Damage detection accommodating nonlinear environmental effects by nonlinear principal component analysis

Abstract

Damage detection in structural health monitoring should accommodate the variation caused by varying environmental conditions such as temperature, humidity, loading, and boundary conditions. A structural damage detection technique is proposed to deal with the continuous monitoring data of a structural system subjected to the complex nonlinear behavior caused by varying environmental conditions. Based on the identified or measured target features of the structural system under varying environmental conditions, e.g. stiffness of the structural components, the nonlinear principal component analysis (NLPCA) using autoassociative neural network is performed to extract the underlying environmental factors. Then a prediction model for NLPCA is proposed to estimate the damage extent. This proposed technique is capable of dealing with not only the non-increasing features, e.g. stiffness, but also the non-decreasing features, e.g. damage index, after damage is introduced without measuring the environmental factors directly. A numerical study is performed to demonstrate the advantages of the proposed technique over the technique using principal component analysis. A synthetic bridge model is simulated with the consideration of a specific element stiffness reduction together with the change due to environmental conditions including temperature, gradient of temperature, humidity, and frozen supports. Results show that the extent of stiffness loss can be quantified accurately and promptly after the damage is introduced.

A damage detection algorithm integrated with a wireless sensing system

“…In order to accommodate environmental effects when applying the proposed approach in this paper, data normalization techniques are necessary. Many approaches to tackle the environmental effects when performing damage detection can be found in the literature [21,22]. The methods which deal with environmental effects by implicitly modelling the underlying relationship between the environmental factors and structural stiffness are recommended to be integrated with the proposed approach.…”

A State of the Art Review of Modal-Based Damage Detection in Bridges: Development, Challenges, and Solutions

“…Hsu and Loh [111] implemented a NLPCA method based upon the use of an Auto-Associative Neural Network (AANN), originally developed by Kramer [112], which uses a five-layer mapping-network whose output pattern is identical to its input pattern. The five layers consist of an input layer of measured variables, a sigmoid nodal mapping layer that projects the inputs onto the following feature-space bottleneck layer of linear transfer nodes which act as the nonlinear principle components, followed by a second sigmoid de-mapping layer that finally projects the non-linear principle components back into their original form.…”

Machine learning algorithms to damage detection under operational and environmental variability

“…Several statistical methods have been reported in the literature related to the data normalization procedure [18,19,20,21,22]. These methods are also known more generally as machine learning algorithms because they are designed and developed in such a way that their performance is improved based on the analysis of normal condition (i.e.…”