Data regarding dynamic performance predictions of an aeroengine.

Area of Science:

• Aerospace Engineering
• Computational Fluid Dynamics
• Machine Learning

Background:

• Real-time aeroengine control systems increasingly require machine learning techniques.
• A significant limitation in developing these algorithms is the scarcity of in-flight aeroengine performance data.
• Dynamic aeroengine models offer a viable solution to generate necessary performance data.

Purpose of the Study:

• To present a dataset of turbojet performance metrics predicted by a dynamic engine model.
• To provide data for training machine learning models for in-flight aeroengine performance prediction.
• To make available the predictions from a Nonlinear AutoRegressive with eXogenous inputs (NARX) neural network.

Main Methods:

• A dynamic turbojet engine model was developed using Gas turbine Simulation Program (GSP) software.
• The dynamic model generated performance data, including Exhaust Gas Temperature (EGT).
• An Artificial Neural Network (ANN), specifically a NARX model, was implemented using the generated data.

Main Results:

• The dynamic engine model successfully predicted key turbojet performance parameters.
• The NARX neural network was trained on this simulated data to predict in-flight performance.
• The dataset includes both the simulated engine performance and the ANN predictions.

Conclusions:

• The presented data facilitates research in machine learning applications for aeroengine control.
• Dynamic modeling provides a crucial method for overcoming data limitations in aeroengine research.
• The dataset supports the advancement of accurate in-flight performance prediction algorithms.