Nonlinear proportional integral-derivative controlled rehabilitation robotics by neural network / 中国组织工程研究

ABSTRACT

This paper presents a motion rehabilitation-training instrumentfor stroke patients.Taking linear motion mode and circinal motion mode as training trajectory. we are aimed to help patients to accomplish nerve rehabilitation according to preset motion trajectory. Rehabilitation-training instrument takes pneumatic artificial muscle as a motion driver.Pneumatic artificial muscle has the following characteristics,such as,light weight,strong power,good flexibility,safety in use, easiness and cleanliness in source, etc. and it is suitable for rehabilitation medical treatment.The compression of gas,the nonlinearity of pneumatic artificial muscle and the existence of inherent hysteresis in the joint make the driving joint position in pneumatic artificial muscle be poor controlled and the response time be delayed,and thereby,the trajectory planning problem will be influenced directly.Therefore,we adopts proportional-integral-derivative (PID) controller by neural network to control the position of pneumatic artificial muscle.