Fast-Sensing System of Permanent Magnet Magnetization Using Matrix-Arranged Hall Sensors Combined With Deep Neural Network

Permanent magnets (PMs) are essential for designing high-performance electromagnetic machines, such as PM synchronous motors (PMSMs), in the industrial field. Because the performance of the PMSMs is substantially dependent on the magnetization distribution of the PM, a fast estimation system for the PM magnetization intensity is indispensable after the magnetization operation in the manufacturing process of PMSMs. In this letter, a fast-sensing system combining a special circuit board to which matrix-arranged Hall sensors are applied, and a deep neural network is proposed and practically implemented. Consequently, compared with the conventional approach, the proposed sensing system can estimate PM magnetization with high-speed and high accuracy.