Power Load Probability Density Forecasting Method Based On Big Data Analysis And LASSO Quantile Regression

Power load forecasting plays an important role in power system scheduling,planning and production planning.Accurate load forecasting helps us ensure safe and stable operation of the power system.The power load forecasting is a challenging task,because the predictive accuracy is easily affected by multiple external factors,such as society,economics,environment,as well as the renewable energy,including hydro power,wind power and solar power.These potential external influence factors and the interference of random factors often make power load forecasting complicated increasingly.Particularly,in the smart grid with large amount of data,how to extract valuable information of those external factors timely is the key to the success of power load forecasting..In order to solve the high-dimensional data problem in power load forecasting,the paper presents a probability density forecasting method based on LASSO quantile regression(LASSO-QR)and a probability density forecasting method based on LASSO neural network quantile regression(LASSO-QRNN)to reduce the uncertainty in the prediction process.Firstly,the LASSO regression algorithm is used to screen out the important explanatory variables from the potential impact factors of power load forecasting,and the LASSO-QR model and LASSO-QRNN model are established.Then,combined with the kernel density estimation method,the power load probability density forecasting is performed.The prediction results can obtain not only the probability density curve of the future load intact,but also the predicted value and fluctuation range of the future load.For verifying the effectiveness of the proposed method,from a statistical point of view,this paper evaluates the point prediction results on the probability mean,median and mode of the probability density curve,and estimates the prediction interval using the relevant prediction interval evaluation criteria.In this paper,the probability density forecasting method based on LASSO-QR is proposed for short-term and medium-term load forecasting.Moreover,the probability density forecasting method based on LASSO-QRNN is presented for power consumption forecasting.In the process of forecasting,the external influence factors are taken into account to select important influence factors for constructing a suitable mathematical model.This paper adopts five cases to carry out simulation experiments,including shortterm power load forecasting in winter and summer in Ontario,Canada,medium-term power load forecasting in a sub-provincial city in eastern China,electricity consumption forecasting in Guangdong Province,China and electricity consumption forecasting in California,USA.Through comparison experiments with the state of the arts,it is further shown that the probability density forecasting method proposed in this paper can significantly reduce the uncertainty in the prediction process and improve the accuracy of power load forecasting.On the basis of science,the method proposed in this paper can not only meet the requirements of power system decision-makers to avoid major economic losses,but also find an effective way to solve the load forecasting problem in big data environment,which has important practical significance.