Research On Short-term Load Forecasting Based On Hourly Weather Factors

Short Term Lord Forecasting (STLF) is one of the most important contents of running and dispatching power system, and important part of power market. It is the premise and basic of safe, economic and reliable operation of power system. The forecasting precision will greatly affect the economic benefit of power system. Therefore, how to raise the accuracy of load forecasting has been a hotspot of research at home and abroad.Research shows that power load is influenced by many factors, between these factors, meteorological factors affect the power load most significantly. Therefore, study how the load power is affected by weather factors and how to establish a appropriate forecasting model is the key to improve load forecasting accuracy.In this thesis, firstly, the load characteristic of Hangzhou area is analyzed by monthly, weekly and daily, and influence of temperature and humidity was qualitative analysised.Secondly, through the study of ANN, three different ANN models are carried out by using three different input: using weather data directly, using Human Body Amenity Indicator, and using Temperature-Humidity Index. And also three different optimization algorithms of ANN are used in modeling. By comparing these results, the way of using THI and historical load data as input was picked. Test results show that the proposed model has a high level precision in forecasting summer days.Finally, on the basis of the studies above, by analyzing the co-relationships between weather factors and weather-sensitive load, a correction method considering the accumulation effect of temperature is proposed. This method not only considers the influence of the temperature of a few days ago, but also considers the influence of the temperature of a few hours before the to-be-forecasted hour. Correction method corresponding to the humidity is also developed. Above all, a regression forecasting model based on load decomposition and correction of accumulation effect of temperature is proposed. The availability of the model was proved by forecasting simulation of Hangzhou area in summer 2007, it can improve the forecasting accuracy remarkably.