Intelligent identification of the flow regimes of two-component particles in a fluidized bed with the optimized fuzzy c-means clustering algorithm

Flow regime identification is important in the application of fluidized beds. This paper provides a method for deciding flow regime number by objective criterion. The optimized fuzzy c-means clustering algorithm was used to cluster the flow regime classification of two-component particles in a fluidized bed. The genetic algorithm was applied to optimize the initial center clusters of fuzzy c-means clustering. Hilbert-Huang transform was applied to analyze pressure fluctuation signals and extract the characteristic parameters. Three clusters were found and respectively ascribed to three flow regimes: bubbling bed, slugging bed, and turbulent bed. A multilayer neural network was used to train and test the identification system of the flow regimes. The identification accuracies of bubbling bed, slugging bed, and turbulent bed can reach 91.67%, 92.85%, and 91.30%, respectively.