AI and condition monitoring

The rise of artificial intelligence (AI) has unlocked multiple efficiencies and scaled-up processes for manufacturers and is seen as particularly useful in the field of condition monitoring.

Condition monitoring — the process of monitoring changes to a piece of equipment’s condition — is primarily used to detect developing issues or faults and prevent a possible breakdown. Applying AI to this process has allowed condition monitoring to become increasingly sophisticated, with some researchers estimating that AI-enabled condition monitoring can predict equipment failures with a 92% accuracy rate.

More accurate condition monitoring is enabled through AI’s advanced data analysis capabilities, integrating and analysing data from multiple sources like sensors, temperature gauges and cameras, all in real time. This ability to process multiple data streams simultaneously allows AI to extract insights and trends from the data, as well as allowing for the swift detection of anomalies.

AI-enabled condition monitoring also improves predictive maintenance — a crucial benefit when considering that unplanned downtime costs manufacturers an estimated US$50 billion annually.¹ Predictive maintenance is intended to reduce the risks of unexpected machinery failure or breakdown by flagging when time-sensitive maintenance is needed.

With the help of AI-enabled condition monitoring, manufacturers can not only prevent unexpected breakdowns, but also better understand the root causes of potential issues. This can also keep costs down by scheduling maintenance only when needed, avoiding unnecessarily time-consuming and expensive repairs.

We’re also seeing an increasing number of manufacturers employing AI tools to estimate the remaining lifespan of a particular asset, which enables better budgeting and inventory management, as well as ensuring minimal downtime while newer models are ordered.

The benefits unlocked by integrating AI into condition monitoring processes are further enhanced when applied to a condition monitoring tool. A good condition monitoring tool will include hardware and integrated software that can automatically collect and evaluate data from the IO-Link sensors.

Since any IO-Link sensor can be used, and the base unit doesn’t require a connection to the machine controller, a manufacturer employing condition monitoring systems shouldn’t have to adapt their existing systems, only supplement them. The onboard AI can be trained to learn ‘normal’ machine operation and to detect deviations and anomalies beyond simple thresholds.

So, while we’ve touched on the real-life applications of AI in condition monitoring and the tools currently available that allow for the seamless and fast integration of AI onto your factory floor, allow me to get futuristic for a moment. I’m talking, of course, about automated maintenance.

As AI capabilities continue to advance, we may soon start seeing an increase in equipment fitted with AI-based tools that enable automated maintenance — allowing equipment to self-repair without human intervention.

The possibilities of this are endless, ranging from tools that allow for self-diagnosis of machine condition and possible issues, automatic resetting of parameters as needed and, in the most extreme example, self-repairing of components. IIoT-enabled machines will also be able to communicate with other machines on the network, requesting assistance if necessary.

While fully automated maintenance through AI is still a way off, it will eventually be an extremely effective tool in reducing the probability of human error, and maintenance and downtime costs.

^{1. IndustryWeek and Emerson 2016, Unlocking Performance, <<https://partners.wsj.com/emerson/unlocking-performance/how-manufacturers-can-achieve-top-quartile-performance/>>}

*Jim Wallace is the Sales Director of Balluff Australia & New Zealand and a member of the Open IIoT cohort. He regularly contributes for features about Industry 4.0 and Smart Automation in Manufacturing.

Top image credit: iStock.com/anandaBGD