Robust Fault Diagnosis Based on Nonlinear Model of Hydraulic Gauge Control System on Rolling Mill

A nonlinear model of a hydraulic automatic gauge control (AGC) system is established for fault detection and isolation (FDI). By analyzing the relationship between faults and load uncertainties, a decoupling subsystem has been derived using a differential geometric approach. An exponential gain observer has been designed based on the observable decoupling subsystem. Diagnosis residual signal is sensitive to designated faults and robust to load uncertainty. Two real data examples verify that the observer is stable and asymptotically convergent. The correctness and superiority are testified by actual data examples.