Processed subset used in our STR-DDPM experiments on the University of Ottawa Electric Motor Dataset
Description
This repository provides the processed experimental subset used in our paper, “STR-DDPM: Residual-domain diffusion-based data augmentation with seasonal-trend-residual decomposition for few-shot motor fault diagnosis”, derived from the publicly available University of Ottawa Electric Motor Dataset. Only the subset and operating conditions used in this study are included here. The files were reorganized and processed for our experimental pipeline. This record is intended to support reproducibility of the associated paper and does not replace the complete original public dataset. The original University of Ottawa Electric Motor Dataset is publicly available from its official data record. According to the original dataset page, the data contain vibration, acoustic, and temperature measurements collected from 8 different D396 Marathon Electric 3-phase motors, with raw data provided in CSV and MATLAB formats, and the dataset is distributed under the CC BY 4.0 license. Users should cite the original dataset and acknowledge the original source when reusing this processed subset.
Files
Steps to reproduce
1. Use the processed files provided in this record. 2. Follow the data organization and class labels described in the associated paper. 3. Apply the preprocessing, segmentation, and split settings reported in the manuscript. 4. Train and evaluate the model using the implementation details and hyperparameters described in the paper.