Model-Free Adaptive Control Based on Local Dynamic Linearization

Abstract:Linearization technique is essential for the controller design and analysis of complex nonlinear plants. At least, the linear affine form of the original nonlinear process is required. However, the traditional linearization methods are approximation-based with unmodeled dynamics involved. In this work, a new local dynamic linearization method is proposed using differential mean-value theorem, which can be estimated by using the I/O data only. Then, a new model-free adaptive control is proposed by using the principle of optimality, where the controller design and analysis is data-driven without using any model information. The effectiveness of the proposed method is demonstrated by the simulation results.

Translation or Not:no