Research On Simultaneous And Continuous Motion Estimation For Multiple Joints Of Upper Limb Based On Muscle Synergy Analysis

The surface electromyography(sEMG)is widely used in rehabilitation robots,intelligent prostheses,human-machine interaction and some other fields for its good anti-interference,easy acquisition and it contains abundant motion control information.The traditional pattern recognition based myoelectric interface can only complete the switch between a few simple movements during the actual control process.In order to realize continuous and smooth control of rehabilitation assisted equipment,the estimation for continuous motion information of joint has become a research hot spot in recent years.However,most of the current studies focus on the sequence motion of single degree-of-freedom(DOF)of single joint,while the natural motion of human body is always the simultaneous motion of multiple DOF.In view of the problem of multiple DOF,some scholars have proposed a muscle synergetic contraction model in recent years.Through muscle synergy analysis,continuous motion information of joint can be estimated during simultaneous movements of multiple DOF,providing more intuitive,flexible and efficient control methods for intelligent rehabilitation assisted equipment.This thesis takes the wrist joint and elbow joint of the human upper limb as the research object.The muscle synergy analysis method was studied for multiple joint simultaneous motion analysis combined with the muscle synergetic contraction theory.And then effective synergetic activation model was established to accurately estimate the angle information during continuous movement of joints.Finally,the interactive control system was designed based on the parallel proportional control strategy for multiple DOF,realizing simultaneous and continuous motion control of the mechanical arm with multiple DOF.Specific research work of this thesis is as follows:(1)The sEMG signals and joint angle signals during the wrist and elbow joint movements were collected,and then preprocessing was conducted for the collected signals.Aiming at the problem that the low signal decomposition stability caused by the unsupervised characteristics of traditional non-negative matrix factorization(NMF)method,a semi-supervised NMF muscle synergy analysis method was proposed combined with Moore-Penrose generalized inverse,which could effectively extract muscle synergy elements and their activation coefficients.At the same time,in order to obtain the muscle activation degree in the muscle synergetic model,various time domain features of sEMG signals were extracted,and the best time-domain features for muscle synergy analysis were selected through experimental comparison.(2)To solve the defect that the activation coefficients extracted through muscle synergy analysis cannot directly represent the motion information,and obtain more accurate and stable control information,muscle synergetic activation model of activation coefficients and joint angle information was established through Support Vector Regression(SVR).Brainstorm algorithm was used to optimize the model parameters,which can reduce the optimization time and effectively ensure the high estimation accuracy of joint angle information and the good generalization ability of the model.The above models were respectively established for the wrist and elbow joint of the upper limb,realizing simultaneous estimation of angle information for multiple joint.(3)The application of simultaneous and continuous motion estimation for multiple joints in myoelectric control was studied.An on-line simultaneous and continuous motion estimation system for multiple joints based on muscle synergy analysis was designed.The motion intentions of the subjects were reflected through the estimated joint angle.Then the trajectory of a mechanical arm was controlled by combining multi-DOF parallel proportional myoelectric control strategy,realizing simultaneous and continuous motion control for multiple DOF of the intelligent equipment,which can effectively verify the effectiveness and practicability of the designed simultaneous and continuous motion estimation system for multiple joints in myoelectric control.