Short term load forecasting is a basic work in China’s power industry,which is of great significance to power system operation control.Therefore,it is necessary to put forward a new short-term load forecasting model with the continuous improvement of people’s living standards.This paper first briefly describes the research background and significance of short-term load forecasting,focusing on the analysis of several methods commonly used in power system short-term load forecasting research,as well as the research status at home and abroad at this stage.Then,the basic concept of power load,load forecasting related principles,factors affecting the forecasting results,data preprocessing,and several short-term load forecasting methods are described in detail.SOFM network is selected for cluster analysis,Elman network is used for short-term load forecasting.In order to further improve the prediction accuracy of the model,genetic algorithm and chaotic whale algorithm are used to optimize Elman,and a short-term load forecasting method based on SOFM-CWOA-Elman intelligent combination model is proposed.Because the short-term load forecasting of power system has strong regional character,this paper takes the historical load of a certain city as the data base,analyzes the load characteristic,fully considers the related factors that affect the load,including the forecast day’s temperature,air condition,date type,wind,humidity and other factors,all factors are quantified to facilitate computer processing of digital information,and all digital information is normalized,the influence of different dimension and range of numerical value on the prediction results is eliminated.Aiming at the problem of uneven classification in SOFM cluster analysis,two solutions are proposed in this paper.The results of cluster analysis can be adjusted by modifying the parameters of SOFM in the combined model,and Ada-SOFM-CWOA-Elman combined model,which uses Ada Boost classifier to classify the forecast days according to SOFM classification standard,and then uses the corresponding class model to forecast.In the end,this paper takes MATLAB 2019 as the platform to carry on the simulation research,through the modeling simulation,the Ada-CWOA-SOFM-Elman combination model presented in this paper has the best performance. |