在炼钢生产过程中,铁水脱硫过程是发生了一系列物理和化学变化的复杂工业过程.为建立各控制量与脱硫剂使用量间的模型,本文利用减法聚类法实现对ANFIS网络的结构识别,并在此基础上对ANFIS进行参数识别,完成铁水脱硫系统的建模任务.最后,通过将该方法与普通的BP算法建模进行比较,说明ANFIS在收敛速度及建模精度方面的优越性. In the process of steelmaking, hot metal desulphurization process is a complex industrial process in which a series of physical and chemical changes have taken place.In order to establish a model of the control amount and the amount of desulphurizer used, this paper uses subtractive clustering method to realize the structure of ANFIS network Identification and ANFIS parameters identification based on this to complete the modeling task of hot metal desulphurization system.Finally, this method is compared with ordinary BP algorithm modeling to illustrate the superiority of ANFIS in the convergence rate and modeling accuracy Sex.