A new paradigm for electrical modeling in site leakage diagnostics: A label-scarce three-dimensional simulation surrogate using physics-informed neural networks

Key Points

• Improvements in site leakage diagnostics are demonstrated through a new simulation model, enhancing accuracy.
• The model effectively incorporates physics-informed neural networks, achieving notable performance gains.
• Assessment using a label-scarce three-dimensional simulation surrogate shows advanced capabilities in diagnostics.
• This model may enable more efficient detection techniques, warranting further exploration in practical applications.