Chapter 11 - Long-term load forecasting in the smart grid framework using time series and econometric approaches

Abstract

The continuous growth of human civilization has been resulting in a sharp increase in electricity consumption. Therefore, efficient power control and load forecasting have been reported as a major requirement for the planning and operation of power systems. Recently, several techniques have been developed to handle load forecasting. However, the erratic nature and the uncertainty characteristics of the electric load render demand forecasting a challenging task for power utilities where no particular technique can be identified as the most efficient. This makes it insightful to identify the technique that best fits a specific dataset. In this chapter, a comparative study of four long-term load forecasting techniques, namely, autoregressive integrated moving average models, multiple linear regressions, and nonlinear autoregressive neural networks with and without exogenous variables, NARX and NAR is proposed. The study is carried out on the Tunisian electricity consumption dataset, and the efficiency of each technique is assessed with respect to the variation of the sampling period and the forecasting time frame. Moreover, quantitative and qualitative evaluations of the results are also presented to highlight the strengths and the shortcomings of the evaluated forecasting techniques.