This is a sample data architecture for a ERP centric landscape for a manufacturing customer . This fictional customer has IOT driven data analytics .
1. IOT sensors capture and stream data which is captured by the data layer
2. This data layer filters, cleanses and aggregates the IOT data
3. It takes out the pertinent data , for example the exceptional cases :
a.Temperature (it takes the records and timestamps when the part overheats)
b. Acceleration (for manufacturing company which makes automobiles , takes the values when the acceleration is not smooth )
c. Electrical outages (Voltage spikes )
4. All this data is identified by unique Equipment ID .
5. This can be co-related against ERP and other applications (On Premise / SaaS ) which stores all maintainance cases and service orders againts the Equipment ID
6. The Data Store aggregates these data into the Historical Data store for future analysis and also derives KPIs
7. The KPIs are sent to the Analytical Layer for visualization
Real Time Dashboards show performance for real time systems in discrete manufacturings . Predictive Analysis uses Machine Learning models to predict future behavior and co-relate failures . Azure / AWS / Google Cloud / IBM Watson all provide excellent ML functionalities
Why do we send the data to Historical Data store ? Maybe we think of a new KPI or a need to look at the old data , for which we dont have analytics right now .
Example : A new equipment type B is releases which is similar to product line A 3. This has a lot of electrical issues after releasing the products . Thus we want to look at the data of product line A which is stored in the historical data store
Comments
Post a Comment