Research On Short-term Power Load Forecasting

China’s modernization level is constantly increasing,and the electricity consumption in various industries is also growing rapidly.Faced with rapid growth in electricity consumption,the power system needs to have precise supply and demand relations for the electricity market,and short-term power load forecasting is an essential reference for the reasonable control of supply and demand in the power system.Effective and accurate short-term power load forecasting can save resources,improve economic efficiency,and maintain the stability of social development.Firstly,this paper expounds the relevant theory of short-term electric load fore casting.According to the characteristics of historical load,this paper introduces the related theoretical method about Ensemble Empirical Mode Decomposition(EEMD)and uses this method to decompose historical load data and use the decomposed vector as the input part of the prediction model.Secondly,the least squares support vector machine model(LSSVM)is studied and the parameters of the model are optimized by the standard particle swarm optimization(PSO)algorithm.Aiming at the local convergence premature problem in PSO algorithm,an improved PSO algorithm is proposed and an improved PSO-LSSVM model is established.Considering the influence of the load influencing factors,the selected seven influencing factors are reduced in dimension and added to the input of the forecasting model.The simulation experiments show that the prediction accuracy can be improved by considering the load influence factors,and the dimension of the input variables is reduced through principal component analysis.The improved PSO-LSSVM prediction model solves the problem of local optimal solution,and the prediction accuracy of the model is improved.Finally,the Extreme Learning Machine Network Model(ELM)is studied and the Bio-geography-based Optimization(BBO)algorithm is used to optimize the ELM model parameters.For the local convergence problem in the BBO algorithmite ration,an improved BBO algorithm is proposed and an improved BBO-ELM model is established.The simulation experiment is carried out under the premise of the original model input.The results show that the improved BBO-ELM prediction model solves the problem of convergence premature,improves the prediction precisely and the prediction accuracy is slightly higher than the improved PSO-LSSVM model,and this paper uses the MATLAB GUI to design Human-computer interface of the short-term load forecasting.