This repository contains the implementation and demonstration code for the webinar "Enhancing Machine Condition Monitoring Using Order Spectrum Analysis" presented on November 13, 2024.
This project demonstrates the implementation of order spectral analysis for detecting bearing faults in rotating machinery. The study focuses on identifying bearing faults by monitoring abnormalities in:
- Outer race (BPFO - Ball Pass Frequency Outer)
- Inner race (BPFI - Ball Pass Frequency Inner)
- Fundamental train frequency (FTF)
- Rolling elements (Ball Spin Frequency)
The implementation uses the open-source HUST bearing dataset, which provides practical data for ball bearing fault diagnosis under different load conditions:
- I400: Motor at 0W load
- I402: Motor at 200W load
- I404: Motor at 400W load
- I600: Motor at 0W load (different bearing type)
- I602: Motor at 200W load (different bearing type)
- I604: Motor at 400W load (different bearing type)
Sampling frequency: 51,200 Hz
- Signal Processing: Bandpass filtering and envelope spectrum analysis using the Hilbert transform
- Bearing Modeling: Theoretical calculation of bearing characteristic frequencies
- Order Analysis: Conversion from frequency domain to order domain for variable speed analysis
- Fault Detection: Identification of bearing defect frequencies in the order spectrum
main.ipynb # Main analysis notebook
data/ # HUST bearing dataset
models/ # Bearing models and configurations
requirements.txt # Python dependencies (currently empty)
sample.slides.html # Presentation slides
README.md # This file
- Clone the repository:
git clone <repository-url>
cd 2024ecc_presentation- Create and activate a virtual environment:
python -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate- Install required dependencies:
pip install numpy scipy matplotlib jupyter- Ensure the HUST bearing dataset is placed in the
data/directory - Launch Jupyter notebook:
jupyter notebook main.ipynb- Run the cells sequentially to reproduce the analysis
- Data Loading: Load vibration data from HUST bearing dataset
- Bandpass Filtering: Apply FIR filter (22.6-23.3 kHz) to isolate bearing frequencies
- Envelope Analysis: Calculate envelope using Hilbert transform
- Order Conversion: Transform frequency spectrum to order spectrum using shaft speed
- Fault Detection: Identify peaks at theoretical bearing defect orders
The bearing characteristic frequencies are calculated using standard formulas:
- BPFO:
(Nb/2) × (1 - (Db/Dp)) - BPFI:
(Nb/2) × (1 + (Db/Dp)) - FTF:
(1/2) × (1 - (Db/Dp)) - Ball Spin:
(Dp/Db) × (1 - (Db/Dp)²)
Where:
- Nb = Number of balls
- Db = Ball diameter
- Dp = Pitch diameter
- Event: Webinar
- Date: November 13, 2024
- Duration: 30 minutes
- Target Audience: Engineers and maintenance professionals
- Link: Webinar Details
This project is for educational and demonstration purposes related to the webinar presentation.
For questions about the implementation or webinar content, please refer to the original webinar materials or contact SeaQuest Solutions.