Abstract: The features of the lead-zinc imperial sintering process include strong nonlinearity, time variance, large time delay, and so on. Based on an analysis of heat state, the gas temperature profile for the sintering apparatus was investigated; a soft-sensor model of the burn-through point (BTP) was developed. Technological-parameter-based and time-sequence-based predictive models that take the dynamic features of the BTP into account were established; they were designed using neural networks and grey theory, respectively. Then, based on the concept of intelligent integration, the synthesis and coordination of these two models was implemented through a fuzzy classifier. The results of actual runs show that intelligent integration provides a practical and effective way of predicting the BTP, which, in turn, serves as a basis for implementing state optimization in the lead-zinc sintering process.