Fast analysis for large-scale electromagnetic scattering problems by a hybrid approach. Issue 8 (24th May 2017)

• Record Type: Journal Article
• Authors: Li, Chenlu; Sun, Yufa; Wang, Guohua; Wang, Zhonggen
• Is Part Of: Journal of electromagnetic waves and applications. Volume 31:Issue 8(2017)
• Page Start: 808
• Page End: 819
• Publication Date: 2017-05-24
• Subjects: Electromagnetic scattering -- fast dipole method (FDM) -- adaptive cross approximation (ACA) algorithm -- multilevel characteristic basis function method (MLCBFM); Electromagnetic waves -- Periodicals; Electromagnetic devices -- Periodicals; 621.3
• DOI: 10.1080/09205071.2017.1317608
• Languages: English
• ISSNs: 0920-5071

Abstract

Abstract: In this paper, a fast and accurate hybrid approach based on the multilevel characteristic basis function method (MLCBFM) is presented to analyze the electrically large scattering problems. First, a modified fast dipole method (MFDM) is proposed to improve the accuracy and efficiency of the conventional fast dipole method (FDM). Then, the MFDM is combined with the adaptive cross approximation (ACA) algorithm and the equivalent dipole method (EDM) to enhance the efficiency of the MLCBFM. In the proposed hybrid method, the use of MFDM makes the construction of secondary characteristic basis functions and reduced matrix of the MLCBFM more accurate and efficient. The impedance matrices representing the interactions of the middle-field pairs are efficiently compressed by using the ACA in order to enable the hybrid method to further reduce the time and the memory significantly when the criterion for the far-field pairs becomes stricter to obtain relatively high accurate solutions. The numerical results of one hundred PEC cubes have demonstrated that the MFDM is capable of effectively reducing the CPU time and the relative errors. Furthermore, another results have shown that the proposed hybrid method has reduced the CPU time and the memory requirement by 27 and 32%, respectively, when the same criterion for the far-field pairs are chosen.