Acceleration Datasets

Acceleration datasets for anomaly detection and predictive diagnosis in industrial automation.

Asset Description

The asset consists of datasets containing accelerations (along three axes) measured by means of a sensor board while the AI-REDGIO 5.0 E2mech experiment was running.

Asset Details

Dataset Information

Each dataset is represented in a csv file, where each column contains the acceleration measured for a time horizon equal to 4 times the mechanism's rotating period. Also, this data corresponds to a given angular speed and a healthy or faulty condition. The healthy and faulty information is included (in two separate datasets). This corresponds to different control algorithms adopted to steer the experiment mechanism. Conditions corresponding to the mechanism running at different angular speeds are provided (each with its own healthy and faulty scenario). All the information concerning the axis along which the measurement has been taken, the system’s angular speed, and the healthy or faulty situation is provided in the file name.

Usage

We provide a set of files concerning a pretty specific system’s condition (in terms of speed, measurements, healthiness), so that the interested user can either perform analysis, grouping them as needed (e.g., by selecting only a given acceleration axis or a system’s speed) and without the need of downloading a huge file (e.g. in json format) to be then parsed and split according to the desired work of analysis. In this respect, given the description of the data above, it should be pretty straightforward to import the files in one’s preferred data analytics tool (python, Matlab, R, etc…) and test condition monitoring and anomaly detection algorithms.

Maturity

The datasets concerning the 2nd iteration of the E2Mech experiment in the AI-REDGIO 5.0 project have been completed in September 2025.

Licence

Open source

Resources

• Datasets available in GitHub
• Provided by ACTEMA

Acknowledgement

This work was funded by AI REDGIO 5.0 (101092069). The dataset has been created in the context of the AI REDGIO 5.0 E2Mech Experiment.