Spanish technology to try to make you forget the mobile charger
When we charge the mobile or start an electric car we assume that they will work. That they will have the necessary autonomy to allow us to surf the internet or move from one place to another. We are unconcerned about what happens inside these products. We forget how electromagnetics and technology keep batteries from overheating or phones burning. Behind this apparent magic there are companies like the Spanish Frenetic, which has revolutionized the manufacture of both chargers and batteries with the help of an artificial intelligence system. The goal: that these take up less space. Make them lighter. Ultimately, they go completely unnoticed.
Frenetic was born in 2018 with the purpose of changing the design of transformers, transistors and plates – key elements when managing the electric current that flows through a charger or a battery – to reduce their size. As its founder, Chema Molina, explains, applying the technique of machine learning, that is, getting machines to improve with experience, in these components allows their manufacture to be more efficient. “The transformer represents 50% of the electronics in a charger. It is what allows it to be more manageable, smaller and more reliable. Determine its characteristics. We have trained a system that optimizes its design and even helps to predict its useful life and operation ”, he says.
Programming the algorithm that reduces the size of electromagnetic components has required years and years of research and building a database with enough information to allow these parts to be optimized. When Molina created the company, it was clear to him that he needed a laboratory to create this data. In this center he began to train artificial intelligence through trial and error. This process, repeated almost ad nauseam, allowed the machines to learn and arrive at the solutions that are currently being sold to Airbus, HP, Siemens, Indra and other large global companies. It has managed to forge a large repository of information on electromagnetic behavior.
It is not by any chance that he has used the technique of trial and error. As suggested by Macià Capó, a project engineer at the CIPCEA-UPC Power Electronics Research Center, he could not have reduced the weight and size of these elements if machine learning had been dedicated to solving Maxwell’s equations, which determine the behavior of chargers and batteries. “The artificial intelligence system would have required greater computational capacity, which would have triggered consumption and, therefore, its weight and size. With this type of predictive model, such as Frenetic’s, it is easier to establish the optimal point of a transformer or a transistor ”, he says.
If Molina’s research continues at current rates — and technology does its job hand in hand with data science — the weight and size of these components can still be greatly reduced. A very attractive effect for the aerospace and automotive industries. Both sectors are looking for the perfect transformer. The one that is tiny, does not overheat and does not lose electricity. “If you have the appropriate data, which is one of the most complicated parts, and sensors, you can optimize the performance and size of the transformers,” says Macià.
Safer space missions
Another possibility that Molina is exploring is the fit of the artificial intelligence system in space missions – some international agency has already knocked on his door to inquire about his project. Take the example that we send a probe to Mars. During the journey and also upon arrival, there is no type of network with constant power supply values, as is the case on Earth. Solar panels and batteries would power the probe, but based on variable parameters. And that’s where artificial intelligence would establish how much power the artifact requires from the panel, from the battery, or from a charger. “Currently we only use the technology in a specific component. The next step will be that machine learning design the electromagnetic systems completely ”, he predicts.
Frenetic’s birth arose almost by chance. Molina received his doctorate in electronics for electric vehicle chargers and power supplies. He knew first-hand the problem he wanted to solve. However, it was not until he met a roommate, a student of artificial intelligence, that he was able to provide a solution, in its theoretical moment, to the problem of weight and size in these internal parts. “In our talks I learned a lot about your field, about automation. And I wanted to apply this knowledge in my field in some way. This is how the company arose ”, ditch.
According to Molina, more and more companies are interested in his solution, which currently employs 35 people. Although he does not want to publicize them, in order to maintain the confidentiality of the contracts, he assures that they are very cutting-edge companies, first-rate technology. “In an artificial intelligence system, a wrong data kills the whole process. Innovative development invalidates you. I hope that our system is becoming more precise and formal. That captures the attention and solves the problems of more organizations ”, he concludes.