Analyses for the component effect on atmospheric corrosion of low alloy steel in Guangzhou by artificial neural network prediction

摘要

In this article, a Back-Propagation (BP) neural network model is developed for prediction of the corrosion loss of low alloy steel in the atmosphere. The model is built on the corrosion database of 17 low alloy steels exposed in the atmosphere in Guangzhou. An artificial fish-swarm algorithm (AFSA) in dealing was modified for optimization of the BP neural network. Using the optimized model, the effects of chemical elements on the corrosion resistance of low alloy steel in the atmosphere was analyzed. The involved elements were Cu, Cr, Ni and Mo for commonly-used ones to enhance corrosion resistance and the other ones of Mn and Mo. The results indicate that the model can well predict the corrosion loss of low alloy steels. It is also demonstrated that the elements of Mo, Ni, Cu and improve the corrosion resistance, while Cr abnormally has detrimental impact continuously. Owning to its prediction capacity, the model can serve as the method for optimum design of low alloy steel for service in a similar environment.