Scholarlink

This paper presents a rapid Deep Neural Network (DNN)-based methodology for predicting the reachable domain of re-entry vehicles. The proposed approach employs a virtual target method rooted in Sequential Convex Programming (SCP) to generate boundary samples, constructing a dataset closely resembling the authentic reachable domain boundary. By dividing the prediction model for the reachable domain boundary, the method reduces problem complexity, resulting in improved prediction accuracy. Numerical simulations show that the average prediction time of the proposed method is only 0.0153 s, which is significantly better than the 346.2 s of the traditional SCP. In addition, the DNN method can quickly increase the density of boundary points compared to the traditional SCP method.

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

A Rapid Prediction Method for Re-entry Vehicle Reachable Domain Based on Deep Neural Network

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Changshuo Zhu,
  • Yang Ni,
  • Binfeng Pan,
  • Jixiang Jiang,
  • Keyuan Yue

摘要

This paper presents a rapid Deep Neural Network (DNN)-based methodology for predicting the reachable domain of re-entry vehicles. The proposed approach employs a virtual target method rooted in Sequential Convex Programming (SCP) to generate boundary samples, constructing a dataset closely resembling the authentic reachable domain boundary. By dividing the prediction model for the reachable domain boundary, the method reduces problem complexity, resulting in improved prediction accuracy. Numerical simulations show that the average prediction time of the proposed method is only 0.0153 s, which is significantly better than the 346.2 s of the traditional SCP. In addition, the DNN method can quickly increase the density of boundary points compared to the traditional SCP method.