Learning in intelligent systems for process safety analysis

Abstract Process safety analysis is necessary for analyzing and assessing in detail the inherent hazards in chemical processes. We have developed a tool (called PHASuite) to assist experts conducting process safety analysis. PHA is knowledge intensive, and the analysis capacity and quality of PHA Suite depend exclusively on the quality of domain knowledge. It is, however, impossible and impractical to encode all kinds of knowledge into the knowledge base during development phase of PHASuite. Thus, the major aim of this work is to address the important practical learning needs. The learning-from-experience strategy using case-based reasoning methodologies and learning from data using Bayesian learning, are investigated.