Research On Model And Method Of Medium And Long Term Combined Load Forecasting In Regional Power Grid

The medium and long term load forecasting play a fundamental role in developmentplanning of the regional power grid. Their accuracy has a direct influence on power gridsecurity, economy, reliability and reasonable operation and investment. Though the model andmethod of medium and long term load combination forecasting are varied, but each predictionmodel has its own scope of application, one model is not suitable for all cases. Thereforecombination forecasting which contains multiple individual forecasting models becomes aresearch hotspot. The combination forecasting model can make use of the various forecastingsample information which is more systematic and comprehensive than a single forecastingmodel. Thus, the combination forecasting can reduce effectively random factors occurred in asingle forecasting and improve the forecasting accuracy.The single medium and long term load forecasting model is established on the basis ofZhaoqing power load characteristics including regression forecasting, gray system forecastingmodel and exponential smoothing forecasting models. Then the combination forecastingmodel is established on the basis of single medium and long term load forecasting model. Thecombination forecasting model consists of non-optimal positive weights combinationforecasting model and optimal positive weights combination forecasting model. Thenon-optimal positive weights combination forecasting model consists of equally weightedaverage, the reciprocal of sum of squares due to error (SSE), the reciprocal of root meansquare error (RMSE), binomial coefficient method and grey relational analysis models. Theoptimal positive weights combination forecasting model consists of the minimum of SSE andthe minimum of sum of absolute error models.According to annual power supply data from2004to2011in Zhaoqing power grid, theresults of single medium and long term power load forecasting model are analyzed, the resultsobtained from these combination forecasting model are compared and concluded that theoptimal positive weights combination forecasting model can improve the forecasting accuracygreatly compared with the single forecasting model and non-optimal positive weightscombination forecasting model in Zhaoqing power grid.