Control And Analysis Of Primate Bionic Mechanical Limbs

Bionic mechanical limbs are used in many fields,such as military,medical and industrial manufacturing.Its structure and control method are also very different due to the different user.Based on the analysis of the kinematic features of primates and the way of biological signal transmission,this paper designs a set of easy-to-wear bionic mechanical limbs,which are mainly targeted at the aged groups from 60 to 80 years old.This paper first analyzes the kinematic features of primates,and at the same time from the structure of the mechanical limb,the design of the control system of the mechanical limb,including the signal acquisition system and multi-sensor human behavior recognition system.The signal acquisition system designed in this paper is small in size,light in weight and easy to wear,which can meet the daily tasks of sEMG signal and inclination signal acquisition.The sEMG signals of walking,sitting down and standing up and the body Angle signals of daily behaviors of four aged 60 to 80 years old were collected by the designed signal acquisition system.The threshold value of balance state and fall state was divided by the body inclination information collected,and the balance state and imbalance state were classified and verified by the prototype.By using BP network and SVM,the feature vectors of the collected sEMG signals are recognized by three kinds of behavior:walking,sitting down and standing up.When the sEMG signal data collection interval is 300ms,the average classification accuracy of BP network classification is 93.3%,and the average classification accuracy of SVM is 98.9%.Then,the classification accuracy of four data acquisition intervals of surface sEMG signals for three behaviors was analyzed.The results show that when the collection interval is 280ms and 290ms,the average classification accuracy is 98.9%,which is the same as the classification accuracy when the collection interval is 300ms.Therefore,when the SVM classification is used and the data collection interval is 280ms,the real-time performance and classification effect of the mechanical limb are the best.