An Adaptive Optimization Algorithm Using Kriging Model And Its Application To The Optimal Design Of PMLSM

The global optimization methods for optimal design of electromagnetic devices are simulated annealing algorithm, genetic algorithm, tabu search algorithm, ant ant colony algorithm and so on. Although these algorithms can be used to obtain the approximately global optimal solution, for more complex inverse problems in the electromagnetic field analysis, they have some disadvantages such as large calculation size and long computation time. To solve the problem that analysis of inverse problem of electromagnetic field is over-reliance on computer resources, this paper proposed an adaptive global optimization algorithm instead of the traditional optimization algorithms, which have to analyze electromagnetic field in each iteration. The optimal design of a permanent magnet linear synchronous motor (PMLSM) was taken as an example to verify the algorithm. The detailed works are as follows:First of all, based on the analysis of Kriging interpolation technique, an adaptive optimization algorithm was proposed, in which Kriging model and genetic algorithm are combined, and the corresponding program was compiled. In the proposed algorithm, Kriging model is used to approximate the objective function and Genetic algorithm is to optimize this objective function. Through decreasing the size of design space and inserting new sampling points the objective function was approximated step by step, the operational efficiency was improved and simulation accuracy was achieved.Secondly, Latin Hypercube Sampling (LHS) technology was introduced in this paper, and several LHS design criteria was presented from the perspective of uniform sampling. An optimal LHS design based on Pareto multi-objective optimization was proposed, and coupled with the above adaptive optimization algorithm so that the number of finite element analysis was reduced and sufficiently accurate Kriging surface using sampling points as little as possible was obtained. The proposed adaptive optimization method has been validated using several analytic function tests. Numerical results show that the algorithm has the property of fast global searching.Finally, the magnetic characteristics of a 9-pole 10-slot PMLSM was calculated. The cogging force and the end force of the motor were analyzed and compared with the traditional 8-pole 12-slot motor. To reduce the end force of the 9-pole 10-slot PMLSM, the design parameters was chosen and the adaptive algorithm was applied. The end force is greatly reduced after optimization, and the optimizing results verify the effectiveness of the adaptive algorithm proposed in this paper.