Comprehensive Model for Mid-Long Term Load Forecasting Basing Three-Target Quantities and RBFNN

Mid-long term load forecasting is a critical component in power system planning and operation. Aiming at the characteristic of it bring forward a comprehensive model basing on three target quantities which involve target gross quantity and target increase quantity and target increase rate. At first construct an analytic hierarchy process (AHP) model to analyze and appraise three target quantities dividedly, and pick up best prediction models for them. And then synthesize three selected models´ forecasting results to the final prediction result using radial basic function neural network (RBFNN), whose input data involve GDP of forecasting object in addition. AHP model integrates the forecasting error and fitting accuracy of models and introduces man´s influence successful which are expert trusting of models and trend reliability of model forecasting results. AHP model analyzes history data which are most nearby to the forecasting time and therefore is better real-time. Experiment results show that the comprehensive model is provided with high forecasting precision and achieves satisfactory results in application.