Recently, the development of digital measurement technology has enabled the instantaneous acquisition of numerous data. Image data of digital cameras and time series data recorded over a long period by high-speed A/D converters are some examples.

For this study, Primitive Pattern Decomposition (PPD) was developed to extract arbitrary data patterns easily and quickly from a large digital data sequence. The primitives use nine patterns. PPD with the primitives is applied for the detection of crack locations from torque logging data obtained during drilling. In a rotational drilling machine, the torque directly reflects the drilling energy. It is also sensitive to discontinuities such as layer boundaries, layer separations, and cracks in the rock mass. Torque logging data are obtained from a roof bolter with thrust, rotation, and stroke during drilling for the roof rock of the roadway in underground coal mines. Results show that the crack distribution along the drill hole is obtainable by application of PPD to the torque data. Moreover, by compiling the crack distributions of roof drillings in the same area, a 3-D roof rock geostructure can be reconstructed. Comparing core sample inspection and drill hole observation using an optical fiber scope to results estimated using this system, the accuracy of discontinuity detection is probably less than 45 mm, with a fitting ratio higher than 60%.

Results of these experiments demonstrated that the PPD method can easily and flexibly produce a pattern to be extracted. It is possible to extract it rapidly from original signals. Furthermore, PPD is useful to obtain a rough approximation of complex signals and for the rough estimation of signal characteristics as a type of spectrum analysis.

This content is only available via PDF.
You can access this article if you purchase or spend a download.