Font Size: a A A

Research On Saturation Load Forecasting Considering Electricity Substitution

Posted on:2020-11-06Degree:MasterType:Thesis
Country:ChinaCandidate:X J MaoFull Text:PDF
GTID:2392330623963522Subject:Electrical engineering
Abstract/Summary:PDF Full Text Request
China is in the period of energy revolution,and works hard to accelerate the electrification process of energy consumption.The government promotes a slogan of “replace coal and oil of electricity” to increase the proportion of electricity consumption in terminal energy consumption.Thus,it puts forward new requirements to the forecasting of saturation load time and scale.This paper first introduces the research status of load forecasting,and focuses on the research status of saturation load forecasting and the application of artificial intelligence in load forecasting.Then the factors affecting the saturation load level are analyzed,and ten key factors that can be quantified are extracted according to the analysis.The correlation degree of these factors to the saturation power demand is quantitatively calculated by three analysis methods: correlation coefficient method,grey correlation degree and K-L(Kullback-Leibler)information quantity.Secondly,a medium and long-term load forecasting model based on combined forecasting is proposed.The sub-model includes a forecasting model based on energy scale and structure predictions,a forecasting model based on the calculation of electricity substitutes and basic electricity consumption,and a forecasting model of support vector machine considering electricity substitution factors.The weighting method of the combined forecasting is based on Shapley value.Meanwhile,considering the disadvantage of high correlation requirement of input factors in traditional models,a neural network model based on long short-term memory(LSTM)is constructed to realize medium and long-term load forecasting with multiple factors.Finally,based on the basic concept of saturation load,the social and economic characteristics of the saturation load stage are analyzed through the study of typical developed countries where the foreign electricity demand has entered or nearly entered the saturated stage.Then,a method for judging the saturation time point considering the potential of electricity substitution is proposed.The main criterion is the growth rate of electricity demand,and the auxiliary criteria are economic and social indicators and energy indicators.Then,the saturation time and scale can be obtained by the criterion,and a complete set of saturation load analysis method is finally set up.The example shows that the prediction accuracy of the model is good,which has a certain guiding significance for network planning.
Keywords/Search Tags:electricity replacement, LSTM neural network, saturation load forecasting, support vector machine, combined forecasting
PDF Full Text Request
Related items