Short-term Load Forecasting Based On Least Squares Support Vector Machine

With the development of China’s economy,power demand is growing rapidly.The study of short-term load forecasting model is conducive to the security,stability and economic operation of power grid.Based on the study of short-term power load,this paper expounds the influencing factors of load,selects least squares support vector machine(LSSVM)as the forecasting method,and uses swarm intelligence optimization algorithm to set its super parameters to build a short-term power load forecasting model.In this paper,the influencing factors are divided into date type,meteorological elements and historical load,which are used as the input variables of the model.In order to improve the scientificity of input variables,grey relational analysis-entropy weight method(GRA-EWM)is used to select the main influencing factors from all the influencing factors.In order to eliminate the correlation between the input variables,principal component analysis(PCA)is used to reduce the dimension of the main influencing factors,and finally a new input variable is obtained.Studing LSSVM and using a two-dimensional vector to represent the combination of the two super parameters.Using particle swarm optimization(PSO)algorithm and seting the fitness function according to the prediction error to select the super parameters of the model.Aiming at the problem that PSO has many parameters,a LSSVM model based on Salp Swarm Algorithm(SSA)is proposed.However,there is a contradiction between the search speed and ergodicity.Therefore,an improved Salp Swarm Algorithm(ISSA)is proposed.In this algorithm,the initial population is evenly distributed by tent mapping,the leader uses dynamic weights to improve the convergence speed,the followers use random crossover strategy to increase diversity,the rebels use reverse learning strategy to jump out of the local optimum,and the Beetle needs to s Beetle Antennae search(BAS)near the food location to improve the prediction accuracy.Through the simulation comparison of three prediction models,it is proved that the improved model has advantages in stability and prediction accuracy.