Using AI to optimize magnet and improve performance of motor magnet
Research and development aimed at improving the performance of motor magnets is currently gaining momentum. This is because motors used in electric vehicles (EVs), which are increasingly in demand, are also urgently required to improve performance under the trend of decarbonization on a global scale. Japan intends to use artificial intelligence (AI) to effectively identify the best composition of magnets in order to maintain Japan’s industrial competitiveness and compete with countries such as China.
“We found that by using AI, the performance of magnets can be improved with a small number of experiments,” said Yasuke Sasaki, a principal researcher at the Institute for Materials Science and Technology in Japan.
The agency has developed a method to improve the performance of “neodymium magnets” used in EV motors, etc., using machine learning, one of AI. This is part of the Japanese government’s research project on magnet materials.
Neodymium magnets, known as the strongest permanent magnets, were developed in 1982 by Masato Sagawa, a consultant at Daido Special Steel Co., Ltd. Currently, it is widely used in various equipment such as electric motors and wind power generation motors, electronic components, and magnetic resonance imaging (MRI).
High-performance magnets used in EV motors
High-performance magnets used in EV motors need to withstand high temperatures of nearly 200°C
Drive motors for EVs need to operate at temperatures close to 200°C. In general, the performance of magnets such as magnetic force decreases at high temperatures. The research team focused on the “thermal processing” method, which is a “thermal processing” method that can produce magnets that do not easily degrade in performance at high temperatures even if the same materials and proportions are used, although the manufacturing process is complicated. The pressure and temperature adjusted during the manufacturing process are optimized using AI.
For a magnet with a certain composition, even if only 6 values are adjusted, “if one does not fall to the ground, all corresponding to various possibilities, there will be 66 million manufacturing conditions” (Taisuke Sasaki). In this regard, magnet experts first streamlined 18 conditions and trial-produced magnets. This data is then used by machine learning experts to train the AI to predict performance based on manufacturing conditions.
In addition, with regard to the temperature and pressure of the manufacturing process predicted by the AI after its performance was improved, the research team repeated trial production under about 40 conditions and let the AI learn the results, thereby improving the prediction accuracy of the AI.
After only dozens of trial productions, the value indicating heat resistance has increased by about 1.5 times compared to before machine learning. Magnets comparable to existing magnets with optimal values can be manufactured with the same composition.
With higher performance, magnets can be made smaller, allowing for more efficient use of resources. Different applications require different magnet properties. Using AI is expected to develop a magnet manufacturing method that satisfies the required properties such as high strength and rust resistance. In order to promote the practical application of the new method, the Japan Material Research Institute plans to cooperate with companies in the future to investigate in detail whether it can cope with different characteristics.
Neodymium magnets are composed of iron, rare earth element neodymium, and dysprosium to improve heat resistance. Many of these rare earth elements are dependent on Chinese supplies. The R&D will also help reduce dependence on Chinese imports.
Toyota developed a “neodymium-saving magnet” with reduced neodymium content in 2018 with funding from the New Energy and Industrial Technology Development Organization (NEDO) project. Using AI to optimize magnet and improve performance of motor magnet
“Neodymium-saving magnets” reduce the content of neodymium by 20 to 50% by mixing rare earth lanthanum and cerium that exist in large quantities on the earth, but the heat resistance is basically the same as that of conventional neodymium magnets. Efforts are currently underway to further improve performance.
In this project, Denso Corporation is developing magnet materials that do not use rare earths such as neodymium at all. That’s where the company got its inspiration from the magnets contained in iron meteorites, which are made of an iron-nickel alloy. When nitrogen is blown into a substance in which iron and nickel are randomly arranged and then removed, the iron and nickel are aligned at the atomic level to form a magnetic material.
For high-performance magnets that hold the key to decarburization, fierce competition has been launched internationally. Koyo Ozaki, director of the Magnetic Powder Metallurgy Research Center of the Institute of Advanced Industrial Technology, which leads the NEDO project, analyzed: “China not only has advantages in the output and price of magnets, but also has significantly improved performance and research levels.” Using AI to optimize magnet and improve performance of motor magnet
Other countries, too, are using AI to develop higher-performance magnets. It is crucial to achieve practicality as soon as possible.
Square Mini Smco Rare Earth Permanent Magnets
Sintered SmCo Segment Rare Earth Permanent Magnets
Customized Permanent ARC Motor Rare Earth Magnets
Custom Small Permanent Plated Rare Earth Magnets
Rare Earth Disk Sm2Co17 Samarium Cobalt 5G Circulator Permanent Magnets
Permanent Rare Earth Samarium Cobalt Rectangular Magnets 50x10x10mm
Self Adhesive Powerful Disc Permanent Rare Earth Magnets w/ Epoxy Plating
Laminated Permanent Segment & Block SmCo Rare Earth Magnets
Bullet Shape Permanent Rare Earth SmCo High Energy Magnets
Wholesale Permanent Rare Earth Smco Plate Magnets
Permanent Rare Earth Magnets SmCo Bars 50x12x3mm
Rare Earth Smco Permanent Rectangular Magnets for Motor
