Elbow point-based adaptive signal segmentation (EPASS) for aerial ropeway transportation systems

Abstract Aerial ropeway transportation (ART) systems operate at high altitudes to transport passengers and goods, where even minor malfunctions can lead to severe casualties. However, vibration signals acquired from ART systems are heavily contaminated by irregular, non-periodic shock waves caused by variations in gondola load, wind speed, ambient temperature, and transient operational control, which fundamentally degrade the reliability of conventional anomaly detection algorithms. To address this challenge, this paper focuses on separating non-shock wave intervals from vibration signals dominated by irregular and non-periodic shock waves. An Elbow Point-based Adaptive Signal Segmentation (EPASS) method is proposed to autonomously isolate non-shock wave intervals. EPASS consists of three key steps: (1) preprocessing envelope signals using a sliding window to characterize energy distributions, (2) applying an elbow point-based adaptive threshold to separate shock wave and non-shock wave signals without relying on fixed criteria, and (3) validating the proposed method using real vibration data collected from an in-service ART system. Experimental results demonstrate that EPASS extracts non-shock wave intervals more accurately than conventional signal segmentation techniques across varying window sizes. Furthermore, anomaly detection models trained on EPASS-segmented signals achieve significantly higher detection accuracy than those trained on raw vibration signals. These results confirm that EPASS effectively enables reliable anomaly detection in ART systems by mitigating the influence of irregular shock waves.