Proposal of a Practical Ensemble System for Multiple Feature Selection in Damage Judgment of Bridge

Abstract

Recently, it has become more and more important to establish a rational maintenance program for infrastructures. In bridge structures, concrete slabs are suffering from damage, because they directly resist applied loads. In this study, an attempt is made to develop a new damage detection method for concrete slabs with cracks, by using pattern recognition techniques. In the pattern recognition, the feature extraction is a quite important process, because the recognition rate depends on the characteristic of the recognition object. The previous research proposed the method which can select efficient features among various features. However, the method can't apply the problems with many input values. The proposed method can deal in those problems by introducing ID3 with classifiers consisting of many feature vectors. Numerical examples are presented to demonstrate the efficiency and applicability of the proposed method.