Open AccessCalculation method of line loss rate of photovoltaic station based on PCA-GRNN
Author(s) -
Yuan Li,
Jing Liu,
Hongwei Tan,
Yajie Li,
Xinping Diao,
Li Yao,
XU Zhi-guang
Publication year - 2021
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 - 10.1088/1742-6596/1754/1/012201
Subject(s) - principal component analysis , line (geometry) , artificial neural network , generalization , computer science , regression , regression analysis , grid , feature (linguistics) , linear regression , identification (biology) , photovoltaic system , pattern recognition (psychology) , reliability engineering , algorithm , statistics , artificial intelligence , engineering , mathematics , machine learning , electrical engineering , mathematical analysis , linguistics , philosophy , geometry , botany , biology
Line loss rate of substation area is a comprehensive economic and technical index of power companies. With the continuous expansion of grid connected scale of distributed generation, accurate calculation of line loss rate of substation area with distributed generation is imminent. In this paper, considering the basic operation attributes and the grid connection attributes of distributed energy, a calculation method of line loss rate of substation area based on principal component analysis and generalized regression neural network is proposed. Firstly, the influence factors of line loss are extracted for the area containing photovoltaic distributed generation, and the feature dimension is reduced through principal component analysis; secondly, the reduced features are used as the feature input of generalized regression neural network for training, and the error between the training results and the real line loss is analyzed; finally, the ability of the proposed algorithm in the area identification of high line loss rate is tested. The results show that the generalized regression neural network trained in this paper has good generalization ability, the calculation error of online loss rate is small, and the identification of high loss area has high accuracy.