Abstract: After analyzing the engineering characteristics of highway bridges in mountainous areas, thispaper used intuitionistic fuzzy analysis to find out four engineering characteristics which have greater in-fluence on the construction cost of highway bridges in mountainous areas, which are in turn adopted asthe inputs to the model to construct a prediction model of highway bridge cost in mountain area based onfuzzy logic and GA-BP neural network. It also designed a program with MATLAB neural network toolbox.and select 36 sets of completed engineering data to train, test, and verify the model. The results showthat the prediction precision of the model meets the requirements. By comparing the results of the CA-BPneural network model and the BP neural network model, it shows that the CA-BP neural network modelhas faster convergence speed, higher accuracy and better stability. It has further verified the feasibilitand validity of the model based on fuzzy logic and GA-BP neural network.

Key words:  CA-BP neural network, highway bridges in mountainous areas, cost prediction