Novel glassbox based explainable boosting machine for fault detection in electrical power transmission system

Area of Science:

• Electrical Engineering
• Artificial Intelligence
• Power Systems

Background:

• Reliable electrical power transmission is vital for uninterrupted supply.
• Faults in transmission systems cause disruptions, economic losses, and safety hazards.
• Effective fault detection and classification are crucial for grid protection.

Purpose of the Study:

• To develop and evaluate an advanced artificial neural network methodology for fault detection and classification in electrical power transmission systems.
• To distinguish between single-phase, three-phase, and three-phase symmetrical faults.
• To propose an optimized, interpretable Explainable Boosting (EB) model for enhanced fault detection.

Main Methods:

• Utilized MATLAB for modeling and simulating fault scenarios using line currents and voltages.
• Applied time series analysis on signal data and SMOTE-based oversampling for dataset balancing.
• Developed and compared four machine learning and one deep learning model, focusing on an optimized Explainable Boosting (EB) approach.
• Incorporated hyperparameter optimization, k-fold validation, and eXplainable Artificial Intelligence (XAI) for model assessment.

Main Results:

• The proposed Explainable Boosting (EB) model achieved 99% efficiency in detecting and classifying transmission line faults.
• Demonstrated superior performance compared to traditional fault detection methods.
• XAI analysis provided insights into the EB model's decision-making process.

Conclusions:

• The optimized Explainable Boosting (EB) model offers a highly efficient and interpretable solution for fault detection in power transmission.
• The research contributes a scalable and adaptable method for advancing smart grid technology.
• The findings pave the way for more secure and efficient electrical power transmission systems.