Skip to main content
S

SIGMA

SIGMA offers a physics‑informed reliability engine that calculates real‑time fault probability for microgrids using power‑quality signals. By extracting five fast‑changing features—voltage sag, total harmonic distortion, power‑factor deviation, and the rates of change of voltage and current—the system delivers interpretable, millisecond‑scale predictions of incipient faults up to 12 seconds before they occur, enabling proactive protection of renewable‑rich, dynamic grids.

Updated 1 month ago

Funding

Funding not disclosed

Funding rounds are not available yet.

Founders

Founder details are not available yet.

Product

Problem

Microgrid operators often rely on traditional protection schemes that react only after a fault occurs, missing brief sub‑threshold voltage sags, current spikes, and harmonic bursts that precede permanent failures. These precursor signatures are too short and low‑magnitude for conventional relays, leading to undetected incipient faults and reduced system resilience.

Solution

SIGMA delivers a physics‑informed reliability engine that continuously computes a probabilistic fault score from real‑time power‑quality measurements. By extracting five interpretable features—voltage sag, total harmonic distortion, power‑factor deviation, and the rates of change of voltage and current—the system feeds a constrained logistic model that maps these indicators to a calibrated fault probability every millisecond. The probability stream is processed through a state machine that classifies operating conditions (NORMAL, AHEAD, FAULT, RECOVERY) and generates alerts up to 12 seconds before a fault manifests. The model is trained with ℓ₂‑regularized maximum‑likelihood on labeled data and updated online via a sliding window, ensuring monotonic risk assessment and adaptability to dynamic microgrid conditions.

Target Audience

Primary customers are operators of renewable‑rich microgrids and data‑center power systems that require early fault detection to maintain reliability and avoid equipment damage.

Features

  • Lightweight logistic regression model constrained to non‑negative coefficients for monotonic risk interpretation
  • Real‑time extraction of five physics‑based power‑quality features (voltage sag, THD, power‑factor deviation, dV/dt, dI/dt) at 1 Hz sampling
  • Kalman‑filter preprocessing and Pandapower digital twin integration for noise‑robust feature estimation
  • Continuous probability stream evaluated by a state machine with defined thresholds for proactive alerting
  • Millisecond‑fast inference and alert generation, providing ~12 second lead time before fault onset
  • Open API for integration with existing microgrid monitoring and control platforms
This profile is AI-generated and may contain inaccuracies.