During the long-distance transportation of wet-gas, the dominant frequency is of great significance for the study of pipeline fatigue and damage, and the safety production. Therefore, the theoretical and experimental researches for dominant frequency are carried out increasingly. However, most of the current prediction correlation of dominant frequency are mainly applicable to atmospheric pressure conditions (0.1 MPa), and the prediction accuracy is not accurate enough. The paper obtains the time series signal of liquid film thickness by near-infrared (NIR) sensor, and then calculates the wave frequency by the power spectrum density (PSD). The performance of typical predictive correlation is evaluated and analyzed by utilizing the experimental data at different flow and pressure conditions (0.1–0.8) MPa. The structure of Strouhal number and Lockhart-Martinelli (L-M) parameter are optimized reasonably, the mean velocity of the liquid film surface, the density increment of gas core, the gas core mass flow and average liquid film velocity are considered in the L-M parameter, a modified interfacial wave frequency correlation is proposed. The results indicate that the mean absolute error of the predictive correlation is 9.06% (current data) and 25.64% (literature data). The new correlation has a better predictive accuracy.