Publikation

Sliding Window Symbolic Regression for Predictive Maintenance using Model Ensembles

Outline:

J. Zenisek, M. Affenzeller, J. Wolfartsberger, M. Silmbroth, C. Sievi, A. Huskic, H. Jodlbauer - Sliding Window Symbolic Regression for Predictive Maintenance using Model Ensembles - Lecture Notes in Computer Science 10671, Las Palmas de Gran Canaria, Spanien, 2017, pp. 8

Abstract:

Predictive Maintenance (PdM) is among the trending topics in the current Industry 4.0 movement and hence, intensively investigated. It aims at sophisticated scheduling of maintenance, mostly in the area of industrial production plants. The idea behind PdM is that, instead of following xed intervals, service actions could be planned based upon the monitored system condition in order to prevent outages, which leads to less redundant maintenance procedures and less necessary overhauls. In this work we will present a method to analyze a continuous stream of data, which describes a system's condition progressively. Therefore, we motivate the employment of symbolic regression ensemble models and introduce a sliding-window based algorithm for their evaluation and the detection of stable and changing system states.