Adaptive System Modelling for Active Attenuation of Sound

Active Noise Cancellation (ANC) is the process of using a secondary sound source to cancel an undesired primary sound. Using a system modelling approach, the parameters representing both the process model and the disturbance model are obtained for the overall ANC configuration. The acoustical elements in these two models are thoroughly investigated, including analysis based on a novel transfer function technique. Parametric system identification methods are developed to estimate the process and disturbance models. After verification by computer simulations, these algorithms are implemented in a practical set-up both in real and non-real time. The controller, which regulates the process generating the secondary sound, is thoroughly investigated using feedforward and feedback design. Adaptive processing is also incorporated to accommodate the time-varying nature of the overall system. Extensive computer simulation verifies the ANC controller design for both broadband and periodic sound. Finally, a digital signal processing (DSP) environment is developed, incorporating both high-level software and DSP hardware. From this set-up, a real-time adaptive noise controller is successfully implemented giving significant noise attenuation. There is an attenuation up to 25dB for periodic sound, while for broad-band sound, an attenuation up to 10dB can be achieved. This work, which has blended together acoustic modelling, adaptive control and design of a DSP environment, proposes one solution to the ANC problem.