According to Gartner research, more than 50% of analytic implementations will make use of event data streams generated from instrumented machines, applications or individuals – the essence of IoT in manufacturing. In an earlier article, Intelligent machines depend upon machine data analytics it was described how the data generated by machines is used by other machines to take further decisions. As more devices get connected to each other, they will need to develop their own sense of “intelligence” to take certain actions and not rely on human intervention.
What is IoT in manufacturing all about?
All these intelligent machines will make use of a real time stream data, which comes from events generated by machine sensors. Enterprises are recognizing the economic value of information coming from such streaming data. As more machines get connected, they are finding opportunities to capture and harness even greater volumes of detailed data. But simply collecting this data for the sake of collection is clearly pointless. Analyzing this data, will require analytics technologies capable of making sense from this event stream data. This goes beyond traditional and mainstream BI to a breed of technologies capable of producing autonomous insights and inferences quickly.
To produce and harvest this data from physical assets and other event sources, the market will expand for flexible, multipurpose sensors for temperature, humidity, vibration, pressure, sound, light/color, electrical or other utility flows, motion, facial expressions, voice inflection, health monitoring and other systems. Moreover, such event data from physical assets (operational technology [OT]) is sometimes combined with event data from administrative information systems (information technology [IT]) to develop richer, more powerful holistic systems (creating an IT/OT convergence). In addition, technology and consumer product vendors are hastening to enable their wares to capture and emit more consumption and environmental data. Several BI application vendors in particular, have already intensified their ability to collect more usage data and are devising quid-pro-quo arrangements with customers that allow leveraging their de-identified data for alternate commercial purposes.
Here are three basic steps a business which is on the verge of generating significant machine data can take:
- Create an inventory of the range of current data collected by products, services and machines
- Identify what additional high-value information could be captured through further instrumentation or existing inventory
- Ensure that the data collected from IT systems, applications, devices and users is maximized with equal consideration for performance implications and probable future business relevance
Preparing for IoT in manufacturing require the following further steps: gather data, build predictive models, test the models for their predictive ability and deploy the models via a dashboards to allow real time (or near real time) interaction for process or plant engineers to help with their goals. For example, this article on how predictive modeling can leverage machine data analytics provides more background on the potential use of such data and models.
This post was contributed by Vaibhav Waghmare on Mon, Apr 14, 2014 @ 06:43 AM