Internal Modular Control Of Magnetic Suspension Spherical Induction Motor Based On Fuzzy Neural Network Inverse System

This thesis conducted a research of the magnetic levitation spherical induction motor in case of improving on static and dynamic performance such as transmission efficiency, precision, retainability, quick response of motor. This kind of motor has many excellent characters like small and compact structure, high operating precision and six spatial degrees of freedom. These characters can just meet the needs of high efficiency and high precision of today’s industry robots and multi-degrees of freedom equipments. Thus, the research of the magnetic levitation spherical induction motor owns great significance no matter in theoretical exploration or in practical application.In fact, magnetic levitation spherical induction motors own very complex systems. Thus, in case of solving the control problem of the motor, this thesis used FNN (Fuzzy Neural Network) theory to do the research of this motor. Decoupling the motor’s six stator currents and three angles and three displacements into six independent fake linear systems and using IMC (internal model control) to complete the control of six output variables of the motor system.So far, most decoupling and control methods concerning neural network or fuzzy neural network aim at unidirectional motor or other unidirectional industrial system, which seldom concerns multi-degrees of freedom systems. Thus, this thesis tries using this modern control theory and method to study the magnetic levitation spherical induction motor, which has received quite good results.Following is the research contents of this thesis.Made a deep study of the working principle of magnetic levitation spherical induction motor, deduced the electromagnetic torque expression from the voltage equations and magnetic linkage equations and established the electromagnetic levitation force mathematical model of the motor by air gap magnetic energy and electromechanical energy conversion relations.Established the inverse system of motor’s original system according to FNN, which skillfully simplified the complex multivariable coupling relationship of the motor. Established the ANFIS model and made the motor’s pseudo linear system. Controlled the motor by using the internal model controller, designed the control plan and determined the control structure based on the fuzzy neural network inverse system.Tested the control model of the magnetic levitation spherical induction motor using the tool of MATLAB/Simulink and adjusted the parameter of internal model controller until having got the most stable simulation result. Established a test bed of motor and verified the previous theoretical analysis.The simulation results of the rotor’s characteristics of displacement and speed showed that the FNN inverse system of motor’s original system and IMC method had very good dynamic response and robustness. The test based on the control structure this thesis had proposed showed very quick dynamic response and strong robustness.