The minimum spouting velocity, U-ms, is a critical topic, and an exact prediction of this parameter certainly can be useful in design and scale-up of the conical spouted beds. In the present study, a number of intelligence methods was applied to predict U-ms in conical spouted beds equipped with non-porous draft tube. Six dimensionless moduli comprising the essential operating and geometric parameters, namely, the gas density, the gas viscosity, the particle diameter, the particle density, the cone angle, the nozzle diameter, the static bed height, the length of the tube, the entrainment height, and the tube diameter were taken as models inputs. The total number of data samples is 1004 that includes 572 data points to train the smart models and 432 data points to test those models. The self-organizing map (SOM) was used to examine the effect of inputs correlation on the performance of the chosen smart models. Among the different models, a multi-layer perceptron (MLP) trained by Bayesian Regulation (BR) incorporating SOM i.e. (MLP-BR-SOM) predicted the best results with mean relative error of 9.71 and 12.45% for train and test data, respectively. In addition, sensitivity analysis (SA) of the proposed model was performed and it was shown the Ar, D-T/D-0, and H-0/D-0 have most influential parameters in prediction of U-ms. (C) 2018 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.