A Matching Pursuit Approach for Ultrasonic Signal Enhancement and Characterizing Corrosion Damage of Buried Anchor Rods, 18-R9845

Printer Friendly Version

Principal Investigator
Adam C. Cobb

Inclusive Dates:  07/28/08 – 11/28/08

Background - In the United States, there are thousands of miles of aboveground, high-voltage transmission lines supported by tower structures. Typically, these structures are held in place with several guy wires attached to anchor rods, all of which are susceptible to corrosion. It is prohibitively expensive and impractical to disassemble the structure for inspection, necessitating the development of in-service inspection methodologies for ensuring the safety of the structure. Prior work at SwRI has developed the use of Magnetostrictive Sensing (MsS®) generated guided waves for anchor rod inspection. In an undamaged anchor rod, the ultrasonic energy propagates along its length before reflecting back from the end. In the damaged cases, the corrosion corresponds to a localized volumetric change, effectively resulting in an acoustic impedance change in the structure. This impedance change produces ultrasonic reflections prior to the energy reaching the end of the rod.

As currently implemented, there are several limitations of the existing inspection. First, the ultrasonic waves propagate along the anchor rod and interact with the surrounding material, such as concrete. These interactions yield a decrease in the apparent signal-to-noise ratio (SNR), effectively defining the minimum defect size able to be characterized. Furthermore, given the inherent complexity of ultrasonic waves, proprietary data analysis software developed by another organization is often required to interpret the waveforms for discriminating corrosion responses, and the currently reported results are qualitative in nature (i.e., good anchor rod, moderate corrosion, or excessive corrosion). These primary limitations combine to reduce both the efficiency and usefulness of in-service inspection of anchor rods.

Approach - An initial signal analysis strategy was proposed for quantifying the extent of corrosion damage based on changes in MsS-generated ultrasonic responses. The proposed signal processing approach combines a time-frequency decomposition, in particular the matching pursuits algorithm, with a model of the guided wave response. Matching pursuits is a signal analysis method for decomposing waveforms into a linear combination of basis functions known as decomposition atoms. By choosing these atoms to correspond to damage alone, the result will include only the effects related to damage, providing large improvement in the SNR. In previous strategies using matching pursuit for ultrasonic signal analysis, the decomposition atoms were chosen to closely match waveform responses without correlation to underlying physical mechanisms. For this study, a model of the guided wave responses as a function of volumetric loss was employed. Thus, the matching pursuit decomposition not only provides improved SNR, but also localizes and characterizes the damage directly from the decomposition parameters.

Accomplishments - A number of experimental mock-ups of anchor rods with different types of symmetric, corrosion-type damage were analyzed using this signal processing strategy. The approach was to use repeat applications of a matching pursuit-based algorithm for iteratively identifying arrivals. The dictionary for decomposing the waveforms was created by modeling the expected ultrasonic reflection and transmission coefficients. The primary conclusions and contributions were:

  1. Model-based basis functions allow for an efficient decomposition of ultrasonic waveforms, providing both improvements in the apparent SNR as well as excellent characterization.
  2. A simple modification of a prior guided-wave model to allow for computation of transmission coefficients along with the existing reflection coefficient.
  3. A methodology for combining the positive- and negative-directed ultrasonic responses in finite length structures to identify arrivals of interest while suppressing the unwanted information.

Initial work for automating the signal analysis was completed, enabling the MsS technology to be more successfully used as an inspection tool. Thus, the urgent need for signal processing methodology has been met.

2009 Program Home