Inspiration

We want to build E2E use case for showing how data from Physical Asset can be ingested into Predix Timeseries, analytics and visualizations to predict MW Power and optimize the process to control the Asset.

What it does

It reads data from Intel Edison Chip Board using Predix Machine.data for Turbine Speed, Inlet and Outlet temperature. RPM is simulated by using different sensor. All tags are ingested into time series. With a time interval, analytics run and calculate the Predictive power in MW that would have been generated in those operating conditions. Using MQTT, a real time graph is displayed which shows predicted power vs power demand. The delta between the Predicted Power and the Actual Power in this graph will represent the scenarios that need tuning of the Asset like increasing the RPM or outflow/inflow to generate more power.

How we built it

Flow of application is 1.We used Edison Board >> Timeseries >> Analytics (Catalog/Runtime) >> Messaging MQTT >> Schedulers >> Predix Seed User interface and Intel IOT Software to control the Step Motor 2.Edison Board : We used Predix Machine and Intel IOT Java libraries to run on same machine by reading and writing data. 3.Timeseries & Analytics : Integrated Time series with Power Curve Analytics that predicted the MW generation 4.MQTT: Messaging backbone. Processing time based on 2 event streams & predicted MW Power from Analytics and Demand Simulator by simulating the power demand 5.Creating a Web socket based implementation on MQTT to connect User interface and show the graph on run-time.

Challenges we ran into

1.Local WIFI Setup and connectivity with Intel Edison board 2.Issues with PostgreSQL Connection issues.

Accomplishments that we're proud of

  1. Step Motor that can be controlled from Cloud - show how Predix Asset can improve and optimize its operations
  2. Complete setup of Intel board and changes done in the Pre Processor to capture & change different sensors to simulate the running turbine. like Rotary switch value & simulate he RPM
  3. Creating a scalable architecture using MQTT based messaging system that allow different data sets to be treated like stream of events for easy consumption on UI layer as well as implementing run time business(Demand Vrs Predictive) rules as data comes in

What we learned

1.Never limit your ideas. Using Predix and Predix Machine as a PaaS, there are ample opportunities to change world 2.Show how quick a use case can be converted from an idea to actual implementation- Predix power 3.How to use Intel Edision Hardware to program and control multiple sensors

What's next for DigitalTwin-GasTurbineProfile – controlled & predictive

Built With

Share this project:
×

Updates