Data Diven Based Extraction and Recognition of Space Target’s Radar Micromotion Features

Abstract

The micromotion features of the space target is an important information source of radar target recognition. Multiple micromotions due to certain complex features can further characterize the specific targets and then improve the accuracy of the target recognition. With regard to the extraction and recognition of space target complex micromotion features, the data-driven method was used to conduct effective extraction and training recognition. On the one hand, the EMD method was used to effectively separate the complex micromotion components of the space target, and then the motion parameters of the target are estimated with high precision. On the other hand, we take the motion parameters of the target in various forms of micromotion as input and learn typical object movement characteristics by a machine learning algorithm, thus improve the precision and accuracy of target detection. The simulation results show that EMD method can accurately extract various motion parameters of space complex micromotion targets, and the neural network model can complete the target classification and recognition task with high accuracy.