Skip to main navigation Skip to search Skip to main content

Techniques avancées de traitement de signaux d’émissions acoustiques pour améliorer la détection de défauts dans des vaisseaux sous pression

Translated title of the thesis: Advanced signal processing techniques to improve detection of defects in pressure vessels by acoustic emission testing
  • Konstantin Polyakov

Student thesis: Master's thesisMaster in Engineering: Mechanical Engineering

Abstract

Acoustic emission testing is a well-known and versatile inspection method for structural health monitoring and non-destructive evaluation of several components found in the petrochemical industry, such as pressure vessels, pipelines, and storage tanks. The sensitivity and reliability of the method is affected by spurious signals (background noise) from technological processes or the environment. The project objective concerned the study of the state of development of method and the advancements of signal processing aimed at improving the quality and reliability of control under industrial conditions, and also evaluation and comparison of the efficiency of four filtration techniques. The experimental work aimed first at recording acoustic emission signals with the presence of natural background noise. The test bench was composed of a system of pipes of diameters up to 6 inches with the system of controlled circulation of liquid. The other type of noise source, friction, was created with the application of the rotating abrasive disc. During the tests 10 signals from Hsu-Neilsen source were recorded with two types of background noise and saved in two files. Signal acquisition was performed with a piezoelectric sensor, a preamplifier and an oscilloscope connected to the computer. A recording application was developed in the LabVIEW graphical programming environment and contained basic acoustic emission system functions with continuous signal recording to a non-threshold file for further processing. Signal processing was performed in the application also developed in LabVIEW and included finite impulse response (FIR) adaptive filter techniques with the recursive least squares (RLS) algorithm and least squares means (LMS) algorithms, the time-varying filter (TVF) based on the Gabor transform and the wavelet threshold filter. The effectiveness of these techniques was evaluated by filtration run time, number of signals detected and signal-to-noise ratio of output signal.
Date24 Jan 2022
Original languageFrench
Awarding Institution
  • École de technologie supérieure
SupervisorMartin Viens (Supervisor) & Xavier Maldague (Co-supervisor)

Cite this

'