The sintering parameters, such as sinter trolley speed, stuff thickness and ignition temperature etc. have important influence on sinter section image character distribution, and the sinter chemical ingredient have complicated nolinear relation with sintering section image characters. The paper applied Mixed-F statistic analysis to determine the optimal class number of Fuzzy cluster of sintering section radiant image, and fuzzy partition entropy analysis to verify whether the class number is optimal. It is be proved that the algorithm is right and effective, and its application is helpful for the online forecasting sintering FeO range.