Multiple models robust adaptive control with reduced model

The traditional robust adaptive control broadens the application of the routine adaptive control because of considering the uncertainty of the practice plant. However, traditional robust adaptive control solves the problem that the plant would be unstable under some uncertainties and it often ignores the dynamic property and steady-state property, which makes the traditional robust adaptive control unfit for practice plant, whose working condition changes randomly. To solve these problems, this paper designs multiple models robust adaptive controller for the plant which can be describeled by Autoregressive Moving Average (ARMA) model and with unmodeled dynamics. First, the normalization factor is used to convert the system unmodeled dynamic to bounded disturbance, and then, many fixed controller and two robust adaptive controllers are designed based on system reduced model according to the changing range of the working condition and the best controller computed by the performance index is chosen as the working controller. Simulation illustrates that the proposed method is preferable when there are unmodeled dynamics and changing working condition.