Muscle Fatigue Detection And Training Control Of Upper Limb

With the annual growth of stroke patients,rehabilitation treatment for stroke patients is becoming more and more important.At the same time,using rehabilitation robots for rehabilitation treatment has become one of the hot topics at home and abroad.Compared with traditional rehabilitation therapy,robot-assisted rehabilitation training can not only save labor,but also can obtain more accurate recovery information.It is also helpful for doctors to understand the patient's recovery and treat the disease,so as to achieve better rehabilitation training effect.This paper mainly focuses on the issue of muscle fatigue that affects arm rehabilitation training.The fatigue signal is obtained through the new method,and the signal is analyzed then the rehabilitation training is controlled.Patients may become fatigued when they are undergoing rehabilitation training.This fatigue can manifest itself as tremors in the arm muscles.Through the detection of arm muscle tremors,it is possible to assist patients in rehabilitation training to avoid secondary injuries.Detecting the degree of muscle expansion can determine the effectiveness of the patient's training,and looking for the best training intensity of the patient.The research work of the thesis mainly has the following three aspects.Firstly,myoelectric detection platform is built,and the myoelectric signal is detected and analyzed.By comparing the characteristics of myoelectric signals,surface electromyography signals and myodynamia signals,the advantages of selecting myodynamia signals as signal sources are described.Then the myodynamia detection platform consisting of sensors,amplifying circuits,microcomputer and computer is built.The pressure generated by the expansion of the arm muscles is collected by the Flexiforce sensor.After amplification,A/D conversion is performed by the single-chip microcomputer and transmitted to the upper computer through serial communication.Useing Matlab to analyze myodynamia in the time and frequency domain.By comparing the fatigue of muscles in the arms with or without weight,muscle fatigue characteristics are obtained and the factors affecting muscle fatigue are summarized.A new method for determining the optimal training intensity of a patient by obtaining arm muscle fatigue is proposed.Secondly,Bayesian regularization BP neural network predicts muscle fatigue.In order toimprove the accuracy of gradient descent neural network for predict muscle fatigue,Bayesian normalized back propagation was proposed to improve.Establish neural network to predict muscle fatigue model.The formulae of Bayesian regularization neural network are deduced to elaborate the feasibility and advantages of the model.The feasibility of the model was verified by Matlab simulation to predict muscle fatigue.Finally,the auxiliary training is completed through the force/position mixing control of the six-degree-of-freedom rehabilitation robot.During the training,force control is performed on the constraint direction of the robot terminal,and the position deviation in this direction is neglected.The end of robot is considered to be redundant,and the redundant degree of freedom is used to optimize the force control.By setting the force difference performance index,the following forces and positions can be achieved in training.Using Matlab simulation,the feasibility of the control method is proved.