关键词: 雷达目标识别, 高分辨一维距离像, 一维卷积神经网络, 灰狼优化算法

Abstract: As providing the detailed information of target’s features, High resolution range profile (HRRP) is predominantly employed in the field of wideband radar target recognition. The traditional method of Radar HRRP recognition of ballistic midcourse targets is hindered by the influence of environmental noise interference, atmospheric radiation and penetration strategies, resulting in low accuracy. Furthermore, the intelligent optimization algorithm faces the challenge of an extensive number of parameters, coupled with manual parameter adjustment tuning to extract local features of the target. Aiming at this issue, a high resolution range profile recognition method for ballistic targets based on the improved grey wolf optimizer and one-dimensional convolutional neural network (IGWO-1DCNN) is proposed. The method entails the construction of an improved 1D convolutional neural network for the feature extraction from HRRP samples of wideband radar targets. This is achieved through the incorporation of the improved grey wolf optimizer (IGWO) algorithm, which serves to accelerate the convergence speed of the model and enhance its recognition performance. The support vector machine (SVM) is used as the classifier to facilitate the recognition processes. The experimental results demonstrate that this novel method is capable of accurately identifying the ballistic targets, automatically optimizing parameters of neural network, and reducing the burden of manual training and exhibiting higher robustness.

Key words: radar target recognition, high resolution range profiles, 1D convolutional neural network, grey wolf optimizer