A Fuzzy Neural Network Technique for Crack Assessment of RC Bridges

As cracking is one of the leading and most critical types of deterioration for reinforced concrete (RC) bridges, how to properly assess cracking condition is of great importance for effective concrete bridge management. However the current assessment techniques are subjective, time-consuming, and rely mostly on human visual inspection. In order to improve the effectiveness of cracking condition assessment, a fuzzy neural network (FNN) system was developed in this study, as an alternative to traditional assessment methods. Based on the comprehensive analysis of RC bridge cracking, this paper also set up the assessment criteria for RC bridges for both flexural cracking and shear cracking, and gives the form of membership function for each criterion. Finally the system was verified using statistical crack samples of several RC bridges.