Condition Monitoring for Hydraulic Systems in Rolling Mills Using Unscented Kalman Filter

Abstract

This contribution describes a predictive maintenance strategy through a modelbased condition monitoring scheme for hydraulic systems in rolling mills, with specific application to the wrapper roll control of a hot strip coiler. For this purpose, a diagnosis strategy is developed using the Unscented Kalman Filter (UKF) technique based on typically available measured signals at the investigated plant. The considered hydraulic system is simulated under the development tool MATLAB/Simulink®. Comprehensive nonlinear modeling is necessary to improve the accuracy of the developed monitoring scheme. Due to the highly nonlinear behavior of the considered system, the UKF is suitable to estimate the hydraulic system states and parameters compared to other approaches, such as the Extended Kalman Filter (EKF). Investigated faults in this work are internal leakage between the two cylinder chambers, raised friction in cylinder, raised hysteresis in valve, and eroded control edges. The main contribution and challenge of the proposed approach is not only to identify the type of the possible occurred fault(s), but also to provide an estimation of the extent of the detected fault(s).