一类工业运行过程多模型自适应控制方法

针对一类动态未知的工业运行过程，提出一种基于神经网络补偿和多模型切换的自适应控制方法.为充分考虑底层跟踪误差对整个运行过程优化和控制的影响，将底层极点配置控制系统和上层运行层动态模型相结合，作为运行过程动态模型.针对参数未知的运行过程动态模型，设计由线性鲁棒自适应控制器、基于神经网络补偿的非线性自适应控制器以及切换机制组成的多模型自适应控制算法.采用带死区的递推最小二乘算法在线辨识控制器参数，克服了投影算法收敛速度慢、对参数初值灵敏的局限.理论分析和仿真实验结果表明了所提方法的有效性.

Multi-model Adaptive Control Method for a Class of Industrial Operational Processes

In this paper, for a class of industrial operational processes, an adaptive control method based on neural networks and multiple models is proposed. In order to fully consider the influence of tracking errors in bottom closed-loop system on the performance of the operational process, the bottom pole-assignment control system and the dynamic model in the upper layer are combined as generalized dynamic model of the whole operational process. For the dynamic model with unknown parameters, a multi-model adaptive control method, composed of a linear robust adaptive controller, a neural-network-based nonlinear adaptive controller and a switching mechanism, is designed. Since the recursive least square algorithm has a faster convergence rate and less sensitivity to the initial values of parameters than the projection algorithm, the recursive least square algorithm with deadzone is used to identify the unknown controller parameters. Results of theoretic analysis and simulation experiments demonstrate the effectiveness of the proposed method.