Soft Sensor Transferability: A Survey

Soft Sensors (SSs) are inferential dynamical models employed in industries to perform prediction of process hard-to-measure variables based on their relation with easily accessible ones. They allow implementation real-time control and monitoring the plants present other advantages terms costs efforts. Given complexity industrial processes, these generally designed data-driven black-box machine learning (ML) techniques. ML methods work well only if data which is performed share same distribution one model was trained. This not always possible, since can often show new working conditions. Even similar different distributions, making SSs scalable between them. Models should then be created from scratch highly time-consuming procedures. Transfer Learning (TL) a field that re-uses knowledge task learn different, but related, one. TL techniques mainly used for classification tasks. Only recently have been adopted SS field. The proposed survey reports state art nonlinear design. Methods applications discussed directions this research depicted.