Condition monitoring and temporal‐spatial assessment of composite pipeline transporting potable water

• Published in: Structural control and health monitoring Vol. 29; no. 10
• Main Authors: Wang, Jun‐Fang, Zhang, Lin‐Hao, Lin, Jian‐Fu, Wang, Baoxian, Ni, Yi‐Qing, Ren, Wei‐Xin, Wang, You‐Wu, Xie, Yan‐Long
• Format: Journal Article
• Language: English
• Published: Pavia John Wiley & Sons, Inc 01.10.2022
• Subjects: Algorithms; Bragg gratings; Condition monitoring; condition monitoring and assessment; Data acquisition; Drinking water; fiber Bragg grating; Fiber reinforced plastics; potable‐water composite pipeline; Public concern; random forest; Sensors; Skeletal composites; state classification; Water pipelines
• ISSN: 1545-2255; 1545-2263
• DOI: 10.1002/stc.3020

Summary Water pipeline condition monitoring, especially that of potable water pipeline, has attracted more research interest due to the increasing public concern in pipeline serviceability. This study focuses on the condition monitoring and temporal‐spatial assessment of potable‐water‐filled composite pipeline by using a network of specially designed fiber Bragg grating (FBG) sensors. In responding to the challenges imposed by the operational environment in the pipeline, the sensory network development is guided by application‐specific design with three phases, including suitable sensor development, applicability validation, and implementation. A 200‐m‐long buried fiber reinforced plastic (FRP) pipeline for conveying potable water is instrumented and the in‐service pipeline has been monitored mainly by the properly designed and fixed strain sensors for nearly 2 years, forming a database with its data acquired in different conditions. Furthermore, targeting at the difficulty in the online anomaly identification of longitudinally extended structure, this study proposes a temporal‐spatial condition assessment scheme for identifying the anomalous time slots and locations of pipeline. This scheme consists of an offline modeling and validation process taking advantages of random forest (RF) algorithm for serviceability‐state classification, an online temporal‐spatial assessment process featured by a two‐stage assessment strategy which employs the temporal and then the spatial RF offline models for online state identification, and a reporting mechanism based on the analysis of anomaly types. The performance of the proposed state assessment scheme is examined by using the database, and its effectiveness is demonstrated by the high accordance between the identified results and the real conditions of the pipeline.