Lost data recovery for structural health monitoring based on convolutional neural networks

Abstract

Summary Signal transmission loss of using wireless sensors for structural health monitoring is a usual case, which undermines the reliability of the sensors for monitoring the structural conditions. The measured vibration data with a high data loss ratio can hardly be used for the analysis, that is, modal identification, as it will lead to significant errors in the results. This paper proposes a novel approach based on convolutional neural networks for recovering the lost vibration data for structural health monitoring. The used network is a fully feed‐forward convolutional neural network with bottleneck architecture and skip connection, which constructs the nonlinear relationships between the incomplete signal with data loss measured from the sensors with the transmission loss and the complete true signal. The trained network extracts the robust higher representation features of the measured incomplete signals using the compression layers and expands those features gradually throughout the reconstruction layers to recover and obtain the complete true signals. The long‐term vibration data from Dowling Hall Footbridge are employed to validate the effectiveness and robustness of the proposed approach for the lost data recovery. Two case studies are conducted to validate the recovery accuracy for single‐channel and multiple‐channel cases, respectively. The effect of sampling rate on the recovery accuracy is also investigated. The proposed approach exhibits the outstanding capability of lost data recovery, even when the signals have severe data loss ratios up to 90%. To further demonstrate the reliability of the recovered signals for data analysis, modal identification results by using the recovered signals with different data loss ratios show a very good agreement with those obtained from the complete true data.

Damage detection in beams from modal and wavelet analysis using a stationary roving mass and noise estimation

“…Based on this fact, vibration-based damage detection techniques try to detect the presence of damage by analyzing the change in natural frequencies, mode shapes, and/or damping ratios. Some pioneering damage detection techniques [1] were based on the analysis of changes in natural frequencies, which are the most simple dynamic parameters to be measured. However, the natural frequencies are not sensitive to damage.…”

“…There is a significant number of papers that propose different techniques and damage detection parameters to analyse the information provided by mode shapes. [1] The wavelet transform is a rather new mathematical tool that has been developed from the 90s for signal processing and information encoding. [4] The wavelet transform is sensitive to local changes in the original signal.…”

Mass-normalized mode shape identification method for bridge structures using parking vehicle-induced frequency change

“…As important dynamic property of bridge structures, mode shapes are commonly used in the field of structural health monitoring, such as finite element model (FEM) updating, [1][2][3] damage detection, [4][5][6][7][8] and so on. Thus, considerable attention has been given to mode shape identification techniques during the past decades.…”

An enhanced energy vibration-based approach for damage detection and localization

“…Extensive review papers provide an easier access to vibration-based DI; see, for instance, several studies. [5,[10][11][12][13][14][15] Many DI methods rely on modal parameters, that is, natural frequencies, damping coefficients, and mode shapes, which need to be identified before applying the method itself. Output-only structural identification methods, usually referred to as operational modal analysis (OMA), deal with modal parameter identification based on structural responses induced by unmeasured so-called ambient excitation sources; see, for example, Cunha et al [8] From a mathematical point of view, according to Reynders and De Roeck, [16] the unmeasured, ambient forces are usually modeled as stochastic quantities, such as zero-mean Gaussian white noise times series and unknown covariances.…”