Implementation of unscented Kalman filter for nonlinear state estimation in hydropower plant

There is an inverse response of the pressure in the penstock when manipulating the guide vane of a hydraulic turbine in hydropower plant. Considering safety issues, it is of interest to estimate this dynamic state to compensate the uncertainties in the pressure measurements. Besides, the mathematical model of a hydro power plant is nonlinear, which may make classic extended Kalman filter (EKF) theoretically lose accuracy or not capable of estimation due to its linear approximation way. The purpose of this paper is to achieve fast predictions and more precise estimations by introducing an unscented Kalman filter (UKF) to a hydropower system. The derivation of the UKF is presented. Different scenarios with real measurements are introduced to see the effects of UKF.