Hybrid Model-Insensitive Control Using a Piezoelectric Sensoriactuator

A hybrid control system based on a piezoelectric sensoriactuator is presented. The system uses the adaptive feedforward Time Averaged Gradient (TAG) descent algorithm to achieve persistent disturbance rejection. No system modeling is required for the TAG descent algorithm. However, this algorithm is not well suited for control of reverberant plants with slow modes, since the convergence rate is limited by the slowest time scales of the system. It is shown that Rate Feedback (RFB) control can be applied to a system to minimize the transient response and enhance the convergence characteristics of the TAG algorithm. An experimental demonstration of persistent and transient disturbance rejection in a cantilevered beam is presented. Results show that increasing the system damping by a factor of 4-6 increases the convergence time of the TAG algorithm by as much as 20 times.