A self tuning regulator for nonlinear time varying control systems based on evolving linear models

Although evolving models have found different applications in real world systems, they have not been developed significantly for nonlinear control systems. In this paper, Evolving Linear Models (ELMs) are proposed for self tuning regulation of a main class of nonlinear time varying control systems. First, the structure of ELMs is introduced and then it is shown that nonlinear time varying control systems-under some constraints-can be represented as ELMs. It is demonstrated that if the ELM of a nonlinear system is known and if the system is minimum phase, a tracking control law can be extracted, easily. Next, an indirect self tuning regulator is proposed based on online identification of the ELM and using the tracking control law. The proposed control approach is applied to an artificial control system and a flexible joint robot. The results show clearly that ELMs can be used as basis linear models for control analysis and design of many nonlinear time varying systems.