Researches On Forecast Of The Rate Of Steel Corrosion Based On Grey Relational And Artificial Neural Network

Problems about the breakage of concrete structures induced by corrosion of steel in concrete have already become increasingly outstanding. The direct and indirect loss annually caused by it during different countries is much beyond expectation, and more and more concerned about. It is very important to study the question of steel corrosion in concrete for evaluating the resistance and dependability of the existing structures, accurately estimating service life and the surplus life span of structures, etc.This thesis reviews the steel corrosion's development process and various analysis theories' development. Pass to various compare within the steel corrosion fix quantify, certainty, experience and theoretical analysis, thought to adopt these two kinds of methods of grey system theory and the BP neural network method to carry on the analysis to the steel corrosion is a more convenient on the engineering. Basic principle, basic theory, basic method, main contents and mathematics theory of grey system theory are introduced. For the factors of the influence steel corrosion, not only need to know each factor's importance that affects steel corrosion, but also to know the degree of importance. This thesis introduces the grey relational analysis which can turn the factors' important to sequence and quantity.This thesis makes use of the grey relational analysis to find out the relational sequence of the steel corrosion influence factors and gives corresponsive sequence. Estimates the steel corrosion using BP network of artificial neural network in MATLAB toolbox . It can be gained weighting coefficient by using grey relational analysis. Simulates this data that calculated again because of coefficient using BP net work . It is reasonable that make use of mufti-method in complex problem of steel corrosion.