Wind Turbine Sensors
Industrial telemetry arrives through MQTT topics.
Current Project
Wind-Turbine Sensor Data Pipeline
A live industrial telemetry platform that moves wind-turbine sensor data through Kafka, stores time-series history in InfluxDB, and serves realtime plus historical monitoring to a Shiny dashboard.
Project Story
From raw turbine telemetry to usable monitoring.
The platform is designed around a practical data-engineering split: Kafka carries the live stream, InfluxDB keeps historical time-series data queryable, and the dashboard uses each system for the thing it does best. Live views connect to the Kafka live stream API over `/ws/live`, while historical charts call `/history`, which queries InfluxDB.
Architecture
Purpose-built flow for live and historical telemetry.
Kafka
Receives normalized windfarm telemetry and makes it available to storage and live API consumers.
Kafka Live Stream API
Consumes Kafka, keeps a live buffer, broadcasts `/ws/live`, and serves `/history` from InfluxDB.
store
Consumes Kafka messages and writes clean time-series points into InfluxDB.
InfluxDB
Stores historical readings for range queries and sensor comparisons.
consume
Shiny + Plotly dashboard combines `/ws/live` updates with Influx-backed historical charts.
Design Decisions
Realtime and history are treated as separate jobs.
Kafka as the stream boundary
Kafka is the realtime spine. The live API subscribes to the Kafka topic instead of coupling the dashboard to upstream telemetry ingestion.
InfluxDB for historical analysis
InfluxDB stores time-series data for historical queries, comparisons, and asset-level sensor analysis.
Websocket for true live UI
The dashboard receives snapshot, status, and event payloads through `/ws/live`, backed by the Kafka consumer in the live API.
Engineering Work
Problems handled so far.
This list is intentionally editable as the project grows.
Realtime transport
Replaced brittle live polling with websocket-based updates.
Stream boundary
Kept live updates on `/ws/live` while historical query charts use `/history` backed by InfluxDB.
Dashboard usability
Added separate live selection, multi-sensor history comparison, and asset filtering.
Operational packaging
Docker Compose keeps the pipeline services runnable as a full local system.
Use Cases
Where this architecture fits.
Industrial telemetry dashboards
Wind farm monitoring
Asset-level sensor analysis
Predictive maintenance visibility
Realtime data-engineering showcase
Tech Stack
Tools chosen for the data path.
Python
MQTT
Kafka
InfluxDB
Kafka WebSocket API
Docker Compose
Plotly
Shiny