Showing 1–1 of 1 results for author: Wilhelm, J P

Small jet engine reservoir computing digital twin

Authors: C. J. Wright, N. Biederman, B. Gyovai, D. J. Gauthier, J. P. Wilhelm

Abstract: Machine learning was applied to create a digital twin of a numerical simulation of a single-scroll jet engine. A similar model based on the insights gained from this numerical study was used to create a digital twin of a JetCat P100-RX jet engine using only experimental data. Engine data was collected from a custom sensor system measuring parameters such as thrust, exhaust gas temperature, shaft s… ▽ More Machine learning was applied to create a digital twin of a numerical simulation of a single-scroll jet engine. A similar model based on the insights gained from this numerical study was used to create a digital twin of a JetCat P100-RX jet engine using only experimental data. Engine data was collected from a custom sensor system measuring parameters such as thrust, exhaust gas temperature, shaft speed, weather conditions, etc. Data was gathered while the engine was placed under different test conditions by controlling shaft speed. The machine learning model was generated (trained) using a next-generation reservoir computer, a best-in-class machine learning algorithm for dynamical systems. Once the model was trained, it was used to predict behavior it had never seen with an accuracy of better than 1.8% when compared to the testing data. △ Less

Submitted 15 December, 2023; originally announced December 2023.