Abstract: Due to the influence of regional seismic source structures, the actual observed ground motion has significant regional differences. In the current structural dynamic time-history analysis, although the synthetic method can provide the spectra-compatible ground motion input, it cannot reflect the actual ground motion characteristics in the target area. In order to synthesize the ground motion, which contains the characteristics of the target region and is perfectly compatible with the target spectrum, a new method for ground motion synthesis based on intelligent algorithm is proposed. In this method, the principal component analysis algorithm was used to obtain the actual regional ground motion characteristics, the maximum variance theory was used to extract the seeds ground motion in the target area, the response spectrum of seed ground motion was adjusted by time domain response spectrum matching method optimized by particle swarm optimization algorithm, and the spectra-compatible ground motion time history containing regional characteristics was finally constructed. The rationality and effectiveness of this method are verified by taking Sichuan area as an example, and the characteristics of the synthesized ground motion in time domain and frequency domain are consistent with the actual ground motion in this area. The advantage of the proposed method is that the synthesized ground motion has significant regional characteristics, and has better matching effect and higher matching efficiency with the target spectrum, which can provide low uncertainty ground motion input considering regional differences for structural dynamic response analyses.