Prediction of ash content and yield percent of clean coal in multi gravity separator using artificial neural networks

Abstract

ABSTRACT In coal processing, ash content and yield percent of clean coal are the two performance parameters commonly used in measuring the performance of the plant. Estimation of these parameters gives an idea about the amenability and extent of separation taken place and also indicates the overall plant efficiency. In the present article, a model developed through neural network envisages the ash content and yield percent of clean coal based on the parameters of multigravity separator (MGS) such as drum inclination, stroke length, and drum speed for the beneficiation of coal fines. Different methods of sensitivity analysis have been adopted to find out the significance of the most important parameter which influences the product quality and quantity. Elucidation was based on neural interpretation diagram, from the weights associated with the input to examine the direct or indirect influence of parameters of MGS on the ash and yield. A predictive model equation has been proposed for ash content and yield percent of clean coal based on connection weights as model parameters.