According to the broad-band characteristics of a combined D-V-A spectrum and the limitations of traditional methods in artificially generating earthquake ground motion on the fitting of peak values, this paper proposes a new method for matching a combined D-V-A spectrum and all the peak ground values. Based on wavelet principles, a natural ground motion record was discomposed into wavelet components in a time-frequency domain. The wavelet components that mostly affected the peak values of acceleration (PGA) and peak values of displacement (PGD) were identified to modify the natural ground motion′s peak values and preliminarily match the combined D-V-A spectrum. Then, the time domain wavelet synthesizing method was applied to modify the generated artificial ground motion to match the combined D-V-A spectrum with the peak values of velocity (PGV). After iteration, the artificial ground motion precisely matched the combined D-V-A spectrum with the target PGA, PGV and PGD values. The results of a numerical example showed that the generated ground motion, which includes acceleration, velocity and displacement time histories, well fit the target spectrum and every peak value with little impact on the natural earthquake ground motion. This suggests that the proposed method can be used to generate high-quality artificial earthquake ground motion.