The Prediction Of Daily Peak Load Of Summer In Hunan Province Considering The Effects Of Typical Meteorological Factors

Short-term load forecasting is a technology of giving an advance estimate to electricitydemand in the future considering some influencing factors and the load data in the past. It isan important work to power system, which will influence production arrangement, theeconomic scheduling and safety operation of power system. Due to the superabundantinfluencing factors, there is still no perfect prediction method. With the further developmentof power market and the modernization of power grid, accurate load forecasting is more andmore important to planning and economic efficiency of power system.According to the typical characteristics of regional climate in summer in Hunan province,the article will make a forecast according to daily maximum load in summer considering theeffects of typical meteorological factors. Firstly the article analyzed the variationcharacteristics of daily maximum load in summer, considered the influence that typicalmeteorological factors such as heat accumulation effect, rainfall lag effect, temperaturemutation makes to the change of load, revised relevant meteorological factors, whichimproved the relevance between meteorological factors and load. Secondly the article gave anintroduction about the unique climate of Hunan province in summer and development ofsmall hydropower station. The meteorological characteristics are that temperature mutationoccurs frequently and the temperature difference is large, meanwhile continuous hightemperature and continuous precipitation are also typical characteristics of Hunan in summer,Therefore it is necessary to consider them properly and add the the peak output of smallhydropower station to prediction model because electric energy production of the smallhydropower station was directly sent to main grid,which must affect unit adjustment of load.According to neural network has good ability of self-learning and nonlinear approachingand can learn the relationship between typical meteorological factors with the maximum load adequately, the prediction model of daily maximum load in summer based on neural networkwas established, in which the typical meteorological factors and the peak output of smallhydropower station were considered. By adjusting the number of hidden layer nodes, theoptimal number of nodes that minimized the error was found. By selecting representativesamples that can reflect the change rule of daily maximum load in summer, the creatednetwork can learn the rule. Using the trained network to predict the load considering andignoring the typical meteorological factors, we can get a conclusion that adding the typicalmeteorological factors and the peak output of small hydropower station to the prediction model can improve the accuracy of prediction and have certain practical significance to theprediction of daily maximum load in summer in Hunan province.