Predictive multi-channel Feedback active noise control for HVAC systems

Abstract

An important application of active noise control (ANC) system is to eliminate the acoustic (sound) noise pollution in modern living environments. In this paper, we present an ANC method and its implementation for reducing the acoustic noise generated mainly by the compressor of an actual HVAC (Heating, Ventilation, and Air-conditioning) system used in a commercial or residential environment. To achieve cost effectiveness and robustness of the ANC system, we propose a “Feedback” ANC method which avoids the use of reference signals thereby removing additional (source or reference) microphones. Furthermore, using the periodic property that is inherent in the compressor's noise, a “Period Aware Linear Prediction” filter is proposed and implemented in the Feedback ANC. The proposed technique is used to form a multi-channel Feedback ANC having Multiple Input and Multiple Output (MIMO). We refer to this new architecture as the MIMO-PALP ANC system. The MIMO structure enables us to achieve larger effective canceling zone around the surface of the compressor. Because of the fact that additional information has been used in the prediction part, MIMO-PALP gives higher noise attenuation level compared to the conventional Feedback ANC in MIMO ANC system designs.