You are here

Sensors and smart algorithms: keys to predictable maintenance

August 2, 2017

In the next 5 to 10 years, 100% predictable maintenance for rotating equipment will be the new norm. Developments in sensor technologies and artificial intelligence are accelerating the pace at which it becomes possible to accurately predict when and why equipment fails. These technologies enable smart condition monitoring services that currently act as information filters. Simon Jagers, founder of Semiotic Labs, says that soon, however, these will evolve into decision support systems, assisted decision mechanisms and fully automated processes.

Current developments in condition monitoring enable Condition Based Maintenance (CBM) regimes. CBMs address challenges at both the demand and supply side of the maintenance equation.

On one hand, asset owners can schedule maintenance before breakdowns occur or when performance desists. In doing so, they minimize both planned and unplanned downtime. On the other hand, original equipment manufacturers and maintenance service providers can offer uptime guarantees and “power by the hour” models based on remote online condition monitoring systems. Both benefit from insight into the condition of assets.

The new generation of condition monitoring solutions relies predominantly on sensors and smart algorithms. Sensors are constantly becoming cheaper, better and smaller due to innovations in the smart phone industry. The resulting influx of data provides fuel for the development of better algorithms that predict when and why equipment fails. Over the next couple of years, we’ll see these systems evolve from smart filters via decision support systems and assisted decision applications into fully automated production lines.

Over the next couple of years, we’ll see these systems evolve from smart filters via decision support systems and assisted decision applications into fully automated production lines.

How smart filters work

Smart filters enable maintenance professionals to focus on assets that require attention. These filters use sensor data and turn it into patterns of behavior. Using self-learning algorithms, these systems indicate which assets exhibit healthy behavior patterns and which don’t. Maintenance professionals use this information to focus on assets that are suspect instead of spending time on inspections of healthy assets.

Over time, smart filters evolve into decision support systems. Algorithms underpinning smart filters learn to recognize patterns and label them according to failure causes. Further analysis uncovers the time to live for these assets. With this information about when and why equipment fails, maintenance professionals can decide which actions to take to minimize downtime and operational risk. 

The future of human intervention

Converging systems lead to assisted decision applications. Suppose your decision support tool identifies a failing motor and the cause and time of the upcoming failure. If that application is integrated with your scheduling tool, is aware of demand forecasts and can verify spare part availability from your supplier’s system, the application will be able to schedule the optimal time for mitigating actions.

A human operator subsequently reviews and approves these suggested actions before they are initiated. With the ever-increasing quality of sensors and algorithms, human intervention will soon prove to be unnecessary. When that time comes, we’ll see the advancement of fully automated systems that operate production lines.

Investing in sensor technology

Filters, decision support systems and fully automated shop-floors – all of these rely on the availability of quality data. From a condition monitoring point of view, data from PLCs (Programmable Logic Controllers) and SCADA (Supervisory Control and Data Acquisition) systems or processes are almost always insufficient to deliver high accuracy and timely warnings.

To capitalize on the smart condition monitoring developments, it is essential to start investing in sensor technology for existing assets and purchase new equipment from vendors that equip assets with high quality sensing systems. It might be the best investment you’ll make this year.


Simon Jagers

  • Founder, Semiotic Labs
  • His company specializes in reducing unplanned downtime by combining sensors, machine learning algorithms and domain knowledge on an online platform that predicts when and why equipment fails.
  • One of the leading voices in the field of artificial intelligence-driven maintenance.