TripGuard

TripGuard is a web app with an AI layer that helps fossil and transition power plants reduce forced outages by predicting unit trips and surfacing the most likely root causes. It ingests historian/SCADA tags, DCS alarms, maintenance work orders, and operator logs, then builds a “trip fingerprint” library for each unit (boiler, turbine, HRSG, balance-of-plant). The app flags leading indicators (e.g., valve stiction, vibration drift, burner instability) and recommends targeted checks before the next start or ramp. It also auto-generates a post-event package: timeline, alarm rationalization hints, and evidence for internal reliability reporting and regulator/insurer questions. This is not a magic black box—models are constrained to plant-specific patterns, with confidence scores and traceable signals so engineers can trust (or reject) recommendations. The MVP focuses on one unit type and a small set of high-value trip modes.