Jianming Tsai

Publication Date


Document Type


First Advisor

Kuo, Sen M. (Sen-Maw)

Degree Name

M.S. (Master of Science)

Legacy Department

Department of Electrical Engineering


Noise control; Adaptive signal processing; Electro-acoustics


Active noise control has been the subject of considerable research in recent years. A large portion of this work has been directed toward the active control of noise in ducts because of potential widespread industrial applications and the feasibility of building experimental systems. In this thesis, a complete analytical model using the acoustical transmission line theory is developed which allows the calculation of the transfer function of individual paths in a duct ANC system for estimating the performance of the system. Three different secondary source arrangements were evaluated in the duct ANC system. Measurements of plant and error path transfer functions agree well with theoretical predictions. For the tree-shaped system, it is shown that the significant performance improvement is achieved at the difficult node frequencies due to the symmetry arrangement of the secondary sources. For the T-shaped and Y-shaped systems, an audio equalizer can compensate for the imperfections in the error path, thus increasing the noise reduction of the systems. This approach is made from a hardware point of view to improve the performance and control the spectral contents of the residual error in the systems. In this thesis, a new algorithm called the filtered-E LMS algorithm is developed to use the filtered error to adapt the coefficients of the FIR filter. The application of this algorithm to active noise control is discussed, by which the suitably chosen shaping filter is able to give the residual noise a desired spectral content. The analysis shows that, in some applications, a slow convergence may be expected. Computer simulations demonstrate that this method can not only attenuate the noise level, but can also effectively reshape the spectrum of the residual noise. It performs well for both narrow band and broad band noise. Most active noise control systems may be treated as system identification problems in which adaptive filters are used to model the responses of unknown electroacoustic plants. For the system identification, it is demonstrated that the signal with the lower peak factor and more concentrated power band can be used as a training signal. The analysis also demonstrates that, for a chirp signal, the time duration affects the performance of system modeling. Some music is able to work in modeling an unknown system and a more stable convergence can be achieved. In addition, an integrated 3D ANC with audio system benefits from the use of music on an on-line basis while the system is operating.


Includes bibliographical references (pages [86]-89)


x, 102 pages




Northern Illinois University

Rights Statement

In Copyright

Rights Statement 2

NIU theses are protected by copyright. They may be viewed from Huskie Commons for any purpose, but reproduction or distribution in any format is prohibited without the written permission of the authors.

Media Type