The safety issues posed by blasting vibration to surrounding buildings urgently need to be addressed, while traditional vibration-reduction methods rely on empirical parameters and lack statistical support based on small sample sizes. By conducting blast vibration monitoring at a certain mine, the peak vibration frequency was extracted using the fast Fourier transformation (FFT). The characteristics of the peak vibration frequency were then analyzed using a comparative approach that incorporated the Bootstrap method, Bayesian analysis, and maximum likelihood estimation. Based on this, the inter-hole delay time was adjusted to half a cycle and verified through field tests to confirm the accuracy of this method. The results show that as the distance from the explosion source increases, the vibration velocity of the particle gradually decreases, and the attenuation characteristics of vibration velocity vary with direction. The Bootstrap method does not require the assumption that the population follows a specific distribution, instead drawing samples directly from the original sample. This approach better reflects the characteristics and distribution of the samples, resulting in statistical outcomes that are more representative of reality. An 8-ms half-cycle is determined to be the optimal inter-hole delay time. Field tests show that, compared to a 20-ms inter-hole delay time, an 8-ms inter-hole delay time reduces peak vibration velocity in the X, Y, and Z directions by 14.81%, 12.68%, and 36.51%, respectively, while the peak vibration velocity in the combined velocity direction decreases by 22.55%, demonstrating a significant vibration reduction effect. The research findings provide a basis for optimizing parameters in medium-deep holes underground blasting, effectively addressing safety issues caused by blasting vibrations.