Design Of Finger Rehabilitation Device Based On SEMG

With the aggravation of aging,stroke has become one of the three major diseases threatening human health.Most stroke patients are accompanied by hand motor dysfunction,which seriously affects the quality of life of patients and their families.With the development of robot technology,rehabilitation robot is used in the rehabilitation treatment of hand function,but the clinical performance of passive rehabilitation training is not satisfactory.Therefore,the active finger rehabilitation device is researched and developed,and the surface electromyography(sEMG)of the patient’s arm is used as the control source,which plays a promoting role in improving the patient’s awareness of active participation and the effect of rehabilitation training.Firstly,the biological characteristics of human hand are analyzed,and a 3-DOF finger rehabilitation device structure is designed based on the coupling relationship of finger joints.Its transmission mode is the hybrid transmission of gear rack and steel rope;At the same time,the metacarpophalangeal joint is designed as the central projection mechanism to solve the interference and slip between the affected finger and the mechanism.The kinematic model of finger rehabilitation device is established,and the rationality of mechanical structure design is verified by software simulation and mathematical model simulation.Secondly,in order to obtain the patient’s action intention,six channel sEMG signals of the patient’s arm are collected,and the intention is recognized by BP neural network.The sEMG signal acquisition,signal filtering,active segment detection and eigenvalue extraction of four gesture actions are completed,the BP neural network model is built,the number of hidden layer nodes and sliding window size of the model are optimized,and the accuracy of off-line gesture prediction is90.2%.In order to meet the requirements of portability,BP neural network is transplanted to the embedded platform after model training by PC.Then,a distributed master-slave control scheme is proposed,and the software and hardware design of the control system of finger rehabilitation device are completed.Stm32f407igt6 is used as the main controller to realize sEMG signal processing and pattern recognition algorithm,and is responsible for the communication with the host computer and slave controller;Stm32f103c8t6 is selected as the slave controller to drive the stepping motor and collect the finger posture.In terms of hardware,the minimum system circuit of master-slave controller,external flash circuit,motor drive circuit and power supply circuit are designed.In terms of software,taking Free RTOS operating system as the system platform,the embedded software development of control system and the upper computer software development for testing and experiment are completed.Finally,the physical prototype of finger rehabilitation device is made and tested,including manipulator function test,control system performance test and active rehabilitation training experiment.The test results show that each module of the finger rehabilitation device has good function,the control system can complete the information acquisition,information analysis and decision-making control process in 200 ms,and the accuracy of online gesture prediction is more than 82%.