Abstract: Sodden aging method was used to deal with BSSF steel slag, and the content of free calcium (f-CaO) in BSSF steel slag was traced and checked.Soft sensor modeling based on gaussian process regression was established by gaussian process regression to forecast the f-CaO content in steel slag by sodden aging.The results show that f-CaO content in BSSF steel slag is effective reduced by sodden aging method to meet the requirements of the safe use, namely f-CaO content in A-BSSF steel slag is about 3.98%-4.03% after 120 d and f-CaO content in B-BSSF steel slag is about 9.84%-10.03% after 105 d.Soft sensor modeling based on gaussian process regression real value agree well with the forecast value, the relative error is-1.493%-0.748%, which increases the prediction accuracy efficiently of the f-CaO content in steel slag by sodden aging.

Key words: sodden aging;BSSF steel slag;gaussian process regression;free calcium;soft sensor