On the neural learning and harmonic control of a planar complex

Abstract

This paper presents results on the harmonic control, in contrast to the fixed-point control as in the cases of inverted pendulum balance and truck backer-upper. In general, the harmonic control for driving a plant trajectory to a periodic orbit or a limit cycle is more difficult than the fixed point control. We use four a priori trajectories as well as a neural-based system identifier to train the controller. During the performance test on a planar complex, the controller is shown capable of driving the test plant to follow the predefined periodic orbit (i.e., limit cycle stability), given an arbitrary initial state and external disturbance added incidentally. The resulting planar model can be applied to real-life examples, such as the stability control of shipping-building, and smoothness control of rocking chair design.