Adaptive techniques for active cancellation of periodic sounds
MetadataShow full item record
This thesis describes an active noise control system designed to cancel periodic noise, or APNC (Active Periodic Noise Control) System, as it is called throughout the thesis. The thesis begins by briefly describing the acoustic environment (a cylindrical duct) in which the system works. After that, the APNC System itself is described in detail with various simulations to illustrate the behavior of the same. Various alternatives for the system are explored, between them the use (or not) of compensation for the error path and the type of compensation that should be used (FIR or HR filter). A new on-line modelling algorithm is also derived, and its effectiveness in modelling the error path is investigated. The behavior of the system for a sweep in frequency is also analyzed. The thesis also describes a graphical approach to a stability analysis of the APNC system. Through this stability analysis is not only demonstrated the effects of the compensation for the error path in the stability of the system, but also the effects of the step size p and system order N in the stability of the system. The implementation of the APNC system in real time is also discussed. Matters of practical consideration (like inversion of the error signal for use in the LMS algorithm and error averaging) constitute the main focus of this section of the thesis. Performance of the real-time compensated system and uncompensated system are compared and analyzed. It is also demonstrated that the real time compensated system presents a reasonable overall cancellation for the spectrum of frequencies in which it works. The thesis is concluded by a brief assessment of the material included in the same. It is deliberated that the new on-line modelling technique and the graphical stability analysis are two of the main contributions of the thesis. It is also deliberated that the APNC system in question is inherently more stable than other active noise control systems that use an input microphone. It is also mentioned that, due to the small number of computations required by the APNC system, it can be implemented in a more inexpensive chip, therefore yielding a cheaper active noise control system.