Materials Performance

AUG 2018

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

Contents of this Issue

Navigation

Page 62 of 92

60 AUGUST 2018 W W W.MATERIALSPERFORMANCE.COM CM CORROSION MANAGEMENT corrosion rates. Typically, the corrosion of CS in the atmospheric environment was uniform. With time, the curves also showed a reduction of the metal corrosion rate, and the rate of that reduction diminished as well. However, if there were chloride ions (Cl – ) present in the environment (e.g., in the coastal area), then the occurrence of crev- ice c orro sion or pittin g n eed ed to b e considered. MLR and PCA— Reducing Input Variables Factors were selected to conduct a step- wise regression analysis. After sorting and testing, these variables were entered into the industrial, coastal, and rural models to meet the basic hypothesis of regression analysis. The independent variables for CS were average dew temperature (DEW), exposure time (EXP), solar radiation (SUN), rainfall time (RF), chloride (Cl – ), average temperature (TEMP), average wind velocity (WS), and average wind direction (WD). Th e select ed variables for each atmo- spheric zone are shown in Table 1. The independent variables selected from the stepwise regression analysis were then used as input variables for subsequent ANN construction. Variables selected from the different atmospheric zones underwent a PCA analysis where the KMO values of the industrial, coastal, and rural zones were 0.51, 0.52, and 0.56, respectively (as shown in Table 1), which all met the acceptance condition of ≥0.50. Therefore, the data were suitable for principal component analyses. The components with eigenvalues greater than 1 were then selected, and two to four components were chosen for each atmo- spheric zone. The accumulated explana- tor y total variations reached were 82.3, 88.5, and 91.8% for the coastal, industrial, and rural zones, respectively. The compo- nent scores of the selected three principal components were saved and used as the i n p u t v a r i a b l e s f o r su b s e q u e n t A N N construction. Construction of Artificial Neural Forecasting Model In this study, three analysis methods— ANN, MLR - ANN, and P C A- ANN—were adopted to predict the atmospheric corro- sion rates of industrial, coastal, and rural FIGURE 2 Trend change of the CS surface over time for the industrial test locations. FIGURE 3 Trend change of the CS surface over time for the coastal test locations. FIGURE 4 Trend change of the CS surface over time for the rural test locations.

Articles in this issue

Links on this page

Archives of this issue

view archives of Materials Performance - AUG 2018