NASA Turbofan Engine Degradation Simulation (CMAPSS Dataset)

Dataset Overview

The Commercial Modular Aero-Propulsion System Simulation (CMAPSS) dataset is the benchmark for Remaining Useful Life (RUL) prediction research. Contains simulated degradation trajectories from 100 turbofan engines until failure.

Dataset Structure (4 Subsets)

• FD001: 100 engines, 1 operating condition, 2 fault modes
• FD002: 120 engines, 6 operating conditions, 1 fault mode
• FD003: 100 engines, 2 operating conditions, 2 fault modes
• FD004: 248 engines, 6 operating conditions, 2 fault modes

Sensor Measurements (21 Sensors)

• Temperature Sensors: 7 total air temperature sensors
• Pressure Sensors: 6 pressure measurements
• Rotational Speed: 2 rpm sensors
• Dynamic Sensors: Fuel flow, vibration, ratio metrics
• Operational Settings: 3 control parameters

Key Characteristics

• Total Cycles: 20,000+ cycles per engine
• RUL Labels: Exact remaining cycles until failure
• Operating Conditions: 6 different flight profiles
• Fault Progression: Realistic gradual degradation

Research Impact

Used in 1,000+ research papers. Benchmark for deep learning prognostics, LSTM, CNN, and attention-based RUL prediction models.

Data Format

4 CSV training files + 4 test files + RUL ground truth. Each row: unit_number, time_in_cycles, 3 settings, 21 sensors.