DocumentCode :
69321
Title :
Design Optimizations of Electromagnetic Devices Using Sensitivity Analysis and Tabu Algorithm
Author :
Lin Yang ; Ho, S.L. ; Fu, W.N.
Author_Institution :
Dept. of Electr. Eng., Hong Kong Polytech. Univ., Hong Kong, China
Volume :
50
Issue :
11
fYear :
2014
fDate :
Nov. 2014
Firstpage :
1
Lastpage :
4
Abstract :
This paper presents a novel implementation of the sensitivity analysis with an improved Tabu search algorithm for the optimal design methodology of electromagnetic devices using finite-element method (FEM). A direct sensitivity formulation is proposed to calculate the sensitivity versus design variables by postprocessing the FEM result only once, and hence, it is expected to be very well received by the industrial users. The sensitivity analysis is also used as an indication of step vector to enhance the search performance during the optimization process. Further improvements, including a transition criterion for different current states and the introduction of a Tabu list with binary space partitioning tree, have been introduced to help obtaining the global optimum rapidly, as these are particularly useful for time-consuming optimal design problems. Optimal design examples of electromagnetic devices are reported to positively demonstrate the feasibilities and advantages of the proposed algorithm in practical engineering.
Keywords :
electromagnetic devices; finite element analysis; optimisation; search problems; sensitivity analysis; trees (mathematics); FEM; Tabu list; binary space partitioning tree; current states; design variables; direct sensitivity formulation; electromagnetic device design optimization process; finite-element method; global optimum; improved Tabu search algorithm; sensitivity analysis; step vector indication; transition criterion; Algorithm design and analysis; Finite element analysis; Linear programming; Optimization; Sensitivity analysis; Vectors; Optimal design; Tabu search (TS) algorithm; sensitivity analysis;
fLanguage :
English
Journal_Title :
Magnetics, IEEE Transactions on
Publisher :
ieee
ISSN :
0018-9464
Type :
jour
DOI :
10.1109/TMAG.2014.2322625
Filename :
6971439
Link To Document :
بازگشت