嵌入式經驗模態分解及其在即時生理訊號 處理應用之實作

Design of Embedded EMD and its Real-Time Implementation for Biomedical Signal Processing

林義軒(Yi-Hsuan Lin)

指導教授 ： 葉經緯 ； 林泰吉

國立中正大學/工學院/電機工程學系/碩士(2012年)

摘要

經驗模態分解（empirical mode decomposition；EMD）已證實在「非線性」與「非穩態（non-stationary）」訊號的分析有卓越的效果，但其運算量極大，故尚未見其應用在嵌入式即時訊號處理中。本論文提出適用於嵌入式系統之EMD設計並用以成功實作一嵌入式即時心電訊號雜訊消除器。首先，本論文針對缺乏數學模型之EMD提出嵌入式最佳化設計環境：經由「合成雙頻（dual-tone）測試訊號之拆解還原」及「心電訊號雜訊消除之雜訊容忍度」有效量化及評估EMD之設計參數與最佳化演算法對其拆解能力之影響；另外，針對合成雙頻訊號我們提供一套快速設計空間探索方法可大幅降低參數及演算法評估之時間。本論文同時提出「結合類小波濾波之混合式分解法」與「多率EMD」進一步降低計算複雜度：前者可有效減少EMD中遞迴篩選演算法之疊代次數，同時有效改善EMD常見的模態混雜（mode mixing）現象；後者則有效利用EMD拆解過程中訊號頻寬逐步降低的特性，動態降低訊號取樣頻率大幅降低資料儲存及運算量。最後，針對嵌入式硬體實現，本論文進一步提出「切割式（segmented）立方雲線（cubic spline）計算」有效降低系統緩衝記憶體需求。我們所提出之「結合類小波濾波之混合式分解法」、「多率EMD」、及「切割式立方雲線計算」等EMD最佳化設計均以FPGA完成設計驗證，可即時濾除心電訊號雜訊。相對於原始EMD演算法，本論文所提出之設計大幅降低30%運算量，並減少59.7%記憶體使用。

關鍵字

希爾伯特-黃轉換 ； 經驗模態分解 ； 立方雲線 ； 去除心電雜訊

並列摘要

Empirical mode decomposition (EMD) has outstanding performance in non-linear and non-stationary signal analysis. But it is not widely adopted in embedded and real-time signal processing applications due to its high complexities. This thesis proposes an embedded EMD design, which has been integrated into an embedded real-time ECG de-noising system successfully. First, an embedded optimization framework has been constructed, where “the decomposition of synthetic dual-tone signals” and “the noise removal capability of ECG” have been proposed to quantitatively evaluate the various design parameters in EMD and optimization schemes. A compact test signal has also been presented to improve the design space exploration time. Then, a hybrid decomposition scheme with DWT-like filtering and a multirate EMD algorithm have been proposed to significantly reduce the computational complexities. The former effectively decreases the required number of siftings in EMD, while mitigating mode mixing effects. The latter exploits the incremental bandwidth reductions during EMD to dynamically down-sample the signal for dramatic savings in storage and computations. Finally, segmented cubic spline computation has been proposed to reduce the buffer size in embedded hardware realizations. All design techniques have been implemented and verified using FPGA and a real-time ECG de-noising system has been demonstrated. Compared with the original EMD algorithm, 30% computations and 59.7% memory space have been saved with our proposed design techniques.

並列關鍵字

Hilbert-Huang transform ； empirical mode decomposition ； cubic spline ； ECG de-noising

參考文獻

computational efficiency of empirical mode decomposition algorithm by downsampling,” International Journal of Advanced Information Technologies, pp. 56-70, Dec. 2010.

[1] J. A. Antonino-Daviu, M. Riera-Guasp, M. Pineda-Sanchez, and R. B. Perez, “A critical comparison between DWT and Hilbert–Huang-based methods for the diagnosis of rotor bar failures in induction machines,” IEEE Trans. Ind. Appl., pp. 1794-1803, Sep./Oct. 2009.

[2] N.E. Huang, et al., “The empirical mode decomposition and the Hilbert spectrum for non-linear and non-stationary time series analysis,” in Proc. R. Soc. Lond. A, pp. 903-995, 1998.

[3] A. Zeiler, et al., “Empirical mode decomposition: an introduction,” in Proc. IJCNN, 2010, pp.1-8.

[4] M. Blanco-Velasco, B. Weng, and K. E. Barner, “ECG signal denoising and baseline wander correction based on the empirical mode decomposition,” Comput. Biol. Med., pp. 1-13, Jan. 2008.