A convolutional neural network‐based full‐field response reconstruction framework with multitype inputs and outputs

Abstract

Summary Structural health monitoring (SHM) systems evaluate the state of the infrastructures by analyzing the monitored responses. As measuring all target responses is difficult to accomplish due to technical or economic limitations, converting other easy‐measuring responses to the target one is a popular way. Relative approaches are separated into data‐driven and model‐driven ones. This paper proposes a deep learning‐based framework to reconstruct multitypes of full‐field responses. The adopted architecture is a convolutional neural network (CNN) with an autoencoder structure and skip connections. Varied from other data‐driven approaches, the training set in this paper is the responses computed by a finite element model (FEM), with which the CNN can learn the full‐field mapping relationships among varied response types. Therefore, the proposed framework is data‐model‐co‐driven. In the numerical simulation section, a simply‐supported beam and a continuous beam bridge have been adopted to discuss the influence of hyperparameters (training epoch, kernel size, skip connection, and bottleneck size), sensor arrangement, modeling error, and measurement noise, which indicates that the framework applies to the in‐field structures. Furtherly, a laboratory experiment has been conducted to validate the framework using a two‐span continuous bridge with obvious FEM error. All results have shown that the deep‐learning‐based response reconstruction algorithms can obtain the training set from not only in‐field measurements, but also numerical models to improve the diversity of training data.

Multi-End Physics-Informed Deep Learning for Seismic Response Estimation

“…Except for the network width (i.e., the kernel numbers), the network depth, kernel sizes, skip connections, and bottleneck size can all influence the fitting ability of the network in a non-monotonic way [ 40 ]. Herein, we only focused on the width of the network, as increasing the width is a relatively simple way to strengthen the network’s learning ability.…”

Full-field static and dynamic strain measurement by an inverse conjugate beam method with two-type sensor placement

“…A virtual sensor model has been obtained to measure partial vibration based on a CNN [20]. A training set from a FEM is used to learn the full-field mapping relationships [21]. Generative adversarial network (GAN)-based methods have been developed to improve the reconstruction accuracy.…”

Structural seismic response reconstruction method based on multidomain feature-guided generative adversarial neural networks

“…Obtaining significant amounts of real-world seismic response data is difficult because of the relative rarity of severe seismic events for actual buildings and high cost of shaker test operations. Therefore, some scholars have used numerical time-range analysis of structural finite element models to provide datasets for structural response reconstruction models [31][32][33]. For example, Li et al [32] generated seismic response data for training neural networks by performing nonlinear time-course analysis on a 3D finite element model, and the trained model could be used for real-time reconstruction of the seismic response of a high-rise building.…”

“…Therefore, some scholars have used numerical time-range analysis of structural finite element models to provide datasets for structural response reconstruction models [31][32][33]. For example, Li et al [32] generated seismic response data for training neural networks by performing nonlinear time-course analysis on a 3D finite element model, and the trained model could be used for real-time reconstruction of the seismic response of a high-rise building. Pan et al [33] proposed a deep neural network-based model for real-time reconstruction of seismic response of bridge structures, the training data of which was generated by an established finite element model.…”