Applicability of SVM and LVQ to evaluation of bridge integrity

Abstract

The importance of maintenance of existing bridges rises, and the researches on the automation of the comprehensive evaluation of the visual inspection result have been performed. In some researches, Neural Network is used for finding out the criterion from an actual evaluation of bridge integrity by an expert. In this research, the applicability to the evaluation of bridge integrity is examined by the comparison of SVM and LVQ that is skillful in the pattern recognition. Though it is a result of dependence on the sample data, SVM shows high classification ability with complex parameter setting, and LVQ shows high robustness. In addition, the usability is considered by using SVM and LVQ on R that is statistical analysis tools.