期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2014
卷号：65
期号：2
出版社：Journal of Theoretical and Applied
摘要：Electromyography (EMG) is a study of muscle function through analysis of electrical activity produced from specific muscle of interest. This electrical activity which is displayed in form of signal as the manifestation of neuromuscular activation associated with muscle contraction. The most well-known technique of EMG signal recording are by using surface (non-invasive) and needle/wire (invasive) electrode. This research focus on surface electromyography (sEMG) signal. During sEMG recording, there are several problems had to be encountered i.e. noise, motion artifact, signal instability, cross talk and signal inconsistency. Inconsistency here refers to the variation of the quantity of EMG features with respect to the quantity of force produce by the muscle. In addition, inconsistency of features to force mapping occurs across different person as well as across different reading of an individual. Inconsistency is due to muscle strength and size, cross talk, signal to noise ratio (SNR), signal bandwidth and fatigue condition. Inconsistency causes nonlinearity or linearity relationship between features to force mapping. Previous method introduce to solve the inconsistency across different reading of an individual but the problem lies on reading across different person. Thus, this paper presents a method to solve the inconsistency of EMG signal across different person by normalizing the EMG based on percentage of maximal voluntary contraction adaptive with muscle endurance (pre-fatigue), %MVCPF. This method is based on a hypothesis that Integrated EMG (IEMG), Mean Absolute Value (MAV), Root Mean Square (RMS), Sum Square Integral (SSI) and standard deviation features are directly proportional to %MVCPF of all person. There are 2 indicator to measure the inconsistency problem which are p-value must be less than 0.05 and the root mean square error of regression must be less than 10%. The results show that p-value of each person signal after normalizing is 0.0126 which is less than 0.05. In addition, the results for root mean square error of regression show IEMG, MAV, RMS, SSI and standard deviation features are less than 10%. These shows the improve normalizing method give the better results to solve inconsistency reading problem across different person. For the conclusion, an objective which is to solve inconsistency features to force mapping issue of different person successful solve.
