"Holy Grail of predictive maintenance"
Creating knowledge platform and use cases TinyML
Since September 1, 2022, André Gerver has been working at Techport as a Quartermaster. In fact, André will be working on setting up use cases in the field of Tiny Machine Learning. Together with companies and the educational institutions Hogeschool van Amsterdam, Hogeschool Inholland, Vrije Universiteit and Nova College, he will work on mapping the possibilities of Tiny Machine Learning for SMEs.
Tiny Machine Learning is a new development. Machine Learning (ML) and Artificial Intelligence (AI) is already used in large companies. The knowledge and experience with ML and AI is useful and needed for this new development. André: "What is so great about Tiny Machine Learning is that the possibilities are unprecedented. For example, it offers opportunities to reduce energy consumption of companies, such as manufacturing companies. It also offers opportunities to make smart maintenance more accessible and accessible to SMEs. Another area of application concerns safety and nuisance."
Tiny Machine Learning is still in its infancy
André joins Techport to work on bringing together SMEs, education, research institutions and technology. "We want to lend a helping hand by offering value to companies with Tiny Machine Learning. The data provided by this technology can provide energy, time and cost savings. It is a fairly new development that is still in its infancy. That is precisely why it is important for companies to take the first steps now, learn the process and develop it for the long term."
How exactly does Tiny Machine learning work?
You can think of this technology as mini computers all together, also called microcontrollers. These microcontrollers can be placed on machines and devices. On these controllers, in turn, all kinds of sensors can be pasted, such as sound and motion sensors. And with these, you can monitor when a machine's system deviates from normal behavior.
Holy Grail of predictive maintenance
Microcontrollers are also stuck on older machines such as a fan, motor or pump. In addition, controllers consume little energy which is good for the environment and wallet. As an operator, you can easily read if the device shows an abnormality. Also consider remote reading of wind farms and oil rigs. It is possible to detect damage to a rotor or blades. André: "This technology can also be called the Holy Grail of predictive maintenance."
Possibilities are unprecedented
"The beauty of this technology is that the possibilities are unprecedented": tells André. Especially in manufacturing companies. To keep up with future developments, it is important for SME companies to take the first steps with Tiny Machine Learning now. To be successful with this, you have to experiment.
André advises, "Go with it, start with a small project, start learning from it and expand. If you want to keep your competitive edge for 5 to 10 years from now, start now. With this technology you can produce cheaper, faster and more flexible, save energy and reduce your CO2 footprint. It's a technology you need to get started with. Consider also the rise of the Internet. There were companies that went way too fast and fell over. But the companies that started small kept developing and gaining knowledge and, as a result, they have become indispensable. It is a long-term development. It's not only technological, but this technology requires the right implementation. Make sure the people in your company learn to think data-driven. This is already happening at Tata Steel, among others."
Setting up cases and disseminating knowledge
A case has already been conducted at the company Montarent. André will soon visit a number of Techport companies to discuss cases. Based on a case, he will ask students from the Hogeschool van Amsterdam who have taken the Data Science minor from different disciplines to run a pilot. "We can help companies and think along in the developments at hand. This way we bring knowledge and experience together. This requires many specialists from business. So I am going to work on the one hand with the use cases and in addition bringing knowledge together in the network."
Responding to industry needs
Energy is one of Techport's themes. Of course, this is also an item of great concern. Tiny controllers can be used to map the energy consumption of companies and we can work on making companies more energy neutral. With Tiny Machine Learning, you can see where energy is leaking out. André: "Google has already embraced this years ago and made their data centers more energy neutral, they have lost over 30% in energy. We live in a time where so much is possible, many people still have no idea what is already possible."
Sparring over a case?
Do you have questions about Tiny Machine Learning or want to spar about a case? Feel free to contact André at andre.gerver@techport.nl. He would love to help you!