Modelfitting-Chapter9.PDFVIP

Chapter 9 Model Fitting to Power-Plant Steel Data 9.1 Data and ?tting procedure The output from the BN measurement software is a text ?le containing only the forward and reverse RMS noise data. From the BN unit control software, it was known that the applied current amplitude was 0.7 A. A program was  written to extract the two sets of data and assign current values to them, assuming that BN data points were acquired at equally spaced intervals of current. It was further assumed, as in Chapter 5, that the current was proportional to the applied ?eld experienced by the domain walls. Models 1 and 2 were ?tted using the programs described in the Appendix. 9.2 Results Figure 9.1–Figure 9.6 show examples of ?tting using Model 1 and Model 2. Both models give a close ?t, although Model 2 is rather better at correctly ?tting the leading edge. Table 9.1 gives the errors calculated by the programs for all the data sets examined. The mean error of all the data sets, shown in the ?nal row of the table, is smallest for Model 2 and largest for empirical ?tting. The di?erence between the goodness of ?t of Model 1 and Model 2 is less pronounced than for the data ?tted in Chapter 5. The double peaks in those data increase the di?culty of ?tting using a model which is physically – 239 – Chapter 9 Model Fitting 35 Real data Model 1 30 Model 2 25 20 15 10 Voltage percentage (5 V = 100%) 5 0 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 Applied current / A 1 Figure 9.1: 2 4 Cr1Mo steel, 500?C, 8 h inaccurate, whereas a single peak with a slope c