Health Index Estimation Through Integration of General Knowledge with Unsupervised Learning

arXiv:2405.04990v1 Announce Type: new
Abstract: Accurately estimating a Health Index (HI) from condition monitoring data (CM) is essential for reliable and interpretable prognostics and health management (PHM) in complex systems. In most scenarios, complex systems operate under varying operating conditions and can exhibit different fault modes, making unsupervised inference of an HI from CM data a significant challenge. Hybrid models combining prior knowledge about degradation with deep learning models have been proposed to overcome this challenge. However, previously suggested hybrid …

abstract arxiv complex systems cs.ai cs.lg data general health index inference integration knowledge making management monitoring monitoring data systems through type unsupervised unsupervised learning