Data-Based Structural Analysis, Modeling and Fault Diagnosis in Complex Industrial Plants

Monitoring and control of complex industrial systems have become exceptionally difficult due to their permanent increase in size and complexity. For that reason, data based methods for plant and processes analysis are gaining on importance. The size of the available plant or process data to be analyzed raises difficulties in application of data based methods. It is necessary to analyze as much as possible of available measurements, since they contain valuable information about the observed plant or process. The analysis in acceptable time regardless the size of analyzed data remains the main requirement. In this paper a systematic approach to structural analysis, modeling and fault diagnosis which can cope with large quantities of plant and process measurements has been presented. An application example of the discussed approach for monitoring and fault diagnosis in a tandem cold rolling mill in steel industry is illustrated through experimental results.