On-boarding and data flow
In Plant Performance Analytics, there are two methods for onboarding the data, onboarding for master, event and maintenance data, and onboarding parametric data.
On-boarding Master and Event Data
The master data and event data are loaded by placing the respective .csv file in the specified inputdatain directory where the PPADataLoader application is running. An orchestration xml file defines the flow of triggering the analytical models and the maintenance optimization model from the model library.
On-boarding Parametric Data
The parametric data is loaded on an Apache Hadoop File System in Hbase. The Plant Performance Analytics solution uses the Apache Phoenix capability of interacting with the Hbase with JDBC calls. The Parametric table is defined in Apache Phoenix.
Once the table is defined, the data is loaded with a MapReduce job. The parametric data in .csv format is processed by a MapReduce job to filter out the required data fields for the analytical model. The MapReduce job then stores the data in Apache Phoenix Table that is in turn persisted in HBase.
The parametric data that is referenced here is the Torque data for a specific piece of equipment, in this case, a robot.
The data stores and output
Plant Performance Analytics stores the analytical model in the PPA data store. The analytical model stores the current and historical predicted data in the PPA operational data store.