Abstract:
The purpose of this study is to analyze the existing control algorithms and propose the effective control algorithm for passive type rehabilitation training. Proposed technique will help to perform passive type rehabilitation exercises effectively to help stroke patients. The key parameter in passive type rehabilitation is position control. To track the exercise trajectory, a precise position control is needed. Due to human coupling, external disturbances and sensory feedback noises it’s difficult to control the position trajectory of the exercise. This may damage the patients muscle instead of recovering. A mathematical model of a rehabilitation robot is developed using a Lagrangian approach and merged with the actuator model to get the composite transfer
function of the system. Various control algorithms have been analyzed such as conventional PID, Ziegler Nichols PID, Model Predictive Controller (MPC) and fuzzy PID for position control on the developed system model using MATLAB as a simulation environment. The system is subjected to various external disturbances and feedback noises to analyze the controller reliability and robustness. Based on the analysis a new hybrid MPC-PID control technique is proposed to control rehabilitation robot to perform passive type rehabilitation effectively. Hybrid MPC-PID interface is designed based on error computation which switches between Model Predictive Control (MPC) and Proportional Integral Derivative (PID) control based on error computing range.
Proposed technique rejects the external disturbances as well as eliminates noise from sensory feedback and keeps the system stable under extreme perturbed conditions.
