TY - GEN
T1 - A hybrid contextual deep learning model to predict renewable energy generation
AU - Kanneganti, Deepak
AU - Mistry, Sajib
AU - Rajakaruna, Sumedha
AU - Krishna, Aneesh
AU - Beheshti, Amin
PY - 2025
Y1 - 2025
N2 - Renewable energy sources are widely adopted as they are safer for generating energy with less atmospheric harm. Solar power prediction depends on sunlight to forecast solar power effectively, and one of the significant challenges is weather dependency. Weather conditions, especially during overcast and rainy days, significantly impact power generation and can lead to infrastructure challenges and disruptions in the energy supply. We explore state-of-the-art deep learning algorithms to analyze contextual time series data, crucial for understanding the impact of weather uncertainty on power generation. To address this, we propose a novel deep learning-based hybrid contextual model built using Long Short-Term Memory (LSTM), Bidirectional Long Short-Term Memory (Bi-LSTM), and Gated Recurrent Unit (GRU) networks. This model leverages big data collected over ten years (from 2013 to 2023) to account for the variability and uncertainty inherent in weather conditions. Experimental results demonstrate that the proposed hybrid contextual model exhibits a 4.23% performance improvement over the individual deep learning models in predicting power under weather uncertainty conditions.
AB - Renewable energy sources are widely adopted as they are safer for generating energy with less atmospheric harm. Solar power prediction depends on sunlight to forecast solar power effectively, and one of the significant challenges is weather dependency. Weather conditions, especially during overcast and rainy days, significantly impact power generation and can lead to infrastructure challenges and disruptions in the energy supply. We explore state-of-the-art deep learning algorithms to analyze contextual time series data, crucial for understanding the impact of weather uncertainty on power generation. To address this, we propose a novel deep learning-based hybrid contextual model built using Long Short-Term Memory (LSTM), Bidirectional Long Short-Term Memory (Bi-LSTM), and Gated Recurrent Unit (GRU) networks. This model leverages big data collected over ten years (from 2013 to 2023) to account for the variability and uncertainty inherent in weather conditions. Experimental results demonstrate that the proposed hybrid contextual model exhibits a 4.23% performance improvement over the individual deep learning models in predicting power under weather uncertainty conditions.
KW - Renewable energy prediction
KW - Long Short-Term Memory (LSTM)
KW - Bidirectional Long Short-Term Memory (Bi-LSTM)
KW - Gated Recurrent Unit (GRU) networks
KW - Time series generator
UR - https://www.scopus.com/pages/publications/105009962666
U2 - 10.1007/978-981-96-6606-5_17
DO - 10.1007/978-981-96-6606-5_17
M3 - Conference proceeding contribution
AN - SCOPUS:105009962666
SN - 9789819666058
T3 - Lecture Notes in Computer Science
SP - 243
EP - 257
BT - Neural Information Processing
A2 - Mahmud, Mufti
A2 - Doborjeh, Maryam
A2 - Wong, Kevin
A2 - Leung, Andrew Chi Sing
A2 - Doborjeh, Zohreh
A2 - Tanveer, M.
PB - Springer, Springer Nature
CY - Singapore
T2 - 31st International Conference on Neural Information Processing, ICONIP 2024
Y2 - 2 December 2024 through 6 December 2024
ER -