Dataset of Vibration, Temperature, and Speed Measurements for Spherical Roller Bearings with Single, Dual, and Multi-Bearing Defects Under Variable Loads and Speeds

Description

This dataset has been created by the ISED (Industrial Systems Engineering and Design) research group at Politecnico di Torino to extend the available data for the development of fault detection models and predictive maintenance strategies for medium/large-sized spherical roller bearings commonly used in industrial applications, by resorting to a test rig with spherical roller bearings simultaneously monitored..

Building upon the previously released dataset focused on single localized defects, this new dataset introduces multiple defect conditions, including:

Bearings with two defects on the inner race.

Bearings with simultaneously defected in the test rig.

These additions allow for a more comprehensive evaluation of diagnostic models under more complex and realistic fault scenarios.

Experimental Setup

The data were collected using SKF 22240 CCK/W33 spherical roller bearings on the medium/large-scale bearing test rig developed by the ISED research group at Politecnico di Torino. The test rig is capable of independently applying both axial and radial loads to the bearings and can test bearings with an outer diameter of up to 420 mm. Tests are conducted at 10 nominal spin speeds and 4 load conditions (including one with axial load).

Dataset Structure

The dataset is organized into three main folders, each corresponding to a different fault configuration:

InnerRaceDualDamage – The fourth bearing (sensor 4) has two defects on the inner race, positioned 180° apart:

One defect: 2 mm diameter, 0.5 mm depth.

Second defect: 1 mm diameter, 0.5 mm depth.

OuterRaceDamage + InnerRaceDualDamage – The test rig contains:

A bearing at sensor 3 with an outer race defect (2 mm diameter, 0.5 mm depth), positioned at 0° with respect to the radial load.

A bearing at sensor 4 with the same dual inner race damage as in the first condition.

OuterRaceDamage + RollerDamage – The test rig contains:

A bearing at sensor 3 with an outer race defect (2 mm diameter, 0.5 mm depth) at 0° with respect to the radial load.

A bearing at sensor 4 with a rolling element defect (2 mm diameter, 0.5 mm depth).

File Naming Convention

Each test file is stored as a .mat file with the following naming format:

(Nominal_Rotation_Speed)rpm_(Radial_Force)kN_(Axial_Force)kN.mat

Where:

Nominal_Rotation_Speed refers to the machine's nominal rotational speed.

Radial_Force represents the applied radial force.

Axial_Force represents the applied axial force.

For some tests, “ramp” files are included, containing data where the rotational speed was varied linearly during the test.

Signal Structure and Sensor Data

Each .mat file contains multiple structures, depending on the type of test. Each structure is labeled as Signal_ followed by a number (0, 1, 2, 3, 4), where each represents a specific signal extracted during the test. There is no fixed correspondence between the signal number and the type of measurement. For example, Signal_2 does not always represent the accelerometer signal. Users are encouraged to inspect the y_values.quantity field to identify the signal's unit and nature. For instance, if y_values.quantity.label shows "g", the signal corresponds to an accelerometric measurement.

All signals have been exported using the MKS system, so y_values.values contains data in units of m/s² for acceleration signals. To convert the values to the unit indicated in y_values.quantity.label, users can apply the multiplication factor and offset provided in y_values.quantity.unit_transformation. In the case of accelerometric data, the multiplication factor is 0.1020, converting y_values.values from m/s² to g.

A more detailed description of the data structure can be found in the Test.Lab documentation.

In addition to acceleration data, the files include temperature signals, tacho sensor signals measuring shaft angular speed, and tacho impulse signals. In some cases, there is also a signal measured in "N", representing the frictional force generated by the bearings (as detailed in this Paper). This friction force signal is not always present, as the load cell could go into overload during certain tests.

Sensor Data Organization:

Acceleration and temperature data are presented in tables with four columns, each corresponding to one of the four bearings. In the damaged condition tests, the damaged bearings correspond to sensor 3 and 4 (i.e., y_values.values(:, 3), y_values.values(:, 4)).

Usage and Applications

This dataset can be used for:

Developing and benchmarking fault detection algorithms for industrial bearings.

Studying the effects of multiple damage interactions on vibration signals.

Testing machine learning and deep learning models for predictive maintenance.

Exploring transfer learning techniques in industrial applications.

A detailed analysis of the dataset and test conditions can be found in related publications by the ISED Research Group at Politecnico di Torino.