This project focuses on optimizing aircraft engine maintenance operations by utilizing statistical and machine learning techniques. Regression and classification models are used to predict component failures, estimate Time-To-Failure (TTF) or Remaining Useful Life (RUL), and enable proactive maintenance planning. The project aims to enhance maintenance operations, reduce downtime, and improve overall aircraft performance by leveraging sensor data analysis and telemetry.
The related notebook regarding this analysis can be found here.
- Elastic Net Regression
- Random Forest Regressor
These regression models are applied to predict component failures and estimate the Time-To-Failure (TTF) or Remaining Useful Life (RUL) of aircraft components.
- K-Nearest Neighbors Classification
The classification model is used to classify the health condition of engine components and predict failures during operation.
The dataset used for this project can be found in data folder. It is extracted from a dataset that simulates the run-to-failure scenario of engine operation. The dataset constitutes the anonymized portion of a larger dataset generated by Microsoft, which was extensively used in one of the projects within the Springboard DS Career Track Bootcamp.
-
train_selected.csv: contains 4 simulated sensor measurements for 100 aircraft engines running until failure. It is assumed that the engine's degradation pattern is reflected in its sensor measurements (shape = (20631, 8))
-
test_selected_ttf.csv: contains 4 simulated sensor measurements for 100 aircraft engines running until failure. It is assumed that the engine's degradation pattern is reflected in its sensor measurements (shape = (100, 8))
The dataset used in this project was generated by Microsoft and made available through the Springboard DS Career Track Bootcamp. We acknowledge their contribution and appreciate their support in providing the data for analysis.