Active noise control in a duct system based on a frequency-estimation algorithm and the FX-LMS algorithm

In this paper, an indirect feedback active noise control (ANC) scheme, based on the fundamental frequency-estimation method, is proposed for systems with multiple tonal noises. The engine noise consists of the harmonic components of the rotation of the crankshaft in the engine. When it is difficult to obtain reference signals, which are necessary for a feedforward ANC, conventional ANC algorithms do not work. In this paper, a new method is proposed to generate reference signals with estimated frequencies. The proposed algorithm consists of two parts: the frequency-estimation (FE) algorithm for the estimation of the fundamental frequency of the rotating machinery and then the conventional filtered-x LMS (FX-LMS) algorithm. In the second algorithm, the reference signal is generated using the fundamental frequency as estimated in the first algorithm. The FE algorithm uses a second-order adaptive notch filter, which is insensitive to impulse noise. In addition, the FE algorithm has good tracking capability and a lower variance of frequency-estimation error for a constant sinusoid signal and chirp signal. The performance of the proposed ANC method is verified through simulations and an experiment using a DSP board (DS-1104) inside a short duct.