SIZE DISTRIBUTION PREDICTING FOR WELL-GRINDED PARTICLES USING NEURAL MODEL

摘要

Purpose of this paper is to develop a formal neural model which can be used to predict grinded product distribution for some arbitrary feed distribution inside the distribution ranges used for the training of the neural net. Neural model is realized by General Regression Neural Net (GRNN). GRNN is a hybrid neural network capable to perform general nonlinear multivariable regression.The experiments were carried out on laboratory bead mill, (Willy A. Bachofen Maschinenfabrik, Basel, Switzerland) type KDL, with 600 ml volume cylinder. All tests were done in steady state. To check the prediction, number of comminution tests for given particle distribution input and given concentrations of solid were done.