Modifications on base isolation design ranges through entropy-based classification

A modified base-isolated design technique is proposed and shown to effectively protect structures against extreme earthquakes without sacrificing performance during the more frequent, moderate seismic events. In general, optimal designs can be found through numbers of iterations which are determined by the modification of initial properties, excitation loading and feedback adjustments. However, it is very time-consuming while applying such procedures. If the ranges of inputs (design parameters) and outputs (structural response) could be reasonably estimated beforehand, then the computational efforts could be reduced. The authors introduce a data mining technique, entropy-based classification, to improve preliminary base-isolated design procedures. More specifically, the present study is to extract advisable ranges of design parameters from information of limited preliminary design results. These modified ranges are then used to verify the base-isolated machinery system designs for acceptable level of lower displacement responses. The base-isolated machinery systems investigated are located at Central Science Park, Taiwan. The time-histories records from Chi–Chi earthquake (1999) are collected from 10 different record stations surrounding the park. The modified design data ranges from this study render rational reductions of structural responses.