Filtered-error recursive least squares optimization for disturbance feedforward control in active vibration isolation

This paper addresses the filtered-error recursive least squares (FeRLS) algorithm for disturbance feedforward control in active vibration isolation systems. The controller structure consists of a generalized finite-impulse response (FIR) filter to include a set of pre-determined poles and self-tuning zeros. In addition, residual noise shaping is included to add frequency weighting and improve robustness. Compared to existing filtered-error least mean squares (FeLMS) algorithms, two major improvements are distinguished. First, faster convergence is obtained without the necessity of pre-whitening and an orthonormal basis. Second, the parameters are estimated without steady-state variance. These improvements are demonstrated using simulation studies, which show the potential of the algorithm in active vibration isolators.