Short term electrical load forecasting for mauritius using Artificial Neural Networks

The Central Electricity Board is the sole utility responsible for the generation, transmission, distribution and sale of electrical power in Mauritius. The country´s highest peak demand increased from 353.1 MW in 2005 to 367.3 MW in 2006 and corresponding annual consumptions increased from 2014.9 GWh to 2091.1 GWh and these figures are continuously increasing every year. In this paper, different Artificial Neural Network models are proposed for Short Term Load Forecasting (STLF) of the Mauritian electrical load. It is shown that models based on a combined supervised/unsupervised architecture provide better forecasting abilities compared to those relying on supervised architectures only. This is achieved by clustering of data.