Attacking Unplanned Downtime Through Predictive Maintenance
Isaac Maw posted on September 25, 2017 |
A new testbed project will use machine learning to analyze IIoT data and predict failures.

A new Industrial Internet Consortium testbed aims to tackle the inefficiency of scheduled preventative maintenance. With the real-time data flow of IIoT-enabled machines and components, predictive maintenance has the potential to make preventative maintenance obsolete.

What is the Industrial Internet Consortium?

The Industrial Internet Consortium (IIC) was founded by AT&T, Cisco, General Electric, IBM and Intel in 2014. The IIC is a non-profit organization working to set standards, define best practices and accelerate the adoption of IIoT technology. Their testbed program enables real-world manufacturers to test, research and tinker with industrial internet applications. The resulting case studies and data helps bring those applications to market faster. Examples of IIC testbeds include smart airline baggage management, connected vehicle urban traffic management and connected medical care.

Stay up and running: downtime costs some manufacturers as much as $22k per minute.
Stay up and running: downtime costs some manufacturers as much as $22k per minute.

Smart Factory Machine Learning for Predictive Maintenance

The testbed is led by two companies that specialize in Industry 4.0 solutions: Plethora IIot and electronics company Xilinx. The main goal of the testbed is to evaluate machine learning techniques for predictive maintenance on real-world, high-volume production machinery.

The challenge is to detect and understand failures, to ultimately reduce downtime and increase energy efficiency. IIoT sensors can gather enormous amounts of real-time data, but critical analysis is required to maximize the usefulness of that data.

Machine learning is the solution to this problem. AI can identify, monitor and analyze system variables during operation, then alert operators before a system failure occurs. Ideally, this could eliminate costly unplanned downtime.

“Downtime costs some manufacturers as much as $22k per minute. Therefore, unexpected failures are one of the main players in maintenance costs because of their negative impact due to reactive and unplanned maintenance action. Being able to predict system degradation before failure has a strong positive impact on machine availability: increasing productivity and decreasing downtime, breakdowns and maintenance costs,” said Plethora IIoT team leader Javier Diaz.

“We’re excited to lead this testbed with Xilinx and work alongside some of the leading players in IIoT technologies. This is a unique opportunity to test together machine learning technologies with those involved in the testbed at different development levels starting from the lab through production environments, where a real deployment solution is utilized. As a result, from these experiences, we can significantly reduce the time-to-market of Plethora IIoT solutions oriented to maximize smart factory competitiveness,” Diaz added.

First, the plan is to develop the machine learning techniques and algorithmic approaches in a lab setting in Spain, then try it out in a controlled production environment. Ultimately, the final test will be in deployment at an automotive OEM manufacturing facility.

Predictive maintenance has the potential to add significant value to production processes by increasing efficiency and reducing unplanned and redundant costs. Additionally, the capability for better analysis of IIoT data makes IIoT devices more valuable, as more and more uses for the data are discovered.

To learn more about the Industrial Internet Consortium’s Testbed Program, click here. For more information about the Industrial Internet of Things, click here.


Recommended For You