Open AccessDeep learning architecture for direct probability density prediction of small‐scale solar generation
Author(s) -
Afrasiabi Mousa,
Mohammadi Mohammad,
Rastegar Mohammad,
Afrasiabi Shahabodin
Publication year - 2020
Publication title -
iet generation, transmission and distribution
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.92
H-Index - 110
eISSN - 1751-8695
pISSN - 1751-8687
DOI - 10.1049/iet-gtd.2019.1289
Subject(s) - computer science , deep learning , artificial intelligence , convolutional neural network , probability density function , dropout (neural networks) , machine learning , probabilistic logic , photovoltaic system , intermittency , kernel density estimation , parametric statistics , density estimation , artificial neural network , network architecture , engineering , mathematics , statistics , meteorology , turbulence , physics , computer security , estimator , electrical engineering
With the increasing penetration of photovoltaic (PV) systems, the problems posed by the inherent intermittency of small‐scale PVs are becoming more severe. To address this issue, it is critical to involve the uncertainty of PV generation in the look‐ahead periods in a comprehensive framework. To this end, a direct deep learning architecture for probabilistic forecasting of solar generation is proposed in this paper. An end‐to‐end deep learning architecture as a novel mixture density network (MDN) is designed based on the combination of a convolutional neural network and a gated recurrent unit. Furthermore, a new loss function and training process based on adversarial training is proposed to enhance the accuracy in direct contracting of the probability density function. Then, several deep and shallow networks are implemented, and the results are compared with the proposed architecture. The effectiveness of the proposed MDN in providing complete statistical information is verified through comparison with Monte Carlo dropout, non‐parametric kernel density estimation, and the proposed MDN without adversarial training.