ARX model based fault detection of rolling mill´s automatic gauge control system

In this study a fault detection system is proposed for the automatic gauge control (AGC) system used in hot rolling mill, based on autoregressive estimator model. In hot rolling mills the dynamic control of rolling gap and force is realized with help of electro-hydraulic servo actuators - so-called hydraulic cylinder, capsule - which can adjust the rolling gap at great rolling force dynamically. The dynamic behaviour of the AGC system is mainly depends on the construction of the mill, the work and back-up rolls, hydraulic cylinders and AGC controller. The dynamic behaviour of the hydraulic cylinder is failing by-and-by because of the leakage flow due to abrasion of the seals. This leakage flow can be detected by the examination and fluctuation monitoring of the dynamic parameters of the complete electro-hydraulic servo system. The calculation of parameter fluctuations is realized by an autoregressive parameter estimation system.