Profile monitoring based on transfer learning of multiple profiles with incomplete samples

Profile monitoring is an important tool for quality control. Most existing profile approaches focus on a single profile. In practice, multiple profiles also widely exist and these contain rich correlation information that can benefit the of interested/target this article, we propose transfer learning framework to extract profile-to-profile inter-relationship improve performance. framework, are modeled as multi-output Gaussian process (MGP), specially designed covariance structure proposed reduce computational load in optimizing MGP parameters. More importantly, contains features dealing with incomplete samples each profile, which facilitates sharing among different data collection costs/availability. The method validated compared various benchmarks extensive numerical studies case study ice machine temperature profiles. results show successfully knowledge from related performance target R code paper would be available on-line supplementary materials.