UROP Project

Implementation of Machine Learning Algorithms to compensate for Volumetric Errors of Machine Kinematics

Excellence Cluster: Internet of Production

Instead of static compensation data sets, which lose their validity under changing environmental conditions, online-retrievable information can be used as input data for machine learning algorithms to predict the volumetric error machine kinematics. This information can be collected, for example, by temperature, force and vibration sensors as well as optical displacement measuring systems or control-internal data sources and thus does not become invalid. The self-learning algorithm can recognize such complex interactions independently, the regarded machine can be specified as a black box or gray box. The aim is that the developed algorithm can be applied equally to different machine types in different environments. The auxiliary scientists have access to various experimental machines and state-of-the-art measuring technology on a well-structured 600 sq.m. laboratory space

Task

- Implementation of Machine Learning Algorithms - Experimental Tests/Training phase of neuronal network - Validation of test - Working together with other scientific researchers

Requirements

Advanced knowledge of Python programming as well as basic machine tools and G-code programming skills would be helpful. Ability to work in team and independently.