A new fuzzy c-means clustering-based time series segmentation approach and its application on tunnel boring machine analysis. (1st November 2019)

• Record Type: Journal Article
• Authors: Song, Xueguan; Shi, Maolin; Wu, Jianguo; Sun, Wei
• Is Part Of: Mechanical systems and signal processing. Volume 133(2019)
• Publication Date: 2019-11-01
• Subjects: Tunnel boring machine -- Time series segmentation -- Fuzzy c-means clustering -- Prior information; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2019.106279
• Languages: English
• ISSNs: 0888-3270

Abstract

Highlights: A fuzzy c-means algorithm-based time series segmentation approach is proposed. A decision making method is proposed to determine the appropriate number of segments. The proposed approach has promising applications to TBM time series. Abstract: Tunnel boring machine (TBM) is a complex engineering system widely used for tunnel construction. In recent years, massive in-situ time series data of TBM has been recorded, which can provide important references and useful information for TBM designers and operators. In this work, a new fuzzy c-means clustering-based time series segmentation approach is proposed for TBM time series data, where the prior information of attributes is incorporated to facilitate effective segmentation. In this approach, the segmentation objective function is formed by multiplying the time distance and the spatial distance between data. The prior information, i.e. the torque of cutterhead, is correlated with the penetration rate, is described by a linear model and included in the part of spatial distance between data. A new decision making method based on the distance between the joint segment prototypes is proposed to determine the appropriate number of segments. The application on TBM time series data from a tunnel in China shows that the proposed approach can accurately identify different excavation status of the TBM, and help the other data mining tasks of TBM as well. The proposed approach also has promising applications to other complex engineering systems.