Application Of LS-SVM Forecasting Model Based On Bayesian Evidence Framework In Space Power Load Forecasting

Due to the obvious randomness of the short-term power system load,the load forecasting work is not easy to find the inherent development rule,so it is very difficult to improve the accuracy of short-term load forecasting.The traditional load forecasting based on simple basic theory did not take full account of the key factors influencing the development of power system load.In recent years,based on the improvement of national policy,the government makes the urban development more and more standardized and reasonable,which provides a superior basis for a broader research space.So that many researchers also studied the space load forecasting methods in order to obtain high short-term load forecasting accuracy.The prediction method of short-term load forecasting is the least squares support vector machine(LS-SVM).With the extension of its research,the research scholars have found the LS-SVM prediction method more and more profound.Based on the deep understanding and awareness of the nature of short-term load forecasting theory and profound significance,this paper chooses an optimized LS-SVM forecasting model to carry out a more profound study on short-term load forecasting.This paper comprehensively considers the external factors.The predictive model sets a number of factors to correct and adapt to self-learning,make it more intelligent,and more in line with the actual environment,but make the model more complex and difficult to simulate.The robustness of the prediction model is reduced by the various factors,and the stability of the prediction model is destroyed by some influencing factors.Therefore,this paper applies the weight coefficient factor Modify its forecasting model to improve robustness and stability.In addition,this paper also uses the Bayesian evidence framework optimization model to build the forecast model faster and more accurately.Through this model optimization,this paper predicts that the model is more practical and efficient.We first introduce the research principle of power load forecasting,and the second studies the important platform of GIS for load data acquisition,and also analyzes the relationship between the data pre-processing and the spatial load classification importance.In the next chapter,the third chapter provides a theoretical basis for constructing the data model based on the forecasting model.In the fourth chapter,we consider a variety of external factors,and build a weighted least squares support vector machine forecasting model.The Bayesian evidence framework is used to optimize the parameters.At the end of this paper,the whole forecasting process is improved,and an administrative function load area is selected on the power GIS layer.The forecasting model is used to forecast the load for 10 days,and the load of the model is compared with that of the model.The advanced reliability of the optimized model have the value.Can be promoted in society to adapt to the development of electricity.