Research On A Forecasting Method Of Short Term Load Based On VMD And ELM

Short-term load forecasting is an indispensable basis for maintaining economic operation and achieving safe and stable control of power grid.It plays a key role in the economy,security and reliability of power grid operation by formulating the most economical security requirements,determining operation constraints,and limiting operation schemes of external equipment,and providing data for power generation planning procedures.Because the accuracy of traditional forecasting methods is generally not high,it has always been a research hotspot for scholars to put forward better forecasting methods and theories.In view of the above problems,this paper proposes a combined forecasting model based on ELM and VMD.Firstly,using VMD to decompose the power time series of historical loads,a relatively stable set of modal components can be obtained.Then,ELM models are established to predict each modal component separately,and finally all the forecasting results are added up to be the final forecasting results.The experimental results show that the model fully embodies the advantages of VMD and ELM,and can achieve both accurate and fast prediction.Firstly,the advantages of VMD in multi-scale mode decomposition are analyzed and compared.Compared with EMD,simulation results show that VMD can suppress modal aliasing more effectively.Moreover,VMD combined with Gram-Schmidt orthogonalization can improve the performance of VMD in feature extraction.Therefore,we can conclude that VMD is more suitable for hybrid prediction model.Through the simulation tool of MATLAB,for the proposed prediction model,the simulation analysis of ultra-short-term load forecasting is carried out by using actual load data,and the results are compared and analyzed with those of various forecasting methods.Because short-term load mainly requires high forecasting accuracy and forecasting speed,this paper verifies the proposed combined forecasting from the speed and accuracy.The feasibility of this method in ultra-short term load forecasting.The research contribution of this paper is that it has good research significance and reference value for the theoretical research of short-term load forecasting method.