Statistical parametric techniques for power residential demand forecasting

Quantifying the dynamic and the evolution of an electrical power distribution system is an important task in operational and planning studies, where some of the issues addressed is to measure the growth of the electrical demand, with special attention to residential consumption due to its randomness. An adequate representation will ensure valid results for future prediction of the system behaviour. In this context, analytical statistical methods are widely used, because of the availability of real data and the ability of these types of models to represent aleatory processes. This paper reviews the statistical parametric methods of Fourier analysis and stochastic differential equations, and implements them as a hybrid model to analyse real measure data of power consumption from Chilean systems. The construction of the model is explained, and the performance of the reviewed methods are compared with the proposed hybrid model and with two classical methods based on regression techniques.