Without processing other optical measurements and completing a full analysis protocol, these force–displacement evaluations were intended to guide test settings and to provide a quick sanity check of the test set-up. However, due to data contamination, the raw data plot (Fig. 19.5) appeared completely unusable. . . or maybe not. As shown next, proper use of advanced DSP techniques can salvage data like this, often resulting in useful and credible results. This data set will be used as an example for how DSP may be used to recover data that has been fouled by noise. To better understand DSP techniques Stearns and David have a useful book [6] which explains various aspects of signal processing algorithms, while Diehl has well explained, practical examples of DSP to solve challenging data problems [7] as well as a valuable training course notebook [8] with many examples, tutorials and explanations of DSP theory written for mechanical engineers. 19.4 Cleaning and Salvaging Data with DSP When applying DSP to any data, the analyst will typically utilize some type of software to apply digital filters to remove the unwanted frequency content from the raw data. Digital filters come in several forms, but a commonly used workhorse for this type of effort is the lowpass Butterworth filter. This section explores several attempts at cleaning-up the previously presented data using sixth-order Butterworth filters from three different software packages; namely Mathcad®15 [9], MATLAB® 2013b [10], and Kornucopia®1.6 [11]. Such DSP tools can be very powerful in cleaning/salvaging data, but these tools can produce undesirable results also, especially if they are not well understood by the user. 0.1 0.05 0 0 2x10−4 4x10−4 Time [s] Displacement [in] 6x10−4 8x10−4 Fig. 19.4 Raw displacement–time response from a tensile test at 100 s 1 400 300 200 100 0 −0.05 0 0.05 Displacement [in] Force [lfb] 0.1 Fig. 19.5 Raw force–displacement response from a tensile test at 100 s 1 156 W.J. Briers III
RkJQdWJsaXNoZXIy MTMzNzEzMQ==