Patent No. 6581014 Apparatus and method for analysis of guided ultrasonic waves

 

Patent No. 6581014

Apparatus and method for analysis of guided ultrasonic waves (Sills, et al., Jun 17, 2003)

Abstract

Method and apparatus for using a guided wave to determine the location of one or more flaws in an inspected object are disclosed. An ultrasonic guided wave is launched into the object using conventional ultrasound methods, and the reflected/received guided wave is sampled to capture a series of individual reflected waveforms. The individual reflected waveforms are then partitioned according to the sampling time. Each of the partitioned acquired waveforms is compared with a selected time-varying dispersion-modeled reference waveform associated with the unique geometry of the inspected object, a multiplicity of "theoretical" flaw locations, and the characteristics of the original ultrasonic guided wave. To make a comparison, the reference waveform is also generated as a series of partitioned waveforms which model the shape of a wave that may be expected to be reflected from a multiplicity of theoretical flaws located in the object. The shape of the reflected waveforms is correlated with the shape of the reference waveforms, and a high level of correlation indicates the presence of a real flaw at the theoretical flaw location.

Notes:

This application claims the benefit of U.S. Provisional Application for Patent No. 60/219,762 titled "Time-Varying Matched Filter Method for Analysis of Guided Ultrasonic Waves" filed on Jul. 20, 2000, is related thereto, is commonly assigned therewith, and the subject matter thereof is incorporated herein by reference in its entirety.

BACKGROUND OF THE INVENTION

1. Technical Field

This invention relates generally to nondestructive testing and, more particularly, to a method and apparatus for analyzing guided waves in inspected objects using time-varying matched filters.

2. History of Related Art

Ultrasonic wave inspection techniques are useful for many Non-Destructive Evaluation (NDE) applications. These techniques typically involve transmitting a narrow band ultrasonic frequency interrogation signal down the length of an object and analyzing the reflected or "inspection" signal for the presence of material boundaries or flaws (e.g., surfaces, joints, welds, cracks, etc.) in the object. Defects in the object that cannot be seen by visual inspection can often be detected by analyzing the inspection signal. Thus, ultrasonic wave inspection techniques can provide a cost effective solution for detecting defects in many objects such as railroad rails, stranded cables, pipes, and the like, from a single set up location.

Generally, in the field of acoustics, there are two fundamental types of waves that propagate through material: pressure waves, and shear waves. These waves are called "bulk" waves and they propagate through the material at a constant velocity over all frequencies, including ultrasonic frequencies. An incident ultrasonic bulk wave transmitted along an object will be reflected from one end of the object so as to arrive at a fixed time at the transmission location according to a predictable, fixed travel time period.

Ultrasonic bulk waves are typically used as the incident waves in non-destructive evaluation applications. As the bulk waves enter an object and propagate along the length thereof, they reflect between the surfaces of the object. In objects of continuous cross section, the interaction of the bulk waves with one another and the object's surfaces produces envelopes of disturbance, called Lamb waves or guided waves, which also propagate along the object. Guided waves, unlike bulk waves, have velocities that vary depending on the frequency components of the waves. Thus, the time of arrival for a guided wave envelope reflected from the end of a pipe is often different for each envelope.

Furthermore, whereas there are only two types of bulk waves, there are an infinite number of guided waves that can exist for an object of a given geometry, such as a pipe. These different types of guided waves are distinguished by their modes; each mode has its own velocity vs. frequency relationship. Moreover, in a typical guided wave inspection, it is virtually impossible to ensure that only one mode will propagate. To the contrary, it is more likely that two or more modes will be present, thereby producing multiple reflections from the same material boundary or flaw, each having different velocities and, therefore, different times of arrival.

The phenomenon of the velocity of a signal being dependent on its frequency is called "dispersion." The effect of dispersion on guided waves is to cause their waveforms to change with time, generally becoming more elongated as they propagate down the length of the examined object. Guided waves have varying amounts of dispersion depending on the wave mode. For example, "longitudinal-one" or L(0, 1) guided waves are dispersive over virtually all frequencies, whereas "longitudinal-two" or L(0, 2) guided waves have a short band of frequencies over which they are not dispersive. Within this short band of frequencies, the velocity of the L(0, 2) guided wave is essentially constant and, therefore, the distance traveled over a given time period may be more readily determined. For this reason, L(0, 2) guided waves are commonly used in locating flaws and defects in piping.

Distinguishing guided wave modes among multiple reflections, however, can be quite a complicated process. As mentioned previously, guided waves of many different modes are produced by the interaction of bulk waves with object boundaries and flaws. The presence of these modes can lead to multiple detections of the same boundary or flaw within the object and different levels of sensitivity to the boundary or flaw based on the specific reflected mode. The presence of noise in the inspection signal can mask reflections to make the task of identifying individual modes even more difficult. Analysis of the inspection signal can therefore become a very complex task that requires extensive knowledge and time.

