Dynamic structure change of a multichannel FXLMS-based active noise control system

Despite the intensive research in active noise control (ANC) done for twenty years in many respected institutions around the world, industrial applications are still limited to several types only. This is caused by the fact that most solutions are inappropriate for reducing noise in industrial conditions, for instance in very large halls, where many machines generate noise, and many people work in or move inside the halls. Local noise control appears to be a suitable solution for this problem. However, for successful implementation several conditions must be met. As people move, the zones of quiet should follow their heads. Due to the movement, contribution of noise generated by subsequent machines changes. Likewise, while the person is getting closer or farther from a particular secondary source, it's contribution to the overall noise reduction is getting stronger or weaker, accordingly. In a response to such situation classical adaptive algorithms try to compensate the gain loss by increasing the controller's gain, what can be dangerous for system stability and involve high power consumption. The aim of the paper is to propose a modification to the classical multi-channel FXLMS algorithm in order to protect against unpleasant sound effects due to the above problems. It is obtained by applying a time-variant, mutually dependent weight parameters in the multichannel system. Additionally, when elements of the system are far enough from the position where the zone of quiet should be generated, their adaptation is stopped in order to reduce the computational load. The paper presents also results of noise reduction obtained for experimental data acquired in a large room simulating industrial conditions.