Today, batteries are usedeverywhere, from small electronics to cars. The development and manufacture of power supplies takes a lot of time and money, and most of all resources require their testing - before selling, you need to identify their service life and distribute them into classes to meet the needs of different customers. Until now, the service life has been determined by multiple charge and discharge cycles, but with increasing battery capacity, this takes longer. As always, artificial intelligence came to the rescue - they taught him to make accurate predictions based on just five cycles.
The development of artificial intelligence engagedResearchers at the Massachusetts Institute of Technology and the Toyota Research Center. Instead of numerous cycles of replenishment and consumption of battery power, they proposed to spend only five cycles, and to give this data to the processing computer algorithm. It uses hundreds of millions of data points to detect service life, and draws attention to voltage drops and other factors that signal a complete discharge. According to the researchers, the prediction accuracy reaches 95%.
According to Toyota researcher PatrickHerring, in this way, machine learning significantly speeds up the development of new batteries and significantly reduces the costs of both research and production. Moreover, the researchers suggest that the technology can help optimize the charging process so that it replenishes as quickly as possible - in about 10 minutes.
It is noteworthy that the Massachusetts technologicalThe institute often conducts research in the field of batteries. For example, in September 2018, she developed a power source that absorbs carbon dioxide.
You probably have something to say about the newworks of scientists - you can share your opinion in the comments. Do not forget to join our Telegram-chat, where there are always lively discussions on the topic of science and technology!