Electromagnetic Analysis And Optimization Design Of Permanent Magnet Spherical Motor

With the continuous development of modern industrial technology,high-performance equipments such as robotic arms and intelligent flexible manufacturing systems that require multi-degree-of-freedom motion in space have been widely used.Those devices are usually combined with multiple single-degree-of-freedom motors by transmission gears,which resulting in increased size and reduced stiffness.For this reason,the permanent magnet spherical motor(PMSM)that can realize multi-degree-of-freedom motion has attracted widespread attention of scholars.Different from the traditional motor,the motor is still in the theoretical research stage due to the particularity of the mechanism.In this thesis,in-depth research on electromagnetic analysis,torque prediction modeling and optimization design of PMSM have been carried out.The specific work can be divided into the following two parts:1.Firstly,the basic structure of PMSM is introduced,and the initial design parameters of the motor structure are given.Then,the magnetic field of the stator,permanent magnet poles and the magnetic field in air gap are analyzed by magnetic equivalent circuit(MEC)method.The correctness of the scheme is confirmed by comparison experiments with finite element analysis(FEA)results.In addition,the Maxwell stress tensor method and the virtual work method are used to obtain the torque of the motor during the rotor have the rotation and yaw motion.The torque results obtained by the two methods are compared to verify the accuracy of the analysis modeling.2.In order to improve the performance of PMSM,in this thesis,BP neural network,BP neural network optimized by genetic algorithm(GA-BP)neural network and BP neural netwo rk optimized by particle swarm optimization algo rithm(PSO-BP)neural network are used to predict the torque of the motor.Firstly,the factorial design is used to obtain the sample space,and the neural network modeling of PMSM is established.By comparing the mean absolute error(MAE)and the mean absolute percentage error(MAPE)of the three methods,the conclusion that the PSO-BP neural network modeling accuracy is higher than the other two methods.Finally,optimization design of PMSM is carried out.Particle Swarm Optimization(PSO)that the inertia factor is dynamically adj usted is applied to improve the torque performance.Considering the torque performance and the power consumption of coils,the adaptive grid algo rithm multi-objective particle swarm optimization(AGA-MOPSO)algo rithm is used to search the best structure parameter the motor based on the PSO-BP neural network modeling.Then the optimal solutions are obtained.The optimized PMSM and the initial PMSM are analyzed using the F EA method.