Open AccessThe Load Forecasting Method based on Characteristic Analysis and Combination Learning
Author(s) -
Zhijun Zhang,
Ang Li,
Xin Qi,
Qiang Li
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1549/5/052112
Subject(s) - computer science , sort , stability (learning theory) , artificial neural network , ranking (information retrieval) , artificial intelligence , machine learning , adaboost , extreme learning machine , random forest , feature (linguistics) , data mining , basis (linear algebra) , support vector machine , mathematics , linguistics , philosophy , geometry , information retrieval
Load forecasting is an important basis for power system planning and safe and economic operation. This paper presents a load forecasting model based on feature ranking and combination learning. In view of the great difference of regional load, firstly, the random forest algorithm is used to sort the factors which have great influence on the prediction target. Then, the model selects the characteristic attributes with high characteristic contribution, dynamically combines the prediction results of the extreme learning machine, AdaBoost and neural network model, and updates the weights in a certain period through Lasso algorithm to obtain the final prediction results. Finally, the actual load data of Tianjin Power Grid is used for example verification. The results show that the prediction model established in this paper has good prediction accuracy and stability.