Research Of Short-Term Load Forecasting Based On Data Mining

Short-term load forecasting of the power system is an important work of power departments, it plays vital role in the reasonable arrangement of units running and stopping,the power transaction,the maintenance of power equipments. It's prime task is forecasting power loads in several hour,one day even several days , based on certain supposition condition .It is an important part in the energy management system of power system ,and plays very vital role in secure,stable and economical running of the power system . With development of the power enterprises marketability reform, higher requirement is brought up to precision of the short-term load forecasting by the power departments.It is an extremely complex work to obtaining higher precision forecasting load , because of kinds of complex limit of the power load, for example: economy,politics,climate and power system itself kinds of active status etc .Because the forecast load precision is influenced by the sample datum, in order to obtaining higher precision forecasting load, it needs to obtain the higher precision sample datum . Therefore in this paper power load is analyzed, based on these, according to the power load datum characteristic, the data mining method is proposed to deal the sample datum .Because the simulation annealing algorithm has the strong overall situation optimization ability, therefore in this article one load forecasting model , which connect simulation annealing algorithm and the BP neural network algorithm ,is proposed, to overcome the flaw running into the partial minimum of the BP neural network . In this article, the validity and superiority of this algorithm has been confirmed, through simulation take the Urumqi actual data as an example.