Efficient embedded system design for machine fault diagnosis using vibration features
Abstract ? The early detection of failures in machinery equipment is one of the most important concern to industry. The fault diagnosis of machine consists of feature extraction and classification of faults. Every industry wants a compact device or embedded system of low cost
2025-06-28 16:32:19 - Adil Khan
Efficient embedded system design for machine fault diagnosis using vibration features
Project Area of Specialization Mechanical EngineeringProject SummaryAbstract
• The early detection of failures in machinery equipment is one of the most important concern to industry. The fault diagnosis of machine consists of feature extraction and classification of faults. Every industry wants a compact device or embedded system of low cost for the early fault diagnosis of machinery.
• This project is focused at efficient embedded system design for early fault diagnosis of machine and classification of faults based on each machine fault having a unique vibration feature.
• The embedded system will be deployed on a workbench to analyse machine faults and system efficiency.
• The test rig will be designed for the experiment to collect vibration signals and diagnose the machine faults. Advanced signal processing techniques like Empirical Mode Decomposition (EMD), Hilbert Transforms (HT’s), Wavelet features and fault classifiers will be used.
Aims and objectives
A Test Rig will be designed consisting of a two-stage Transmission System using Gears, Pulleys and Shafts.
A Data Acquisition System will be designed for collecting Machine Vibration Signals and MATLAB will be used for advance signal processing techniques.
In addition to Time Waveform, Frequency spectrum, Power Spectrum Density Technique, Empirical Mode Decomposition and Artificial Intelligence Toolbox using MATLAB classifiers will be deployed for early fault detection of machines.
Steps:
- Machine Vibration signals will be collected using tri-axial accelerometer placed at three different positions on the Test Rig. (Vertical, Horizontal and Axial).
- Time Waveform and Spectrum Analysis will be used to identify different faults within Machine Transmission System.
- Vibration Signals of healthy and defected gears are to be obtained using the experimental facility working under different operating conditions.
• Efficient Embedded System will be designed which analysis data in real time.
• Most Techniques used previously like Fourier Transform detects the fault at 70% build-up of failure, whereas techniques used in our System like Time Waveform Frequency Spectrum, Power Spectrum Density, Empirical Mode Decomposition and Hilbert Transform which detects the faults at 30% Build-up.
• Failures of the gearbox may cause injury to human beings and important machinery losses. To avoid the consequences of any harmful accidents we apply early fault detection.
• Condition monitoring and system health management are the primary tools used to describe the maintenance scheme of machine components ensuring the operational safety of the systems.
• Most failures in rotating machinery are related to the mechanical transmission system, which contributes 30% of the machine’s total maintenance cost.
• The gearbox is a vital transmission system composed of gear bearings and drive shafts that form a complex system. Gears contribute to 60% of gearbox failures.
• Gear analysis is an important activity in the field of Condition Monitoring and Machine Fault Diagnosis.
• Early detection of local gear faults in industrial environments is very important to optimize the maintenance schedule and reduce the operating cost of gearbox damage.
Condition monitoring of gearbox using advanced signals processing techniques through vibration analysis at multiple operational speed range and constant load.
Time waveform and Power spectral density techniques will generate ideal vibrating system amplitude behavior. Broken tooth gear system will be installed for varying percentage of fault and vibration features will be extracted for early fault diagnosis.
Gearbox defects are evaluated using Vibrations features like:
1.Time Domain: Fluctuates around X-axis and increases with shaft speed.
2.Root Mean Square
3.Crest Factor
4.Kurtosis
5.Frequency spectrum
6. Empirical mode decomposition (EMD)
Fluctuating speeds will be used to distinguish variations in time domain and frequency spectrum for ideal and defected gears.
Artificial intelligence module of MATLAB will be used for early fault diagnosis of machine high frequency faults.
Generated vibration from gearboxes result in system resonance and will be transmitted to the structure in the form of noise and vibration.
Signals will be transferred through MPU6050 sensor module which is an integrated 6-axis Motion tracking device and accelerometer, is connected to the Raspberry Pi Model 4B in which we have developed a GUI which will display the analysis of the signals captured at different conditions.
Final Deliverable of the Project HW/SW integrated systemCore Industry OthersOther Industries IT Core Technology Internet of Things (IoT)Other Technologies Artificial Intelligence(AI)Sustainable Development Goals Industry, Innovation and Infrastructure, Responsible Consumption and ProductionRequired Resources| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 61930 | |||
| Raspberry Pi Model 4B | Equipment | 1 | 18200 | 18200 |
| arduino MPU6050 | Equipment | 1 | 3600 | 3600 |
| Wires+connectors+technical accessories | Miscellaneous | 1 | 2300 | 2300 |
| Display+mouse+keyboard | Equipment | 1 | 7000 | 7000 |
| belts | Equipment | 4 | 370 | 1480 |
| shafts | Equipment | 1 | 1600 | 1600 |
| bearing | Equipment | 8 | 500 | 4000 |
| gear | Equipment | 8 | 500 | 4000 |
| wooden base | Equipment | 1 | 1750 | 1750 |
| Manufacturing and assembly Costs of Rig | Equipment | 1 | 18000 | 18000 |