Materials Performance is the world's most widely circulated magazine dedicated to corrosion prevention and control. MP provides information about the latest corrosion control technologies and practical applications for every industry and environment.

Issue link: https://mp.epubxp.com/i/1007306

59 MATERIALS PERFORMANCE: VOL. 57, NO. 8 AUGUST 2018 the collinearity test before it could be used to simulate atmospheric corrosion rates. Principal Component Analysis Model Structure The main objective of principal compo- nent analysis (PCA) is to reduce the num- ber of independent variables. It is a process that does not reduce the capability of esti- mating the dependent variable. In this pro- cess, it is expected that most of the original independent variables used to explain the variance of the dependent variable should be kept. Then, the independent variable used to explain the same trend of variance is combined using a linear formula to form a new variable ( principal component); please refer to Huang. 4 The principal com- ponents of Kaiser-Meyer-Olkin (KMO) test values ≥0.5 were analyzed and sorted out for use in the construction of subsequent ANN models in this study. Advantages and Disadvantages of Forecasting Models This study was divided into three types of forecasting models for atmospheric cor- ro si on rat e s , a n d R MSE a n d R 2 w ere adopted to evaluate the advantages and disadvantages of the models. Then, the atmospheric corrosion rate predicted by the model was classified. Finally, the mean absolute percentage error (MAPE) was used to obtain the optimal forecasting model. The calculation method of MAPE is shown in Equation (1): (1) where Y i is the observed value of ith sample, is the predicted value of ith sample, and n is the number of samples. Results and Discussions In this study, the test period ran from September 2011 to September 2013. The correlation between the metallic corrosion rate and environmental factors was found through the classification ranking of metal- lic corrosion, surface analysis and trend change characteristics, and the use of MLR, FIGURE 1 The map of Taiwan shows the locations of the test stations as well as their atmospheric zone classification. PCA, and ANN; and a forecasting model was determined. Trend Change of the Metallic Corrosion Rate The corrosion rate curves for samples from dif ferent atmo sph eric zon e s are shown in Figures 2 through 4. The corro- sion rate values in the industrial zones were all higher than the values in the other two zones, and the corrosion rates in the industrial and coastal zones were all higher than those of samples in rural zones. Addi- tionally, even after 24 months of exposure, the corrosion rates of samples in the indus- trial zones all approached 200 μm/y. All the surfaces showed sloughing of the corrosion product that exposed the underlying metal. Since the CS was exposed in the form of wire material wound in a helical shape, re- sidual stress in the wire material in creased the surface corrosion rate. 5 In all locations, the initial three months of exposure generally showed the highest

- MTP_L1
- MTP_991
- MTP_992
- MTP_1
- MTP_2
- MTP_3
- MTP_4
- MTP_5
- MTP_6
- MTP_7
- MTP_8
- MTP_9
- MTP_10
- MTP_11
- MTP_12
- MTP_13
- MTP_14
- MTP_15
- MTP_16
- MTP_17
- MTP_18
- MTP_19
- MTP_20
- MTP_21
- MTP_22
- MTP_23
- MTP_24
- MTP_25
- MTP_26
- MTP_27
- MTP_28
- MTP_29
- MTP_30
- MTP_31
- MTP_32
- MTP_33
- MTP_34
- MTP_35
- MTP_36
- MTP_37
- MTP_38
- MTP_39
- MTP_40
- MTP_41
- MTP_42
- MTP_43
- MTP_44
- MTP_45
- MTP_46
- MTP_47
- MTP_48
- MTP_49
- MTP_50
- MTP_51
- MTP_52
- MTP_53
- MTP_54
- MTP_55
- MTP_56
- MTP_57
- MTP_58
- MTP_59
- MTP_60
- MTP_61
- MTP_62
- MTP_63
- MTP_64
- MTP_65
- MTP_66
- MTP_67
- MTP_68
- MTP_69
- MTP_70
- MTP_71
- MTP_72
- MTP_73
- MTP_74
- MTP_75
- MTP_76
- MTP_77
- MTP_78
- MTP_79
- MTP_80
- MTP_81
- MTP_82
- MTP_83
- MTP_84
- MTP_85
- MTP_86
- MTP_87
- MTP_88
- MTP_993
- MTP_994