Physics-Informed Machine Learning for Structural Health Monitoring

The use of machine learning in structural health monitoring is becoming more common, as many the inherent tasks (such regression and classification) developing condition-based assessment fall naturally into its remit. This chapter introduces concept physics-informed learning, where one adapts ML algorithms to account for physical insight an engineer will often have structure they are attempting model or assess. demonstrate how grey-box models, that combine simple physics-based models with data-driven ones, can improve predictive capability SHM setting. A particular strength approach demonstrated here capacity generalize, enhanced different regimes. a key issue when life-time requirement, data do not span operational conditions undergo. provide overview ML, introducing number new approaches modelling Bayesian main tool discussed be Gaussian process regression, we assumptions/models incorporated through constraints, mean function kernel design, finally state-space range applications demonstrated, from loads off-shore aerospace structures, performance long-span bridges.