Power Distribution Phase Identification for Mixed Wye-Delta Loads

The patent uses inexpensive smart meters that are already deployed at customer premises instead of expensive advanced sensors with phasor measuring capability. The smart meters provide sufficient measurement data (real power, reactive power, voltage magnitude) for phase identification through information theory-based analysis of event signatures, eliminating the need for costly specialized equipment at each customer location.

The patent transforms the approach by changing the analytical parameters from direct correlation/regression of event signatures to mutual information analysis of probability distributions. By computing marginal probability distributions of event signatures and applying information theory metrics, the method effectively handles mixed wye and delta load connections without the cross-interference problems that plague traditional correlation-based methods.

The patent extracts only the essential features needed for phase identification from the measurement data - specifically, event signatures defined by current exceeding fluctuations above a threshold. By focusing on these extracted event features rather than processing complete high-rate measurement data, the method achieves accurate phase identification while significantly reducing computational burden and data processing complexity.

The patent enables phase identification using only the measurement data already collected by smart meters at customer premises for billing purposes. The method extracts event signatures from existing smart meter measurements (real power, reactive power, voltage) and uses information theory analysis to identify phases, eliminating the need for additional measurement devices or infrastructure at both customer and utility sides.