The hierarchical merger model of galaxy formation and evolution predicts that galaxies are assembled through hierarchical mergers along mergers of dark matter halos. A natural consequence of galaxy mergers is the merger of SMBHs from progenitor galaxies. The mechanism or dynamics that drive SMBHs to merge is unknown due to the lack of observational evidence. In this work, we propose to combine the (future) gravitational wave measurements of SMBH binary merger events with the galaxy merger rate distributions from large-scale surveys/cosmological simulations, to infer the delay time of SMBH binary mergers, as a function of binary mass. The future space-based gravitaitonal wave detector LISA or Tianqin/Taiji is optimal to do this work since it will provide us with the population information of the merging binaries (assuming major mergers) across a wide mass range and almost all redshift space, thanks to the large single to noise ratio. We also test different delay time models associated to various dynamical scenarios via the Bayesian model selection method.