H2electrolyzer
A real-time monitoring dashboard for hydrogen electrolyzers with AI predictive maintenance, built for industrial hydrogen operators (ACWA Power). It tracks 10 sensor metrics, detects 5 failure modes, and answers plain-language questions about the data.
Operators watch live electrolyzer telemetry — H₂ output, temperature, voltage, pressure, purity — with anomaly detection across five failure modes, and an AI chat that answers plain-language questions about the sensor data.
What it solves
- Industrial hydrogen operators need real-time visibility into electrolyzer health across many sensor metrics.
- Predictive maintenance requires detecting failure modes before they cause downtime.
- Engineers want to query sensor data in plain language, not dashboards alone.
Impact
H₂, temp, voltage, pressure, purity
Predictive detection
60× playback, 7-day analysis
Architecture
Data flow
- Sensor → CSV → Supabase (day1-7 tables)
- User selects Day N + playback speed
- Query day-N table by timestamp range
- Recharts renders 10 metrics + safe zones
- Alert system detects anomalies (rule / AI)
- User chats with AI → Vercel AI SDK → GPT-4
- Dashboard shows predictions + maintenance advice
Engineering decisions
Supabase PostgreSQL for time-series
Relational time-series tables give fast range queries over 7-day windows.
Vercel AI SDK with graceful fallback
Optional GPT-4 analysis falls back to rule-based alerts when AI is unavailable.
React Three Fiber for 3D schematics
Premium 3D electrolyzer visualization paves the way for future mobile AR.
Recharts for animated time-series
Responsive, animated rendering of the 10 metrics with visible safe-zone bands.
Gallery
