Learning and Adaptation in Neural Control of Higher-Order Linear Systems

We have developed a dynamic learning and adaptive scheme, named Inverse-Dynamic Adaptive Control (IDAC), for unknown linear dynamic plants. The IDAC scheme provides parallelism, in the sense that the learning and control actions are performed simultaneously [4]. In addition, it provides continuous learning and adaptation capability. In this paper, we extend the IDAC pinciple for unknown higher-order dynamic plants. The necessary learning and control algorithm is derived. A dynamic neuro-controller structure developed in this paper can be made applicable to multi-input-multi-output (MIMO) plants, though in this paper we discuss only SISO cases.