Neural Network For Power Load Forecasting

Power industry is the important basic industry in China, and it is vital for nationaleconomy and people’s livelihood, and it is the lifeblood of the national economy. Correct andaccurate power load forecasting is the main basis for electric power system planning,marketing, market trading, scheduling of work departments etc, and it has very importantsignificance for the economic optimization of generation planning, the reasonable deploymentof electric energy, reasonable arrangement of unit operation, obtain a feed-in tariff advantages,to achieve maximum economic benefit and social benefit. However, power system loadforecasting is an extremely complicated and huge project, process is tedious, difficult, and hastoo much social attribute and natural attribute (for example, forecast is often influenced by alarge number of complex factors such as weather, temperature, date, type, economic policy,market, competition, and multiple emergency), therefore, study of advanced and practicalpower load forecasting method is particularly important.This paper mainly studied the short-term power load forecasting model establish-ed bythe related affecting factors, proposed dividing the power load forecasting model into baseload and incremental load caused by impact factors, and proposed using the differencebetween impact factors of different periods to predict the incremental load, so as to simplifythe structure of network model, enhance the prediction speed and prediction accuracy.In the article, I made a detailed analysis of the principle knowledge of BP neural networkand RBF neural network, and BP neural network and RBF neural network are analyzed in themodel simulation by considering influence factors directly and indirectly (considering thedifferential impact factors). Simulation results showed that the proposed load forecasting ofneural network based on differential impact factor can effectively improve the predictionprecision.