Upper Limb End Stiffness Prediction And Its Robotic Arm-Assisted Rehabilitation Variable Impedance Control

Stroke usually causes different degrees of dysfunction,and about 85% of patients have different degrees of upper limb motor function injury.One of the consequences of stroke is that the motor coordination of muscles is interrupted.Intermuscular coupling is a stable spatiotemporal activity mode in which muscles participate in movement performance at the same time.The analysis of intermuscular coupling of patients’ upper limb muscles is helpful to design rehabilitation exercise and guide the normalization of intermuscular coupling characteristics of patients.The rehabilitation robot has good autonomy and flexible operation performance,and can integrate and cooperate with people to complete the rehabilitation task.For specific rehabilitation tasks,a single trajectory teaching cannot meet the needs of impedance change and achieve the expected rehabilitation effect.Facing the needs of upper limb rehabilitation of stroke patients,this topic deeply studies the prediction of arm end stiffness based on upper limb surface electromyography(s EMG)and the mechanical arm assisted rehabilitation control based on intermuscular coupling model.The main research work includes:(1)Prediction of human upper limb end stiffness based on s EMG.Establish the Cartesian impedance model of human arm,master the impedance size and direction attributes of upper limb,and obtain the expected value of stiffness prediction.This paper studies the acquisition and preprocessing algorithm of s EMG signal,extracts the features of s EMG signal through automatic encoder,predicts the stiffness of human upper limb end by using LSTM neural network,and puts forward the prediction model of upper limb end stiffness based on automatic encoder and automatic encoder long short term memory network(AE-LSTM)neural network.(2)Intermuscular coordination analysis based on coherence analysis.The end traction rehabilitation robot can rehabilitate patients’ upper limbs as a whole.Muscle compensation may occur in the process of rehabilitation.It is necessary to analyze the synergy between muscles.A method of intermuscular synergy analysis based on rshift non negative matrix factorization(rshift-NMF)is proposed.The nonnegative matrix decomposition method is introduced into the intermuscular collaboration analysis to construct the collaboration patterns under different conditions.By introducing the time parameter and considering the time delay between muscle electrical signals,the evaluation effect of muscle cooperation is improved.(3)According to the results of collaborative analysis,we extract the feature points of collaborative action in the patient’s motion trajectory,plan the rehabilitation trajectory of the next stage.The kernelized movement primitive model is used to realize the teaching trajectory planning,so that the planned trajectory passes through the peak point of the coupling characteristics in the historical trajectory to guide the patients to normalize the intermuscular coupling characteristics.Auxiliary variable impedance control strategy of manipulator based on upper limb end stiffness.Combined with the stiffness value of human upper limb predicted by EMG signal,the coordination of mechanical arm impedance parameters on human arm stiffness is realized.