A robust model reference adaptive control for nonminimum phase systems (Q5957033)

The authors present a direct model reference adaptive control (DMRAC) method for non-minimum phase systems. They define a good data model by filtering and normalizing raw input data. The data filtering uses two filters: the first is a low-pass filter that reduces the effect of the high frequency modes of unmodelled dynamics, and the second is an exponential filter that generates a plant estimated model with minimum phase. The second filter relocates the estimated plant model zeros inside the unit circle. The data model, based on these two filters, is used for the control synthesis. The authors' approach achieves robustness (with respect to unmodelled dynamics) by using this adaptive input-output data modelling in association with a least squares estimation algorithm with parameter projection and signal normalization. Performance of the control algorithm is illustrated with simulation results.