DESIGN OF SELF-ADAPTIVE OPTIMIZED DECISION FEEDBACK FUNCTIONAL LINK ARTIFICIAL NEURAL NETWORK BASED ACTIVE NOISE CONTROLLER

Kuo-Hsien Liang ； Wan-De Weng ； Ting-An Chang

《電機工程學刊》 22卷5期 (2015/10) Pp. 165-172

https://doi.org/10.6329/CIEE.2015.5.02

並列摘要

In this paper, we mainly probe into the design of an active noise controller (ANC). The principle of an ANC is based on the superposing of two destructively interfered waves to subdue noise. The ANC system equipped with a functional link artificial neural network (FLANN) structure with filtered-s least mean square (FSLMS) algorithm is used in our design. To improve the system performance, decision feedback (DF) mechanisms are commonly used in the FLANN structure. In this paper, an improved structure, which is known as optimized decision feedback-FLANN (ODF-FLANN), is proposed to further speed up the learning of the controller. Simulation results show that ODF-FLANN exhibits much faster rate of convergence than traditional DF-FLANN. For instance, under the assumption that the input noise is a sinusoidal wave plus additive Gaussian process, it takes 22 iterations for ODF-FLANN to reach stability. To achieve the same condition, however, it takes 374, 94, and 125 iterations for FLANN, DF-FLANN and RDF-FLANN, respectively. To demonstrate the tracking ability of the system, sudden changes of noise amplitude and frequency have been added to the noise. It can be seen that ODF-FLANN possesses excellent performance under each of the circumstances.

並列關鍵字

neural network ； active noise controller ； self-adaptive ； FLANN ； FSLMS