NEWS 28 February 2023
A mechatronic Machine Learning case
To date, Machine Learning (ML) has found only relatively limited implementation in high-end mechatronic systems. To test its capabilities in a real industrial application, the challenging case of friction estimation was investigated.
A model was developed for predicting frictional properties of linear ball bearings, for very small displacements, as explained in this Mikroniek article. The resulting ML model performed well during training and validation, but rather less so in stand-alone operation. ML is a promising tool for friction estimation, but clearly there is room for improvement in algorithm development. (Image courtesy of Jorn Veenendaal)
