Hello Muhd,
I understand that you want to know whether we should extract several features before using the Wrapper method for feature selection or can it be directly used on the vibrational signal dataset.
In the case of vibration signals, you will need to extract relevant features from the signals before applying the wrapper method. By doing so, you can represent the vibration signals in a more compact and meaningful way, which can then be used for further analysis and classification.
There are several techniques you can use to extract features from vibration signals for bearing condition analysis. Some commonly used features include
- statistical features (e.g., mean, standard deviation)
- time-domain features (e.g., root mean square, crest factor)
- frequency-domain features (e.g., spectral entropy, peak frequency)
- time-frequency features (e.g., wavelet coefficients)
Once you have extracted the relevant features from the vibration signals, you can later apply the wrapper method to select the most informative subset of features.
Here are some helpful references for more complex signal features:
- Signal Features - MATLAB & Simulink - MathWorks India
- Feature Extraction Using Signal Labeler - MATLAB & Simulink - MathWorks India
Hope this helps!
