Expediting life cycle cost analysis of infrastructure assets under multiple uncertainties by deep neural networks

Abstract

Deteriorating and at-risk infrastructure assets should be maintained at acceptable conditions by asset management systems (AMSs) to ensure the safety and welfare of communities. Project-level AMSs have been proposed to optimize maintenance interventions in the life cycle of assets by incorporating probabilistic and complex models but at the expense of relatively high computation time. To make complex project-level AMSs computationally applicable to all assets in a network, this paper presents a methodology to replace the time-consuming simulation modules of optimization algorithms with a trained machine learning model estimating life cycle cost analysis (LCCA) results. Deep neural network (DNN) models were trained on LCCA results of more than 1.4 million semisynthesized bridges based on the US National Bridge Inventory considering different intervention actions and uncertainties about condition ratings, hazards, and costs. Our findings show that the trained DNN models can accurately estimate the complex LCCA results five order of magnitudes faster than simulation techniques. The proposed methodology helps practitioners reduce the optimization and LCCA computation times of complex AMSs to a feasible level for practical utilization.