Investigation of Procedures for Automatic Resonance Extraction from Noisy Transient Electromagnetics Data. Volume I. Investigation of Resonance Extraction Procedures

摘要

This volume presents major results and conclusions of research in resonance extraction techniques from transient electromagnetics data. Section 2 addresses the question of what the best methods of estimating the resonances for this application are. Among the estimation procedures studied are the conventional technique of using an inhomogeneous set of equations, eigenvalue decomposition techniques, Jain's method, and iterative techniques. Section 3 studies procedures for selecting the proper order and eliminating extraneous resonances. The maximum likelihood procedure for order selection is presented in Section 3. This procedure is capable of automatically determining the proper order without any prior knowledge of the noise level. Also, in Section 3, Henderson's procedure for eliminating extraneous resonances is tested by numerical example. The procedures presented in Section 3 allow an efficient, automatic procedure for order selection and resonance estimation to be constructured by using eigenvalue decomposition of the data. Section 4 discusses various practical problems that must be addressed before resonance extraction procedures for a practical radar target recognition system can be designed.