Scholarlink

In the trajectory correction process of a high-spin tail-controlled corrective projectile, it is necessary to quickly and accurately obtain control commands, which requires predicting its correction capability. In this paper, a model based on Convolutional Neural Network and Long Short-Term Memory Network with an introduced self-attention mechanism (CNN-LSTM-ATT) is proposed. A 7°-of-freedom projectile dynamics model was established, and external trajectory simulations were conducted using the Runge-Kutta method to obtain a projectile trajectory database. The performance of the CNN-LSTM-ATT model was compared with that of BP, LSTM, and CNN-LSTM models. The results show that the CNN-LSTM-ATT model exhibits excellent prediction performance for correction capability and achieves higher accuracy compared to other models.

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

Correction Capability Prediction of High-Spin Tail-Controlled Projectiles Based on CNN-LSTM with Self-Attention Mechanism

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Cunbin Lin,
  • Jian Fu,
  • Liangming Wang,
  • Jie Zhou,
  • Yifu Zhao

摘要

In the trajectory correction process of a high-spin tail-controlled corrective projectile, it is necessary to quickly and accurately obtain control commands, which requires predicting its correction capability. In this paper, a model based on Convolutional Neural Network and Long Short-Term Memory Network with an introduced self-attention mechanism (CNN-LSTM-ATT) is proposed. A 7°-of-freedom projectile dynamics model was established, and external trajectory simulations were conducted using the Runge-Kutta method to obtain a projectile trajectory database. The performance of the CNN-LSTM-ATT model was compared with that of BP, LSTM, and CNN-LSTM models. The results show that the CNN-LSTM-ATT model exhibits excellent prediction performance for correction capability and achieves higher accuracy compared to other models.