Deep generative Bayesian optimization for sensor placement in structural health monitoring

Abstract

Optimal sensor placement (OSP) is essential for effective structural health monitoring (SHM). More recently, deep learning algorithms have shown great potential in sensor-based SHM. However, existing optimization frameworks, such as population-based algorithms, are often not suited for data-driven SHM. Evaluating a number of sensor layouts includes training on large datasets, which is computationally expensive. This paper proposes deep generative Bayesian optimization (DGBO) as a solution for a parallel optimization of black-box/expensive OSP objective functions. Conditional variational autoencoders are leveraged as generative models that transform the OSP problem into a lower-dimensional latent space. Additionally, DGBO utilizes a surrogate neural network to capture the probability distribution of the objective function space. The proposed method is validated on two case studies on a nine-story reinforced concrete moment frame. The first one serves as a proof of concept to show that DGBO can find the global optimum configuration. The second case study aims to maximize the semantic damage segmentation (SDS) accuracy using a fully convolutional neural network. Transfer learning is proposed in training the vibration-based SDS model, which reduces the evaluation times by more than 50%. Without compromising the performance, the number of accelerometers can be reduced by 52% and 43%, respectively, for damage location and severity predictions. It is also shown that DGBO can outperform genetic algorithm with the same number of function evaluations. DGBO can serve as a scalable solution to address the highdimensionality challenge in OSP for large-scale civil infrastructure.

Erythemato-Squamous Disease Detection using Best Optimized Estimators of ANN

“…It allows itself to concentrate on those regions of the dimensional space that it considers will provide the most hopeful validation scores by selecting its value, also known intelligently. This method often needs fewer iterations to reach the ideal set of hyperparameter values [16]. Most significantly, it ignores those regions of the dimensional space that it thinks won't contribute anything.…”

Uncertainty‐aware convolutional neural network for explainable artificial intelligence‐assisted disaster damage assessment

“…Zou et al 28 explored deep neural network approaches for detecting pixel‐wise patterns of ancient buildings in China. Sajedi and Liang 29 proposed the use of deep generative Bayesian optimization for searching optimal sensor placement in the sensor‐based SHM. Case studies showed that generative machine learning achieves approximately 10% better performance compared to a benchmark genetic algorithm (GA).…”

Large‐scale structural health monitoring using composite recurrent neural networks and grid environments

“…It can potentially provide information regarding the existence, location, type, and extent of the damage. The proposed supervised SHM methods include artificial neural networks (de Lautour & Omenzetter, 2009;Gonzalez & Zapico, 2008;Oh et al, 2017), support vector machines (SVM; Kim et al, 2013;Liang et al, 2018;Sajedi & Liang, 2020a), decision trees (Alves et al, 2015), convolutional neural networks (CNNs; Abdeljaber et al, 2018;Sajedi & Liang, 2020b, 2021a, 2021b, 2021c, and vision-based SHM methods (Gao & Mosalam, 2018;Liang, 2019;Sajedi & Liang, 2019;Sirca & Adeli, 2018). However, acquiring damaged condition data is practically difficult, as structures tend to have different structural properties and site conditions.…”