Fast Estimation System of Permanent Magnet Magnetization Using 2D-Arrayed Hall Sensors Combined with Deep Neural Network

Loading permanent magnets (PMs) is essential for realization of high-output of electromagnetic machines such as the permanent magnet synchronous motors (PMSMs). In the most of the PMSMs, the torque performance is dependent on the magnetization distribution of the PM. Therefore, on the manufacturing process of PMSMs, a fast estimation instrumentation is substantially required. In this paper, to estimate the PM magnetization quickly and accurately, a special instrumentation comprising the 2D-arrayed Hall sensors and deep neural network is developed.