A strategy for on-line forecasting of nodal active and reactive power injections in electric power systems

In this paper a strategy for on-line forecasting of nodal active and reactive power injections in electric power systems is presented. In this strategy a stochastic model of the nodal active loads is initially developed from historical data using the Box and Jenkins procedure. This model is then used on-line in the forecasting of nodal active loads. Simultaneously, an adaptive Kalman filter is used to forecast the values of the loads power factors. The results of these two procedures are then combined using the system dispatch strategy and an on-line load flow program to generate the nodal active and reactive injections.