摘要：In this contribution, a data-driven approach towards the prediction of maintenance for the critical component of an injection molding machine is presented. We present our path from exploring and cleaning the data towards the implementation of a prediction algorithm based on kernel density estimation. We give first analytical evidence of the algorithms potential. Moreover, we compare the approach described here with our previous work where we went a model-based approach and present advantages and disadvantages of the two approaches. We try to contribute to a non-comprehensive guide on the implementation of predictive maintenance systems for industrial mass production facilities.
