A Machine Learning Approach for Turbofan Jet Engine Predictive Maintenance

Maintenance of any equipment is crucial for any field revolving around machines. Majorly there are three types of maintenance: Corrective, Preventive and Predictive. A basic idea of predictive maintenance is to plan a maintenance using some type of data analysis techniques to determine a potential failure. There are various techniques for analysis and one such very popular data analysis technique is machine learning (ML). ML techniques can be used to predict any machine failure beforehand preventing various assumable costs related as well as accidents related consequences. This work focuses on predicting Remaining Useful Life (RUL) of Turbofan Jet Engines using ML techniques. Prediction of RUL is highly beneficial, however the most critical task, for predictive maintenance of any component. In this work’s context, remaining useful life is equivalent to the remaining number of flights possible for the engine. The method proposed in this work uses Kaggle version of the NASA’s FD001 turbofan jet engine dataset. The methods include scaling the data using standard scaling, feature engineering using polynomial features; this work adopts Linear Regression (LR), Support Vector Regression (SVR) and Random Forest regress (RFR) with hyper parameter tuning using Random Search CV and 5-fold cross validation. The model has shown 0.789 R2 score with 19.077 RMSE values, which is higher than other machine learning based models in RUL prediction.