RECOGNITION OF SEMG HAND ACTIONS BASED ON CLOUD ADAPTIVE QUANTUM CHAOS IONS MOTION ALGORITHM OPTIMIZED SVM
In this study, in order to improve the accuracy of human hand motion pattern recognition, a novel pattern recognition method for optimizing the support vector machine (SVM) by using a cloud adaptive quantum chaos ions motion optimization (AQCIMO-SVM) algorithm is proposed. The maximum values of wavelet coefficients were extracted as feature samples from the de-noised surface electromyography (sEMG) signals, which were collected from the forearm muscles of several subjects, and then the extracted feature was inputted into an SVM to classify action recognition. In addition, the AQCIMO algorithm was applied to optimize the penalty parameters and the kernel parameters of the SVM, which are used to avoid the uncertainty and complexity of parameter selection and improve the recognition precision of the model, thus improving the model recognition accuracy. The simulation results demonstrated that the two types of movement, which included basic gestures (rest, hand grasp, hand extension, wrist down, and wrist up) and object grabbing gestures (pre-grab, grab, transport and place, release hand, and return to the original position) were successfully identified by the SVM method combined with the AQCIMO algorithm. Compared to mainstream and classic classifiers, namely, GA-SVM, PSO-SVM, and AFSA-SVM, the accuracy of the proposed method was higher by 4.2% to 8.2% than that of the aforementioned classifiers. Therefore, the AQCIMO-SVM algorithm can efficiently solve the problem of the classification of the action pattern of the sEMG signals, which has a very important practical value.