A belief network model for forecasting within the electricity sector

This paper presents a scheme amenable to simulating the decision-making process associated with forecasting the demand and required generating capacity of electricity. The model comprises a belief network based on a relatively simple influence diagram of the demand formation and supply decisions. A dynamic simulation algorithm, which takes into account the feedback effects of decisions to demand, is applied to the belief network. The basic assumption behind the scheme is that the dependencies guiding the decision process remain relatively stable even in a turbulent marketplace. To illustrate the advantages of this approach the model is validated by reconstructing the demand and supply changes in Finland over the period from 1971 to 1987. A series of case studies are examined where the simulation is extended to projections in the future. It is concluded that the simulation scheme provides a convenient tool for examining both past decisions and their extrapolations into the future.