The operational brain: A new paradigm for intelligent data management in the industrial IoT
The Industrial Internet of Thing (IIoT) promises to allow organisations to deploy ever-increasing volumes of machine and sensor data to optimise myriad production processes, enhance security, and improve the worker experience (whether that employee is on the factory floor or in the office).
Industrial businesses, says Christian Lutz, CEO, Crate.io, are finding out that old paradigms of data processing don’t help their teams keep up with the speed of data, don’t match new analytic algorithms and, perhaps most critically, don’t enable the competitive need for real-time data queries.
An approach for solving this problem is combining modern distributed (open source) database architectures with Machine Learning/Artificial Intelligence, and IIoT networks. Together, these technologies form a rather new data management paradigm – what I’d call the operational brain – that goes beyond the traditional notions of databases and solves increasing data issues acute to industrial and manufacturing businesses.
Defining the operational brain
Traditional relational databases (such as Microsoft, SQL Server and Oracle) are technically incapable, usually, of processing the massive volume of data that must be handled for IIoT applications to be successful. These databases really weren’t designed to create the kind of backbone required to develop smart factories, smart cities, or driverless vehicles; use cases like these demand faster and more intelligent data processing. A comprehensive database management strategy is ultimately measured by the added business value of its use – not by its amount of memory or the hard disk’s speed.
I call this type of comprehensive IIoT data management “the operational brain.” The brain is the organ that can receive, structure, and make decisions based on this data. The data management system of the future will invariably function like our central nervous system, connecting directly to sensory impressions and using artificial intelligence to monitor, predict, and control systems in real time.
Data acquisition and enrichment
The modern, networked factory incorporates a diverse array of machines from different manufacturers. The challenge is thus to capture dissimilar data structures, analyse them in the cloud, and derive actions from them. Modern data management systems already start here. This simplifies the implementation and reduces error rates, since the machine and the database do not communicate via a third instance.
Without context, the collected data is useless for further processing. A recorded value is initially just a number: 108. Is that a temperature? If so, is it Celsius or Fahrenheit? Is it a product count? If so, when was the counter reset and what does it actually count? Data needs to be enriched to be meaningful. This enrichment requires three components: a database, a runtime that executes certain rules, and knowledge about the data’s meaning.
The operational brain combines all of these necessary steps into a single model. It saves industrial organisations from writing algorithms in order to make the data available for further processing. Instead, rules can be set up to interpret the stream of processed information. The operational brain is essentially the machine responsible for executing rules that automate processes, improving overall equipment effectiveness (OEE) in factories. It uses real-time data collection from potentially tens of thousands of sensors on equipment deployed across hundreds or thousands of product lines, in one or more connected and remote factories.
The operational brain’s centralised “mission control” processes and analyses sensor data, then provides predictive alerts about relevant maintenance that’s required on the factory floor, such as informing an employee that a particular machine needs to be cleaned every X hours, or alerting an engineer about an error rate in the manufacturing process. In short, the information it provides is far more efficient than anything possible with visual inspections.
The value of data-driven automation
Intelligent data management is much more than just a database. It describes a comprehensive process, from fast acquisition to intelligent data retrieval. Data-driven automation will become the key to success of IIoT projects. It will allow facilities like smart factories to enable real-time data analysis, maintain consistent uptime, ensure rapid development and time-to-value, and ensure low IT operating costs for hosting, integration, and administration.
Take, for example, ALPLA – a manufacturer of plastic packaging for brands like Coca-Cola and Unilever. The company has an operational brain in use to optimise its OEE. Data collected from tens of thousands of sensors on 900 different factory-specific sensor-types is enriched. It then informs the cloud for automated processing, and also a central control room – which in turn monitors plant performance in connected (but remote) factories. From these insights, ALPLA is able to identify trends at an earlier stage and its machine operators can be guided to necessary adjustments, including predication use cases.
With visual inspection systems in almost every production line at their factories, it’s challenging – if not impossible – to have staff on the production floor to react to changes. The operational brain strategy allows for real-time sensor data collection and analysis, directing employees to critical spots (and resulting in lower scrap rates and better efficiency).
The transition is emblematic of the IIoT’s broader shift, as it matures: data collection is not enough. The operational brain – or whatever one prefers to term it – will become an IIoT prerequisite to keep up.
The author is Christian Lutz, CEO at Crate.io, developers of the CrateDB open source real-time SQL DBMS and the Crate.io Machine Data Platform.