Incipient anomaly detection for railway vehicle door system based on adaptive mean shift clustering
Geun-Seok Han, Yunshi Zhang, Ningyun Lu, Bin Jiang, Zhigang Xu, Jingzhi Cao, Xiaohu Shi · 2017
Reliability of door system is directly related to the operation safety of railway vehicles. However, door system is with high failure rate. Accurate detection of the early-stage incipient anomalies is an urgent need in the field of railway transportation. This paper proposes an Adaptive Mean Shift Clustering (AMSC) algorithm, which can successfully detect and isolate common and frequent anomalies in railway vehicle door systems. At the same time, the proposed AMSC algorithm has superior ability for outlier rejection, hence makes it suitable for practical applications. Comparative experiments show that, the proposed AMSC based incipient anomaly detection method outperforms the density-based spatial clustering of applications with noise (DBSCAN) based method for railway vehicle door systems.