Object-Based Diagnostic Network Based on Statistical Learning

摘要

This study proposes an effective diagnostic framework that can be constructed using historical data readily available from the process. The whole system is decomposed into its local diagnostic models based on the direct and local causalities of process variables and the statistical learning model for each local relation is developed using data available from the process. The decomposed local models and the underlying fault assumptions compose of an object-based diagnostic network to perform on-line fault diagnosis. The diagnostic performance of the proposed method has been successfully illustrated in CSTR process.