CNN-LSTM for Turbofan RUL Prediction (FD001)
This repository contains a CNN-LSTM model trained to predict the Remaining Useful Life (RUL) of turbofan engines based on time-series sensor data. The model employs a hybrid architecture where a 1D Convolutional Neural Network (CNN) acts as a feature extractor to identify local patterns from sensor readings within a time window, and a Long Short-Term Memory (LSTM) network then processes these features to capture long-term temporal dependencies indicative of engine degradation. The model takes a window of 40 time steps with 22 features as input and outputs a single value representing the predicted RUL in cycles.
This particular model checkpoint was trained exclusively on the FD001 sub-dataset from the NASA CMAPSS collection. This is done to showcase how this model works with the proposed XAI Framework on this specific condition/machine.
This dataset represents a single operating condition and a single fault mode (HPC degradation).