openconmo 0.0.4

A Python package for vibration signal based condition monitoring

OpenConMo

Python library for vibration signal–based condition monitoring, developed at Aalto University, Finland.
The objectives of the library are:

1. Provide easy access to reproducing signal based condition monitoring papers
2. Enable comparison of AI/ML based techniques with conventional signal processing tools

Installation

Install the latest version from PyPI:

pip install openconmo

Install and load datasets

If you want to download the original CWRU dataset in .mat format in separate files, you can use the code in tools/CWRU_download.py. Just run the code and the files will be downloaded to data/CWRU/RAW (the directory will be automatically created if missing). This is somewhat slow (~7 min).

The same code will create a CWRU.feather file, which holds the data in a format more easily read with python & pandas. The dataframe is formatted as follows:

measurement location	fault location	fault type	fault depth (mil)	fault orientation	sampling rate (kHz)	torque (hp)	tags	measurements
string	string	string	int	string	int	int	list[string]	np.array[np.float64]
DE/FE	DE/FE	OR/IR/B	0/7/14/21/28	C/OR/OP	12/48	0/1/2/3	see below	measurement samples

mil = 0.001 inches

Shorthand explanations: DE - drive end FE - fan end OR (fault type) - outer ring IR - inner ring B - ball / rolling element C - center () OR (orientation) - orthogonal OP - opposite

Possible tags (from Rolling element bearing diagnostics using the Case Western Reserve University data: A benchmark study): electric noise - measurement is has patches corrupted by electric noise clipped - measurement is clipped identical_DE_and_FE - measurements from DE and FE sensors are identical except with a scaling factor

Run notebooks

• Notebook 1: "smith_randall_example.ipynb", reproducing results of "Rolling element bearing diagnostics using the Case Western Reserve University data: A benchmark study" by Smith & Randall
• Notebook 2: "matlab_example.ipynb", reproducing results of "Rolling Element Bearing Fault Diagnosis"

Documentation

See openconmo documentation for documentation.

Requirements

• Numpy
• Pandas
• Scipy
• PyArrow (for loading datasets)

Authors

This software is authored and maintained by Sampo Laine, Sampo Haikonen, Aleksanteri Hämäläinen and Elmo Laine, Mechatronics research group, Aalto University. Please email questions to [e-mail]