Skip to Main Content
Surface electromyogram (sEMG) signals, a noninvasive bioelectric signal, can be used for the rehabilitation and control of artificial extremities. Current sEMG pattern-recognition systems suffer from a limited number of patterns that are frequently intensified by the unsuitable accuracy of the instrumentation and analytical system. To solve these problems, we designed a multistep-based sEMG pattern-recognition system where, in each step, a stronger more capable relevant technique with a noticeable improved performance is employed. In this paper, we utilized the sEMG signals to classify and recognize six classes of hand movements. We employed an adaptive neuro-fuzzy inference system (ANFIS) to identify hand motion commands. Training the fuzzy system was performed by a hybrid back propagation and least-mean-square algorithm, and for optimizing the number of fuzzy rules, a subtractive-clustering algorithm was utilized. Furthermore, this paper employed time and time-frequency domains and their combination as the features of the sEMG signal. The proposed recognition scheme utilizing the combined features with an ANFIS classification provided the best result in identifying complex hand movements. The maximum identification accuracy rate of 100% and an average classification accuracy of the proposed ANFIS system of 92% proved to be superior in comparison with relevant studies to date.