The inversion of ground motion, a strong stochastic process with both amplitude and frequency dual nonstationary characteristics, is very difficult. Thus, finding a nonstationary ground motion modeling method that can simultaneously simulate ground motion characteristics and determine actual ground motion time-varying distribution characteristics has become an important endeavor in ground motion research. A genetic algorithm and quadratic optimization identification technique based on the Kanai-Tajimi power-spectrum filtering method proposed by Du Xiuli et al. are employed to improve the bimodal time-varying modified Kanai-Tajimi power spectral model and parameter identification method proposed by Vlachos et al. Additionally, a method for modeling time-varying power-spectrum parameters for ground motion is proposed. This method is ideal for improving the Kanai-Tajimi spectral model of earthquakes because it satisfies the requirements of high-and low-frequency power spectra by filtering the Kanai-Tajimi spectrum with a series of high-and low-pass filters. The nonstationary ground motion simulation method uses two random variables to accurately capture the second-order statistics of the original stochastic process by Liu Zhangjun, thereby providing an efficient and convenient approach for subsequent verification. The results of a Chi-Chi ground motion example verify that the improved bimodal time-variable Kanai-Tajimi nonstationary stochastic model shows good feasibility and effectiveness. The results of the present research provide an important reference for designing seismic waves during seismic analysis of major engineering structures.