Identification and control of a jumping ring system using dynamic neural networks

The intent of the paper is to describe a neural unit called the dynamic neural unit (DNU), and to use this structure to identify a jumping ring system. The DNU is developed based on the concept of state space which is in contact with a conventionally structured artificial neural unit. Application to a jumping ring system, the dynamics of which are nonlinear and controlled by an AC voltage supply. An aluminum ring is levitated by electromagnetic induction and the levitated ring is measured by a photo sensor. Real motion data is acquired by a 12 bit analog digital converter interfaced to a PC486, D/A converter, and peak sample holder. Experimental results are given to verify that the proposed DNU is useful in a practical sense