Design of adaptive control system based on linearization error model

Abstract

In order to overcome the nonlinearity and time-varying uncertainty of actual industrial processes, an adaptive control system based on linearization error model is proposed. In this system, first, a composite model consisting a ARX model and a linearization error model based on the neuro-fuzzy system is constructed to describe the nonlinear process. Then, by employing a single-neuron controller and by considering the error between the ARX model output and the system output, as well as the gradient information of the composite model, the controller parameters are adjusted online with high control performance. Simulated results indicate that, as compared with the conventional PID controller, the proposed adaptive controller based on linearization error model is of higher response speed.