三維密閉聲場之多通道適應之強健主動式噪音控制系統

Abstract

本研究探討了適用於三維密閉聲場之主動式噪音控制(ANC)系統的 兩種控制架構及兩種控制器演算法。我們使用多通道的最小均方 法(Least-mean-square,LMS)以及H強健控制理論來合成控制器，同時比較前饋式 (feedforward)及回饋式(feedback)控制架構對噪音控制效果之影響。 我們應用了尤拉參數法(Youla'sparameterization)來推導多通道回饋 式最小均方演算法。本研究所提出的方法均以一浮點數位訊號處理器 (fioating-poimt DSP)來加以實現，並且以一矩形木箱之內部空間作 為實驗用的聲場。致動器(actutor)和感測器(sensor)的數目及位置對 控制效果之影響亦在我們探討之列。我們使用了寬頻隨機噪音、引擎 噪音、及衝擊式噪音作為實驗中的主要噪音源，實驗結果顯示前饋架 構比回饋架構有較寬頻的消音能力，而最小均方法和H控制則有相 當的噪音抑制效能；H控制方法亦能有效地抑制暫態衝擊式噪音。

Active noise control (ANC) techniques for a three-dimensional enclosure are compared in terms of two control structures and two control algorithms. The multiple-channel least-mean-square (LMS) algorithm and the H∞ robust control algorithm are employed for controller synthesis. Both feedforward and feedback control structures are compared. The Youla's parameterization is employed in the formulation of the multiple-channel feedback LMS algorithm. The algorithms are implemented by using a floating-point digital signal processor (DSP). Experiments are carried out to validate the ANC approaches for attenuation of the internal field in a rectangular wooden box. Position and number of actuators and sensors are also investigated. A broadband random noie, an engine noise, and an impact noise are chosen as the primary noises in the experiments. The experimental results indicate that the feedforward structure yields broader band of attenuation than the feedback structure. The LMS control and H∞ control achieve comparable performance. The H∞ control is also effective in suppressing the transient impact.