The Research On Intelligent Tracking Control Of Artificial Pneumatic Muscle

Nowadays,the trend of aging is increasingly severe in China,the aging problem and chronic disease has spawned enormous rehabilitation medical market.There is a growing need for rehabilitation care and the professional therapists are getting scarce.Rehabilitation robots can help even replace the therapists so that human costs can be saved significantly.The pneumatic artificial muscle actuators work very similar to human skeletal muscle and they are suitable for rehabilitation robots.However,the pneumatic artificial muscle actuators have complex nonlinear dynamics and slow response time,which make it difficult to achieve high-accurate control and restricted the widespread applications of pneumatic artificial muscles in many years.In the process of rehabilitation,the therapists provide patients with physical rehabilitation training based on the patient's condition.This process can be abstracted into a tracking control problem,which can be realized by the pneumatic artificial muscle-driven rehabilitation robot.This thesis aims to the research on intelligent tracking control algorithm and experiment of the pneumatic artificial muscle actuators.In this paper,phenomenological model theory is used to identify and model the pneumatic artificial muscles experiment platform.Aiming at the problem of poor self-adaptability and poor control accuracy of pneumatic artificial muscles,firstly,this paper proposed an adaptive fuzzy sliding mode control algorithm which approximate the equivalent-type control term of sliding mode control law by a fuzzy controller.The proposed strategy sets an adaptive law to improve the approximate performance of the fuzzy system,which make it possible for the pneumatic artificial muscles to overcome the uncertainties and disturbances,so that the controller is able to catch up with the expected trajectory with different weights.Secondly,we proposed an adaptive fuzzy sliding mode control algorithm in which the switching-type control term in the sliding mode control law is approximated by a fuzzy system,and the chattering in the sliding mode control can be weaken greatly.Lastly,we proposed a echo state network based predictive control algorithm with particle swarm optimization which is self-learning and can realize the high-accurate trajectory tracking control of the pneumatic muscle system even though the plant model is unknown.This paper not only achieve these algorithms in the simulation environment but also applied them to the actual pneumatic artificial muscle experiment platform.The experimental results show that the three control algorithms perform well in tracking accuracy,anti-jamming performance and self-adaptability.There is some deficiencies in the pneumatic muscle experiment platform due to experimental conditions,while the proposed algorithms still make a contribution to the wide application of the pneumatic muscle.