Stress corrosion damage evolution analysis and mechanism identification for prestressed steel strands using acoustic emission technique

Abstract

Stress corrosion is a major failure type of prestressed steel strands damage. Currently, no effective monitoring method exists. This paper is an analysis of the acoustic emission (AE) characteristic signal from the stress corrosion damage to prestressed steel strands using the ant colony optimization and self-organizing feature mapping. First, AE characteristic signals at the different stages of the stress corrosion were obtained through the stress corrosion experiments on prestressed steel strands, which can primarily present the corrosion mechanism and different corrosion sources. Subsequently, the ant colony optimization was applied to analyze the AE characteristic signals of stress corrosion. This resulted in the identification of the four main types of AE sources of stress corrosion on prestressed steel strands. The AE ant colony optimization cluster analysis, based on the principal component analysis technology, can separate the four types of damage sources totally and judge the evolution process of corrosion damage and broken wires signal easily. Finally, the self-organizing feature mapping neural network technology applied to the pattern recognition of stress corrosion on prestressed steel strands. The AE characteristic parameter distribution of different clusters can be realized.