Load Modeling and Forecast based on a Hilbert Space Decomposition

Many energy markets have experienced extreme changes in load dynamics - due, for instance, to energy shortages or changes in market rules. Under these situations, new history may correspond to a mere few years, leaving insufficient amount of data to be explored by classic models, from statistical to neural networks. This paper addresses this problem - modeling under lack of data - and proposes a new method based on functional analysis, applied as a sequential iterative procedure. A real case- study enlightens the approach advantages.