Explainable probabilistic deep learning framework for seismic assessment of structures using distribution‐free prediction intervals

Abstract

A new probabilistic framework is proposed for providing a distribution-free prediction interval (PI) of seismic responses required for various earthquake engineering applications. The framework overcomes the limitation of point prediction models and avoids the complexity of traditional probabilistic methods. The framework utilizes a few assumptions of probability distributions and requires no prior assumed statistical distribution for the PI. Ensemble probabilistic deep learning models (DLMs) are used to provide quality-driven PIs of seismic responses for low-to mid-rise buildings with limited irregularity. Considering these systems and ground motions with the aid of Monte Carlo simulation and nonlinear time-history analysis (NLTHA), huge datasets are generated for training. To have an insight into the probabilistic DLM, explainable artificial intelligence techniques are used. The superiority of the proposed framework in quantifying uncertainties is validated by comparison with the conventional Bayesian method. In addition, its applicability is investigated by providing bounds of seismic fragility curves, life cycle cost, and resilience index obtained by NLTHA for a benchmark case study model. The results showed that the proposed framework is robust and outperforms the conventional Bayesian method in uncertainty quantification for the considered dataset.

A Feasibility Study on Microseismic Monitoring of Rock Burst in Traffic Tunnels

“…In optimizing the distribution of geophones, reference should be made to the theories of the optimal distribution of seismic networks. Noureldin et al [37]. considered the theory and proposed a Monte Carlo algorithm for the numerical calculation of the monitoring capacity of seismic networks and a method for the design of microseismic networks based on D-value and C-value optimization design theory.…”

Machine learning–assisted drift capacity prediction models for reinforced concrete columns with shape memory alloy bars

“…In the context of predicting the strength of RC members using ML, extreme gradient boosting model specifically tailored for predicting shear strength of squat RC walls (Feng, Wang, Mangalathu, & Taciroglu, 2021), shear strength prediction of RC deep beams using seven ML models and interpretation of feature impor-tance (Feng, Wang, Mangalathu, Hu, et al, 2021), and ensemble approach to predict the shear capacity of RC deep beams (Pak et al, 2023) have been conducted. For the case of system-level application of ML, multidimensional fragility of skewed bridges using surrogate model made of artificial neural network (ANN) , steel frame buildings under fire using data augmented by employing both Monte Carlo simulation and random sampling (Fu, 2020), rapid damage classification of RC buildings using three ensemble models (Mangalathu, Sun, et al, 2020), investigation on the effect that each input feature has on the RC slabs under blast loading using ML algorithm (Almustafa & Nehdi, 2020), seismic performance assessment of structures using deep learning (Noureldin et al, 2023), nonlinear flutter behavior of bridges using long short-term memory network (Li & Wu, 2023), and dynamic response prediction of structures using a physics-driven support vector machine (SVM) (Luo & Paal, 2023) have been conducted. ML has also been employed in the prediction of concrete compressive strength (Rafiei et al, 2017b), as well as the analysis of how concrete mixtures (Rafiei et al, 2016(Rafiei et al, , 2017a and type and gradation of coarse aggregates influence the compressive strength.…”