Research On Pole Extraction And Target Recognition Technology Of Stealth Aircraft In Resonant Region

The radar target is in resonant region when the size of the target is similar to the wavelength of incident electromagnetic wave.The scattering characteristics of military aircraft target in resonance region are obvious,which are beneficial to extract pole features used for target recognition.Pole feature is only related to the inherent properties such as the shape and size of the target.And pole feature has better azimuth consistency and is independent of incident waveform,attitude and polarization mode.These features provide theoretical support for resonant region radar target recognition based on pole feature.In view of the research background,research is carried out around three aspects which are resonant region target modeling and scattering characteristic RCS acquisition,pole feature extraction in resonant region and object recognition based on pole features.1.The geometric modeling and RCS frequency domain data acquisition of stealth aircraft B2 and F22 are studied.Due to the confidentiality of military targets,no actual measurement RCS data can be used.Fortunately,the wavelength corresponding to the target scattering characteristics in the resonant region is relatively large,so the influence of geometric modeling size and data error on the overall target scattering characteristics can be ignored.Thus,RCS data can be obtained through geometric modeling and electromagnetic simulation.Firstly,according to the characteristics of complex target in resonance region,a surface geometry modeling method based on Hypermesh software is adopted.And the steps of geometric modeling,the influencing factors of modeling,mesh generation criteria and geometric cleaning method of target model are discussed.Secondly,the FEKO software is used to calculate the resonant region B2 and F22 target RCS,and the specific steps are given.Finally,taking the F22 stealth aircraft as an example,the RCS under the conditions of different mesh sizes and azimuth angles of electromagnetic wave incidence are analyzed.The effects of RCS to the target’s flight attitude and mesh size are discussed in detail,and the reasonable mesh size is given.The relationship between the distribution of RCS azimuth diagram and the structure of each part of the target is also studied.2.The azimuth consistency and robust pole feature extraction method of typical stealth target based on RCS data is studied.Firstly,the pole extraction is based on time domain data,so a scheme to obtain time domain transient response from RCS frequency domain data is studied.Specifically,the time domain transient response is the inverse Fourier transform of the RCS conjugate symmetric data.Secondly,the start time of the late-time stage which can be used for pole extraction is determined from the time domain transient response.Combined with B2 and F22 physical model,the process of incident electromagnetic wave irradiating the target is analyzed.Different late-time stage start time is calculated when the target’s flight attitude changes relative to radar,and the late-time start time used for pole extraction is determined.Thirdly,a new sliding window matrix pencil method(SWMPM)is proposed to achieve reliable pole extraction,which effectively solves the difficulty of order setting and time domain transient response aliasing in traditional matrix pencil method(MPM).Through simulation analysis,the performance of SWMPM is better than that of MPM.Fourthly,in order to further improve the robustness of poles,a method of extracting principal poles based on residues and azimuth consistency is designed.The number of main poles is small,but their energy contribution is large and azimuth consistency is strong,which is more conducive to the target recognition based on pole features.Finally,the setting range of the key parameters of the proposed method is analyzed in detail by simulation,and the main poles of azimuth consistency for B2 and F22 targets are given.3.The target recognition in resonance region of B2 and F22 based on main poles is studied.Firstly,the B2 and F22 pole feature database based on residue order frequency selection and feature extraction are established.On the one hand,residue order frequency selection can find the poles with large energy contribution;On the other hand,feature extraction based on maximizing the average feature distance between different poles can effectively distinguish different targets.Secondly,since the main poles of B2 and F22 are limited,it is a small sample classification problem to identify them.Therefore,a target recognition method based on support vector machine and pole matching is proposed,and then target recognition classifier of B2 and F22 is designed.Finally,the actual radar system may face with low SNR of RCS data when detecting and identifying long-range targets,and the effect of noise on pole extraction and target recognition is analyzed in detail.The simulation results show that the proposed method can still get better target recognition results under the condition of low SNR.