Research On On-Orbit Status Detection And Imaging Methods Of Space Targets Based On Phased Array Radar

Establishing a system consists of on-orbit space target detection,inversion of the inverse synthetic aperture radar(ISAR)imaging and image enhancement based on phased array radar is important for preventing and responding to space emergencies.However,the phased array radar usually tracks monitoring of multiple objects in actual missions.Therefore,the radar observation resources acting on single target will be limited,resulting in the negative effects on on-obit status detection and imaging of the targets.Focusing on the problems in the detection and imaging of space targets based on phased array radar,this paper carries out specific researches as follows:In accordance with the difficulties of artificial feature extraction which caused by the low data rate of RCS under the phased array radar system,an RCS anomaly detection method based on a convolutional neural network(CNN)and bidirectional gated recurrent units(Bi GRU)is proposed in this paper.1-D CNN is adopted to extract high-dimensional features of the RCS sequence;BIGRU helps to find out time-series dependency features of the high-dimensional feature vectors.After that,a fully connected layer is used for the classification and recognition of anomaly targets’ RCS.The experimental results indicate that the proposed method presents higher detection accuracy and stronger noise robustness than traditional methods in conditions of low data rate.Moreover,the model trained by simulation data has high detection performance under the condition of low data rate for the measured data of different sources,indicating that the model has strong generalization ability.In view of ISAR imaging degradation caused by sparse aperture of single target echo and estimated error of spinning speed,an ISAR imaging method of spinning targets based on minimum image entropy criterion and compressed sensing is proposed in this paper.Iterative optimization of the optimal sampling interval based on ISAR image entropy helps to eliminate the negative effects caused by the estimated error of spinning speed.Based on the sparsity of the resampled one-dimensional range profile,the measurement matrix is constructed,which is used to reconstruct the target ISAR images by compressed sensing algorithm.The experimental results indicate that the proposed method has strong adaptive compensation ability for different degrees of speed estimation error,and the reconstructed ISAR image maintains high similarity with the ideal result in conditions of different echo sparsity.Contraposing to the extant CNN-based ISAR image enhancement methods,which have poor scene adaptability and lose detailed information while reconstructing images,a modified ISAR image enhancement framework based on a recursive residual network is proposed in this paper.Image filters capable of simulating ISAR imaging degradation are adopted to generate training dataset,which helps to minimize the influence of the fixed imaging scenarios.The local residual connection units are applied to retain the original image information,while the parameter sharing mechanism is adopted to accelerate the training speed.A combined loss function,composed of the multi-scale structural similarity loss and the smooth L1 loss,is adopted to retain the high-frequency information and the edge details of the ISAR image.Experimental results demonstrated that compared with extant methods,the proposed method provides reconstructed ISAR images with higher resolution,and effectually enhances the detailed information of the images.