การศึกษาผลการเปรียบเทียบตำแหน่งและจำนวนอิเล็กโทรดของระบบควบคุมอุปกรณ์ด้วยสัญญาณไฟฟ้ากล้ามเนื้อจากข้อมือ

Abstract

In general, myoelectric control system is controlled by forearm EMG signals, but from literature review it has been mentioned that wrist EMG signals can be used for myoelectric control system as well. Therefore, this engineering project focuses on the used of wrist EMG signals for myoelectric control system. The objective of this study was to compare the results of using wrist EMG signals by varying number and site of electrodes. The signal was collected from 7 subjects while they randomly performed 10 hand gestures (not including resting) in 3 hand positions and compare the classification accuracy of each model, such as 4 channels model, 2 channels model A and 2 channels model B. Linear Discriminant Analysis (LDA) was used as a classifier and gave the classification accuracy at 7 6.7 6 %, 6 7.93% and 69.19% respectively. To increase the efficiency of classification, The One vs One Multiclass Classification (OVO) was applied to the LDA and gave the classification accuracy of 81.10%, 6 9.76% and 70.96% respectively. The results of statistical tests showed that the 4-channel model had a statistically significant difference with the two 2-channel models the statistical test in comparison between using either LDA or LDA(OVO) showed that when using LDA as the classifier 4 channels models was statistically significant different from 2 channels model and while compare between LDA and LDA(OVO) found that 3 models were statistically different.