Real-time Auto-regressive Modelling of Electric Power Network Frequency

Active network management is an essential part of future electric power networks with high-concentrations of renewable and energy storage assets. Several active network management techniques utilise predictions of network variables, and in this research, an optimal real-time predictor for grid frequency evolution is analysed. In particular, the analytical and numerical characteristics of auto-regressive estimation for the prediction of the frequency evolution in electrical systems are taken into account. Two case studies are considered to evaluate the performance of the frequency predictors; an isolated small-scale microgrid system and the UK public grid; both are representative of future power networks. As for the public network case, the auto-regressive approach guarantees an improvement of 20% with respect to the zero-order hold estimator and up to 70% with respect to the linear approximation.