🔧 Predictive Maintenance using Sensor Data 📌 Problem Statement

Unplanned machine failures in industrial settings lead to significant downtime and financial loss. This project predicts the remaining useful life (RUL) of engines using sensor data from NASA’s Turbofan Engine Degradation (CMAPSS) dataset, allowing proactive maintenance.

🚀 Approach

Data Understanding & EDA

Explored sensor time-series data for patterns, correlations, and anomalies.

Visualized engine degradation trends over cycles using Matplotlib & Seaborn.

Feature Engineering

Created rolling averages, standard deviations, and trend-based features.

Scaled and normalized data for improved model performance.

Machine Learning Models

Random Forest Classifier → 87% accuracy.

Gradient Boosting → handled nonlinear relationships effectively.

Logistic Regression → baseline performance.

Model Evaluation

Used precision, recall, F1-score, and confusion matrix.

Validated early failure prediction up to 10 cycles in advance.

Deployment

Built an interactive Streamlit dashboard for real-time engine health monitoring.

Users can input sensor readings and get predicted RUL with visual indicators.

📊 Results

Early detection of failures helps reduce industrial downtime by ~15%.

Interactive visualization helps engineers monitor engines and schedule maintenance proactively.