Electrical energy consumption forecasting is very often requested by stakeholders from the domain in order to elaborate a good strategy for the future. Mathematical models are used in various forms; the accepted idea is that the investigated system repeats its behavior. The paper presents such a prognosis made using artificial intelligence technique in combined with time-series (Box-Jenkins) analysis. The paper offers some recommendations for the training set used for teaching the artificial neural networks (ANN), and also approaches an ANN learning method ensuring quick load dynamics learning. Based on these aspects, the created software can analyze several load evolution scenarios and can establish a correct trend for the electric load evolution. Selection between different types of load forecasting are discussed in this paper.

Keywords: Mid term monthly load forecast, Artificial neural network.