Comprehensive Knowledge-Driven AI System for Air Classification Process

Abstract

Air classifier devices have a distinct advantage over other systems used to separate materials. They maximize the mill’s capacity and therefore constitute efficient methods of reducing the energy consumption of crushing and grinding operations. Since improvement in their performance is challenging, the development of an efficient modeling system is of great practical significance. The paper introduces a novel, knowledge-based classification (FLClass) system of bulk materials. A wide range of operating parameters are considered in the study: the mean mass and the Sauter mean diameter of the fed material, classifier rotor speed, working air pressure, and test conducting time. The output variables are the Sauter mean diameter and the cut size of the classification product, as well as the performance of the process. The model was successfully validated against experimental data. The maximum relative error between the measured and predicted data is lower than 9%. The presented fuzzy-logic-based approach allows an optimization study of the process to be conducted. For the considered range of input parameters, the highest performance of the classification process is equal to almost 362 g/min. To the best of our knowledge, this paper is the first one available in open literature dealing with the fuzzy logic approach in modeling the air classification process of bulk materials.