High-Dimensional Phase Space Reconstruction with a Convolutional Neural Network for Structural Health Monitoring.

Yen-Lin Chen,Pei-Hsin Chiu,Yuan Chiang,I-Chen Huang,Chang-Wei Huang,Yu-Bai Xiao,Shu-Wei Chang

doi:10.3390/s21103514

Abstract

In order to accurately diagnose the health of high-order statically indeterminate structures, most existing structural health monitoring (SHM) methods require multiple sensors to collect enough information. However, comprehensive data collection from multiple sensors for high degree-of-freedom structures is not typically available in practice. We propose a method that reconciles the two seemingly conflicting difficulties. Takens’ embedding theorem is used to augment the dimensions of data collected from a single sensor. Taking advantage of the success of machine learning in image classification, high-dimensional reconstructed attractors were converted into images and fed into a convolutional neural network (CNN). Attractor classification was performed for 10 damage cases of a 3-story shear frame structure. Numerical results show that the inherently high dimension of the CNN model allows the handling of higher dimensional data. Information on both the level and the location of damage was successfully embedded. The same methodology will allow the extraction of data with unsupervised CNN classification to be consistent with real use cases.

Highlights

Our results demonstrate that reconstructed attractors of different damage cases can be discriminated by the convolutional neural network (CNN) model
In most practical in situ measurements with signal noises and measurement errors, such differences are not identified from the measured time-history data. These results reveal the limitation of the frequency-based structural health monitoring (SHM) method and the superiority of the proposed SHM method
It is desirable to construct a representative attractor of the underlying dynamical system from such a sequence. This is achieved using Takens’ embedding theorem; time-delayed versions of this output signal can be gathered to form an attractor with the number of DOF matching that of the original system

Summary

Introduction

Structures suffer from varying levels of damage during their lifespan due to aging, external forces, and environmental changes. The implementation of damage detection for structures, known as structural health monitoring (SHM), is key for both preventing unexpected failures and optimizing the maintenance of existing structures [1,2,3]. SHM is indispensable for the sustainability of structures and attracts extensive attention in both the industry and academia. The goals of SHM can be classified into four categories [4]: (1) to determine whether the structure is in a healthy state, (2) to identify the location of damage, (3) to quantify the level of damage, and (4) to estimate the remaining service life of the structure

Objectives

Methods

Discussion

Conclusion