Load forecasting using nonlinear modelling

The best-possible prediction of the gas consumption represents highly valuable information for gas producers. Using an appropriate model of the time-dependent and nonlinear gas consumption process, a prediction scheme can be set up for all the nodes in a gas network. In this paper, a nonlinear model with several inputs using noisy observations is built up as a combination of a pre-specified dynamical term and an unknown static nonlinear component realized by region-wise base point driven interpolation. Also, the details of the practical application for a real gas network are shown.