Model Fittng and State Estimation for Thermal Model of Synchronous Generator

Abstract

The power factor of a synchronous generator, in a European hydropower generation, is constrained to [0.85, 0.95], and for Norway, it is generally below 0.86. A higher power factor means more exploited power (active power), and currents through the generator; and vice-versa. However, relaxation on the constrained power factor for higher active power generation should be balanced with thermal heating due to more winding currents through the generator which eventually influences the generator lifetime. Thus, the measure of temperature evolution in generator metals are vital if the relaxation of power factor is allowed. In this thesis work, we are developing several thermal models for air-cooled synchronous generator based on constant, and temperature dependent resistance and specific heat capacity for metals and fluids (air, water). The developed models are then analyzed, and state estimation algorithms are applied for air and metals temperature estimation. The state estimation algorithms, Unscented Kalman Filter, and Ensemble Kalman Filter are compared. Based on computational speed Unscented Kalman Filter performs better, however, based on estimation accuracy Ensemble Kalman Filter performs better.