Sensor fault diagnosis based on a new method of feature extraction in time-series
Du Jingyi, Lu Wang · 2010
This paper presents a new method of how to choose the key points of monotone sequences based on the basic theory of time series segmentation algorithm, which is to select the key points from monotone sequences by calculating the curvatures. With such method, time series can be well linear-fitted. This method is also used for fault diagnosis of sensor. Key point sequence of the maximum difference can be achieved by comparisons among different time series of sensors, thus the fault sensor can be determined.