【摘要】 提出了一种近地面爆源参数的贝叶斯声震联合反演方法，推导了多类型数据的贝叶斯声震联合定位理论，建立了爆源多参数的综合反演方法，探讨了声单传感器测点到时、声阵列方位角和到时及地震波到时等多类型数据的联合定位精度和特点，分析了格点搜索法和MCMC方法求解起爆时间、爆源位置等源参数的定位精度和求解效率，讨论了爆源多参数综合反演方法的性能。研究表明，贝叶斯声震联合反演方法得到的爆源参数分布范围集中、离散度小、相对声或地震等单一反演方法的源参数估计更稳定、估计偏差始终位于较小水平。贝叶斯MCMC方法在有限步长情况下可以快速搜索到真实解附近，相对格点搜索法定位精度和求解效率更高。起爆时间、爆源位置和爆炸当量等源参数综合反演具有较高的估计精度，其中爆源位置和当量真实值位于可信区间，95%可信区间可以有效估计爆源位置，90%可信区间可以有效估计爆炸当量。更多还原

【Abstract】 The inversion of near-surface explosion source parameters is of great significance to explosion monitoring, weapon power evaluation, and effectiveness testing. In this paper, a Bayesian method for acoustic-seismic joint inversion of near-surface explosion parameters is proposed, including the derived Bayesian acoustic-seismic joint localization theory of multiple types of data and the integrated inversion method for explosion source parameters. The joint localizaiton accuracy and characteristics of the mutiple types data, such as the arrival time of the acoustic single measurement point, the azimuth angle and arrival time of acoustic array, and the arrival time of the seismic measurement point, are explored. Moreover, the localization accuracy and solution efficiency of the grid search and MCMC methods solving explosion time and location are analyzed to compare the performance of the grid search and MCMC methods. Finally, the performance of the integrated inversion method for multiple explosion source parameters is also discussed. The results show that the source parameters obtained by Bayesian acoustic-seismic joint inversion method are distributed in a narrow range with a small discrepancy level, and can improve the stability of the source parameter estimation with a relatively small estimation error. Furthermore, the Bayesian MCMC method can quickly search near to the real solution in the case of finite steps, and has higher accuracy and efficiency than the grid search method. The integrated inversion method of source parameters, such as explosion time, source location, and yield, has high estimation precision. The real values of explosion location and yield are located in the confidence interval with the 95% confidence interval estimating the explosion position location effectively and the 90% confidence interval estimating the explosion yield effectively.更多还原