Development of a compound intensity measure using partial least-squares regression and its statistical evaluation based on probabilistic seismic demand analysis

Abstract

The primary motivation of this study is to develop a compound intensity measure by partial least-squares regression (PLSR) model and evaluate the compound intensity measure based on probabilistic seismic demand analysis. To illustrate this, the compound intensity measure, which can be obtained by a linear combination of six intensity measures in the log-scale, is assessed by using the nonlinear time-history analyses of the single-degree-of-freedom (SDOF) systems represented by four types of hysteretic models under a suite of 98 natural ground motion records. The compound intensity measure is developed by a multivariate regression model to investigate the potential for practical implementation offered by SDOF systems in the entire period ranges. It is observed that the efficiency of the compound intensity measure for SDOF systems represented by different hysteretic models is found to be approximately of the same magnitude, and the compound intensity measure is proficient and sufficient in response estimation with good confidence. This validates the robustness of the presented method in estimating the predictability of the compound intensity measure on structural responses.