RSW gun fault prediction benchmark data set (demo)

<p>The resistance spot welding (RSW) welding gun fault prediction benchmark data set has 72 multivariate time series in the training set and 8 in the testing set. Each time series length 604800 sampled at 1 Hz with missing values and has 20 dimensions (c1-c19 and the error code). We retain the missing value and the outliers of the welding gun time series for the potential of imputation research in the future.<br> This data set supports an academic paper named &#39;benchmark study for welding gun fault prediction&#39;.</p> <p><strong>Feature name and explanation:</strong></p> <p>c1 : &nbsp;Electrode cap offset;</p> <p>c2 : &nbsp;Electrode force;</p> <p>c3 : &nbsp;Electrode position;</p> <p>c4 : &nbsp; Force build-up;</p> <p>c5 : &nbsp;Balance pressure;</p> <p>c6 : &nbsp;Friction;</p> <p>c7 : &nbsp;Maximum aperture;</p> <p>c8 : &nbsp;Maximum electrode force;</p> <p>c9 : &nbsp;Mtart friction;</p> <p>c10 : &nbsp;US2;</p> <p>c11 : &nbsp;Welding point count;</p> <p>c12 : &nbsp;Position count;</p> <p>c13: &nbsp;Setpoints of counterbalance pressure;</p> <p>c14: &nbsp;Setpoints of electrode force;</p> <p>c15 : &nbsp;Setpoints of electrode position;</p> <p>c16: &nbsp;Setpoints of sheet thickness;</p> <p>c17 : &nbsp;Setpoints of velocity;<br> c18: &nbsp;Setpoints of force build-up;<br> c19 : &nbsp;Offset value in robot.</p> <p><strong>Machine Learning Task:</strong><br> This dataset is suitable for a time series forecasting&nbsp;task, where machine learning models can be trained to predict future welding parameters based on the provided welding&nbsp;parameters time series in history.&nbsp;</p> <p><strong>Code for quick start:</strong></p> <p><a href="https://zenodo.org/record/7655025">https://zenodo.org/record/7655025</a></p> <p>If you want to have an overview of the data before downloading all of it, you can download only the files with the word &quot;Damo&quot; in the file name.</p> <p>For any question, please contact 1910633@stu.neu.edu.cn</p>