The use of additive decomposition and deep neural network for photovoltaic power forecasting
The use of additive decomposition and deep neural network for photovoltaic power forecasting
Predicting photovoltaic (PV) power production is indispensable for security and reliability of the grid. In this article, a short-term forecasting method, namely trend decomposition two-dimensional convolutional neural network based on additive decomposition and convolution neural network (CNN) is proposed. Firstly, the additive decomposition model is deployed to decompose the PV power generation series to the long-term trend (LT), the seasonal trend (ST), and the random component. Then, three independent two-dimensional convolutional neural networks are designed to extract daily and hourly dependencies among the decomposed components. Finally, the prediction results of these networks are summed for the final forecast. The one-day-ahead forecasting capability of the presented method is evaluated with two case studies using real datasets gathered from Limburg and Luxembourg, Belgium. Analysis of the prediction's results indicates that the proposed method has higher accuracy compared to individual multi-layer perceptron, two-dimensional convolutional neural network, long short-term memory (LSTM), gated recurrent unit, and bidirectional LSTM networks.
Additive decomposition, Convolution neural network, Deep learning, PV power forecasting
Dehghan, Fariba
e0863ef3-0a6c-467f-87ed-824cbd16408c
Parsa Moghaddam, Mohsen
fade89c8-c47d-497a-b52f-8631de238f0d
Imani, Maryam
7175b856-1aa2-4b2d-9c30-74c7fd30ae07
11 December 2023
Dehghan, Fariba
e0863ef3-0a6c-467f-87ed-824cbd16408c
Parsa Moghaddam, Mohsen
fade89c8-c47d-497a-b52f-8631de238f0d
Imani, Maryam
7175b856-1aa2-4b2d-9c30-74c7fd30ae07
Dehghan, Fariba, Parsa Moghaddam, Mohsen and Imani, Maryam
(2023)
The use of additive decomposition and deep neural network for photovoltaic power forecasting.
In 2023 31st International Conference on Electrical Engineering (ICEE).
IEEE..
(doi:10.1109/ICEE59167.2023.10334838).
Record type:
Conference or Workshop Item
(Paper)
Abstract
Predicting photovoltaic (PV) power production is indispensable for security and reliability of the grid. In this article, a short-term forecasting method, namely trend decomposition two-dimensional convolutional neural network based on additive decomposition and convolution neural network (CNN) is proposed. Firstly, the additive decomposition model is deployed to decompose the PV power generation series to the long-term trend (LT), the seasonal trend (ST), and the random component. Then, three independent two-dimensional convolutional neural networks are designed to extract daily and hourly dependencies among the decomposed components. Finally, the prediction results of these networks are summed for the final forecast. The one-day-ahead forecasting capability of the presented method is evaluated with two case studies using real datasets gathered from Limburg and Luxembourg, Belgium. Analysis of the prediction's results indicates that the proposed method has higher accuracy compared to individual multi-layer perceptron, two-dimensional convolutional neural network, long short-term memory (LSTM), gated recurrent unit, and bidirectional LSTM networks.
This record has no associated files available for download.
More information
Published date: 11 December 2023
Keywords:
Additive decomposition, Convolution neural network, Deep learning, PV power forecasting
Identifiers
Local EPrints ID: 505445
URI: http://eprints.soton.ac.uk/id/eprint/505445
PURE UUID: ae29d791-ed95-40fd-954c-dfa412f300bc
Catalogue record
Date deposited: 08 Oct 2025 16:55
Last modified: 09 Oct 2025 02:22
Export record
Altmetrics
Contributors
Author:
Fariba Dehghan
Author:
Mohsen Parsa Moghaddam
Author:
Maryam Imani
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics