Abstract
The fundamental of structural inspection is a visual examination of representative parts of the structure. Several accidents on bridges in Japan and USA two years ago have shown that not only the bridge inspection but also the estimation of structural safety by bridge experts are necessary. But it is almost impossible to perform that kind of estimation of structural safety due to economical condition of the local governments. In this paper, the synthetic safety estimation of infrastructures by SVM (support vector machine) is proposed. Generally SVM requires the input data of the training data like the estimation of the structural experts. In this paper the input data are the deterioration degrees of the structural parts and the output data is the working year of the structure. As numerical example, 2959 inspection data of the bridges in Hokkaido are used as training data and, for example, the health ranking of the bridges is shown.
