TY - GEN
T1 - Deep Learning based Fault Diagnosis in a Grid-Connected Photovoltaic Systems
AU - Hichri, Amal
AU - Hajji, Mansour
AU - Mansouri, Majdi
AU - Bouzrara, Kais
AU - Nounou, Hazem
AU - Nounou, Mohamed
N1 - Publisher Copyright:
© 2022 IEEE.
PY - 2022
Y1 - 2022
N2 - Photovoltaic (PV) system fault diagnosis is crucial because it helps PV system operators reduce energy and income losses. It also decreases the risk of fire and electric shock from PV system failures. Thus, the implementation of fault diagnosis tool is required. The most well-known techniques are machine learning models. Although these techniques provided well in terms of diagnosis and classification accuracy, they rely on manual feature extraction, which is time consuming, expensive, and requires diagnostic skill. Manual feature extraction is a problem that has to be addressed. This paper therefore presents a PV system fault diagnosis techniques that can automatically extract features from raw data in order to classify PV system faults. The implemented methods include long short-term memory (LSTM) bidirectional long short-term memory (BiLSTM) networks, which are deep learning based algorithms. Finally, the presented fault diagnosis approaches showed a good diagnosis performances when applied to an emulated GCPV system.
AB - Photovoltaic (PV) system fault diagnosis is crucial because it helps PV system operators reduce energy and income losses. It also decreases the risk of fire and electric shock from PV system failures. Thus, the implementation of fault diagnosis tool is required. The most well-known techniques are machine learning models. Although these techniques provided well in terms of diagnosis and classification accuracy, they rely on manual feature extraction, which is time consuming, expensive, and requires diagnostic skill. Manual feature extraction is a problem that has to be addressed. This paper therefore presents a PV system fault diagnosis techniques that can automatically extract features from raw data in order to classify PV system faults. The implemented methods include long short-term memory (LSTM) bidirectional long short-term memory (BiLSTM) networks, which are deep learning based algorithms. Finally, the presented fault diagnosis approaches showed a good diagnosis performances when applied to an emulated GCPV system.
KW - Bidirectional Long Short Time Memory (BiLSTM)
KW - Fault Diagnosis
KW - Grid-connected PV systems
KW - Long Short Time Memory (LSTM)
UR - http://www.scopus.com/inward/record.url?scp=85143750625&partnerID=8YFLogxK
U2 - 10.1109/SSD54932.2022.9955779
DO - 10.1109/SSD54932.2022.9955779
M3 - Conference contribution
AN - SCOPUS:85143750625
T3 - 2022 19th IEEE International Multi-Conference on Systems, Signals and Devices, SSD 2022
SP - 1150
EP - 1155
BT - 2022 19th IEEE International Multi-Conference on Systems, Signals and Devices, SSD 2022
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 19th IEEE International Multi-Conference on Systems, Signals and Devices, SSD 2022
Y2 - 6 May 2022 through 10 May 2022
ER -