A locomotion-mode recognition method for intelligent prosthesis based on step by step machine learning
In order to remedy the shortcomings of feature value selection and single classification model in gait recognition of intelligent prosthesis,and improve the accuracy of gait recognition of intelligent prosthesis wearers,surface lec-tromyography(sEMG)signal is selected as the gait recognition signal source,a feature extraction method based on gray scale model is proposed,and a road condition recognition model based on step by step machine learning is es-tablished.The gray-scale model coefficient is selected as the input characteristic value.The deep fine wave neural network is used to identify the road condition and distinguish the easily confused gait.An extreme learning machine optimized based on pollination algorithm is used to further distinguish the confusing road conditions,and the final recognition accuracy is 98.25％and the recognition time is 70.48 ms.Compared with a single machine learning model,the accuracy of this method is higher than that of a single machine learning model for seven gaits:level ground walking,upstairs,downstairs,uphill,downhill,standing up and sitting down.
万方数据
