Research And Realization On Active Training Of A Lower Limb Rehabilitation Robot

How to use the traditional medical scheme combining modern technology to achieve efficient, accurate and user-friendly recovery of limb motor function of patient is the focus of research and practice, especially considering the stroke and limb disabled patients grow with each passing day. The lower limb rehabilitation robot is a kind of auxiliary rehabilitation training equipment effectively save medical resources, reduce the labor intensity of doctors and improve the patient’s quality of life. Active training is a training strategy for the recovery of limb function, which makes full use of the intelligent human computer interaction function of the rehabilitation robot, and completes the man-machine coupling movement based on the patient’s autonomous movement intention.Firstly, this paper introduced the common clinical treatment method of the motor rehabilitation theory, which is the basis of the active training on the hardware platform of the lower limb rehabilitation robot. Then analyzed the mechanical structure and sensor system of this device, which part was the basis of the follow-up analysis and signal processing.Secondly,when active training based on force / torque signals, first it constructed man-machine integration no-load static model which considering the self-balance of mechanism, then it obtained the mapping relationship between the joint variables, no-load torque, measured torque and the force of the patient’s end, subsequently, completes the mapping function of terminal force and terminal velocity according to the damping control strategy, after that, completes man-machine coupling movement according to the joint angular velocity conveied to the drive system. During this period, the sensor signal was filtered by Calman and the affective factors were considered during the training process.Thirdly, when active training based on sEMG signals,first it selected muscle groups which affect joint movement then completed the multi-channel sEMG signals acquisition, after this compareed common time domain characteristics by using the discrete index to find the most three sensitive of the characteristics of the signal fluctuations. Then the BP neural network is trained by using the best feature weights and determined the optimal network with the minimum of error. Then studied the control strategy, and ntroduced the safety measures in the training process.Finally, this paper introduced lower limb rehabilitative robot hardware platform of active training test. This paper also introduces the situation of the lower limb rehabilitation robot in a hospital based on the actual application environment. and this experiment had good results which can verify the feasibility and stability of the lower limbs rehabilitative robot.