Spatial-temporal associations representation and application for process monitoring using graph convolution neural network

Modern industrial processes generate many dynamic, associated, and multi-scale variables, which are more likely to implicit spatial-temporal associations knowledge for describing irregular changes at different times. Inspired by this, a novel spatial-temporal associations representation method is proposed for process monitoring. Spe- cifically, numerous variables and their associations simultaneously can be utilized to construct a static graph network snapshot. Then, graph network snapshots corresponding to process states at different times are fed into a graph convolutional neural network to implement graph classification. Finally, process monitoring is realized by continuously identifying each snapshot. Monitoring feasibility and applicability are demonstrated by the Ten- nessee Eastman (TE) benchmark and cobalt removal process application.