ISSA-LSSVM Power Load Forecasting Based On Complementary Ensemble Empirical Mode Decomposition

With the increasing global energy shortage and pollution,electricity is gradually replacing fossil energy as the main energy source for people’s daily life.Among them,power load forecasting is the basis of power system operation and maintenance,and short-term power load forecasting plays a key role in the safe dispatch and economic operation of the power system.In order to improve the accuracy of power load forecasting,this paper proposes an ISSA＿LSSVM forecasting model based on complementary ensemble empirical mode decomposition.When the least squares support vector machine is used for load forecasting,the parameter selection will directly affect the forecasting accuracy.This paper uses the improved sparrow algorithm to optimize the parameters of LSSVM(Least Squares Support Vector Machine,LSSVM).The sparrow algorithm is a new intelligent optimization algorithm proposed by Xue Jiankai and others in 2020.While the sparrow algorithm has the characteristics of high solution accuracy,it also encounters the shortcomings of slow convergence speed and easy to fall into local optimum.In order to solve these problems,an improved sparrow optimization algorithm was introduced,use the reverse learning strategy to initialize the population,optimize the diversity of the population,and improve the step size factor.The Levy is introduced into the sparrow position update formula to speed up the convergence speed of the sparrow optimization algorithm and avoid the sparrow algorithm from falling into local optimum,get the improved sparrow optimization algorithm.The improved sparrow algorithm,sparrow algorithm and other optimization algorithms are tested for function optimization.The test results show that the improved sparrow algorithm has the best optimization ability.The improved sparrow algorithm is used to optimize LSSVM,find a set of optimal hyperparameters,and build the ISSA＿LSSVM prediction model.Under the historical data set of a certain area in Henan Province,after the example simulation,the problems existing in the ISSA＿LSSVM prediction model are analyzed,the load sequence is decomposed and then combined prediction is carried out,and the CEEMD(Complementary Ensemble Empirical Mode Decomposition,CEEMD)decomposition method is introduced.A combined prediction model based on CEEMD＿ISSA＿LSSVM is proposed,The model uses CEEMD decomposition technology to decompose the original power sequence,establishes the ISSA＿LSSVM prediction model for each subsequence obtained by decomposing,and trains and predicts each subsequence to obtain the predicted value of each subsequence,and the prediction of each subsequence,The values are superimposed and summed to obtain the final predicted value.And conduct instance simulations with multiple prediction models.The simulation results show that the combined prediction model proposed in this paper has the highest accuracy.