SymbolicPhasor
Power system phasor estimation via Deep Symbolic Regression (NAPS, 2026).
While black-box neural networks can estimate fault current phasors rapidly, protective relays require interpretable, computationally lightweight mathematical expressions for real-time hardware deployment.
Supported by the UM-Dearborn EXP+ Independent Research Grant, SymbolicPhasor leverages Deep Symbolic Regression (DSR) to discover closed-form analytical equations for accurate phasor estimation under fault transients.
Publication:
- S. Mohammadi, W. Su (2026). “SymbolicPhasor: Power System Phasor Estimation via Deep Symbolic Regression,” 58th North American Power Symposium (NAPS) (Accepted). arXiv:2608.08552