Fault diagnosis of industrial systems with bayesian networks and mutual information

The purpose of this article is to present two new methods for industrial process diagnosis. These two methods are based on the use of a bayesian network. An identification of important variables is made by computing the mutual information between each variable of the system and the class variable. The performances of the two methods are evaluated on the data of a benchmark example: the Tennessee Eastman Process. Three kinds of fault are taken into account on this complex process. The challenging objective is to obtain the minimal recognition error rate for these three faults. Results are given and compared on the same data with those of other published methods.