The fourth industrial revolution is accompanied by the actuation and sensorization of production processes, giving engineers access to a variety of data. Data contains hidden, usually nonlinear correlations to process states, product qualities, and optimal control variables, knowledge of which can be used to derive valuable decisions. In most cases, this requires the application of machine learning and deep learning techniques, whose performance is significantly determined by the collected data quality, preparation and transformation.
In order to provide engineers of the future with competencies in dealing with high-dimensional data sets, the PtU is offering master students the tutorial since last semester, which focuses in particular on the processing of time series and image data using supervised learning algorithms. Recently, the first student group successfully completed the tutorial and demonstrated that acceleration signals from a blanking process contain information about the wear condition of the punch. We are pleased about the huge interest in the new course and look forward to future tutorial runs! “Machine Learning in Forming Technology”