Ensemble of differential evolution algorithms for electromagnetic target recognition problem

• Published in: IET radar, sonar & navigation sonar & navigation, 2013-08, Vol.7 (7), p.780-788
• Main Authors: Secmen, Mustafa, Tasgetiren, Mehmet Fatih
• Format: Article
• Language: English
• Subjects: Algorithms; Basis functions; Classification; DE algorithm; differential evolution algorithms; discrimination feature; electromagnetic target classification; electromagnetic target recognition problem; electromagnetic wave scattering; Evolution; Fitness; JADE; jDE; late‐time scattered signal; lossless dielectric spheres; noiseless signals; noisy signals; object recognition; Recognition; resonance scattering region; signal detection; Sonar; Target recognition; target recognition method; target set
• ISSN: 1751-8784; 1751-8792; 1751-8792
• DOI: 10.1049/iet-rsn.2012.0212

In this study, an ensemble of differential evolution (DE) algorithms is presented to classify electromagnetic targets in resonance scattering region. The algorithm aims to synthesize a special incident signal for each target, which is defined as the main discrimination feature in the given target recognition method. In the proposed algorithm, the amplitudes of basis functions and the duration of this incident signal are optimised to give minimum late-time scattered signal's energy, which is the main fitness function of the algorithm. The proposed DE algorithm is applied to a target set consisting of lossless dielectric spheres and correct recognition rates for both noiseless and noisy signals are obtained. The results for both developed DE algorithm and other DE variants of traditional DE, adaptive differential evolution with optional external archive (JADE), jDE are also given to compare the algorithms and show the effectiveness of the proposed one.