Abstract
Several algorithms on the immune network information model, which can be applied to sensor fault diagnosis of process plants, is proposed, and they are evaluated by simulations. The sensor network can eliminate the information of abnormal sensors by mutual recognition.
In cement processes, the reliabilities of sensors, on which the diagnostic system is based, are often far from perfection. The proposed sensor network can be used in such situation that the diagnosis of the diagnostic system itself becomes critical. The usefulness of the sensor network is confirmed by the simulation for the firing section of a cement plant.
