Research On The Method Of Motion Intention Recognition Of Intelligent Lower Limb Prosthesis Based On Mixed Feature Space

Motion intention recognition is an increasingly complete research field in the intersection of robotics engineering and rehabilitation medical engineering.The recognition effect of human motion intention determines the fitting degree of intelligent lower limb prosthesis and lower limb amputee.How to accurately and quickly identify the movement intention of lower limb amputees and help lower limb amputees to achieve natural,smooth and stable walking is the focus and difficulty of this study at the present stage.In the current research on motion intention recognition of intelligent lower limb prosthesis,feature selection has an important influence on the accuracy and time of motion intention recognition.Existing studies often use a variety of types of sensors,according to the sensor embedded in the lower limb prosthesis body and the collection of human biological signals collected on the residual limb,and rely on prior knowledge to select statistical characteristics,to realize the recognition of human movement intention.The traditional statistical features reflect the basic trend of human movement from a global perspective.However,the nature of human movement is continuous,and the statistical characteristics are not considered enough.In addition,it is difficult for sensors embedded in prosthesis to accurately reflect the motion intention of lower limb amputees and the hysteresis problem exists.In this paper,a single type of mechanical signal sensor is used to collect the timing movement data of lower limb amputees' uninjured side in the early phase of swing phase.At the same time,on the basis of the motion intention recognition strategy of the intelligent lower limb prosthesis before the movement pattern transformation of the affected side,the geometric feature,the maximum value slope,which reflects the continuous motion characteristics of the human body is constructed by using the functional data analysis method.The method of motion intention recognition of intelligent lower limb prosthesis based on the first type of mixed feature space and the second type of mixed feature space is constructed by combining geometric feature and statistical feature.This method can predict the next movement type from the whole level and the local change,so as to improve the recognition method.The rationality and feasibility of the proposed algorithm are verified by using the self-collected data set.The experimental results show that the hybrid feature base not only reduces the complexity of the algorithm by dimension reduction,but also effectively reduces the training time of the classification model,and the recognition accuracy is improved to some extent.