Analysis of Acoustic Guided Wave Signal Transmission by Use of Empirical Mode Decomposition and Low Amplitude Mode Mixing Separation for Optimal Time-of-Flight Detection
Abstract
The use of ultrasonic sensor technology is rapidly growing in a wide range of industrial applications from medical diagnostic systems to process and safety measurement in the oil and gas industry. A typical trend in ultrasonic sensor development is increased measurement accuracy for increased measurement reliability, as such sensors become vital parts of autonomous decision systems and process. Adaptive signal processing is a significant contribution to such accuracy improvement strategy. In this thesis, the focus has been on analysis, hence compensation of non-stationary and nonlinear signal effects in guided wave ultrasound signal transmission. XSENS AS, a company developing ultrasonic flow meters for installation on pipeline surface, have made real transmission data available for analysis and optimisation of time-of-flight (ToF) detection in acoustic guided wave transmission. A signal-processing method for reduction of measurement uncertainty, hence more accurate measurement of time-of-flight of ultrasonic signal pulses, is proposed. For test and validation, XSENS raw data is used.
Hilbert Huang Transform (HHT) has been the basis for the study and Empirical Mode Decomposition (EMD) together with the proposed Low Amplitude Mode Mixing Separation (LAMMS) method forms the basis for adaptive signal analysis. Ultrasonic signal pulses may by interference of noise/disturbances and installation variations become nonlinear and non-stationary, hence adding uncertainty onto the ∆t detection through zero crossing detection. Such effects should therefore be quantified for optimum correction of enhancement of signal excitation for minimising ∆t uncertainty. In this thesis ∆t is measured through zero crossing detection, in combination with HHT used for signal filtering. LAMMS method achieves mode mixing control.
The combined EMD and LAMMS analysis method has been thoroughly tested at data sets containing different excitation frequencies as well as time-of-flight signals in both up-stream and down-stream direction (referring to flow meter pipe fluid direction), for validation of the proposed method.
By the use of Hilbert transform and associated phase angle results and envelope generation, results from the method proved the method to be a viable option for signal reconstruction and zero crossing correction, applicable to optimisation of acoustic guided wave time-of-flight detection.