Data-based quality analysis in machining production: Influence of data pre-processing on the results of machine learning models

Quality assurance as a non-value-adding process is constantly reviewed for cost optimisation and potential savings. In the pursuit of utilising advanced data analysis and machine learning methods to improve efficiency of quality assurance in machining processes there are several influencing factors severely impacting the performance and hence the value of said methods. Especially data preparation is a time consuming task requiring both domain and data expert knowledge and yielding various options for data preparation. In this paper, the impact of different input data sets for predicting part quality in a drilling process is investigated, using machine control data.