
تعداد نشریات | 8 |
تعداد شمارهها | 428 |
تعداد مقالات | 5,580 |
تعداد مشاهده مقاله | 6,876,196 |
تعداد دریافت فایل اصل مقاله | 5,845,972 |
Fatigue Diagnosis Utilizing the SVM Classification of Wrist EMG Signals with Feature Selection | ||
AUT Journal of Mechanical Engineering | ||
مقالات آماده انتشار، پذیرفته شده، انتشار آنلاین از تاریخ 17 مهر 1404 | ||
نوع مقاله: Research Article | ||
شناسه دیجیتال (DOI): 10.22060/ajme.2025.23810.6157 | ||
نویسندگان | ||
Farnaz Dehkordi1؛ Majid Sadedel* 2؛ majid Mohamadi moghadam1 | ||
1Department of Applied Design, Faculty of Mechanical Engineering, Tarbiat Modares University, Tehran, Iran | ||
2تربیت مدرس-مهندسی مکانیک | ||
چکیده | ||
Robotic Rehabilitation illustrated advantages over traditional methods for the past decade. Biological signals such as electromyography (EMG) signals, are the perfect description of human intention of movements and they could also be perceptible to robots. Pattern recognition of movements is used to diagnose the fatigue and weakness of the patient's muscles. In this study, by evaluating and processing the EMG signal of the wrist, an attempt has been made to diagnose the wrist's muscle fatigue in terms of the patient's EMG signals without the need for wrist movements. For this purpose, by performing laboratory tests of EMG signals for both normal and fatigued wrist subjects, processing and extracting the appropriate features of each signal, wrist movements are divided into four levels in terms of weakness. Sixteen features for each EMG signal have been computed, and SSC (Slope Sign Change), WAMP (Willison Amplitude method), MMAV (Modified Mean Absolute Value), SSI (Simple Square Integral), and MYOP (Mayopulse Percentage Rate) perform better to separate the different levels. The SVM classification method has been implemented on EMG data to classify them into four predetermined levels. The feature selection improves the total accuracy of classification from 89.8% to 93.57% for flexion movements, from 75.9% to 93.2% for extension movements, and from 95.3% to 96.8% for supination-pronation movements. | ||
کلیدواژهها | ||
Electromyography؛ Feature Extraction؛ Pattern Recognition؛ SVM classification | ||
آمار تعداد مشاهده مقاله: 1 |