Free electricity market: how industrial customers and ESCOs can make the most from load forecasting techniques

The implementation of linear and neural dynamic models to get the load forecasting for an industrial site, on different time frames (hourly, daily and weekly based) is presented in this paper. The identification process of these models is based on the electric energy consumptions sampled on a quarter of an hour basis over the last two years is a single point of acquisition. The performance of the proposed models have been compared: in this paper, results on the weekly-based models reported and the effective cost reduction verified. This well-known tool is investigated from an ESCO's point of view, whose intention is to promote additional services to customers who are keen to invest less on equipment than on value-added services.