Abstract

Control of prosthetic hand was achieved by placement EMG (electromyograph) electrode at specific location on the surface of forearm. EMG signals are no-stationary and noise affected signals, so it can lose their morphology. Good feature extracting algorithm allows the resulting signals in controlling adequately grasp or release of prosthetic hand. In this paper Experimental artificial hand was designed to recognize the right grasp or release timing from forearm EMG signal. New feature extracting method based on wavelet transform (WT) was proposed. One factor of EMG signal was extracted via wavelet transform and gives exact classification of signal characteristics. The resulting EMG grasp classification parameter of wavelet transform is greater than that of release one. The system successes to detect sharp clear grasp or release thresholds signals. It was used to distinguish and detect the two types of signals (grasp or release) with probability error of 0.01. Intelligent fuzzy logic control strategy was constructed to control Releasing, resting and grasping of artificial hand. Each person has own recognition parameters, the system would require to the user to undergo minimal learning to construct his own fuzzy knowledge base, and therefore the control inputs would be natural as possible. . Fuzzy logic is a powerful tool to online control the hand during release, rest or grasp mode.

Keywords Electromyography (EMG), Artificial hand, Continuous wavelets transforms (CWT), Fuzzy logic, Functional electrical stimulation (FES).