Power quality disturbance identification using hybrid deep learning in renewable energy systems

Area of Science:

• Electrical Engineering
• Renewable Energy Systems
• Artificial Intelligence in Power Systems

Background:

• The increasing integration of hybrid wind-solar photovoltaic (Wind-SPV) systems necessitates robust power quality monitoring.
• Ensuring high power quality is crucial for the reliability and customer satisfaction of renewable energy networks.
• Conventional methods for detecting power quality disturbances (PQDs) often have limitations in resolution or computational efficiency.

Purpose of the Study:

• To introduce a novel hybrid deep learning-based approach for the specific diagnosis of PQDs in Wind-SPV integrated networks.
• To evaluate the effectiveness of the proposed method in enhancing power quality within hybrid renewable energy frameworks.
• To demonstrate the superiority of the proposed technique over traditional PQD sensing methods.

Main Methods:

• A hybrid deep learning model combining Continuous Wavelet Transform (CWT) scalograms with deep neural networks (ResNet, VGG-Net).
• Integration of Neighborhood Component Analysis (NCA) and Support Vector Machine (SVM) for classification.
• Simulations conducted using MATLAB/Simulink on customized IEEE 9-bus and IEEE 13-bus test systems.

Main Results:

• The proposed method achieved high accuracy (98.54% on IEEE 9-bus, 97.17% on IEEE 13-bus), precision, and recall.
• Performance significantly outperformed conventional techniques like Fourier Transform and Discrete Wavelet Transform.
• Demonstrated effectiveness in complex, real-world hybrid renewable energy scenarios.

Conclusions:

• The developed hybrid deep learning approach offers an effective solution for PQD diagnosis in Wind-SPV networks.
• This advancement contributes to improved power quality and supports the global shift towards renewable energy.
• The method provides a reliable and computationally efficient alternative to existing PQD detection techniques.