Unsupervised Structural Damage Diagnostic Method Using Judgement of Change of Response Surface by Statistical Tool. Application for Damage Detection of Composite Structure.

Abstract

Most of the structural health monitoring systems adopts parametric method based on modeling or non-parametric method such as artificial neural networks. The former method requires modeling of each structures and latter method requires a large number of data for the training. The modeling and data for the training demands high costs, and it is impossible to obtain training data of damaged state of inservice structure. By the present method, damages are detected by judging statistical difference of data of intact state and present state. The method requires data of un-damaged state and does not require the complicated modeling and large number of data for the training. As an example, the present study deals monitoring of delamination detections of a composite beam. The damages are detected from the change of strain data measured on the specimen surface by the statistical tools such as Response Surface and F-Statistics. As a result, the new method successfully diagnoses the damage without using the modeling and the data of damaged state.