Prediction of Remaining Useful Life for AeroEngines

The project, “Prediction of Remaining Useful Life for AeroEngines”, focuses on developing a data-driven framework to estimate the remaining operational lifespan of aircraft engines, known as Remaining Useful Life (RUL). Accurate RUL prediction is critical for proactive maintenance, safety assurance, and reducing unexpected engine failures in aviation. This work employs machine learning techniques to analyze sensor data collected from aeroengines during operational cycles. The system allows users to upload engine sensor datasets in CSV format, and it utilizes pre-trained regression models, including Extra Trees Regressor, Random Forest, Gradient Boosting, XGBoost, and neural networks, to predict engine health status and RUL. Input features, such as temperature, pressure, and vibration readings from multiple sensors, are processed and fed into the models to generate precise RUL estimates. The predicted RUL is classified into three categories: critical, warning, and normal, accompanied by actionable maintenance suggestions for each class. A web-based interface, implemented using Flask, provides interactive functionalities for registration, login, dataset upload, model selection, and real-time RUL prediction. The system integrates database support through MySQL for secure user management and data storage. Experimental results demonstrate that ensemble learning and neural network models can achieve high predictive accuracy, enabling reliable and timely decision-making for aeroengine maintenance. The approach ensures safety, optimizes maintenance schedules, and reduces operational costs by preventing unscheduled failures.

Keywords: Remaining Useful Life (RUL), Aeroengine, Machine Learning, Predictive Maintenance, Regression Models, Ensemble Learning, Sensor Data, Flask, Extra Trees Regressor, XGBoost.