Single Drive Multi-Training Mode Adaptive Wrist Rehabilitation Device

Wrist rehabilitation devices can effectively help stroke patients recover wrist function. However, wrist rehabilitation robots currently on the market generally have problems such as heavy weight and bulky volume, only single degree of freedom training and single training mode. By analyzing the physiological structure of the human wrist, this paper designs a wrist rehabilitation device that can perform multi-degree-of-freedom training and multiple training mode selection. The configuration of the device consists of a drive and a passive joint. While realizing wrist flexion/extension and radial/ulnar deviation rehabilitation training with a single drive, the system mass is reduced to 4.7 kg, and the overall size is controlled at 500 X 200 X 360 mm3, reducing the weight and volume of the device. Three training modes can be provided according to different user needs: active training, passive training, and active-passive fusion training, and the training speed and angle can be changed according to different needs during training. In the study of active-passive fusion training, a PB neural network prediction model based on genetic algorithm optimization is constructed, and online recognition of movement intention by analyzing surface electromyography (sEMG). The experimental results show that the movement traj ectory of the device is smooth, and the maximum output torque reaches 1.04Nm, which can meet the torque required for rehabilitation, and has a good effect on the prediction of wrist movements with a high accuracy rate. The device is small in size and light in weight, and can provide patients with reliable wrist rehabilitation training. It provides an effective solution to meet the multiple needs of patients for home rehabilitation and has broad application prospects.