28 J.H. Gordis and L.T.K. Papagiannakis 5 10 15 20 25 30 35 40 45 50 55 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Res. Norm e RME MFA QRB ABC Set No. QRB flag RME (slightly offset for clarity) MFA Norm of Residual, e 1 Mode (m=1) Fig. 3.7 Solution metrics and QRB flag for element #1, versus ABC set number, mD1 mode used ABC Set #1: Size= 1,10 Rank= 1 10 Element (Parameter) # x 107 1 2 3 4 5 6 7 8 9 0 0.5 1 1.5 2 2.5 3 3.5 QRB = 0 RME = 0.062 MFA = 0 ABC ID True Fig. 3.8 Localization is exact for as-tested configuration x 107 ABC Set #9: Size= 1,10 Rank= 1 10 Element (Parameter) # 1 2 3 4 5 6 7 8 9 0 0.5 1 1.5 2 2.5 3 3.5 = ABC pin ABC ID True QRB = 0 RME = 0.062 MFA = 0 Fig. 3.9 Localization is exact for ABCset #9 f1; 9; 10; 11; 48 W 56g. Hence, the QRB flag being set “zero” is a condition which can be used to select ABC sets. Figure 3.7 makes it clear that only a subset of all the ABC sets can be used to locate error in element #1, specifically, those ABC sets, cQR 1 , at which the QRB flag is zero. We will plot the solution vector u.c/ for several selected ABC configurations for which the QRB flag is zero (including the as-tested, i.e. c D1), and also for one ABC configuration for which the QRB flag equals one (failed solution). These four solutions are shown in Figs. 3.8 through 3.11. The as-tested configuration produces an accurate localization (Fig. 3.8), as does ABC sets #9 & #48. These three solutions all have QRBD0. The artificial pin locations are indicated by the small triangle. ABC set #33 does not produce an accurate solution, as predicted by QRBD1.
RkJQdWJsaXNoZXIy MTMzNzEzMQ==