Abstract: Abstract: In this paper, the empirical model decomposition and Shannon entropy are integrated to investigate the multiscale nonlinear dynamic characteristics of vertical upward gas liquid two phase flow. Firstly, the gathered fluctuant conductance signals are processed through empirical mode decomposition. Then, a set of index energy is computed using the above intrinsic mode functions. Thus, an index energy based algorithm is proposed to compute multiscale entropy. It is shown from simulation results that the values of multiscale entropy can be not only taken as feature vectors for flow pattern identification, but also used to effectively reveal the complex flow characteristics of gas liquid two phase flow with different regimes.