Sustainability-informed intelligent management of aging civil infrastructure systems with emphasis on bridge networks

Infrastructure systems are subjected to deterioration throughout the duration of their life-cycle from exposure to the environment. The bridge network systems play a crucial role in urban, and their safety are highly related to the carbon emissions within the construction field. This study suggests a deep reinforcement learning (DRL) approach for the sustainability-informed life-cycle management of aging infrastructure systems to meet the goal of dropping the global carbon emissions. The management of aging structures completely considers the environmental, economic, and safety impacts. The management optimization with Markov decision process is achieved with the DRL approach. The effectiveness and efficiency of the approach are validated with a bridge network. The proposed DRL-based management approach maximizes structural conditions while minimizing overall carbon emissions and economic expenses. The proposed approach could also assist stakeholders in efficiently allocating funds to maintain aging structures and understanding the performance, risk, sustainability, and life-cycle of infrastructure assets.