The prediction for paint of steel bridges considering deterioration causes

Abstract

This paper describes a method that estimates some deterioration causes of steel bridge paint. A numerical formula that attaches environment and the deterioration causes that environment is composed the relation is supposed to calculate influence degree of the deterioration causes. As the method that calculates the influence degree, the authors use back propagation method (BP) that is a general learning method of neural network that is one of AT technology. By express the numerical formula with a network and using BP, it becomes possible to calculate the influence degree that matches to data most. Actual inspection data has scatter not a little. The scatter can be considered by using BP. Granted, scatter of the data is scatter in the subjectivity by inspector, the method is able to correspond flexibly.