Publication Date
Spring 2024
Degree Type
Master's Project
Degree Name
Master of Science in Computer Science (MSCS)
Department
Computer Science
First Advisor
Saptarshi Sengupta
Second Advisor
Ching-Seh Wu
Third Advisor
Navrati Saxena
Keywords
Long Short-term memory (LSTM), Prognostics, CMAPSS, N-CMAPSS, Recurrent Neural Networks, Predictive Analytics, Turbofan Engine Degradation Modeling
Abstract
Aircraft engines are susceptible to failure at multiple points over their lifespan and need replacement or repairs. The ability to proactively determine how long an engine will function helps avoid fatalities and build a reliable prognostic system. To accomplish this, predictive models are being developed using various approaches like physics-based and data-driven techniques. Physics-based models need huge computing power for simulations and domain knowledge for understanding and implementing the models. Alternatively, if we have substantial data for prediction, data-driven models can be used. In this research, we use data-driven approach for engine Remaining-Useful-Life (RUL) prediction on the NASA Commercial Modular Aero-Propulsion System Simulation (CMAPSS) dataset, and the New Commercial Modular Aero-Propulsion System Simulation (N-CMAPSS) dataset, containing huge number of samples on real-time flight conditions (~several GBs). As the sensor readings are in the form of
time series, deep Recurrent Neural Networks are suitable for predicting the Remaining-Useful- Life (RUL) of an engine, especially Long Short-Term Memory (LSTM) networks. In this paper,
we propose a Bidirectional LSTM model for RUL predictions on the sub-datasets FD001, FD002, FD003, and FD004 of the CMAPSS dataset resulting in RMSE values of 17.60, 29.67, 17.62, and 31.84 respectively. For the N-CMAPSS dataset, on the sub-datasets DS01, DS02, DS03, and DS07, different architectures of LSTM including Vanilla LSTM, Stacked LSTM, and Bidirectional LSTM are explored with the Stacked LSTM performing best for DS01 with RMSE 6.22 and Bidirectional LSTM performing best for DS02, DS03, and DS07 with RMSE values of 9.08, 10.19, and 11.73.
Recommended Citation
Tippareddy, Samaikya, "Predicting Remaining Useful Life of Turbofan Engines on CMAPSS and N-CMAPSS using Deep Recurrent Neural Networks" (2024). Master's Projects. 1360.
https://scholarworks.sjsu.edu/etd_projects/1360
