Meeting to the precision demand of short term load forecasting (STLF) inelectricity market, the research of intelligent STLF evaluating system based onmulti-layer & multi-agent was presented with creativity. The system consists ofinterface layer, evaluating&decision-making layer, multi-theory forecasting layer,and database management layer. According to the actual load character of powersystem, the operating status of power system and the evaluating result offorecasting, these higher precision forecasting-agents were self-adaptively chosento execute. In addition, the system has extensively open character and scientificcreativity. The new theory agent can be added into this evaluating&decision systemto perfect and develop continuously the STLF system. To avoid the tight coupling of traditional component pattern in STLFsoftware R&D, the system was developed with relax coupling .NET platform. Asthe need of evaluating system for multi-recourse data, heterogeneous databaseswere integrated by means of ADO.NET in the complex Internet environment;theforecasting data warehouse was built;the Web Service providing forecastingservice for other users was afforded. STLF evaluating system with finevisualization and maneuverability can provide multi-form output of evaluatingresult on forecasting performance.The STLF evaluating system has good performance as follows: it can offer theevaluating report of STLF study models on new theory and obtain theirperformance in actual forecasting. The system can adjust itself and realize the bestforecasting target to promote the STLF development and scientific innovationstudy;according clients' request, the system can test the evaluating performance ofSTLF models in practice and find the problems of models to adjust them.Therefore the research on STLF evaluating system has practical value andscientific innovation.
|