Bring Edge computing, machine, APM and Predix catalog services all together
What it does
Application forecasts the temperature for next hour (trend analysis) using the 24 hours data.
How we built it
Using Predix machine data ingestion services application is ingesting data to APM from sensors deployed on Intel Edison board. Data is ingested to TimeSeries storage with tag TemperatureTAE.
Since APM analytics lags scheduling capability we have setup a Azuqua workflow which reads the data from TimeSeris instance and calls a custom microservice to make the prediction and pushes the data to timeseries store every one hour.
(Using Predix analytics runtime was another approach which could have followed)
Challenges we ran into
Accessing APM timeseries Enhancing the APM UI to display the forecast
Accomplishments that we're proud of
Configuring Edison board and working with sensors, ingesting data and working with time series
What we learned
APM was the big learning
What's next for TCSPredixTempAnalysis
Bringing humidity and other factors on board to do more coverage on analytics.