【正文】 過程避免不了的信號干擾，我們稱之為噪音。噪音在肌電信號最終體現(xiàn)為分類標簽的過渡數(shù)據(jù)，如下所示：類向量中兩個箭頭間的數(shù)據(jù)為過渡數(shù)據(jù)，即噪音。為減小噪音影響，使得模式識別的精確率提高，本文將采取以下兩種方法對噪音數(shù)據(jù)進行處理?！馦ajority Vote：由上面可知，測試集通過模式識別得出的預期輸出是一個分類標簽，Majority Vote采取窗口=8的間距對分類標簽進行“多數(shù)表決處理”。即當窗口推移到過渡性數(shù)據(jù)時，采取“少數(shù)服從多數(shù)”的方式對過渡性數(shù)據(jù)進行類別優(yōu)化?！袢ピ牒瘮?shù)：該處理方式相對于Majority Vote較為簡單快捷，它直接將過渡性數(shù)據(jù)刪除，從而提高分類效率。Figure 10Figure 116．結(jié)論提取肌電信號肌電信號的預處理提取肌電信號的特征值模式識別分類標簽優(yōu)化處理圖12處理肌電信號流程圖Figure 12 The Flow hart of sEMG processing如圖12所示，本文通過五個步驟對采集的表面肌電信號進行處理分析，分別是肌電信號的拾取、肌電信號的預處理、肌電信號的特征提取以及模式識別。首先對前臂肌肉群肱橈肌、橈側(cè)腕屈肌、橈側(cè)腕長伸肌、尺側(cè)腕伸肌、尺側(cè)腕屈肌等8個位置進行表面肌電信號的采集，然后采用RMS（均方差）與AR模型兩種方法對采集的肌電信號進行特征分析和特征提取，再通過監(jiān)督式學習、LDA分類算法對肌電信號進行模式識別，區(qū)別屈肘、屈腕、屈指和前臂旋轉(zhuǎn)等多種動作。論文的最后還采用了Majority Vote以及去除噪音函數(shù)對分類標簽進行優(yōu)化處理，并使用MATLAB實驗軟件繪圖直觀表示分類方法的識別率。其中，%；僅對分類標簽進行Majority %；%；同時采用Majority Vote以及去除噪音函數(shù)則使得誤差最低，%。參 考 文 獻[1],SPORT SCIENCE2000,20(4):5660.[2]王健,CHINA SPORT SCIENCE AND TECHNOLOGY2000,36(8):2628.[3]雷敏,CHINESE JOURNAL OF MEDICAL INSTRUMENTATION2001,25(3):156160.[4]羅志增,CHINESE JOURNAL OF SENSORS AND ACTUATORS2003,16(4):384387.[5]蔡立羽,王志中,JOURNAL OF DATA ACQUISITION AND PROCESSING2000,15(2):255258.[6]羅志增,CHINESE JOURNAL OF SENSORS AND ACTUATORS2004,17(2):220223.[7]楊廣映,CHINESE JOURNAL OF SENSORS AND ACTUATORS2004,17(3):355358.[8]王人成,鄭雙喜,CHINESE JOURNAL OF REHABILITATION MEDICINE2008,23(5):410412.[9]羅志增,CHINESE JOURNAL OF SCIENTIFIC INSTRUMENT2006,27(9):996999.[10]盧蕾,殷濤,JOURNAL OF CLINICAL REHABILITATIVE TISSUE ENGINEERING RESEARCH2011,15(22):41034106.[11]胡曉,王志中,（英文版）,JOURNAL OF SOUTHEAST UNIVERSITY(ENGLISH EDITION)2005,21(3):324329.[12]李醒飛,朱嘉,CHINESE JOURNAL OF BIOMEDICAL ENGINEERING2007,26(2):166169.[13]李醒飛,張國雄,CHINESE JOURNAL OF MECHANICAL ENGINEERING2004,40(4):3235.[14],(1):57.[15] 吳冬梅,孫欣,JOURNAL OF CLINICAL REHABILITATIVE TISSUE ENGINEERING RESEARCH2010,14(43):80738076.[16] GRAUPE D．Function separation of EMG signals viaARMR identification methods for prosthesis controlpurposes[J]．IEEE Trans．Syst．Man Cybern．1975，5：252～259．[17] KELLY M．The application of neural networks to my oelectric signal analysis：a preliminary studyl,J]．IEEETran．Biomed．Eng．，1990，37：221—229．[18] UNSER M，AIDROUBI A．A review of wavelets in biomedical applications[c]．Proc．Of the IEEE，1996，84(4)：626638．[19]JANG G CH．Using time frequency analysis techniquein the classification of surface EMG signals[A]．Annual Int．Conf．of the IEEE Engineering in Medicineamp。Biology Society．1994，16：1242—1243．[20] DisselhorstKlug C, SchmitzRode T, Rau G. and muscle force: limits in and new approaches for applications. Clin Biomech(Bristol, Avon). 2009。24(3):225235.[21] Sbriccoli P, Bazzucchi I, Rosponi A, et and spectral characteristics of biceps Brachii sEMG depend upon speed of isometric force generation. J Electromyogr 。13(2):139147.[22] Reddy NP, Gupta V. Toward direct biocontrol using surface EMGsignals: control of finger and wrist joint models. Med Eng 。29(3):398403.[23] 羅志增, HMM 的表面肌電信號模式分類[J].華中科技大學學報:自然科學版，2008,36(4):7275.[24],22(2):6366[25] et al .Real t ime put er control using pattern recognit ion of the electromyogram. Annual Int ernational Conference of the IEEE Engineering in Medicine amp。 Biology Society Proceedings. 1993, 15: 12361237.[26] et simulation of artificial hand motionwith Fuzzy EMG pattern RCIEEEEMBS amp。 14th BMESI1995: .[27]. , 23(2) : 80282.[28] of multifunction surface EMG using advanced AR model of the Northeast , 96298.[29] et cascade classification of myoelectric signals. IEEEEMBS and ,139921400.[30] et al. A dynamic neural network ident ification of electromyography and arm trajectory relationship during plex movements. IEEETrans. Biomed. ,43(5):5522558.附錄1. 人體肌肉結(jié)構(gòu)圖圖13人體肌肉結(jié)構(gòu)圖Figure 13 Human muscle structure diagram