Conventional methods used to analyze guided wave inspection signals apply joint time-frequency analysis techniques in an attempt to observe dispersive behavior in reflected guided waves and then match the behavior to the modes theoretically predicted by such behavior. However, the time when a reflection begins (i.e., the onset of reflection) can be unclear, and dispersion of the narrow frequency band initiation pulse used to produce guided waves tends to decrease resolution due to elongation or "widening" of the waveform over time. Furthermore, a dispersive guided wave mode can sometimes appear to be non-dispersive, such that a portion of the L(0, 1) mode may resemble a portion of the L(0, 2) mode.

One commonly used joint time-frequency analysis technique is the Short-Time Fourier Transform (STFT). The STFT display, or spectrogram, can make evident velocity differences between frequency components of an examined portion of the inspection signal. The STFT provides useful results, but has several limitations. First, due to its limited resolution in both the time and frequency domains, the STFT result becomes difficult to accurately interpret as the distance traveled by the guided wave increases. As mentioned earlier, dispersive guided wave modes elongate in the time domain as they propagate down the length of the examined object. The resulting elongated shape in the STFT can interfere with other reflections. The presence of noise in the signal further complicates STFT interpretation. Therefore, techniques based on the STFT can have difficulty pinpointing the exact onset of a reflection signal due to the limited resolution of the STFT. Thus, analysis of the STFT typically requires tedious labor by skilled analysts with extensive experience, and is difficult to automate.

Accordingly, it is desirable to provide a more reliable and robust method and apparatus for analyzing guided wave inspection signals. Specifically, it is desirable to provide a signal processing method and apparatus that can more effectively accommodate the dispersive nature of guided wave modes so as to aid in the reliable characterization of inspected object geometric boundaries and flaws.

SUMMARY OF THE INVENTION

The present invention is directed to a method and apparatus for analyzing guided waves using a time-varying matched filter to correlate received guided waves with a time-varying dispersion-modeled reference wave to determine the location of a real flaw in an inspected object.

In general, in one aspect, the method includes the steps of selecting a time-varying dispersion-modeled reference signal associated with the geometry of an inspected object, a multiplicity of theoretical flaw locations located within the inspected object, and one or more characteristics of the selected guided wave; launching the selected guided wave signal into the inspected object; receiving a reflected signal generated by the interaction of the selected guided wave and the geometry of the inspected object (including any real flaw located therein); comparing the time-varying dispersion-modeled reference signal with the received reflected signal; and determining the location of the real flaw in the inspected object if the time-varying dispersion-modeled reference signal is substantially similar to the received reflected signal. The comparison can be conducted over the entire waveform for the reference and reflected signals during each sample interval. The comparison can also be conducted using only the detected wave envelope for each reference and reflected signal.

In general, in another aspect, the apparatus includes a means for selecting a time-varying dispersion-modeled reference signal associated with the geometry of an inspected object, a multiplicity of theoretical flaw locations located within the object, and the characteristics of the selected guided wave; a means for launching the selected guided wave signal into the inspected object; a means for receiving a reflected signal generated by the interaction of the selected guided wave and the geometry of the inspected object; a means for comparing the time-varying dispersion-modeled reference signal with the received reflected signal; and a means for determining the location of a real flaw in the inspected object if the time-varying dispersion-modeled reference signal is substantially similar to the received reflected signal.

The means for selecting the time-varying dispersion-modeled reference signal associated with the geometry of the inspected object, the multiplicity of theoretical flaw locations, and the characteristics of the guided wave may include a workstation or desktop computer capable of simulating the dispersive behavior of a guided wave in the inspected object, such as a pipe, as it interacts with the geometry of the inspected object. The computer typically includes a memory unit, a processor unit, and a storage unit for storing one or more program modules to generate individual reference signals which correspond to each sample of the reflected signal to be compared.

The means for launching the guided wave signal into the inspected object and the means for receiving the reflected signal generated by the interaction of the guided wave and the geometry of the inspected object may be an ultrasonic signal generator and a transducer, respectively.

The means for comparing the time-varying dispersion-modeled reference signal with the reflected signal and the means for determining the location of the real flaw in the inspected object if the time-varying dispersion-modeled reference signal is substantially similar to the reflected signal may also comprise a desktop computer, workstation, or other data processing apparatus, as are well known to those skilled in the art.

The method and apparatus operate under the assumption that a theoretical flaw exists at every point in time during which a waveform is acquired. If the waveform of the acquired reflected signal has a high level of correlation with the waveform of the generated reference signal corresponding to that point in time, then the hypothesis or assumption of a flaw at that location in the inspected object is verified. If there is a low level of correlation, then the hypothesis fails, and no flaw is detected.