Approximate Synchronization Control System And Methods For Symmetrical Rehabilitation Of Fingers

With the development of aging society,the number of patients with hand dysfunction caused by stroke is increasing year by year.Hand is an important part of human body.The damage of hand function will seriously affect the normal life of patients.The finger rehabilitation robot can help the hemiplegic patients to recover their hand movement function by bending and stretching,so that the patients can return to normal social life as soon as possible.According to the needs of hand function rehabilitation of hemiplegia patients,a near synchronous control system for symmetrical rehabilitation of fingers is designed in this paper.The real-time finger posture data collected by the angle acquisition system of leap motion controller is used to train the BP neural network optimized by genetic algorithm.The predicted finger posture results drive the finger rehabilitation robot to move,help the hemiplegic patients to complete the symmetrical movement of fingers,so as to restore the movement function of the disabled fingers.In this paper,the experimental prototype has been successfully developed and tested,and the following indicators have been achieved:the accuracy of symmetrical motion recovery has reached more than 95%;the accuracy of attitude prediction has reached more than 96%;the delay from the attitude calculation to the action execution of the upper and lower position machines is controlled within 1 s.The experimental prototype system meets all the design requirements and achieves the expected results.First of all,this paper designs the executive mechanism of finger rehabilitation robot with "three finger" manipulator structure.The mechanism consists of three mechanical structures:thumb,index finger and middle finger,with six degrees of freedom.This design ensures that the palm can complete 90%of the normal hand movement,simplify the control strategy,and quickly complete the symmetrical rehabilitation movement.At the same time,the weight of the manipulator is greatly reduced and the user's rehabilitation experience is improved.Secondly,a set of perfect embedded system is designed for the executive mechanism of rehabilitation robot.Based on stm32f103rct6 platform,the hardware circuit of motor drive and the communication circuit of upper and lower computer are designed to drive the robot finger to complete the action.The embedded real-time operating system(RTOS)architecture is adopted in the system software design,and real-time task scheduling is used to run tasks such as attitude calculation,motor drive,data communication,etc.the embedded real-time operating system can improve the utilization efficiency of the hardware platform,optimize the expansion ability of the software platform,and achieve "low coupling".Finally,based on reducing the random time delay of communication and mechanical transmission,the BP neural network prediction algorithm optimized by genetic algorithm is adopted.BP neural network is optimized by genetic algorithm to predict the angle data of the patient's normal hand,and the predicted results are transmitted to the manipulator to drive the patient's fingers to carry out synchronous rehabilitation training.Finally,the prediction results of BP neural network and BP neural network optimized by genetic algorithm are compared to verify the effectiveness of the latter prediction.