On the basis of studying the concrete durability and the artificial neural network, this paper selects the suitable network configurations and the most effective improved algorithms according to the concrete carbonization and the bar corrosion. Then the paper compiles the program to predict the depth of the concrete carbonization with MATLAB and analyses the rule of the concrete carbonization. Finally the paper creates the different artificial neural network to appraise the degree of the bar corrosion in two conditions and analyses the feasibility by the actual data.
|