Classification of abnormal location in medium voltage switchgears using hybrid gravitational search...

Area of Science:

• Electrical Engineering
• Artificial Intelligence
• Computational Intelligence

Background:

• Accurate and efficient fault diagnosis in power system switchgears is crucial for grid reliability.
• Existing methods may lack the speed or precision required for real-time applications.
• Developing advanced diagnostic techniques is essential for modern power infrastructure.

Purpose of the Study:

• To propose a hybrid Gravitational Search Algorithm (GSA)-artificial intelligence (AI) technique for classifying abnormal locations in switchgears.
• To evaluate the performance of Artificial Neural Network (ANN) and Support Vector Machine (SVM) classifiers optimized by GSA.
• To compare the proposed hybrid approach with other metaheuristic techniques.

Main Methods:

• Utilized measurement data from ultrasound, transient earth voltage, temperature, and sound sensors.
• Employed Artificial Neural Network (ANN) and Support Vector Machine (SVM) as AI classifiers.
• Optimized classifier performance using the Gravitational Search Algorithm (GSA).
• Compared GSA-optimized classifiers against unoptimized versions and other metaheuristic techniques.

Main Results:

• GSA optimization significantly improved the accuracy of both ANN and SVM classifiers.
• The GSA-optimized ANN achieved a slightly higher accuracy (97%-99%) compared to GSA-optimized SVM (95%-97%).
• GSA-SVM demonstrated faster convergence than GSA-ANN.
• The hybrid GSA-AI approach outperformed several other well-known metaheuristic techniques.

Conclusions:

• The hybrid GSA-AI technique offers an effective solution for switchgear abnormal location classification.
• GSA optimization enhances the diagnostic capabilities of AI classifiers for power systems.
• The proposed method provides a computationally efficient and accurate approach to fault diagnosis.