Open AccessTransmission Line Loss Prediction by Cross Validation and Gradient Boosting Decision Tree
Author(s) -
Jicheng Yu,
Feng Zhou,
Kai Zhu,
Changxi Yue,
Jiangchu Wang,
Congzhen Xie
Publication year - 2020
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/440/3/032099
Subject(s) - gradient boosting , decision tree , computer science , transmission line , test set , boosting (machine learning) , line (geometry) , generalization , granularity , mean squared prediction error , tree (set theory) , cross validation , artificial intelligence , data mining , algorithm , statistics , mathematics , random forest , telecommunications , mathematical analysis , geometry , operating system
To solve the problem of the accuracy and generalization of the transmission line loss prediction, a new method for transmission line loss prediction based on cross-validation (CV) and gradient boosting decision tree (GBDT) is proposed. In this method, time granularity matching and statistical feature extraction are firstly carried out to improve the information content of the data. Then the data is divided into a training set and test set by CV method to train the GBDT model. Finally, the actual transmission line loss verification test is conducted. The result shows that for 500kV transmission lines with an average daily supply of 7990MWh, the average line loss error of this model is 35MWh, and the average line loss rate error is 0.093%, which verifies the effectiveness of the method in this paper.