A novel method of searching appropriate ranges of base isolation design parameters through entropy-based classification

Abstract

This paper proposes an idea of selecting the proper ranges of design parameters for modifying the responses of base-isolated systems. Parametric analysis of base isolation designs on seismic protection is generally achieved through multiple adjustments of system parameters based on the observations of system responses, but applying such procedures is very time consuming. If the ranges of inputs (design parameters) could be reasonably estimated earlier to obtain suitable outputs (structural responses), then the computational efforts could be reduced. This study generates limited design cases to analyze the relationships between designs and system responses. Specifically, the Shannon entropy, a Data Mining method, can be used to classify the proper and improper ranges of design parameters. A case study on the lead rubber bearing base-isolated machinery located at Central Science Park in Taiwan subjected to several records of Chi-Chi earthquake from different stations is presented. Entropy-based classification is applied in the case study and the practicability of the analyzed result is verified. Copyright © 2008 John Wiley & Sons, Ltd.