Time Series Data Analysis Technology Based on the Chaos Theory
Takanori Hayashi, Tomoyuki Yomogida · 2016
We developed technology that analyze waveforms of time series data obtained from measurements in plant facilities and also detect waveforms online that are different from ordinary ones. With these technologies, we can numerically grasp changes in trends from past data in order to use the obtained data for fault detection. For the verification of these technologies, we investigated power demand in order to numerically detect changes in trends between weekdays and Saturdays, Sundays, and holidays. Simultaneously, we reconfirmed that we must achieve the theorization of respective data in order to grasp the contents of our investigation. By analyzing a variety of time series data obtained from measurements in actual facilities, we aim to commercialize it so that we can utilize our established technologies for abnormality detection and fault detection technology in the future.