As sensors are attached to more and more industrial machines, these connected devices create a massive glut of...
data. But to unleash the true value of Internet of Things (IoT) data analysis, businesses need to have a clear understanding of its strengths and weaknesses.
In a panel discussion at the 2016 IoT Data Analytics & Visualization conference in Palo Alto, Calif., several speakers said the most important thing to keep in mind is that the industrial Internet should be about improving business processes, not just about implementing cool new technology.
Nauman Sheikh, founder and CEO of analytics consulting firm Asrym Inc., said he recently worked with a large public utility to put sensors on maintenance trucks to do predictive maintenance. Sheikh built the predictive models that analyzed the data coming from the sensors to identify vibration patterns and other signs that a truck might be at risk of breaking down.
Talking IoT in business terms
But when he pitched the idea to the utility's management team, he didn't talk about the algorithms he would build or the sensor and networking technology required. He said the key to getting the project off the ground was to identify with the needs of the business team and "speak their language."
Nauman Sheikhfounder and CEO of Asrym Inc.
"The lesson learned was, if I was pitching to them some nice tools or fancy technology, there would have been no interest," Sheikh said. "If you can connect to their pain the value IoT can bring to the table, the adoption will be fast and very effective."
Similarly, Prakash Iyer, vice president of software architecture and strategy at Trimble Navigation Ltd. in Sunnyvale, Calif., said businesses interested in IoT data analysis should focus on the ultimate goal, which should be automating industrial processes. Trimble makes GPS software and connected hardware for industries such as agriculture and construction.
The point of any IoT or industrial Internet project, Iyer said, is to develop new insights from machinery into which businesses previously had little visibility. By analyzing data coming from machinery, either using simple visualizations or more complex machine learning algorithms, it may be possible to automate industrial processes by developing sets of business rules that kick in whenever an event occurs. Once businesses understand industrial processes, they can develop rules to automate them. But if an IoT data analysis project fails to deliver actionable insights, then the organization should question its investment, Iyer said.
"The most important thing is not the visualization, but how you make it actionable," Iyer said. "We need to be able to automate the next step."
Industrial IoT efforts need flexibility
For Sudhi Ranjan Sinha, vice president of product development, building technology and services at Milwaukee-based Johnson Controls, the key to realizing business value from industrial Internet analytics projects is planning for the future by being flexible.
Sinha said that in the heating and ventilation industry his company is part of, legislated uses of refrigerants and customer expectations of efficiency are likely to change over time. But most of the equipment that Johnson Controls produces has a life expectancy of around 25 years, he noted.
So, Sinha said, when he and his team build predictive models that measure efficiency and identify potential problems in connected heating and ventilation units, it's important to keep in mind that things will change and to build flexibility into the models by not relying on efficiency assumptions made when the unit is produced. "There is no permanency in this space," he said. "Every model we create has to be adaptive."
IoT data analysis still needs time to develop as a discipline
No value to Internet of Things without analytics to make sense of it all
Data architects need to plan ahead for massive volumes of IoT data