I am looking at learning how to complete signal processing for condition monitoring in house on matlab and was wondering if anyone has any resources they would recommend for this?

I have a few resources on signal processing, just less on the actual practical implementation in matlab. Is the signal processing toolbox necessary?


Original Post

I am curious why you choose MATLAB? I would think the Toolbox would be essential for interface with hardware. What DAQ hardware and computer will you use? How many and what kind of machines will you monitor? There are many practical issues for data collection and management that are beyond the basic vibration signal processing that you would program.


I am completing a project for my final year of university where I need to monitor a permanent magnet motor and induction motor. However, I won't be exposed to the actual motors for a few months (when I start an internship) and so currently I am preparing for this by learning as much as I can... I have been told to make sure I understand the processing procedure by making sure I am able to do it in Matlab (as it is widely used at my university) using a dummy fault signal. So I am not yet part of an actual condition monitoring program but will be in the near future.

I do not use MATLAB, but instead use DasyLab. Here are some links to the program and an excellent book on signal processing:

Signals, Processes, and Systems: An Interactive Multimedia Introduction to Signal Processing 3rd ed. 2013 Edition

by Ulrich Karrenberg (Author)

$99.00 527-pages plus includes DVD with DasyLab software and examples

From <https://www.amazon.com/Signals...nalysis+with+dasylab>

DASYLab 2016 - Data Acquisition System Laboratory

From <https://www.mccdaq.com/dasylab-index>


There are several matlab books specializing on vibration analysis. My question may seem trivial but have you done a Google rearch? You may want also to check your university's library. When I was a student, I saw several books on this specific subject (using matlab).


In order for someday in your position to sucseed, it would be beneficial to study three subjects:

1- Using Matlab for general engineering problems 

2- The basics of electric motors, their malfunctions and the vibration generated from motors

3- Practical vibration signal processing


In my opinion,  it would be difficult to directly work on Matlab for a real application if you do not comprehend and connect the three subjects.

Good luck.

Regards- Ali M. Al-Shurafa 

Thanks Walt, I'll definitely check that book out and see what i can do about dasylab.

Ali, I have found a few books on noise and vibration analysis and signal processing, but was just wondering if anyone with experience had any recommendations. I am currently practicing with Matlab and reading about vibration signal processing, so have those covered, but haven't focused on electric motors specifically too much, so I'll definitely look into them more. Thanks for your help.


On the topic of signal processing, I see a lot of information on different processing methods to use for rolling element bearings like envelope analysis etc. but not much info on diagnostic methods for journal bearings. 

Is simply the frequency spectrum gained through fft and things like orbit plots used to diagnose journal bearing faults? Has anyone used more advanced methods like time-frequency analysis using wavelet transforms? 



Signal processing is in fact a wide subject. Based on my limited experience, your observation is correct;  there is more literature recently on the rolling element bearings compared to fluid film bearings.


With respect to the wavelet analysis, it gained a little interest in the common industry compared to fft. Many practionors do not know what a wavelet is or how it could used in practical machinery vibration analysis. 


Having said that, I would recommend to identify the scope of your research /psper. If you are in a bachelor's program, typically students are not expected to come up with "new knowledge". So, maybe focusing on the technology being used in the industry could be better specially if your internship will be in the industry. For example wavelet and many other topics in the signal processing should have a much lower priority for you if my assumptions are correct. To avoid unnecessary hardship in your paper, it would be useful  if you can know the type of machines (rolling element or journal bearings, size, layout etc) operated in your internship's place. Also, focus your study of signal processing on the system they use (what brand, what model, functions offered etc). Aligning your preperation now with what you will face in the internship would be a high consideration.


Again good luck.

Regards- Ali M. Al-Shurafa 

Sorry, I forgot to mention that these motors are run on hydrodynamic bearings.

At the moment the type of motor (likely smaller induction and permanent magnet motors) and the type of bearings is the only information I have been given (and is likely all I will get). So I have been looking at what types of processing is used for fluid film bearings to practice and learn it for myself without the help of a monitoring software system but as mentioned not as lot has been found. 

I agree that trying to use wavelets is probably a bit too much, a lot of it was going straight over my head.

Small motors (AC or DC) generally do not use hydrodynamic bearings (fluid film), but instead use anti-friction (rolling element) bearings. Fluid film bearings are best monitored with noncontact proximity (inductive or capacitance) probes embedded in the bearing (typically 2 per bearing at 90-degrees orientation). The obvious concern for bearing condition is that shaft/journal relative vibrations do not exceed the bearing clearance, or a wipe-out would occur rather quickly. In this case peak-to-peak vibration displacement is measured. Measurements for rotor dynamics include misalignment (shaft and gears), unbalance, rub, whirl (lube oil and steam), loose parts, and cracked shaft. Vibration measurements include shaft position, waveform, orbit (X Vs Y waveform), phase, spectrum, full spectrum, and Nyquist and Bode plots.

Accelerometers and velocity sensors can be used on the journal bearing housing. Shaft rider stick with velocity or accelerometer sensor were used in the past (safety and access issues) to measure absolute shaft vibrations. Both types of measurements have difficulty with bearing condition detection.

It would be best to start your effort with acquiring specific information about the entire machine train and bearing type.


Thanks Walt.

I won't get anymore information until I get there as its in a different country and run through my uni, so all I can do at the moment is practice and gain general info.

You wouldn't happen to know where I could find some journal bearing proximity probe data to practice the shaft position, orbit plots, spectrum and waveform with would you?



I would agree with the comments from Walt about the expected type of bearing used for classic small motors. So analysts monitoring these machines do not really need deep details about orbit and other plots associated with fluid film bearings.


Nevertheless, a very good reference for fluid film bearings is Orbit magazine that is issued by Bently Nevada  (GE). The archived articles are free to download and very helpful as many of them explain the theory in a simple language but with industrial applications.

Youtube also contains many useful clips.

Texas university symposia are a huge help. They have been recently put on the public domain.

But maybe some of these resources could be too much if you are in the undergraduate level. Some of them are more theory than what a practical person should focus on at the beginning of his/her career. 


These are my only suggestions but I could be totally wrong as in many cases.

Regards- Ali M. Al-Shurafa 

Add Reply