Forecasting Electric Energy Demand using a Predictor Model based on Liquid State Machine

Electricity demand forecasts are required by companies who need to predict theircustomers’ demand, and by those wishing to trade electricity as a commodity onfinancial markets. It is hard to find the right prediction method for a givenapplication if not a prediction expert. Recent works show that Liquid StateMachines (LSMs) can be applied to the prediction of time series. The mainadvantage of the LSM is that it projects the input data in a high-dimensionaldynamical space and therefore simple learning methods can be used to train thereadout. In this paper we present an experimental investigation of an approachfor the computation of time series prediction by employing LSMs in the modelingof a predictor in a case study for short-term and long-term electricity demandforecasting. Results of this investigation are promising, considering the error tostop training the readout, the number of iterations of training of the readout andthat no strategy of seasonal adjustment or preprocessing of data was achieved toextract non-correlated data out of the time series