Corrosion detection in pipes by piezoelectric sensors using Artificial Neural Network

Rafezi, H ; Sharif University of Technology | 2012

835 Viewed
  1. Type of Document: Article
  2. DOI: 10.4028/www.scientific.net/AMR.403-408.748
  3. Publisher: 2012
  4. Abstract:
  5. Defect detection in pipes is an essential task specially for sensitive applications such as oil and gas industry where special cares are required. Corrosion is a common defect in pipes which has attracted attention of researchers. In present work a non-destructive methodology for pipe corrosion monitoring is introduced. Polymer of Vinylidene Fluoride (PVDF) Piezoelectric is used as the sensor to measure strain variations affected by internal corrosion. High sensitivity and low cost of piezoelectric materials made them a good candidate for precise industrial applications. Different corrosion conditions (i.e. corrosion location along pipe and corrosion depth) are modeled and sensors voltages in different corrosion conditions are simulated. Finally in order to develop an effective corrosion detection system, an Artificial Neural Network (ANN) is designed to recognize position and amount of corrosion according to sensors voltages. The ANN performed corrosion condition recognition with 91 % of accuracy. This method provides the capability of online implementation for continuous maintenance of pipelines
  6. Keywords:
  7. Artificial Neural Network ; Condition recognition ; Continuous maintenance ; Corrosion depth ; Corrosion detection ; Defect detection ; High sensitivity ; In-pipe ; Internal corrosion ; Low costs ; Non destructive ; Oil and Gas Industry ; Online implementation ; Piezoelectric ; Piezoelectric sensors ; Pipe corrosion ; Sensitive application ; Strain variation ; Vinylidene fluoride ; Condition monitoring ; Corrosion ; Defects ; Gas industry ; Industrial applications ; Piezoelectricity ; Pipe ; Sensors ; Neural networks
  8. Source: Advanced Materials Research, 4 November 2011 through 6 November 2011 ; Volume 403-408 , November , 2012 , Pages 748-752 ; 10226680 (ISSN) ; 9783037853122 (ISBN)
  9. URL: http://www.scientific.net/AMR.403-408.748