Almost disturbance decoupling control of mimo nonlinear system subject to feedback linearization and a feedforward neural network: Application to half-car active suspension system

Abstract

A novel tracking and almost disturbance decoupling problem of multi-input, multi-output (MIMO) nonlinear systems based on feedback linearization and a multi-layered feedforward neural network approach has been proposed. The feedback linearization and neural network controller guarantees exponentially global uniform ultimate bounded stability and almost disturbance decoupling performance without using any learning or adaptive algorithms. The new approach renders the system to be stable with the almost disturbance decoupling property at each step when selecting weights to enhance the performance if the proposed sufficient conditions are maintained. One example, which cannot be solved by the existing approach of the almost disturbance decoupling problem because it requires the sufficient conditions that the nonlinearities that multiply the disturbances satisfy structural triangular conditions, is proposed to exploit the fact that the tracking and the almost disturbance decoupling performances are easily achieved by the proposed approach. In order to demonstrate the practical applicability, a famous half-car active suspension system is investigated.