Neural network for trajectory tracking control of a flexible micro-manipulator

Artificial neural networks may be viewed simply as a wide class of computational models for representing input/output relations. Such networks are attractive for actuator control. This paper clarifies the classification of the neural network controller architecture and its learning process. The micro-manipulator is made of a bimorph piezoelectric high-polymer material (polyvinylidene fluoride). The control system consists of a series combination of both a plant with a feedback loop and a neural network with a feedforward loop. In a direct inverse controller, the neural net learns the inverse dynamics of the system. On the other hand, in a feedback-error-learning controller, the neural network functions as the reference input filter and it organizes a new reference signal to the closed-loop circuit. Experimental results for the positioning control of a piezopolymer manipulator are presented and they show that the proposed control system is effective in accurately tracking a reference signal