Gait Recognition Using GA-SVM Method Based on Electromyography Signal

摘要

To improve the recognition accuracy of the lower limb gait, a classification method based on genetic algorithm (GA) optimizing the support vector machine (SVM) was proposed. Firstly, electromyography (EMG) signals were collected from four thigh muscles related to lower limb movements. Then the values of variance and integral of absolute were extracted as the useful features from de-noised EMG signals. Finally, the penalty parameter and the kernel parameter were optimized by GA. The results show that the GA-SVM classifier can effectively identify five gait phases of the extremity motion, and the average accuracy is increased by 6.56%, higher than the non-parameter-optimized SVM method.