Tapped Delay Neural Network for Active Noise Control
碩士 === 國立中興大學 === 機械工程學系所 === 94 === Tapped delay neural network (TDNN) for active noise control is investigated in this thesis. TDNN can be used to represent a model of a secondary path in an active noise control (ANC) system. TDNN can also be used as an ANC controller to attenuate noise in the ANC...
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Others |
Language: | zh-TW |
Published: |
2006
|
Online Access: | http://ndltd.ncl.edu.tw/handle/92006816328074271172 |
Summary: | 碩士 === 國立中興大學 === 機械工程學系所 === 94 === Tapped delay neural network (TDNN) for active noise control is investigated in this thesis. TDNN can be used to represent a model of a secondary path in an active noise control (ANC) system. TDNN can also be used as an ANC controller to attenuate noise in the ANC system. A filtered-x back propagation (FXBP) algorithm is utilized for TDNN in ANC applications and generally leads to diverge. To overcome this difficulty, two approaches are considered. One approach applies a modified residual error for FXBP. Another approach applies a robustly controlled secondary path for TDNN. Results of computer simulation and experiment show that our proposed approaches can effectively improve the convergence property of FXBP algorithm for TDNN in ANC applications and result in good performance of noise reduction.
|
---|