Neural networks based structural condition assessment of sewer pipelines

Z. Khan, Tarek Zayed, O. Moselhi

Research output: Chapter in book / Conference proceedingConference article published in proceeding or bookAcademic researchpeer-review


There is need to enhance our understanding of the behavior of various infrastructure networks and their components when subjected to different sets of conditions. This study uses artificial neural networks, to investigate the importance and influence of certain attributes of sewer pipes, upon their structural performance. Data on six parameters related to sewer pipeline including: pipe length, diameter, depth/cover, pipe material, bedding condition, pipe age and closed circuit television based pipe condition rating, were obtained from the municipality of Pierrefonds Quebec. Back propagation and Probabilistic neural network models were developed and validated. These models were used to rank the parameters, in the order of their influence on pipe condition. Sensitivity analysis was carried out to simulate the structural condition of pipe at a range of values of each of the above parameter. Results of sensitivity analysis describe the nature and extent of the influence of each parameter on pipe structural condition.
Original languageEnglish
Title of host publicationCanadian Society for Civil Engineering Annual Conference 2009
Number of pages10
Publication statusPublished - 1 Dec 2009
Externally publishedYes
EventCanadian Society for Civil Engineering Annual Conference 2009 - St. Johns, NL, Canada
Duration: 27 May 200930 May 2009


ConferenceCanadian Society for Civil Engineering Annual Conference 2009
CitySt. Johns, NL

ASJC Scopus subject areas

  • Engineering(all)

Cite this