Active learning method for risk assessment of distributed infrastructure systems

Abstract

Event‐based methods are commonly used to assess the risk to distributed infrastructure systems. Stochastic event‐based methods consider all hazard scenarios that could adversely impact the infrastructure and their associated rates of occurrence. However, in many cases, such a comprehensive consideration of the spectrum of possible events requires high computational effort. This study presents an active learning method for selecting a subset of hazard scenarios for infrastructure risk assessment. Active learning enables the efficient training of a Gaussian process predictive model by choosing the data from which it learns. The method is illustrated with a case study of the Napa water distribution system where a risk‐based assessment of the post‐earthquake functional loss and recovery is performed. A subset of earthquake scenarios is sequentially selected using a variance reduction stopping criterion. The full probability distribution and annual exceedance curves of the network performance metrics are shown to be reasonably estimated.

Random fields representation over manifolds via isometric feature mapping‐based dimension reduction

“…Especially, newer and more powerful supervised ML methods have been developed recently, for example, neural dynamic classification algorithm (Rafiei & Adeli, 2017b), dynamic ensemble learning algorithm (Alam et al, 2019), deep reinforcement learning (Chen et al, 2021;Wang et al, 2021), finite element machine for fast learning (Pereira et al, 2019), etc., which enable us to conveniently and rapidly predict the structural behaviors. These methods have already been successfully applied to material properties prediction (Rafiei et al, 2017;Valikhani et al, 2021), hazard early warning (Dong et al, 2021;Rafiei & Adeli, 2017a), construction cost estimation (Rafiei & Adeli, 2018), risk assessment (Tomar & Burton, 2021), etc. The combination of the proposed method and these algorithms may provide a more efficient way to build up surrogate models, which is our future direction.…”

Using machine learning to analyze and predict construction task productivity

“…ML has been used in the design stage (As et al., 2018; Huang & Zheng, 2018; Luo & Paal, 2020; Rodrigues et al., 2017; Valikhani et al., 2020), construction stage (Bai et al., 2019; Poh et al., 2018; Xiong & Huber, 2010; Yu et al., 2019), and the operation and maintenance stage (Farrar & Worden, 2012; Maeda et al., 2020; Okazaki et al., 2020; Yang & Su, 2008; Zhang et al., 2019). ML has also been used in civil engineering applications such as damage assessment of buildings (Cheng et al., 2021), flood warning (Dong et al., 2020), risk assessment of infrastructure systems (Tomar & Burton, 2021). Supervised learning is an ML approach that learns a function that maps input to output based on example input–output pairs and infers a function from labeled training data consisting of a set of training examples (Caruana & Niculescu‐Mizil, 2006; Hastie et al., 2009).…”

Selection of the modeling resolution of infrastructure

“…However, detailed digital twins and their analyses often have prohibitive computational costs and an unnecessary level of details, making digital twins not always the most suitable option to model the performance of infrastructure. On the other hand, past studies (e.g., Ghorbani‐Renani et al., 2020; Liu et al., 2020; Tomar & Burton, 2021) defined simplified infrastructure models at an arbitrary level of resolution based on the available information with limited considerations on the resolution's appropriateness for the intended analyses. The goal of the selection of the model resolution is to allocate computational resources to the model that best delivers the desired information with the desired accuracy level (Arabi et al., 2020; Dao et al., 2017; Mu & Yuen, 2016).…”