本研究針對傳統回授式FXLMS演算法來改善噪音干擾時減噪量不足的情形,提出在管路減噪系統中,將FXLMS演算法搭配運用Haar小波函數轉換與低通濾波器,可達到更好減噪效果。文中首先討論管路所遭受的兩種噪音干擾源:管路系統量測雜訊與外部環境噪音。接著指出,經由模擬與實驗發現,對於上述兩噪音干擾源的各類型噪音(雙頻雜訊、四高頻雜訊週期訊號以及白雜訊訊號),透過結合小波轉換與低通濾波器的分析,可分離出正確的誤差訊號,並將訊號A/D轉換給數位訊號處理器作為第二路徑控制器使用。此時,藉由FXLMS演算法線上更新權重,再透過控制器所計算產生的反噪音訊號,輸出D/A給揚聲器,就可經由此種主動式的破壞性干涉,達到優於傳統方式的減噪效果。
In this study, we examine the use of the conventional feedback filtered-x least mean squares (FXLMS) algorithm to improve noise reduction during noise interference. We recommend the use of the FXLMS algorithm with Haar wavelet function transform and low-pass filters to improve noise reduction in duct noise reduction system. First, we discuss the two noise sources encountered in ducts: internal duct sensor noise and external environmental noise. Next, the conducted simulations and experiments indicate that a combination of wavelet transform and low-pass filter analysis can correctly isolate the error signals from each type of the disturbance noise (dual-tone, four-tone, and white noise). Analog to digital (A/D) signal conversion is performed and the signal is then sent to a digital signal processor that acts as second path controller. The FXLMS algorithm is then used to update the controller weights online. The anti-noise signal calculated and generated by the controller is then sent to a digital to analog (D/A) converter before being output to the speaker. This destructive interference signal can achieve superior noise reduction as compared to that of conventional methods.