To form modeling data with high accuracy which reflect a nonlinear relation between a corrosion velocity and an input variable even when the term to acquire a corrosion velocity and a term to acquire an input variable are widely different.
In an input variable converter 14, physical amount 1D corresponding to a predetermined input variable included in operating data 1 is converted into influence degree data 14A which are weighed based on an influence degree characteristic 15A that shows the relation between the input variable and the corrosion velocity. In a representative value calculation part 12, a representative value 12A to represent the corrosion velocity calculation term is calculated respectively with regard to the influence degree data 14A and the physical amount 1C corresponding to input variables other than the predetermined input variable. In a modeling data forming part 13, the pair of each representative value 12A and the corrosion velocity 11A is output as desired modeling data 2.
KAJIMA TORU
CENTRAL COMP SERVICES CO LTD
YAMATAKE CORP