The Design and Evaluation of Complex Radial Basis Functions Network Adaptive Robust Variable Sliding Mode Controller for Stroke Rehabilitation Robot

Abstract

In accordance with the movement coordination principle of both lower limbs of human beings, a complex radial basis functions network adaptive robust variable sliding mode controller with healthy-dyskinetic side coordination for active stoke lower limb rehabilitation robot was proposed. The technology was specific to stroke patients, the movement information of the patient's healthy side was detected in order to drive the rehabilitation movement. The purpose was to guide patients to the active training driven by autonomous consciousness from passive rehabilitation training, helping patients form gait memory in the cerebral cortex and restoring the connection between the injured central nervous system and the limbs. The variable sliding mode control was proposed to stabilize the system. In order to reduce the problem of chattering, the universal approximation of radial basis functions neural networks was used to approach and compensate external disturbances and uncertainties. The final asymptotic stability was guaranteed with Lyapunov criteria. Compared to PID, the higher performances of the proposed controller were demonstrated by co-simulation.