Open AccessPhotovoltaic output prediction of regional energy Internet based on LSTM algorithm
Author(s) -
Geping Weng,
Chuanxun Pei,
Jiaorong Ren,
Hongzhi Wang,
Qinyue Cui,
Hua Qing,
Yuan Liu,
Xinyu Guan
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 - 10.1088/1742-6596/1732/1/012083
Subject(s) - adaptability , photovoltaic system , computer science , artificial neural network , scheduling (production processes) , modular design , algorithm , recurrent neural network , power (physics) , energy (signal processing) , real time computing , artificial intelligence , mathematical optimization , engineering , mathematics , electrical engineering , statistics , physics , quantum mechanics , biology , operating system , ecology
By the huge development of large scale and modular photovoltaic power generation, accurate photovoltaic (PV) output prediction can help PV power station, scheduling department and power system operate safely and economically. In the process of PV output prediction, the data density is large, and the output data is relatively regular. Therefore, this paper considers the use of long-termed and short-termed memory neural network algorithm to optimize the problem of algorithm gradient vanishment in recurrent neural network, and complete the output prediction of PV power in the regional energy Internet on the basis of historical output data. In this paper, LSTM algorithm is used to analyze the historical output data of PV stations in an industrial zone of a certain city. It can be found that LSTM algorithm has good adaptability for short-term PV output prediction, which can meet the needs of application.