Intelligent predictive maintenance system

Abstract

The machine learning techniques can be efficiently used for optimal maintenance decision making. Currently, most of the companies and manufactures possess huge amounts of sensor, process, and environment data. Combining the data with the information about the failures succeeds in creating useful train data sets for predictive maintenance purposes. In this paper, we propose the approach of efficient data processing in order to maximize the predictive quality of machine learning models. We investigate numerous machine-learning methods and propose the procedure to automatize the predictive maintenance process. The results obtained for the real data were satisfactory and applicable.