A process optimization method based on first principle model for the roasting process

Optimizing the roasting process is essential for achieving high-efficiency production. However, due to lacking key data detection, limited studies are available in this area. To address this challenge, this paper introduces a practical optimization method aimed at operational parameters such as the blast rate to feed rate (RBF) ratio and roasting temperature. First, a first-principles model of the roasting oxidation reaction was established. The model reveals the essential influence of operating parameters on product quality. This model provides a basis for subsequent optimization of operating parameters. Then, to solve the expensive black-box problem, a new optimization framework is proposed. In this framework, agent models are trained via just-in-time learning, which are tailored to the specific characteristics of the incoming mineral sources. The experimental results verify the effectiveness of the model. The proposed optimization method is approved to be adaptable to varying zinc concentrate compositions, facilitating the optimization of more suitable operational parameters.