Analysis of feature extraction methods based on multiple objective genetic algorithms and support vector machine for classification EMG signals arm muscles

For years, researchers tried to rehabilitation and construction of artificial organs, which reduces processing time to be appropriate for use in real time, however in these areas have been successful, but in most previous studies to detect the motion before observing was not concerned. It was Shown to help multi-objective optimization methods, can be contrasted between different objectives were to answer that in fact is a compromise between the various objective. The advantage of this method compared to simple objective methods can be used to identify a greater number of solutions, the ability to add different rules in the future and more realistic models of these issues can be mentioned.
The overall goal of this research is to provide a method for diagnosis of real-time movement in arm muscles based on EMG signal processing surface to control a cybernetic arm using appropriate time-frequency features multi-objective genetic algorithm NSGA-II selected by the As well as the use of support vector machine classifier to real-time motion detection is applied.
The results of processing 100 samples of recorded data from 5 healthy subjects indicated arm muscles that the proposed method can be used, with the optimal window mS256 and 98.43% with high accuracy, process, motion detection and diagnosis of arms used in three different modes. It can be claimed that the use of this method, to achieve optimization of error is less and also the convergence speed than previous methods.