System identification for the prediction of the electric energy consumption of a dairy firm

Abstract

A system identification method based on black-box techniques for the prediction of the electric energy consumption in a dairy firm is presented. This prediction is required by the Italian free energy market where the energy sellers aim at selling energy according to a load flow scheduled some days in advance. The black-box identification is employed as an alternative to an energy investigation of the firm. The inputs of the system are the work shifts of each process unit and the output is the electric energy consumption. Two black-box parametric models have been evaluated-linear and neural-and the principal component analysis method has been employed to preprocess the data.