Active control of vehicle powertrain noise using inverse model LMS algorithm

Current active noise control (ANC) technology cannot yield a balanced performance over broad frequency range when applied to powertrain noise control. It is because most of these ANC systems are configured with the traditional filtered-x least mean squares (FXLMS) algorithm with an inherent limitation in the frequency-dependent convergence behavior. In particular, the phase delay of the secondary path in the FXLMS algorithm will significantly affect the convergence speed and thus lead to a relatively poor tracking ability for the transient event. In this study, a novel inverse model least mean square (IMLMS) algorithm is proposed for active powertrain noise control system with an enhanced convergence speed in order to better track the variation of noise signatures due to unavoidable change in the engine speed. The IMLMS algorithm is realized by utilizing the inverse model of the secondary path to minimize the effect of its dynamics on algorithm's convergence to gain a significant improvement in the convergence speed and tracking ability. Numerical simulation using measured powertrain noise responses is also performed to demonstrate the effectiveness of the proposed algorithm. Results show obvious improvement in the convergence speed and appreciable noise reductions over a broad engine rotational speed range.