Combining measurements with models for superior information in hydropower plants

Abstract

Water flow and pressure measurements play essential roles in the operation of hydropower plants. For all methods of measuring flow and pressure, there is a level of uncertainty with regards to sensor noise and sensor failure. In addition, measurements in key locations are hard to obtain. A combination of measurements with a mathematical model of a hydropower plant can be used to improve information about and operation of the hydropower system. This paper describes the possibility of using nonlinear estimators such as Ensemble or Unscented Kalman filters in order to estimate the states of the hydropower system based on water flow and/or pressure measurements. The implementation of the estimators is done in Python using a Python API for operating OpenModelica simulations, where the hydropower system is modeled using an in-house hydropower Modelica library — OpenHPL.