Combination Of Short-term Load Forecasting Method Based On Cloud Computing Research

In recent years,with the innovation of computer technology,the rapid development of smart grid.Grid electricity sector in order to ensure the safe and economic operation,the result of the short-term load forecasting accuracy,high efficiency and real-time performance put forward higher requirements.At present the research direction of short-term load forecasting mainly focus on the optimization of prediction model,which to a certain extent,improve the prediction precision of the prediction model and predicted.But most of these prediction methods based on overall analysis of the influence factors,on the nature of the various factors to consider not comprehensive,lead to the accuracy of the prediction model is difficult to further improve,less universal.In this paper,according to the correlation of different influencing factors,the combination forecast model has higher prediction efficiency for related research.First,this paper analyzes the short-term load forecasting of the research background and research status at home and abroad,has carried on the induction summary to a variety of traditional intelligent prediction algorithm,in the after comparing the advantages and disadvantages of each method,summed up the future development direction of short-term load forecasting.This study chooses the historical data of a certain region of zhejiang province as the training sample and forecast sample,the load characteristics of the prediction area,economic,meteorological factors such as carried on the thorough analysis,and according to the characteristics of the data itself for data preprocessing,comparison screen repair of abnormal data in both directions,enhanced the reliability of the predicted results.Secondly,in order to further improve the accuracy of the prediction results,this paper deals with the various factors influencing the load size uncertainty classification,bacterial foraging algorithm was used to optimize extreme learning machine forecasting model in view of the uncertainty related affecting factors to predict,optimize nuclear extreme learning machine based on cloud model prediction model for uncertain influencing factors related to forecast,then will the predictive results of the prediction model for the weighted sum,get the ultimate load value.Finally,due to the complexity of the combined forecasting model operation,greatly increasing the difficulty of operation,in order to solve the problem of insufficient of computing resources,this paper introduce cloud computing to parallelize the transformation of combination forecast model,improve the computational efficiency of prediction model,to enhance the application effect of forecast model.Results show that compared with the traditional prediction method,in this paper,by introducing a cloud model optimization model to predict extreme learning machine,increase the consideration of uncertain related influencing factors,improve the accuracy of the prediction results,the prediction precision is increased by 0.23%,with the introduction of cloud computing,increase the prediction model of parallel computing performance,makes the prediction of single time reduced by about 900 s,accelerate the calculation speed,improve the efficiency of the staff.