Where AI is Proving Its Value and What's Coming Next

July 9, 2024
Research from Fluke Reliability and insights from Rockwell Automation show how artificial intelligence is changing predictive maintenance spending and how access to device dark data is poised to expand AI use and manufacturing software capabilities.

Links to the articles referenced in this podcast:


Welcome to the Automation World Gets Your Questions Answered podcast. I’m David Greenfield, editor in chief at Automation World and in this episode, we’ll be looking at where artificial intelligence is proving its value in manufacturing today and take a look at what may be coming next.
So, while the jury may still be out on the value of generative AI to manufacturing—at least until the copilot technologies being developed by several automation technology suppliers are in wider use—artificial intelligence analytics has clearly shown its value.
And the main area where AI has been proving its value is with equipment data analytics that help manufacturers move from reactive or scheduled maintenance to predictive maintenance.
According to Ankush Malhotra, president of Fluke Reliability, a supplier of asset management software, “Predictive maintenance is becoming a need, not a want in industry, especially as skilled labor is hard to come by and retain. He says, AI offers a clear pathway to predictive maintenance applications and there is a strong belief within industry that manufacturers who don’t adapt to these benefits are likely to be left behind.
In this comment, Ankush was referring to a recent study conducted by Censuswide for Fluke Reliability that surveyed more than 600 senior decision-makers and maintenance professionals in the U.S., the U.K. and Germany.
The results of the survey indicate that many manufacturers are planning to implement AI technologies into their day-to-day operations in the very near term for predictive maintenance and related machine learning activities.
According to the survey, only 8% of respondents currently have a predictive maintenance strategy. However, 77% are planning to shift to predictive maintenance with the help of AI technologies.
Overall, respondents said they intend to invest 44% of their technology budgets on AI in 2024 alone. In fact, 30% of those surveyed plan to invest between 50 and 75% of their technology budget on AI this year.
The survey also indicates that 61% of respondents expect to achieve their near-term AI goals within the next year.
The main drivers behind this focus on AI, among the manufacturers surveyed for this study, include: Addressing data processing and analysis requirements, improving production efficiencies and customer service, and compensating for the skilled labor shortage.
Aaron Merkin, chief technology officer at Fluke Reliability, said “It’s no surprise that manufacturers are bullish in their adoption of AI. We know it works and have customers who have seen value from it in as little as three months.”
Digging deeper into the AI trend around manufacturing data and looking to what’s coming in the near future, Paul Brooks, senior manager of open architecture management at Rockwell Automation, points out that, despite the proliferation of smart devices and associated AI technologies, most manufacturers are still a long way from realizing the true potential of smart manufacturing. The reason for this, he says, is because the real work of enabling transformative smart manufacturing won’t happen at the controller level, but at the software level. 
His point is that most of the data generated by a manufacturer’s array of production equipment and devices lies deep in the device firmware and is unseen by the user of the device. Paul calls this data at the firmware level dark data. He claims this data can enable more comprehensive predictive maintenance than is possible now. He says it can also be used to develop smarter machines that can self-configure for new production runs and provide better insights into how machines and plants are performing across an enterprise.
The ability to get at this data is coming to industry based on ongoing work at the OPC Foundation and ODVA to advance the EtherNet/IP protocol and the OPC UA information modeling framework. Paul says the work being done by these groups will make dark data from multi-vendor production environments available in the form of structured, human-readable models that will be used to develop new software applications.
Now, of course, AI-enhanced software exists today to monitor production assets and notify maintenance teams if deviations are detected. The challenge, Paul says, is that many of these applications can only be applied to a vendor’s own technologies. However, most production environments use technologies from multiple vendors—and those vendors naturally don’t want to reveal their proprietary secrets by exposing their design documentation. 
That’s where the work being done by OPC UA and ODVA comes into play. The goal of their work around this is to allow automation vendors to make their devices become self-describing to communicate not only their data, but also their capabilities and functionalities, thereby controlling the way that proprietary information is exposed.
Paul says this will break open the door for data scientists to create software that can finally connect to all of a producer’s control and automation devices. And that software will have much larger datasets than anything that maintenance teams have been able to access thus far, because it will leverage not just the process data that’s widely used today but also massive volumes of device dark data. He says this will allow the software to uncover opportunities to optimize production from the machine level to the enterprise level.
If you’re interested in learning more about this and you’re accessing this podcast from one of the many podcast platforms, visit the page where this podcast resides on the Automation World site at, www.automationworld.com/55090352 . There, I’ve placed links to the two articles we’ve recently posted from which the points highlighted in this podcast are taken. So, check out those links to learn even more about how AI is driving predictive maintenance spending in industry and how access to dark data is poised to expand AI-driven analytics and the operations benefits it provides even further.
So thanks for listening to this episode of the Automation World Gets Your Questions Answered podcast series. And remember to keep watching this space to stay on top of the latest news, trends and insights on the world of industrial automation.