This project is for classifying the Remaining Useful Life (RUL) Time of bearing

1. Data Collection

   • Acquire vibration signals from bearings and label them with Remaining Useful Life (RUL).

2. Data Preprocessing

   • Segment the time-series data into smaller windows.
   • Normalize the signals.

3. Signal Conversion

   • Extract features using methods like Short Time Fourier Transform (STFT) or wavelet transform.
   • Convert extracted features into images in this work simple plot creation is used.

4. Dataset Preparation

   • Split the dataset into training, validation, and test sets.
   • Apply data augmentation to enhance image diversity.

5. Transfer Learning

   • Select a pre-trained CNN model (Exception network).
   • Modify the output layer for RUL classification.

6. Model Training

   • Train the modified CNN on the image dataset, utilizing transfer learning techniques.

7. Evaluation Results

8. Deployment

   • Integrate the model into operational settings for real-time RUL predictions, with a feedback loop for continuous improvement.