Based on the time series experimental data of heated tube wall temperature, which reflect some responses on nonlinear evolution, in vapor-liquid (V-L) and vapor-liquid-solid (V-L-S) flow boiling systems, the genetic algorithm (GA) was modified and used for constructing global explicit models to describe the nonlinear dynamic behavior of the aforesaid systems. The results of comparison and analysis between the calculated values and experimental data indicate that they are accordant with each other. Therefore, the models obtained here can express the nonlinear dynamic behaviors of two-phase and three-phase flow boiling systems finely, meanwhile, the precision of two-phase flow boiling system is higher than that of three-phase flow boiling system. Keywords vapor-liquid two-phase flow, vapor-liquid-solid three-phase flow, genetic algorithm, nonlinear prediction model, flow boiling, heat transfer