A Data-Driven Physics-Informed Method for Prognosis of Infrastructure Systems: Theory and Application to Crack Prediction

Abstract

Infrastructure systems are the backbones of the socioeconomic development of a community. However, after installation, these engineered systems undergo deterioration, leading to a degradation in their condition while in operation. In this work, a generalized modeling framework is proposed and validated for the diagnosis and prognosis of infrastructure systems based on real-time data. A data-driven modeling scheme, dynamic mode decomposition (DMD), is used for prognosis. The novelty of the proposed framework lies in the fact that the developed prognostic model is data-driven and physics informed, and the model works better on problems with unknown/implicit governing equations and boundary conditions. The developed prognostic model provides more accurate predictions based on real-time data and identification of dominant spatiotemporal modes, as evident from the application of mortar cube crack prediction under compressive testing. This framework can be recommended to researchers/practitioners for predicting the remaining useful life of infrastructure components and systems before their maintenance or failure. Such robust predictions of the future condition of existing infrastructure will be beneficial to stakeholders for sustainable development.