Research And Application Of Short-term Power Load Forecasting Based On Support Vector Machine

Short-term load forecasting is an important basis for power dispatching,power supply planning,safety monitoring and unit optimization.Therefore,it is very important to improve the accuracy of short-term load forecasting for the optimal operation mode,maintenance of power system and economic operation.In this paper,the current research status of short-term load forecasting is studied in detail,and the common methods,advantages and disadvantages of forecasting are summarized,and the principle,characteristics and influencing factors of short-term load forecasting are analyzed.Using historical load data and meteorological data of a region of Middle East Gansu province,the influence of various factors on prediction was analyzed.Then,on the basis of statistical theory,the basic principle of Support vector machine(SVM)is introduced,and the improved method of SVM-the least square support vector machine(LSSVM)was established by modifying the historical load data of errors,omissions and affecting factors,and forecasting the load of different date types and analyzing the causes of errors.Moreover,the particle swarm optimization algorithm(PSO)is introduced in the process of modeling LSSVM,which leads to the large error of prediction in the empirical selection parameter.Finally,the actual load data of a region of Middle East Gansu province were collected,after data preprocessing,the prediction model was used for simulation.The error of the predicted value and the actual value is compared and analyzed.Results show that PSO-LSSVM model for short-term load forecasting,can achieve the purpose of the model parameter optimization selection,so as to improve the prediction precision.It proves that the prediction method has good convergence and high prediction precision and faster training speed advantage.This method can be applied to the short-term forecasting system of power system,and it has a summary effect on the forecast region's medium and long term power load variation rules.