Adaptive evolutionary neural control of perturbed nonlinear serial PAM robot

This paper proposes a novel adaptive joint position control system for a highly nonlinear SCARA serial robot using the pneumatic artificial muscle (PAM) actuator. First the new inverse and forward neural NARX (IFNN) models are proposed as to dynamically identify all nonlinear and hysteresis features of the SCARA serial PAM-based robot. Parameters of the new IFNN model are optimized by the modified differential evolution (MDE) algorithm. Secondly, the new IFNN model is applied in the novel proposed adaptive evolutionary neural IFNN-IMC controller that is applied to improve the precision and to reject the steady-state error in the joint position SCARA serial robot control. Finally, the novel adaptive back-propagation (ABP) algorithm based on fuzzy reasoning is applied for online updating the weight values of the IFNN model which helps the novel proposed adaptive evolutionary neural IFNN-IMC controller adapt well to external disturbances and dynamic variations in its operation. Experimental tests confirmed the performance and advantages of the proposed control scheme in comparison with other nonlinear control methods.