the effectiveness and feasibility of the method.