Machine Learning Informed Knowledge Driven Framework supporting Holistic Bridge Maintenance

Bridge maintenance is a complex task, which demands a wide spectrum of factors to achieve multi-objectives, multi-criteria optimum decisions. Physics-informed analysis can simulate complex and closely coupled problems, e.g., bridge structural analysis. However, it cannot account for some loosely coupled discrete factors, which complementarily could be addressed by ontological based semantic inference. This paper presents an overarching machine learning (ML) informed knowledge driven framework, which can enhance existing and static knowledge base via dynamically linking to real-time ML information for bridge structural safety as the governing consideration, to make accurate and holistic maintenance decisions. The framework includes semantic modelling, ML based numerical modelling and the Web OWL based reasoning mechanism for integration. The approach could contribute to some fundamental mindset changes towards maintenance decision making.