Mid-long term load forecast of power system based on data-mining
Min Cui · Dianli zidonghua shebei · 2004
Mid - long term load forecast of power system is affected by various uncertain factors and research shows that clustering method can synthesize numerous relative factors and put them into forecast model.An improved clustering method is put forward ,which unites the advantages of Chameleon algorithm and density - based algorithms to achieve optimal clustering.It makes up the disadvantages of hierarchy method ,which is irreversible and useless for the complex shapes.With ROUSTIDA algorithm as the pre - process ,exact and intact historical data are ensured.Practical examples prove its fast calculation speed ,high forecast accuracy and small error variety.When there are more in - fluencing factors and the historical data are non - integrated or inexact ,this method shows especially its advantages.