Neural network model for predicting deterioration of bridge components using visual inspection data

As a primary component of a Bridge Management System (BMS), prediction models are crucial for planning the maintenance work, making budgetary decisions, life-cycle analysis and optimization of future maintenance works programs. In a BMS, planned decisions for conducting repairs and rehabilitation are based on inspection data collected for the bridges by trained inspectors following a condition rating method developed by the authority. There is a significant need to produce a practical system where these visual inspection data can be converted into a decision tool. To address this need, a study has been conducted to forecast deterioration of reinforced concrete bridge components using an Artificial Neural Network (ANN). Visual inspection data was sourced from Victoria, Australia and utilized to develop an ANN model using Levenberg- Marquardt optimization with the help of Bayesian regularization process. A neural network based model for prediction of bridge condition rating is proposed. The back-propagation algorithm was used to train the network to recognize the pattern of deterioration of bridges. The various factors which influence the deterioration rate are considered as input to the system. Deterioration of three major structural components was modeled and presented in this paper.