Research Repository

Classifying structural condition of deteriorating stormwater pipes using support vector machine

Tran, Huu Dung and Ng, A. W. M (2010) Classifying structural condition of deteriorating stormwater pipes using support vector machine. In: Pipelines 2010 : Climbing New Peaks to Infrastructure Reliability--Renew, Rehab, and Reinvest. ASCE, Reston, Virginia US, pp. 857-866.

Full text for this resource is not available from the Research Repository.

Abstract

Stormwater pipe networks are an important part of the water infrastructure that removes stormwater runoff from urban cities of Australia. This paper presents the attempts made to improve the classification or prediction of structural condition for individual stormwater pipes. Such predictive information can be used to prioritize pipes for inspection and subsequent maintenance actions and thus support water utilities implementing their proactive management strategies. In this paper, a support vector machine (SVM) was developed, tested and compared against the back‐propagation neural network (BPNN) model which has been used in a previous study. The developed model was applied to a case study using a sample of CCTV inspected pipes and corresponding pipe factors. The results showed that with regards to correctly predicting structural condition of individual stormwater pipes, the SVM model outperformed the BPNN significantly in the train dataset and marginally in the test dataset. Several advantages of the SVM model were also found compared to the BPNN model.

Item Type: Book Section
ISBN: 9780784411384
Additional Information:

ASCE conference proceedings

Uncontrolled Keywords: ResPubID19508, deterioration, pipes, stormwater management, classification, support vector machine, neural network, prediction
Subjects: FOR Classification > 0905 Civil Engineering
Faculty/School/Research Centre/Department > School of Engineering and Science
Depositing User: VUIR
Date Deposited: 28 Feb 2013 00:52
Last Modified: 20 May 2014 06:20
URI: http://vuir.vu.edu.au/id/eprint/6770
DOI:
ePrint Statistics: View download statistics for this item
Citations in Scopus: 0 - View on Scopus

Repository staff only

View Item View Item

Search Google Scholar