Putting artificial intelligence to work "Too much restraint in the use of AI in maintenance"
The use of AI, so artificial intelligence, in maintenance is a bit of a hot potato. Opinions concerning the technology range from „future junk“ to „already fully operational“. The magazine Instandhaltung spoke with the two ROBUR experts who are very familiar with the issue: Kyriakos Kosmidis and André Panné.
Kyriakos Kosmidis and André Panné know their way around artificial intelligence. We asked them for their view on the use of AI in maintenance.
Kosmidis has worked in the sector for 15 years and is in charge of the wind energy business at ROBUR, which operates as a buy & build investor to integrate small to medium-sized enterprises. He is responsible for driving internationalisation and digitalisation – including AI solutions – to provide the company with the necessary basis for innovative maintenance.
Before founding his services company Tradum, Panné worked for Apple and BenQ Mobile (Siemens Mobile), for Andersen Consulting (Accenture) and Proxicom, as well as for VW-gedas. He has been involved in the ITC industry for over 25 years, focusing on business development, innovation and digitalisation. At present, Managing Director of GIS, part of ROBUR, operating in the industrial segments of wind, water, energy, industrials and the process industry.
INSTANDHALTUNG: What is the current status of AI technology in your view?
Kyriakos Kosmidis: „Undoubtedly the necessary technology is already advanced, but is still miles away from what will appear perfectly normal in 10 to 15 years. The surprising developments that will fuel our phantasies in the creation of even more ideas, applications and business models will certainly dash our current illusion of progress.“
André Panné: „We need to take a nuanced approach assessing the momentary status of AI. If we look as the use of machine learning, for instance in image recognition, some systems are already able to produce better results than humans. This applies to surface analysis of materials and other areas. But we are still a long way from developing genuine artificial intelligence. Most likely we would not notice or even be able to tell whether a machine is showing ‚real‘ intelligence.“
INSTANDHALTUNG: Do useful AI tools for maintenance technicians already exist?
Kosmidis: „Again it is true that universities, start-ups and company departments experimenting with this topic have already achieved a lot. But what we are definitely lacking is that maintenance companies and data scientists put their heads together and combine their know-how, creating inspiration to bring the potential, services and business models to fruition.“
Panné: „Without a doubt! Especially the image recognition I mentioned earlier has a lot of potential, much of which could be put to good use. Many start-ups and SMEs are already working in this field. GIS is one of them.“
INSTANDHALTUNG: For which use cases would these tools be suitable?
Kosmidis: „In the wind energy sector, they would mainly be issues that deliver better forecasts to reduce loss of earnings, improve the predictability of maintenance and repairs, provide more accurate estimates of cost blocks and optimise organisation in the supply chain. So, basically topics that help to cut costs.“
Panné: „There is another area besides cost cutting, namely error prevention. People become susceptible to making mistakes due to fatigue and loss of concentration, especially when dealing with repetitive tasks. That doesn‘t happen to machines controlled by AI.“
INSTANDHALTUNG: Why are we failing to use AI in maintenance?
Panné: „Failing would be the wrong word. A better way of putting it would be too much restraint. German engineering is legendary, so it is incredibly hard for us to cope with the idea that we will not be able to predict the precise outcome when using a new technology. But AI and many of the trending methods – which tend to be related more to agile and design thinking – e.g. MVP (Minimum Viable Product) – do not possess this product maturity. We need to learn how to handle imponderables and, like the original industrialists in the past century, show greater willingness to roll the dice.“
INSTANDHALTUNG: Which AI tools do you or your partners already use for maintenance?
Kosmidis: „One area that has seen a lot of use in the wind energy sector is the combination of visual inspections, for instance of rotor blades, and AI-assisted damage recognition and classification. But we are not tapping into the full potential here, either, as the software companies working on shaft applications do not collaborate sufficiently with specialists in the development of solutions that deliver more than just the proud claim that AI can tell the difference between damage to the rotor blade and dirt. What else could we achieve – way beyond this simple statement – if we were to harness the principles of swarm intelligence in the development of optimised solutions for our customers.“
INSTANDHALTUNG: Which pain points prompted you to initiate AI projects?
Kosmidis: „Sadly, the objectives for the use of AI, as well as the requirements in the wind energy sector, have not been defined with the necessary precision. It is unfortunate to note that the industry is yet to grasp the importance of sharing data with all stakeholders in such a way that rapid and consistently cost-optimised expansion of this crucial segment is not unnecessarily put on hold, thus impeding the introduction of streamlined structures across the entire value chain.“
Panné: „One reason for using AI-based predictive maintenance and other solutions is the considerable dispersion of the systems across larger distances – as is the case in wind energy. AI helps you to plan more purposeful service and maintenance assignments, for instance if the vibration patterns of structural elements change, depending on the wind strength, temperature or other factors. AI can be used to perform better analyses that any purely structure-based methods.
INSTANDHALTUNG: Given the current hype, companies are often offered dubious benefits of AI projects. How do you cherry-pick the best solutions?
Kosmidis: „Testimonials and contact with customers who may have benefited from the advertised AI are probably the only ways.“
Panné: “Eine immer wiederkehrende Herausforderung der IT-Branche ist die inflationäre Verwendung von Buzz-Words. Auch KI, Industrie 4.0 und Machine Learning fallen darunter. Wenn sie heute über eine Messe gehen, würden sie an jedem Stand derlei Schlagworte finden. Und in den meisten Fällen zu Produkten, die letztes Jahr noch mit anderen Schlagworten verkauft wurden.”
Panné: „One of the recurring challenges within the IT sector is the inflationary use of buzzwords. This applies to AI, Industry 4.0 and machine learning as well. You will find this kind of hollow phrase at every stand when you attend a trade show. And in most cases they will describe products that were sold using a different set of buzzwords just last year.
It is therefore advisable to question just how innovative providers actually are and to put them to the test. Even the ones that ‚just‘ use these terms as buzzwords will often be offering very good – albeit normal – solutions. You need to try and keep these things apart.“
Get fascinating insights into the world of ROBUR and its partner companies and learn regularly about the highlights of the group - conveniently by email.