A Preliminary Method of Abnormal Pattern Recognition (APR) Based on Adaptive Slide Window (ASW) and Template Matching(TM)

Due to the high complexity, large volume, and variability of big data, traditional data detection methods are no longer suitable. Aiming at these characteristics, this paper proposes a preliminary method of APR, which uses ASW and TM to detect change points. This method uses the combination of the TSTKS algorithm and ASW to realize rapid detection of multiple change-points in time series data, extract continuous multi-window fluctuation characteristics, construct a normalized fluctuation vector of time series data, and perform abnormal state detection on large-scale disease signals and analysis. Experimental results such as simulation data and EEG signal analysis show that this APR has the advantages of short specific detection time and higher accuracy rate, and is suitable for the detection and analysis applied to large data.