Dynamic Structural Model with Covariates for Short-term Forecast of Electricity Demand in Angolan Cities

Abstract

Short-term electric load forecasting is an important tool for planning electrical generation systems, operations (scheduling the generation of electric load flow required for consumption) and control. In this study the aim is to show how the effects of covariates can be incorporated into state space models. A structural model based on dynamic processes and a method for generating predictive densities and estimating the maximum flow of the consumed electrical load are examined. The model and method are empirically investigated using actual data of daily electricity demand in Cabinda city to obtain short-term forecasts. The performance of the methodology is validated via comparisons using predicted values and observed values.