Application of data mining techniques to build master plant relationships based on heterogeneous databases

In general, the process relationship between different entities in the plant are not available explicitly to the user in a digitized format. This work attempted to automatically abstract industrial plant's physical process entities and their relationships based on various engineering information available in different systems. Several heuristics and machine learning methods such as association analysis were used to mine the relational connectivity among process entities, in order to form a master plant entity relationship network. This method was applied on a real plant. The precision of the method was above 80%. This method was also used to identify the strength of relationship between different entities and also segregate between different sections of the plant using community detection. This relationship table can be used for various applications such as fault prediction, root cause analysis and personnel training.