Abstract
Material integrity cyberphysical attacks induce defects in parts that compromise their functionality. The socio-economic cost of monitoring-based part disposal and production stoppage necessitates rapid in-process recovery from continued defect formation. But cyberattacks can circumvent existing real-time control by introducing intermittent and a-priori unknown perturbations of exogenous process parameters. This paper presents a novel AI-based framework to address this issue and examines its capabilities for Fused Filament Fabrication (FFF) as an example. We demonstrate real-time recovery from inter-road defects created by attacks on endogenous and exogenous parameters, both trained- and untrained-for, with unprecedented spatial resolution.
