In structural health monitoring (SHM), measuring and evaluating structural dynamic responses are critical for safety management of civil infrastructures. Particularly, online forecasting of the structural responses under extreme external loading conditions (e.g., earthquakes) takes a significant role in SHM to provide early warning and ensure safe operation. In practice, complex causality and intrinsic interactions between seismic excitation and structural response make it challenging to establish a reliable predictive scheme. The present paper proposes a novel deep recurrent neural network (RNN) model implemented in the architecture of a time-series attention-based RNN encoder–decoder (TSA-RNN-ED), for predictive analysis of structural responses under seismic excitation. In the proposed data-driven model, upcoming sequential responses are predicted through sequence-to-sequence learning from historical multivariate time-series signals. A time-series attention mechanism is proposed to exploit the heterogeneous, but directly related, hidden features between the seismic loads and the corresponding structural responses. The proposed architecture can reliably regress excitation-response interactions to predict dynamic responses subjected to future earthquakes while satisfying the need of real-time forecasting for on-site practical implementation. This article systematically evaluates the proposed model by using two real-world structural cases: 1) the tallest building in China, the Shanghai Tower and 2) a woodframe classroom on a shake table at the University of British Columbia in Vancouver, Canada. The experimental results demonstrate the accurate and efficient performance of the proposed methodology in forecasting the seismic responses of the structures under investigation.

This paper studies the effects of long duration ground motion on three conventional low-rise light-frame wood houses with different sheathing configurations. These are two-story residential houses built with an oriented strand board (OSB) wall system and stucco sheathing, an OSB wall system without stucco, and a horizontal board sheathing wall system without stucco. Three-dimensional numerical models of the structures were developed using the Timber3D program and calibrated with full-scale shake table test data. The shear wall systems were simulated with the modified Stewart (MSTEW) model that accounts for cyclic and in-cycle degradation. For this study, seven short duration ground motions from the China Lushan earthquake of 2013 and seven long duration motions from the adjacent 2008 Wenchuan earthquake were selected. Both suites of motions were recorded at the same strong motion stations and were spectrally matched to have similar spectra. Fragility curves for collapse and drift exceedance were derived based on incremental dynamic analysis (IDA). Results indicate that under long duration motions, median collapse capacities for the three houses are reduced by 26–61% compared to that due to short duration motions. A 1% lesser peak interstory drift (ISD) approaching collapse was also estimated under long duration motions. In addition, for a 3% design drift limit, there is a 17% higher probability of exceeding this drift due to long duration shaking. This study demonstrates that from the three types of houses analyzed in this study, house that had blocked shearwalls with OSB sheathing showed significantly higher seismic resistance than the other two housing types analyzed, while house that used horizontal wood boards showed the lowest resistance. Structures with horizontal board sheathing should not be permitted in areas of significant seismic activity.