It is some tech out of GE Digital that, from what I understand, uses sensor data + machine learning against a digital representation of the system to predict failures and tune performance. A friend of mine just recently took a job at a start-up using this technology (https://veerum.com/about/), however I haven't had a chance to catch up with him and really understand how this stuff works.
In large-scale systems, sensor information can be in different locations separated by great distances. For example, a pipeline system may have some assets in a central terminal as well as assets in the field miles away from the terminal - and all of these must be connected to a single asset model to monitor the health of the system as a whole.
Asset data can be ingested through a REST API using files in JSON format or created using a user interface. The data is stored in a graph database that is optimal for searching hierarchies of assets. Tags that indicate pressure or temperature can be associated with each asset. These tags can then be used to bring in and associate time series data and analytics to build digital twins of a complex asset.
For example, you could create an asset model that describes the logical component structure of all pumps in an organization, and then create instances of that model to represent each individual pump. Or you could create custom modeling elements that meet your unique domain needs if a particular pump has a few differences with the generic version.
In large-scale systems, sensor information can be in different locations separated by great distances. For example, a pipeline system may have some assets in a central terminal as well as assets in the field miles away from the terminal - and all of these must be connected to a single asset model to monitor the health of the system as a whole.
Asset data can be ingested through a REST API using files in JSON format or created using a user interface. The data is stored in a graph database that is optimal for searching hierarchies of assets. Tags that indicate pressure or temperature can be associated with each asset. These tags can then be used to bring in and associate time series data and analytics to build digital twins of a complex asset.
For example, you could create an asset model that describes the logical component structure of all pumps in an organization, and then create instances of that model to represent each individual pump. Or you could create custom modeling elements that meet your unique domain needs if a particular pump has a few differences with the generic version.
https://www.predix.io/services/service.html?id=1171