Computationally efficient adaptive algorithms for active control systems

Publication Type:
Thesis
Issue Date:
2021
Full metadata record
As the side effect of urbanization, acoustic noise from various sources is the most common health threat in our day-to-day life. Passive noise control methods are constrained by several factors such as frequency content of noise, absorbing material type, thickness and geometry. Alternatively, active noise control method has emerged as a promising solution to control low-frequency noise cost-effectively. In an active noise control system, the acoustic path from the control source to the error microphone affects the control performance. If the reference microphone is placed in close proximity of the control source, an unwanted acoustic feedback signal from the control source will be captured by the reference microphone, which may lead to system instability. Furthermore, the adaptive control algorithms have a high computational complexity, which limits its application with high sampling frequency and the scalability of a control system for generating a larger quiet zone. Various algorithms have been proposed in literature for modelling acoustic paths, low-complexity implementation of single and multiple channel control systems. However, they are still constrained by factors such as computational complexity, noise reduction performance, causality issue and stability issue. The objectives of this PhD research are to develop low-complexity algorithms for (1) online modelling of acoustic paths without affecting noise reduction performance, (2) achieving improved control performance at transient and steady state, (3) high sampling frequency operation and broadband noise control and (4) multiple channel decentralized algorithm for broadband noise control. […] In summary, online acoustic path modelling methods are proposed using the control signal; an affine combination of adaptive filters are proposed for improved control performance; a time-frequency domain flexible structure is proposed for active control operation for high sampling frequency operation; a decentralized algorithm is proposed to achieve similar noise reduction performance as the centralized one for broadband control.
Please use this identifier to cite or link to this item: