One-step-ahead load forecasting for on-line application


Ali Keyhani, Ahmed El-Abiad
Purdue University, West Lafayette, IN, USA ;

This paper appears in : IEEE Transaction on Power Apparatus and Systems 
Date: July/August, 1975  


Abstract:
The problem of One-step-ahead forecasting of the load demand of an interconnected power system for online application is studied. It is proposed that at any time instant k, a stochastic model with unknown order, ARMA (n,m), be postulated. By using a recursive identification algorithm, the order of the model is identified on-line from the immediate past observations of load data. Having determined the number of parameters to be used (i.e., the order, n and m) at any instant k, these parameters are to be estimated in real time using the past observations. This is done by applying a recursive estimation algorithm.