Modeling metallurgical responses of coal Tri-Flo separators by a novel BNN: a “Conscious-Lab” development

Abstract Tri-flo cyclone, as a dense-medium separation device, is one of the most typical environmentally friendly industrial techniques in coal washery plants. Surprisingly, no detailed investigation has been conducted to explore effectiveness tri-flo cyclone operating parameters on their representative metallurgical responses (yield and recovery). To fill this gap, work for first time processing sector going introduce type advanced intelligent method (boosted-neural network “BNN”) which able linearly nonlinearly assess multivariable correlations among all variables, rank them based model produced responses. These assessments modeling were considered new concept called “Conscious Laboratory (CL)”. CL can markedly decrease number laboratory experiments, reduce cost, save time, remove scaling up risks, expand maintaining processes, significantly improve our knowledge about modeled system. In study, robust monitoring database from Tabas plant was prepared cover various conditions building separators. Well-known machine learning methods, random forest, support vector regression developed validate BNN outcomes. The comparisons indicated accuracy strength over examined traditional methods. sentence, generating novel within apply energy areas, gaps possible interactions, open window plants' fully automotive process.