Augmenting Human Expertise with Machine Learning
By Adam Poole, Product Design Lead
Technology is only valuable when it makes our lives easier, yet that simple truth can be lost in the promotion of the latest technical wizardry. In a work environment, the last thing people need are solutions that leave them with the headache of extra systems to manage. In the current wave of innovation and enthusiasm surrounding smart factories and Industry 4.0, it’s more important than ever to find solutions that support people to build stronger businesses, rather than making their working lives more complex.
Free-flowing data is fast becoming the chief currency in this smart environment. Even so, data is pointless unless it empowers people to make better business decisions. With that in mind, it’s critical that the new generation of industrial systems deliver a user experience that makes people’s working lives better.
Help expertise go further
Get it right and these systems can play a big role in helping to scale scarce human expertise. They can give people more time to make decisions while much of the routine data analysis is carried out automatically behind the scenes.
To truly harness the power of the smart factory, we need to connect humans and machines in the right way. It’s not about artificial intelligence (AI) displacing years of human expertise, it’s about human and machines becoming the ‘smart system’: collaborating and complementing one another.
Set priorities to optimize resources
Senseye PdM is a great example. Our cloud-based Predictive Maintenance system can monitor thousands of connected assets, automatically detecting abnormal behavior and patterns that match the known failure modes of individual machines. The aim is to spot maintenance issues much earlier, enabling users to fix problems before they can disrupt operations.
Busy maintenance teams typically have only a few minutes at the start of each shift to identify which among their thousands of assets most need their attention. Uniquely, Senseye PdM presents the information back to users in the form of a prioritized list, sorted by the Attention Index. This enables users to see immediately where they should be directing their resources.
Moving Beyond Asset Health
In common with most condition monitoring systems, Senseye PdM previously helped users to set priorities by indicating an Asset Health score for every asset. This is now being replaced by the Attention Index, which provides a single way of sorting assets. This new approach is part of a next generation of analytics being deployed using a new range of in-house algorithms.
Attention Index takes into account all the ways that Senseye PdM can detect or predict issues: anomaly detection, trends, thresholds and prognostics. The software, at every level, guides the user towards determining the underlying problem, rectifying it and capturing this in the system. This feeds machine learning algorithms so that failures can begin to be spotted early enough to take action.
Asset Health, as a concept, gives the impression that the system understands everything about an asset. This is never the case. A condition monitoring system can only base this score on the sensors and condition indicators set up for each asset (i.e. vibration level). If this leads to a score of 0, what does that really mean? It’s misleading to users.
Attention Index supports decision making by presenting a prioritized list of assets. It is the expertise of the user that determines the next step. By keeping the design of the software simple and straightforward, valuable time is saved. By harnessing the power of machine learning, informed decisions are made, and by capturing the actual work done, all parts of the system continue to learn.
By helping users to identify where they should be focusing their maintenance resources, Senseye PdM delivers impressive results, with a typical 85% increase in maintenance accuracy accompanied by reductions of 50% in downtime and a boost of 55% in productivity.