An industrial data platform is software that gathers a plant's sensor readings, maintenance records and drawings into one connected model of the site, so that engineers can ask questions of it. This work is also called industrial DataOps. Without it, each kind of data sits in a different system and a different format, and an engineer has to open each system separately and match the results by hand.
Take, for example, an offshore oil platform in the Mumbai High field where a gas compressor stopped suddenly last month. The engineer wants to know why. The sensor readings are in a process historian, the store of a plant's readings over time. The repair history is in the maintenance system. The pipe layout is in drawing files. Each of these refers to the compressor by a different name or code, such as "K-2101" in one and "Gas Comp 1" in another. The platform links all of them to one record for that compressor. The engineer searches for the compressor once and sees its readings, its past repairs and its drawing together.
To do this, the platform needs three things. First, it collects readings as time series data, from the historian, from edge computers near the machines, and from live streams such as event streaming. Second, it reads documents and drawings, finds the equipment names written inside them, and attaches each document to the right machine. Third, it keeps a model of how the equipment is connected, so that a question such as "which machines depend on this compressor?" can be answered.
The buyer is usually the operations or maintenance head of an oil, gas, power or chemical company. The platform does not control the plant. It only reads data and keeps it organised, so the plant's control systems are not disturbed.
Other software then uses this connected data. Asset performance management software reads the history to learn what normal looks like. A plant digital twin shows the live readings on the plant model. Both are parts of predictive maintenance.
A process historian stores readings only. The data platform adds the repair records, drawings and links between them. The same idea is used in other industries. A chip factory, for instance, collects data from its semiconductor manufacturing equipment. For design and test records in aircraft making, see engineering data integration.