Research on PPMCC-PSO-SVR Heat Load Forecasting Combined with Weather Factors
An accurate forecast of heat load is necessary for the improvement of energy utilization and pipe network regulation strategy in heat exchange station.In this paper,five methods support vector regression,extreme gradient boosting regression,grid search optimizes support vector regression,grid search optimizes extreme gradient boosting regression and PPMCC-PSO-SVR are adopted to establish the heat load forecasting models combining weather factors and compare.The characteristic correlation analysis based on Pearson correlation coefficient,five parameters are selected to establish the heat load forecast model,namely hourly tem-perature,hourly humidity,water supply flow,secondary network water supply temperature and return water temperature.The re-sults show that before tuning the parameters of the original model,the evaluation indexes of the extreme gradient boosting regression model are far better than those of the support vector regression model.The PPMCC-PSO-SVR model presents the best performance by combining the weather factors,it can effectively predict the dynamics of heat load system during a short period.The R2,RMSE and MAE of PPMCC-PSO-SVR model are respectively measured 0.996 5,0.151 1 and 0.118 6.
国家科技期刊平台
NSTL
万方数据
