An efficient on-line load-modeling algorithm for short-term forecasting of interconnected power systems

This note addresses itself to the problem of one-step-ahead forecasting of the load demand at all the loading nodes of an interconnected power system. A multivariable state-space combined parameter and state estimation technique is proposed. A normalized stochastic approximation method-which is computationally simple and exhibits good initial convergence-is used to generate estimates of the model parameters and the optimum filter gain matrix. The estimated parameters are then utilized to obtain a one-step-ahead forecast of the load demand. Some results of application of the proposed algorithm to real load pertaining to the Eastern Saudi Arabian power grids are presented.