Fault diagnosis of chemical processes utilizing signed directed graphs-improvement by using temporal information

A fault-diagnosis algorithm using a signed directed graph as a model of the system is presented for the real-time diagnosis of failures. The accuracy of the algorithm has been improved to select candidates which are the most likely to be the real origin of failure, utilizing the time when the measured variable begins to show abnormality as the representation of the dynamic characteristic of the measured variable. The accuracy and speed of the improved algorithm are examined by its application to data obtained in experiments of fault diagnosis in tank-pipeline systems.<<ETX>>