Probability Forecasting And The System Design Of Short Term Load Based On FPA-SVM

Electric power industry is the country's major infrastructure industry in the field of energy,and has become the lifeblood of the national economy.With economy growing rapidly in recent years,the electric power construction in China has to be strengthened constantly.However,in the constructing process of power system,the electric power system becomes more complex,which makes accurate load forecasting more difficult.And accurate and reliable short-term load forecasting is of great significance for sufficient power supply planning,reducing power storage capacity,avoiding major accidents,ensuring the security of production and life,improving social benefits,etc.So it is critically important to emphasize the accurate and reliable short-term load forecasting.Based on the practical background,this paper focuses short-term load forecasting method research and the main research work and innovation are as follows:1.Based on Empirical Mode Decomposition(EMD)of historical data,Flower Pollination Algorithm(FPA)is used to optimize the main parameters of Support Vector Machine(SVM),with the initial value being corrected via Kalman Filter(KF),the EMD-FPA-SVM-KF comprehensive load forecasting model is established.In this paper,the collected historical load series are pretreated and the EMD is used to stabilize decomposition load in different scales.Then,the FPA is proposed to optimize the main parameters of the SVM,and the EMD-FPA-SVM prediction model is built.In order to improve the prediction accuracy of the SVM,Kalman Filter algorithm is introduced to correct the initial predictive values of the model which are used as observation values of KF and are corrected by optimal estimation and adjustment merit of KF.Finally,the combination forecasting mode EMD-FPA-SVM-KF of short-term load is proposed.2.A possibility load forecasting method is presented based on statistical theory and EMD-FPA-SVM-KF model,which provides standard for evaluating the reliability of forecasting method.First,the original load samples are processed to obtain corresponding sets of historical errors.Secondly,the error distribution characteristics both in horizontal level and vertical level are analyzed to obtain the discrete error probability distribution curves of corresponding section.Thirdly,the error probability distribution curves were transformed into load probability density curves to obtain the corresponding confidence interval envelope under confidence level.Finally,the feasibility of short-term forecasting method mentioned in this paper is proved by examples.3.Based on the prediction model method proposed in this paper,each function part of the forecasting software was designed on the VB platform.Therefore the system software of electronic system short-term load probabilistic prediction is proved practical via VB and MATLAB mixed programming.