Distributed predictive modeling framework for prediction and diagnosis of key performance index in plant-wide processes

Abstract

In this work, a distributed predictive modeling framework is proposed for prediction and diagnosis of key performance indices in plant-wide processes. With block division of the plant-wide process, key data information can be extracted more efficiently, based on which the predictive model can then be developed for regression of the key performance indices. In order to determine the root causes of performance degradation for the key performance index, a diagnostic scheme is developed among this framework. First, the critical blocks are identified through definition of the block contribution in the diagnostic model. The contribution of each process variable is then evaluated inside each critical block, based on which the root causes of performance degradation can be successfully located. An example of the distributed modeling method is realized by using the basic Principal Component Analysis and Gaussian Process Regression models, with a detailed case study on the TE benchmark process.