Improved hybrid islanding detection using data fusion, adaptive back propagation neural network and support vector...

Area of Science:

• Electrical Engineering
• Power Systems
• Artificial Intelligence

Background:

• Increasing integration of renewable energy sources (RES) necessitates advanced islanding detection methods in hybrid microgrids (HMGs).
• Traditional islanding detection methods face challenges in accuracy and effectiveness with complex power network structures.

Purpose of the Study:

• To propose a novel hybrid islanding detection method (IDM) for enhanced precision and effectiveness in HMGs.
• To accurately identify islanding events in real-time, minimizing non-detection zones and preventing nuisance tripping.

Main Methods:

• A data-driven approach combining data fusion with an adaptive backpropagation neural network (ABPNN) and support vector machine (SVM).
• Utilizing feature datasets including rate of change of phase angle difference (ROCPAD), intermittent-bilateral reactive power variation (IBRPV), and frequency.
• Offline data preprocessing using k-means clustering and logic operations for ABPNN training, followed by SVM classification of online data.

Main Results:

• Significant improvements in islanding detection accuracy and speed.
• Achieved a zero non-detection zone (NDZ), crucial for grid stability.
• Demonstrated minimal impact on power quality.

Conclusions:

• The proposed hybrid islanding detection method (IDM) is a highly promising solution for islanding detection in HMG environments.
• The IDM effectively enhances the precision and speed of islanding event identification.
• The method ensures reliable operation of HMGs with high RES integration.