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

Area of Science:

• Aerospace Engineering
• Computational Science

Background:

• Accurate thrust prediction is crucial for the design and optimization of cold gas thrusters.
• Traditional methods may lack the precision required for complex operational parameters.

Purpose of the Study:

• To develop and validate a machine learning model for precise thrust prediction in cold gas thrusters.
• To create an accessible tool for real-time thrust estimation and system analysis.

Main Methods:

• Utilized a feedforward neural network (FFNN) trained on key operational parameters.
• Incorporated storage pressure, mass flow rate, nozzle length, exit pressure, and propellant mass density.
• Developed a graphical user interface (GUI) for practical application and real-time predictions.

Main Results:

• The FFNN model achieved high prediction accuracy (0.98) and F1 score (0.981).
• Demonstrated robustness and generalizability across diverse test cases.
• The GUI facilitates simplified design and performance optimization.

Conclusions:

• Machine learning offers a powerful approach for advancing cold gas thruster technology.
• The proposed method enhances the efficiency and innovation in propulsion system development.
• This work provides a practical tool for engineers in the aerospace industry.