We present some discrete-time perspectives on the active vibration control of a simple acausal system subjected to deterministic but broadband excitation. Here, exact vibration cancellation is not possible with bounded control forces. However, on penalizing high control forces using a weighting factor, we obtain a one parameter family of finite-time optimal feedforward (FTOF) controllers that substantially reduce vibration levels through bounded control forces. These controllers require knowledge of the future of the forcing, and are thus acausal. However, only a finite amount of the future needs to be known for practical purposes. These future values are predicted using a neural network. The FTOF controllers perform better than a direct application of the standard LMS algorithm. An LMS algorithm which uses several future values of the forcing performs better still. Some recommendations are made for practical applications.

Date Published: 14 October 2003
PDF: 8 pages
Proc. SPIE 5062, Smart Materials, Structures, and Systems, (14 October 2003); doi: 10.1117/12.514896

Published in SPIE Proceedings Vol. 5062:
Smart Materials, Structures, and Systems
S. Mohan; B. Dattaguru; S. Gopalakrishnan, Editor(s)