Fully labeled dataset for evaluating condition monitoring and other manufacturing related tasks, based on electrical signal analysis.
The dataset contains continuous voltage and current measurements at high sampling rates of two industry-grade coffeemakers, mimicking industrial processes.
The coffeemakers closely resemble industrial machinery and can be used to develop and evaluate manufacturing related algorithms, such as for condition monitoring, event detection, etc. .
We provide the data sampled at 6.4 kSps and additional event information: 370600 expert-labeled component-level electrical events, 1735 machine-generated product events and 3646 machine-generated maintenance-related events.
This repository contains Utility classes and functions for the CREAM Dataset and the corresponding data descriptor.
Folder structure:
- data_collection: Data collection scripts of the MEDAL data acquisition device. Original scripts can be found in the BLOND dataset by Thomas Kriechbaumer.
- data_utility: Utility class for loading and processing the CREAM dataset
- labeling_tools: Jupyter notebook based tools that were used to label CREAM
- manuscript: Scripts for creating the plots in the data descriptor
- technical_validation: Scripts used to perform the technical validation of the data and the labels.
- requirements.txt: Requirements necessary to execute the code in this repository.