<p>To address the adverse effects caused by different types of radar emitters with the same structures during the structural identification process, a radar emitter structural identification method combining compressed sensing (CS) and a neural network is proposed. CS enables sparse signal compression by projecting high-dimensional signals into a lower-dimensional space. A stacked convolutional autoencoder (SCAE) network with strong representation learning capability, along with a loss function that integrates time-domain and sparse-domain constraints (TCS), is employed to extract deep information from the input signals and enhance the correlation between the radio frequency (RF) structural characteristics and the radar emitter structures. A deep neural network (DNN) is applied for structural recognition. Experimental results demonstrate that the proposed method effectively solves the adverse effects caused by different types of radar emitters during the identification process under the same structure, achieving high accuracy.</p>