This paper describes an intelligent integrated state optimization technique for the lead-zinc imperial smelting process, which features strong nonlinearity, strong coupling, a large time delay, time-varying parameters, and multiple constraints. First, the concept of the exponent of synthetic permeability (ESP) is explained. It reflects the permeability state of the process. Next, predictive models of ESP based on time-sequence and technological parameters are established using neural networks, and an intelligent integrated predictive model of the ESP is constructed by combining these two models with a fuzzy classifier. Then, a fuzzy expert predictor is built to predict the location of the burn-through point. Finally, the optimization of the permeability and heat states is implemented using a fuzzy expert controller with a self-study mechanism for those models. Actual runs show this method to be efficient and practical