Research On Neural Network Right Inverse Decoupling Control And Speed-sensorless Technology Of Bearingless Permanent Magnet Synchronous Motor

A bearingless permanent magnet synchronous motor(BPMSM)is a new type of motor which integrates the technology of the magnetic bearing and the technology of the permanent magnet synchronous motor(PMSM).It not only has the advantages of no frictional wear,no lubrication,low noise and long service life of the magnetic bearing,but also has the advantages of high power density and high efficiency of the PMSM.Therefore,it has a broad application prospects in the precision instrument processing,aerospace,flywheel energy storage and other fields.This dissertation takes the BPMSM as the research object,and its working principle,mathematical models,basic structure,decoupling control method,speed-sensorless technology and digital control system are studied.The main works and achievements are as follows:1.The working principle of the BPMSM is analyzed.On this basis,the mathematical models of the radial suspension force and the electromagnetic torque are derived.Finally,two basic structures of the BPMSM,two-degree-of-freedom(2-DOF)BPMSM and five-degree-of-freedom(5-DOF)BPMSM,are introduced in detail.2.Aiming at the nonlinear and strongly coupled characteristics of the BPMSM,a novel active disturbance rejection control(ADRC)method based on the neural network right inverse by combining the neural network right inverse thought with the ADRC theory is proposed in the dissertation,and this method is applied to realize the decoupling control of the 2-DOF BPMSM and the 5-DOF BPMSM respectively.On the basis of discriminating the right reversibility of the2-DOF BPMSM and the 5-DOF BPMSM,the neural network is used to construct their neural network right inverse system respectively.And by series connection with the obtained neural network right inverse system,the 2-DOF BPMSM and the 5-DOF BPMSM are respectively decoupled into several pseudo-linear subsystems.Then,considering the characteristics of the pseudo-linear subsystems,the ADRC controllers are designed to synthesize the pseudo-linear subsystems,which ensure the stability of the 2-DOF BPMSM and the 5-DOF BPMSM.Finally,the effectiveness of the proposed control method is verified by the simulation.3.In order to solve the problems of higher cost,larger volume and lower dependability of the BPMSM caused by the mechanical speed sensor,a speed self-detection method based on the fuzzy neural network left inverse is proposed.Based on the concept of “internal sensor”,the speed subsystem of the BPMSM is established.Then,on the basis of discriminating the left reversibility of the speed subsystem,its left inverse system is constructed by the fuzzy neuralnetwork,and by connecting the constructed left inverse system in series after the speed subsystem,the speed self-detection of the BPMSM is realized.Finally,the feasibility of the proposed self-detection method is verified by the simulation.4.The hardware and software of the BPMSM digital control system are designed.Then,the BPMSM digital control experiment platform is constructed,and the basic experiment researches of the BPMSM are carried out.Finally,the experimental scheme is proposed to realize the speed-sensorless operation of the BPMSM based on the fuzzy neural network left inverse system on the constructed digital control experimental platform.