A modification to filtered-X LMS control for airfoil vibration and flutter suppression

The effects of error path model dynamics and choice of reference signal on the performance of the filtered-X least main square (LMS) algorithm for active vibration and flutter suppression is investigated using a two-degree-of-freedom aeroelastic (pitch-plunge) system. It is shown that the convergence coefficient of the filtered-X LMS algorithm must vary in order to reduce vibration over a broad airspeed range. The control reduces vibration at sub-critical airspeeds and is capable of extending the critical speed by 11%. A modification to the control scheme is presented where the error path model is removed from the feedforward loop. The modified method also reduces vibration at sub-critical airspeeds while maintaining a constant convergence coefficient, and extends the critical speed by 50%. A performance study is conducted comparing the maximum amplitude reduction, settling time, and robustness to measurement noise of all control schemes presented.