Study on robust model predictive control for dynamical systems with constraints of logic and state switchings

在分析混合逻辑动力系统模型与分段线性或分段仿射模型扩展应用适用性的基础上，提出了一类具有状态不连续变化或切换约束和逻辑约束两类混合系统的建模方法。利用该模型，在极大极小控制设计方法的基础上，提出了有限时间后退水平控制设计方法。该设计方法克服了现有显式模型预测控制的缺点。由于计算时间的限制，机械系统、机电一体化系统等快速运动系统难以应用。针对具有强非线性特性的系统，提出了部分外力控制方法。在此基础上，针对采样数据动态系统这一具有约束条件的混合系统，提出了基于模型预测控制和自适应DA （Discrete-Analog）转换器的集成控制设计方法。然后，提出了一些基于混沌噪声的神经网络、遗传算法和进化计算的优化方法，作为所提设计方法的有效优化工具。通过各种仿真和实验，验证了所提设计方法的有效性。

A modeling method for a kind of hybrid systems with two types constraints of discontinuous changes or switching of states and of logics has been developed based on the analysis results of suitability of extended application of the mixed logical dynamical system model and the piecewise linear or piecewise affine model.By using this model, the finite-time receding horizon control design method has been developed based on the minimax control design approach. The developed design method can get rid of drawbacks of pre-existing explicit model predictive control. It's difficult to apply the fast moving systems, such as mechanical systems, mechatronics systems, and so on, due to the computation time. The partial external force control method has also been proposed for the systems with strong nonlinear properties. Furthermore, the integrated control design method based on the model predictive control and adaptive DA (Discrete-Analog) converter has been developed for sampled-data dynamical systems, which is a kind of hybrid systems, with constraints.Then, some optimization methods based on the neural networks with chaotic noise, genetic algorithm and evolutionary computing have been developed as effective optimization tools for the proposed design methods.From various simulations and experiments, the effectiveness of the proposed design methods have been confirmed.