HTS–SLIM design based on Bayesian multi–level, multi–objective optimization and Gaussian process models

Abstract

The performance of the Linear Induction Motor (LIM) is based on the linear force in the moving direction. However, due to the length limitation of LIMs, the end–effects as the barrier force decrease their performance. On the other hand, the primary width is lower than the secondary that causes the non–homolographic distribution of the air gap flux density. These effects result in the reduction of output thrust. In recent decades, the windings are made of High–Temperature Superconducting (HTS) tapes to reduce their power losses and total weight. However, due to their magnetic losses, the HTS–LIMs generally are low efficiency and need an optimal design. In this paper, an optimal design of an HTS Single–sided LIM (SLIM) is conducted by analyzing the proposed equivalent circuit parameters. Also, the output variables of the HTS–SLIM are calculated using the proposed analysis, which is different from the conventional SLIMs. The bayesian optimization algorithm (BOA) is used to optimize the objective functions, which consists of the efficiency, power factor, braking force due to end effects, and AC losses. Moreover, various models of the Gaussian Process (GP) are employed for modeling the objective functions at each step of the BOA process. Since the problem construction is multi–objective, the optimization algorithm is modified as the Multi–Objective BOA (MO–BOA) to optimize the mentioned objectives simultaneously. The problem is also considered as a multi–level problem owing to the various levels based on the sensitive analysis of HTS–SLIM parameters. For increasing the results accuracy, different effects of SLIM such as end-, skin-, and edge- effects are studied in the analytical model. The results of different models of GP and other methods are compared to show the robustness of the proposed method. Also, a 3D model of the motor is built and analyzed with the Finite Element Method (FEM) to verify the results of the analytical approach. The results for both FEM and analytical method show that the proposed structure has high performance and could be applied in practical applications.