Multi-objective Optimization Design Of Double-sided Linear Flux Switching Permanent Magnet Motor

The Double-sided Linear Flux Switching Permanent Magnet(DLFSPM)has been widely attracting attention in the fields of rail transportation,precision machining and electromagnetic ejection because of its high thrust density,high operating efficiency,simple structure and low manufacturing and maintenance costs.However,due to its unique tooth-pole structure,this motor had problems such as high thrust fluctuation,high positioning force,and noise of vibration.With the transformation and upgrading of China’s industrial manufacturing,the performance requirements of linear motors have become more and more strictly,how to output large thrust and low thrust fluctuations and other quality performance has become a research topic of linear motors.Therefore,it is important to develop DLFSPM multi-objective optimization and design.This thesis firstly introduces the basic topology and operation principle of the DLFSPM.According to the theory of divided slot and the power equation and mathematical model of DLFSPM,the initial structure size of this motor is designed.And the 2D finite element model of the motor is established to simulate and analyze the basic electromagnetic performance such as no-load magnet,electromagnetic thrust,detent force and mover’s loss of DLFSPM.Second,the target performance to be optimized for the DLFSPM motor is discussed to determine the target performance and structural design parameters to be optimized.The correlation between the optimization objectives is discussed by combining the design of experiment method with the correlation coefficient calculation,and the key optimization objectives are confirmed.Comprehensive sensitivity analysis,correlation analysis and cross-factor variance analysis are used to study the combined influence of unidimensional and multidimensional design variables on the optimization objectives,and DLFSPM’s structure design parameters are classified into three influence levels according to their influence degree.These works prepare for the multiobjective optimal design of DLFSPM.Finally,this thesis uses the Response Surface Method to complete the fitting work between the four low-influence level parameters and the key optimization objectives.The regression models between the four high-impact parameters and three mediumimpact parameters with the key optimization model are obtained by using the Random Forest regression method,and the generalization ability of the regression models is improved by using the hyperparameter optimization method.The accuracy and robustness of the optimization models were verified by finite element analysis.Based on the above classification results of parameters and the three optimization models,a multi-objective genetic algorithm(NSGA-II)is used to achieve a level-by-level multiobjective optimal design for the target performance of the DLFSPM.The optimal structural dimensions of the DLFSPM are finally obtained by selecting the appropriate combinations of design parameters from the Pareto fronts obtained at each level of optimization design.With the comparative analysis of DLFSPM motor performance such as electromagnetic thrust,detent force,harmonics,and mover’s loss before and after optimization,the feasibility and effectiveness of the multi-objective optimized design method are verified.