ANN-based fault classification and localization with optimized PMU deployment for transmission systems.

Area of Science:

• Electrical Engineering
• Power Systems Analysis
• Artificial Intelligence in Power Grids

Background:

• Modern power grids face challenges in fault detection due to dynamic conditions and diverse fault types.
• Conventional fault diagnosis methods are limited by static assumptions and data requirements, hindering real-time application.
• Accurate and timely fault detection is crucial for grid stability and operational efficiency.

Purpose of the Study:

• To develop a robust Artificial Neural Network (ANN)-based framework for fault classification and localization in transmission systems.
• To utilize only bus voltage measurements for fault diagnosis, overcoming limitations of conventional techniques.
• To introduce an optimized Phasor Measurement Unit (PMU) placement strategy for enhanced observability and reduced hardware costs.

Main Methods:

• A novel Artificial Neural Network (ANN) framework was developed for fault classification and localization.
• Physically meaningful features derived from bus voltage measurements were employed for enhanced robustness and interpretability.
• Simulations were conducted on the IEEE 14-bus system to validate the proposed methodology.
• A new Phasor Measurement Unit (PMU) placement strategy was introduced to optimize system observability.

Main Results:

• The proposed ANN-based method achieved over 98% accuracy in fault classification.
• Fault localization was performed with an error margin of less than 2% of the line length.
• The ANN approach demonstrated up to 5.66% higher accuracy and over 30% lower error rates compared to existing methods.
• The optimized PMU placement strategy improved observability while reducing hardware requirements.

Conclusions:

• The developed ANN framework offers a scalable and resilient solution for real-time fault management in transmission networks.
• Utilizing physically meaningful features from voltage measurements enhances the interpretability and robustness of fault diagnosis.
• The proposed methodology significantly outperforms existing data-driven techniques in fault classification and localization.
• The optimized PMU placement strategy contributes to more efficient and reliable power grid monitoring.