This Repository contains all source code for the time series example that processes telemetry information from 100 simulated engines.
Presentation can be found here
The overal architecture consists of the following components:
- Engine telemetry simulator: netcore simulator that can be used as IoT Edge module or just as a vanilla Docker container. (Details)
- IoT Hub: cloud based telemetry ingestion, having every simulated engine represented as a device
- Time Series Insights (TSI): Timeseries database instance in Azure that ingests all telemetry and allows for time series exploration
- Azure Data Explorer (ADX): The powerful data engine on which TSI is built.
- Azure Machine Learning: Workspace that performs training of predictive maintenance model and hosts it as a service
- Azure Stream Analytics: Streaming instance that can process and perform standing queries in the incoming telemetry stream and uses the above mentioned model
- Event Grid: The event driven service that will be used to publish predictive maintenance events to and seperate handling of events from the detection of events
- Azure Logic Apps: The workflow service that will handle the different predictive maintenance events
- An Active Azure Subscription
- A resource group with the following resources: IoT Hub, Azure Machine Learning Workspace, Time Series Insights environment, etc. (arm template will be provided later)
- For the actual Machine Learning training, following these steps (to do)