Unscented Kalman filter for a coal run-of-mine bin

Abstract

This paper describes the application of an unscented Kalman filter to a coal run-of-mine bin. A dynamic model of the bin is derived using the principle of mass conservation. The dynamic model is nonlinear with unknown parameters that are identified using actual plant production data. The identified dynamic model is used by an unscented Kalman filter to update the states of the system to improve model output accuracy. The derived bin model with and without an unscented Kalman filter is compared with actual plant data. Results show that the unscented Kalman filter can significantly improve the plant outputs estimated by the bin model on its own.