Scholarlink

<p>In this paper, we present a machine learning method to accurately predict thrust in a cold gas thruster using a feedforward neural network (FFNN). The model leverages critical operational parameters, such as storage pressure, mass flow rate, nozzle length, exit pressure, and propellant mass density, to achieve high precision in thrust predictions. To make this technology accessible and practical, we introduce an intuitive graphical user interface (GUI) that allows users to estimate thrust in real-time systems. This tool simplifies design and analysis processes, offering engineers a powerful resource for optimizing the performance of the cold gas thrusters. Based on the simulation results, our proposed method achieves an accuracy of 0.98 and an F1 score of 0.981, showcasing its robustness and generalizability across various test cases. Our work highlights how machine learning methods can be effectively integrated into propulsion system development, paving the way for more innovative, more efficient designs.</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

A neural network based approach for thrust prediction in cold gas propulsion systems

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年7月22日

查看全文
  • Morteza Farhid,
  • Mohammad Reza Ghavidel Aghdam,
  • Moharram Shameli

摘要

In this paper, we present a machine learning method to accurately predict thrust in a cold gas thruster using a feedforward neural network (FFNN). The model leverages critical operational parameters, such as storage pressure, mass flow rate, nozzle length, exit pressure, and propellant mass density, to achieve high precision in thrust predictions. To make this technology accessible and practical, we introduce an intuitive graphical user interface (GUI) that allows users to estimate thrust in real-time systems. This tool simplifies design and analysis processes, offering engineers a powerful resource for optimizing the performance of the cold gas thrusters. Based on the simulation results, our proposed method achieves an accuracy of 0.98 and an F1 score of 0.981, showcasing its robustness and generalizability across various test cases. Our work highlights how machine learning methods can be effectively integrated into propulsion system development, paving the way for more innovative, more efficient designs.