Abstract
The turbofan engine is a pivotal component of the aircraft. Engine components are susceptible to degradation over the life of their operation, which affects the reliability and performance of an engine. In order to direct the necessary maintenance behavior, remaining useful life prediction is the key. This research uses machine learning to provide a prediction framework for an aircraft’s remaining useful life (RUL) based on the entire life cycle data and deterioration parameter data (ML). For the engine’s lifetime assessment, a Deep Layer Recurrent Neural Network (DL-RNN) model is presented. The suggested method is compared to Multilayer Perceptron (MLP), Nonlinear Auto Regressive Network with Exogenous Inputs (NARX), and Cascade Forward Neural Network (CFNN), as well as the Prognostics and Health Management (PHM) conference Challenge dataset and NASA’s C-MAPSS dataset. Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) are calculated for both the datasets, and the values are in the range of 0.15% to 0.203% for DL-RNN, whereas for the other three topologies, they are in the range of 0.2% to 4.8%. Comparative results show a better predictive accuracy with respect to other ML algorithms.
Highlights
The engine is a heart of an airplane
The length from the current time to the failure of a system is characterized as remaining useful life (RUL), and its forecast methodologies are classed as follows: the model-based technique, data-driven technique, and hybrid technique are all examples of model-based techniques [2]
Recurrent Neural Network based on Statistical Recurrent Unit (SRU) was proposed for the RUL prediction of the C-MAPSS dataset of a turbofan engine [23]
Summary
The engine is a heart of an airplane. With the emergence of action cycles, system faults and the deterioration process would inevitably occur. Convolution Neural Networks (CNN) were developed to forecast the RUL of the same engine [22] It uses a time window technique for sample processing, which may yield more deterioration detail. Recurrent Neural Network based on Statistical Recurrent Unit (SRU) was proposed for the RUL prediction of the C-MAPSS dataset of a turbofan engine [23] It can find hidden partners from multivariate time-series sensor data with varied operating state faults and defects, and it outperforms other dynamic deep learning algorithms because it uses linear combinations of few averages to obtain a plurality of historical perspectives. When compared to MLP, the results reveal a higher level of accuracy Various topologies, such as MLP, SVR, RVR, CNN, and LSTM, have been employed for RUL prediction of the turbofan engine, as previously noted. Analysis of similarity enables having the RUL according to the outcome that fits
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have