TY - GEN
T1 - Power Quality Disturbance Classification Using Deep BiLSTM Architectures with Exponentially Decayed Number of Nodes in the Hidden Layers
AU - Gunawan, Teddy Surya
AU - Husodo, Budi Yanto
AU - Ihsanto, Eko
AU - Ramli, Kalamullah
N1 - Funding Information:
The authors would like to express their gratitude to the Malaysian Ministry of Education (MOE), which has provided research funding through the Fundamental Research Grant, FRGS19-076-0684 (FRGS/1/2018/ICT02/UIAM/02/4). The authors would like to acknowledge as well as support from International Islamic University, University of New South Wales, Universitas Indonesia, and Universitas Mercu Buana.
Publisher Copyright:
© 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.
PY - 2022
Y1 - 2022
N2 - In recent years, there is growing interest in automatic power quality disturbance (PQD) classification using deep learning algorithms. In this paper, the average of instantaneous frequency and the average of spectrum entropy were used as time-frequency based feature extraction due to its discriminatory nature. Bidirectional Long Short-Term Memory (BiLSTM) architectures with exponentially decayed number of nodes in deep multilayers were utilized as Deep Recurrent Neural Network (DRNN) classifier. We experimentally generated fifteen classes of synthetic PQD signals. Each class contains 1000 samples divided randomly into training, validation, and testing. Results showed that four hidden layers of BiLSTM with exponentially decayed nodes interleaved with dropout layers provided the best classification accuracy of 99.23%.
AB - In recent years, there is growing interest in automatic power quality disturbance (PQD) classification using deep learning algorithms. In this paper, the average of instantaneous frequency and the average of spectrum entropy were used as time-frequency based feature extraction due to its discriminatory nature. Bidirectional Long Short-Term Memory (BiLSTM) architectures with exponentially decayed number of nodes in deep multilayers were utilized as Deep Recurrent Neural Network (DRNN) classifier. We experimentally generated fifteen classes of synthetic PQD signals. Each class contains 1000 samples divided randomly into training, validation, and testing. Results showed that four hidden layers of BiLSTM with exponentially decayed nodes interleaved with dropout layers provided the best classification accuracy of 99.23%.
KW - Bidirectional long short-term memory
KW - Classification
KW - Power quality disturbance
KW - Recurrent neural network
KW - Time-frequency based feature extraction
UR - http://www.scopus.com/inward/record.url?scp=85112553944&partnerID=8YFLogxK
U2 - 10.1007/978-981-33-4597-3_65
DO - 10.1007/978-981-33-4597-3_65
M3 - Conference contribution
AN - SCOPUS:85112553944
SN - 9789813345966
T3 - Lecture Notes in Electrical Engineering
SP - 725
EP - 736
BT - Recent Trends in Mechatronics Towards Industry 4.0 - Selected Articles from iM3F 2020
A2 - Ab. Nasir, Ahmad Fakhri
A2 - Ibrahim, Ahmad Najmuddin
A2 - Ishak, Ismayuzri
A2 - Mat Yahya, Nafrizuan
A2 - Zakaria, Muhammad Aizzat
A2 - P. P. Abdul Majeed, Anwar
PB - Springer Science and Business Media Deutschland GmbH
T2 - Innovative Manufacturing, Mechatronics and Materials Forum, iM3F 2020
Y2 - 6 August 2020 through 6 August 2020
ER -