Rapid Design of Structural Parameters of Axial-flux Micro-motor Based on Deep Learning

To meet the demand of high-performance micro-motors with size and weight restrictions for micro aerial vehicles, a prediction model for rapid design of structural parameters of Axial-flux Brushless DC Micro-motor (ABDM) was established based on the deep learning BP algorithm. The ABDM is a single-stator dual-rotor configuration, with a diameter of 20 mm for the permanent magnet (PM) rotor. The input parameters of the prediction model include the number of winding turns, embrace, thickness and inner radius of the PM. The output parameters of the prediction model are the electromagnetic torque and efficiency. The 240 groups of data, generated through finite element simulation, were used to train and test the prediction model. Considering the limitations of size and weight, an ABDM prototype was fabricated based on a set of the optimal structural parameters from the prediction model. A torque testing platform for the ABDM prototype was constructed, employing a reaction torque sensor and a dual-blade propeller on the output shaft. The test results show that the average output torque and efficiency of the micro-motor itself both meet the design goals, verifying the accuracy and feasibility of the rapid design of micro-motor structural parameters based on the BP algorithm.