A Study on Classification of Data Pattern for Establishing Anomaly Prediction Method of Machine Tools
Ryo Fujita, Takashi Yoshimi · The Proceedings of JSME annual Conference on Robotics and Mechatronics (Robomec) · 2017
This research aim to establish an anomaly prediction method based only on sensor data acquired from machine tools. In order to realize this aim, first of all it is necessary to classify whether the acquired data is normal or anomaly. This paper show the result of examining the method using distribution and chaos theory as classification method of acquired sensor data. In the distribution method, comparing the anomaly state with the normal state by creating a frequency distribution diagram using the heat map. In chaos theory, comparing the anomaly state with the normal state by attractors reconstructed using Tarkens' embedding theorem. In addition, the result of attempted quantification using qualitative evaluation of chaos theory using topological geometry is shown.