A Hybrid Prediction Model of Recycled Sodium Aluminate Solution Concentration in Evaporation Process

The sodium aluminate solution of the evaporation process is provided for the digestion process to recycle useful resources, decrease alkali concentration in waste solution, and reduce environmental pollution. However, the concentration of recycled sodium aluminate solution as an indispensable indicator for manipulating the evaporation process of the industrial alumina production is acquired offline leading to delayed feedback information. To ensure stable production of the subsequent process, reduce energy, and resource consumption, this article focuses on developing a hybrid prediction model for recycled sodium aluminate solution concentration in evaporation process. First, data reconciliation is utilized to improve the quality of material flow information, and the process mechanism model is established through mechanism research and equilibrium principle to obtain the concentration prediction result. Moreover, an industrial production condition analysis as well as fuzzy expert rules, are introduced for modifying prediction results from the mechanism model. Furthermore, the errors are predicted by the kernel extreme learning machine (KELM), and a concentration prediction model integrated error compensation results and modified predictions are established. The experimental simulations and industrial applications show that the accuracy of the prediction error obtained by the proposed model reaches 90% within ±2%, and the advantages of the hybrid model are particularly prominent under different conditions, which is beneficial for efficient and clean production.