Scholarlink

<p>Accurate ballistic target recognition via high-resolution range profile (HRRP) faces critical challenges, including sensitivity to noise, signal distortion, and the limitations of deep networks with manually tuned parameters. This paper proposes an improved grey wolf optimizer with a 1D-CNN architecture (IGWO-1DCNN). The IGWO employs a Tent chaotic map for initialization, a nonlinear convergence factor, and a Levy flight strategy to efficiently optimize the hyper-parameters of CNN, overcoming local optima and slow convergence, and use support vector machine (SVM) as the classifier. Experimental results demonstrate the method's superior performance, achieving a notable average recognition accuracy of 94.70% across five target types and significantly faster convergence compared to other optimization algorithms, confirming its robustness and optimization efficiency.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Space target based on improved grey wolf optimizer and 1-D convolutional neural network

  • 期刊论文

  • 发表日期：

    2026年2月17日

查看全文
  • Caiyun Wang,
  • Jiaxuan Han,
  • Huiwen Zhang,
  • Jianing Wang,
  • Yida Wu

摘要

Accurate ballistic target recognition via high-resolution range profile (HRRP) faces critical challenges, including sensitivity to noise, signal distortion, and the limitations of deep networks with manually tuned parameters. This paper proposes an improved grey wolf optimizer with a 1D-CNN architecture (IGWO-1DCNN). The IGWO employs a Tent chaotic map for initialization, a nonlinear convergence factor, and a Levy flight strategy to efficiently optimize the hyper-parameters of CNN, overcoming local optima and slow convergence, and use support vector machine (SVM) as the classifier. Experimental results demonstrate the method's superior performance, achieving a notable average recognition accuracy of 94.70% across five target types and significantly faster convergence compared to other optimization algorithms, confirming its robustness and optimization efficiency.