Why Artificial Intelligence in maintenance is required today across industries.
Industrial success is measured by a single metric: Uptime. We know equipment reliability issues are gaining importance across industries’ globally. Today, maintenance and reliability improvements are one of the most cumbersome task because of limited resources (manpower, hazardous in nature, budget), and moreover maintenance is the least activity to be outsourced. At present uptime is the most often used measure of performance of assets and maintenance.
In today’s challenging and competitive business environment along with thin margins, heavy manufacturing industries can’t afford to wait for critical machinery failures. It is also not possible that they can rely anymore on traditional maintenance regimes to minimize downtime, because the present maintenance technologies are seemingly outdated mechanism to meet the need of present and future industry dynamics. Just because the clock or counter says to do the work, you may think that your current maintenance programs are cost effective but repairing or replacing components after they have failed builds up an exponential pressure on production and revenue simultaneously. However, for many companies’ still daily maintenance tasks feels like a routine. It’s absolute critical for companies to realize that the maturity curve of their asset health, so that they can decide;
- What is the current asset health conditions
- Where will they get higher return of their investment in technology and processes
- How to evolve the existing maintenance regime
And, because of so many challenges the art of maintenance is on the tip of a transformation, where we need to evolve from time-based, to condition-based to predictive and move over to futuristic prescriptive maintenance. In our experience, many companies have either plans or are working towards implementing rigorous maintenance programs to reduce cost, optimize operations while improving uptime, and for these companies Prescriptive maintenance is the future, it uses next generation analytics to make predictions of asset health, provide recommendations and empowers to act on such recommendations. Systems like this must be “Artificially Intelligent” and must have the ability to think.
Let’s now examine how a next generation “Artificially Intelligent” maintenance program is headed in 2019 and beyond.
This is what I am focussing on today, because the next generation standard for effective maintenance would be through a prescriptive approach. If we look closely, we will find that prescriptive maintenance is at the intersection of big data, analytics, deep learning and cognitive computing. The use of non-invasive techniques in prescriptive maintenance to monitor the pulse of a structure or machine is unique. This technology can resolve the real-life problems the global industries faces today from unplanned downtimes and unfortunate situations. I am talking of a system which manages health of these critical assets in real-time and provide guided decisions assisting in removing human errors, furthermore, it monitors each asset’s condition to determine its fitness for continued operation and initiates repairs only when the machine or structure itself starts crying for help.
For example, recently we delivered “Autonomous Fatigue Life Assessment” system to a Defense organization, which combines machine learning and physics to prescribe current health and remaining life of an attack vessel under various loading and operating conditions. Ship by itself is a complex structure and subjected to extremely harsh operative conditions, hence there are few sensors to track the structural condition. The solution estimates the current health by continuously observing external and internal loading parameters in real-time and monitors the structure and its member elements behavior, such as strain, deflection, cracks etc., The ship global and local strength is assessed by using physics based numerical models such as computational fluid dynamics and finite element method, furthermore, the system goes beyond the boundary of limitation by using machine learning and deep learning techniques to address the most complex challenge of decisioning of when the ship need to go for repair and adjusting its operational parameters, so that a 100% uptime is maintained and additionally the life of the asset is extended effectively.
Moving forward, by evolving from traditional maintenance methods to prescriptive maintenance, companies can upgrade their maintenance regime from simply efficient to truly strategic, furthermore, by adding the flavor of cognitive computing companies can now integrate their maintenance and operations data with other data sources, such as quality, warranty and engineering data, to become a truly strategic on how the entire company operates.
About the Author
Niladri Dutta is the CEO of Tardid Technologies.