A cascaded modeling approach for comprehensive reaction state perception of a hydrometallurgical reactor

Abstract

A hydrometallurgical reactor is the central production unit in a hydrometallurgy process. Due to the correlation between multiple parallel reactions and feeding condition fluctuations, modeling the outlet technical indicator alone is insufficient to express the highly informative reaction state comprehensively. This paper proposes a cascaded modeling approach for comprehensive reaction state perception of a hydrometallurgical reactor by considering the feeding conditions, operation modes, outlet technical indicators, and their relations. First, propose a perception model based on the information of two adjacent reactors for feeding condition perception. Second, develop an online operating mode perception model, combining dynamic feature extraction methods based on dynamic inner principal component analysis and mechanism knowledge. Finally, an outlet technical indicator perception model integrates mechanism and data features for multiple operation modes. The simulation results reveal that the proposed model can comprehensively describe the reaction state and guide on-site control.