TSPD: A Robust Online Time Series Two-Stage Peak Detection Algorithm

Peak detection (PD) is a valuable technique in time series analysis that can be employed to identify significant local maxima or minima within the time series. Peaks are significant points in a signal or spectrum that are of interest, as they can represent important features in various applications such as chromatography, spectroscopy, biomedical signal processing, and image analysis, to name a few. However, due to the unpredictable and fluctuating nature of many real-time sources, current peak detection algorithms usually fail to detect stable peaks across varied real-time time series. Furthermore, they experience a dip in performance when trying to develop a general algorithm applicable across various application areas. To address this challenge, we have developed a novel and robust online two-stage peak detection algorithm (TSPD). Our TSPD algorithm can adapt to a wide range of application areas by adjusting only two tunable parameters in its computational process. To demonstrate the efficacy of our algorithm, we conducted experiments using seven real-world datasets and demonstrated the improved performance of the TSPD algorithm over other notable existing techniques.