Minghui Cheng
My research
Civil infrastructure greatly shapes our cities and communities. As the changing climate poses an increasing risk, how can we effectively improve the safety, sustainability, and resilience of the infrastructure systems to better serve our communities? To address this question, my research develops efficient decision support tools for civil engineering stakeholders to manage their assets by leveraging big data and artificial intelligence.
System digital twin
I am building a digital twin to manage bridge networks. The digital twin monitors the risk of bridge networks given disparate sources of data. Bayesian network is used to incorporate the data, conduct system analysis, and perform efficient Bayesian updating of annual risk when new data is made available.
Surrogate modeling
I proposed meta-learning-based surrogate modeling (MLSM), which is the first framework to realize the idea of knowledge transfer for surrogate modeling. MLSM is suitable for efficient reliability analysis and optimization at a community level. It has been applied to climate change adaptation of bridge networks and fleet management.
Life-cycle management
Life-cycle management is used to cost-effectively ensure acceptable levels of safety and serviceability of civil infrastructure. Traditional condition-based maintenance strategies employing periodic inspections and fixed replacement thresholds are not cost-effective. Hence, I investigated several adaptive strategies relaxing these constraints. The best strategy (i.e., adaptive inspections with replacement thresholds in governmental guidelines) was then extended to structural systems. I also developed a framework based on deep reinforcement learning to integrate preferences of stakeholders into the strategy.
Measure preferences of civil engineering stakeholders
Behavior patterns affect the selection of optimal life-cycle management plans. Various decision-making models have been developed by economists, but there are no empirical studies to quantify the preferences of civil engineering stakeholders. I conducted two surveys to calibrate the delay and probability discounting models and the cumulative prospect theory to measure them. The calibrated models can better aid in decision-making for the stakeholders.
