A Study on the Classification Effect of sEMG Signals in Different Vibration Environments Based on the LDA Algorithm

Area of Science:

• Biomedical Engineering
• Rehabilitation Technology
• Human-Machine Interaction

Background:

• Myoelectric prostheses aid disabled individuals, but their adaptability to challenging work environments, particularly those with vibration, remains understudied.
• Vibration is a common factor in many industrial and occupational settings, potentially impacting the functionality of prosthetic devices.

Purpose of the Study:

• To investigate the feasibility of using myoelectric prostheses in vibration environments by analyzing grasping intention recognition under varying vibration frequencies.
• To evaluate the performance of Surface Electromyography (sEMG) signal processing algorithms for prosthetic control in the presence of mechanical vibrations.

Main Methods:

• An experimental platform simulating 0-50 Hz vibrations was developed.
• Surface Electromyography (sEMG) signals during gripping actions were recorded using the MP160 system.
• Six time-domain features and Linear Discriminant Analysis (LDA) were employed for sEMG feature extraction and intention recognition.

Main Results:

• The LDA classifier achieved high accuracy, reaching up to 98.4% with optimal feature sets (e.g., RMS, MIN, VAR).
• Recognition accuracy was significantly affected by vibration frequency, with notable decreases around 20 Hz.
• High average accuracies were observed at 0 Hz, 40 Hz, and 50 Hz, indicating potential for use in specific vibration conditions.

Conclusions:

• Myoelectric prosthesis control is feasible in certain vibration environments, but performance varies with frequency.
• Specific vibration frequencies, like 20 Hz, pose challenges for accurate grasping intention recognition.
• Findings provide crucial guidance for enabling disabled individuals to work effectively with myoelectric prostheses in diverse vibrating conditions.