Application of Neural Nets for Optimization of Vibro-Fluidization Drying

In order to intensify the drying process of potassium persulfate the use of vibro-fluidized bed with vertical vibrations was selected. Due to the high mixing degree of solid particles with air in the vibro-fluidized bed, both superficial and internal moisture have been removed. The investigation of the process has been performed through a fractional factorial experiment, repeated for 13 values of the drying time. The process simulation using a neural net was in good agreement with experimental data (for about 80% of data the relative error is less than 10%). On the base of process simulation via neural net, a constrained non-linear programming problem was solved (using MINOS5.4/GAMS solver): identification of the values of process variables that minimize the drying time, with restriction on the desired final moisture content.