Vol.22, No.1, 2022, pp. 3–17
UDC:

The importance of studying corrosive processes is evident. However, researchers are often faced with the lack of sufficient empirical data, especially in the application of advanced modelling techniques in the field of artificial intelligence. In this paper, we applied the technique of inserting synthetic data into empirical databases based on measuring corrosion depth caused by three different marine environments over samples of three different alloys after 12 and 18 months of exposure to the environment. Empirical and extended databases are further used to analyse a linear model of corrosion depth based on the assumption that corrosion processes occur immediately after exposure to the effects of the marine environment. In each observed database, the best two-parameter, three-parameter, and multiparameter continuous distributions are selected by fitting. After a comparative presentation of the obtained results, the influence of the inserted synthetic data is detected. Although an empirical density function can be defined for corrosion data, the practice has shown that the most favourable way to analyse system failure data is to determine a distribution that follows empirical data well and then to calculate other functions from the reliability domain based on the determined probability density function. Based on this fact, starting from the detected most favourable continuous distributions that adequately describe corrosive behaviour of observed alloys in seawater environment, this paper also provides analyses of observed alloys from the standpoint of reliability.

Keywords: corrosion depth, empirical database, extended database, distribution fitting, reliability