Study Of Dynamics Decoupling Control And Electrifying Strategy Of Permanent Magnet Spherical Motor

One kind of Permanent Magnet Spherical Motor (PMSM) is proposed in this paper, and the mechanical structure of each part is discussed first. Dynamical modeling of PMSM is made, and model errors, including structured uncertainties and unstructured uncertainties, are analyzed. According to the fact that there exist lots of inter-axis nonlinear couplings in its dynamic model, dynamic decoupling algorithms based on fuzzy controllers is proposed. In order to eliminate the effects of inter-axis couplings in the dynamic model, the dynamic decoupling algorithm with neural network identifier and ANFIS is proposed then. The two proposed algorithms, in which Computed Torque Method (CTM) structure is employed, can eliminate the couplings and improve the static and dynamic performances of the control system. Simulations verify the effectiveness of the proposed algorithms.A spherical planning based electrifying strategy of PMSM is proposed. First, the Finite Elements Method (FEM) is employed to obtain the static torque model of PMSM. Then the region with stator coils shaded on Stator Spherical Surface is spherically planned and divided into four classes of sub-regions, which can be analytically expressed. All stator coils are labeled according to similar triangle principle. Finally, static torque model is employed and linear equations are solved to obtain the electrifying stator coils and the electrifying currents. PMSM can realize complex trajectory tracking operations under different algorithms with the proposed electrifying strategy applied. The simulations are made, and the effectiveness of the proposed electrifying strategy is verified.With regards to the inverse kinematics problem of PMSM, analytically solving methods are complicated relatively. Solving methods of inverse kinematics based on feed forward neural network and that based on Advanced Ant Colony Algorithm (AACA) is proposed separately. Through the study of training data, neural network can approach the inverse model. According to the learn effects of training data with L-M algorithm applied, the optimum structure of the neural network for the inverse kinematics problem is determined. The AACA is then proposed to solve the inverse kinematics problem. the advantages of AACA are verified, and the parameters'configurations of AACA are discussed according to the solving effects.According to the mechanical structure of PMSM, each main part of hardware circuit are discussed, and the hardware experiment platform based on TMS320F2812 DSP is built. Through the detection of rotor's position and bus current, the hardware platform can realize two closed-loops speed regulations with rapid start and stable operation.