Asset management moves towards AI

11 February 2024

Aaron Merkin explains the benefits that artificial intelligence (AI) can bring to asset management and maintenance tasks.

Continued labour challenges, paired with increased demand and supply chain problems are putting many industry sectors under strain, with managers being pushed to do more, with less.

Emerging technologies do promise some relief. The rise of Artificial intelligence (AI), for example, holds potential for significant change in manufacturing, but it will take some getting used to.

AI holds the potential to increase productivity, reduce downtime, and help scale operations. It can help maintenance teams move faster and work smarter. However, it is just one part of a much bigger puzzle. 

AI has appeared in the maintenance sector, claiming to answer some of the problems maintenance teams have faced for years.

The industry has already embraced Internet of Things (IoT) technologies and predictive maintenance strategies. The affordability of IoT sensors, combined with cloud technology, eases the ability to access and analyse condition monitoring data. Today sensors are now connected to more assets and can instantly serve data for analysis.

The heart of a predictive maintenance program is collecting key asset data and monitoring it across a prolonged period of time to highlight new faults. Vibration and temperature changes, for example, can reveal emerging problems that maintenance teams are then able to correct before they become a fault.

Having more data helps better predict faults so they can be conveniently scheduled and planned for in regards to both having the right resources and spare/replacement parts. 

However, more data can also present a challenge for human workers as more resources are needed to analyse it. As operations grow and additional worksites or assets are added, the amount of data also grows exponentially and teams care not able to effectively sift through it all.

AI can enable data to work faster than ever, sifting through it to give maintenance teams the tools they need to actually use the data that is being collecting.

Machine diagnostics
AI can also go beyond just detecting anomalies and diagnosing the root cause of them. It can also analyse and figure out whether changing vibration levels are caused by a bearing defect, for example, or a misaligned shaft.

From there, maintenance managers can make informed scheduling decisions. It enables them to ask questions, such as ‘Is this a minor issue that won’t impact productivity, or will it cause a shutdown if left unaddressed?’ Having the right data on hand is powerful.

Traditionally, vibration specialists have gathered condition monitoring data, which could be as simple as listening to a motor’s unique rattle and noticing subtle clues. 

This human expertise is still invaluable. Yet, increasingly complex systems combined with expertise constraints mean that maintenance teams can’t keep everything running on their own. The average plant now has more assets than ever and potentially fewer technicians.

Managers are often responsible for numerous sites and can’t manually monitor all of their machines. AI-powered condition monitoring tools will be crucial to help augment their capabilities and achieve scale.

AI can recognise anomalies in condition monitoring data through various training methods. One method involves manually setting thresholds, where the algorithm issues an alert when a certain level is crossed. 

However, AI can also learn to improve its accuracy over time as it takes in more data. It can learn the specific needs of each asset, accounting for differences in vibration signatures or maintenance requirements.

AI's pattern recognition capabilities enable it to diagnose machine faults and communicate effectively with human maintenance teams through generative AI. The more data AI takes in, the more effectively it is able to diagnose the root causes behind changes in condition monitoring data. 

The algorithm continues to fine-tune its diagnostic capabilities over time, so AI could, potentially, learn by studying work order data and the response time of human technicians. 

Implementing AI should be approached carefully – starting with a pilot program and regular testing and employee feedback. It's important to have well-defined benchmarks in place and ensure that everyone on the team has a shared understanding of goals. 

Once the pilot program is over, assess what's worked and begin expanding the program. Facilities that haven't yet moved to predictive maintenance would be well-advised to do so now.

The full potential of AI is still untapped, but early adoption can lead to greater productivity, uptime, and reliability while keeping costs low and making the most out of limited resources. Consulting wit¬h experts in the field can also help avoid pitfalls and accidents. 

Aaron Merkin is Chief Technology Officer at Fluke Reliability.

Contact Details and Archive...

Print this page | E-mail this page