TY - GEN
T1 - Area Control Error Forecasting using Deep learning for an Interconnected Power System
AU - Abdeltawab, Hussein
AU - Radwan, Amr
N1 - Publisher Copyright:
© 2022 IEEE.
PY - 2022
Y1 - 2022
N2 - Area Control Error (ACE) is an essential indicator of the load-generation power imbalance for the transmission system operator. ACE is used to correct the generation dispatch to compensate for frequency deviation. ACE also indicates the required power export or import in an interconnected power system. Unlike wind and solar power prediction, there has been no work to forecast the ACE in the power system. For an interconnected extensive transmission system, the ACE is considered a volatile time-varying signal. For an accurate ACE prediction, this work represents a deep learning-based forecasting model. The model decomposes the ACE signal using the discrete wavelet transform (DWT) and utilizes the bidirectional long short-term memory (BiLSTM) for the prediction. The proposed forecasting technique is trained to capture the deep temporal features of the signal with higher accuracy when compared to other methods. Two ACE datasets with sample times 1-minute and 10 minutes are predicted. The real data is gathered from Pennsylvania, New Jersey, and Maryland interconnection (PJM), USA. To evaluate the proposed technique, it is compared to other benchmark forecasting networks.
AB - Area Control Error (ACE) is an essential indicator of the load-generation power imbalance for the transmission system operator. ACE is used to correct the generation dispatch to compensate for frequency deviation. ACE also indicates the required power export or import in an interconnected power system. Unlike wind and solar power prediction, there has been no work to forecast the ACE in the power system. For an interconnected extensive transmission system, the ACE is considered a volatile time-varying signal. For an accurate ACE prediction, this work represents a deep learning-based forecasting model. The model decomposes the ACE signal using the discrete wavelet transform (DWT) and utilizes the bidirectional long short-term memory (BiLSTM) for the prediction. The proposed forecasting technique is trained to capture the deep temporal features of the signal with higher accuracy when compared to other methods. Two ACE datasets with sample times 1-minute and 10 minutes are predicted. The real data is gathered from Pennsylvania, New Jersey, and Maryland interconnection (PJM), USA. To evaluate the proposed technique, it is compared to other benchmark forecasting networks.
UR - http://www.scopus.com/inward/record.url?scp=85128604786&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85128604786&partnerID=8YFLogxK
U2 - 10.1109/PECI54197.2022.9744044
DO - 10.1109/PECI54197.2022.9744044
M3 - Conference contribution
AN - SCOPUS:85128604786
T3 - 2022 IEEE Power and Energy Conference at Illinois, PECI 2022
BT - 2022 IEEE Power and Energy Conference at Illinois, PECI 2022
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2022 IEEE Power and Energy Conference at Illinois, PECI 2022
Y2 - 10 March 2022 through 11 March 2022
ER -