SMC-PHDF(Sequential Monte Carlo-Probability Hypothesis Density Filter)算法由于不受高斯和线性的限制,在目标跟踪领域有着广泛的应用;然而当系统量测噪声较大,很多样本的归一化权重很小而成为无效样本,最终导致SMC-PHDF算法滤波精...SMC-PHDF(Sequential Monte Carlo-Probability Hypothesis Density Filter)算法由于不受高斯和线性的限制,在目标跟踪领域有着广泛的应用;然而当系统量测噪声较大,很多样本的归一化权重很小而成为无效样本,最终导致SMC-PHDF算法滤波精度较低;针对这一问题提出似然分布自适应调整的SMC-PHDF算法,通过在更新步骤中自适应调整粒子权值,增加先验密度和似然的重叠区,从而达到提高滤波性能的目的;仿真结果表明:在系统量测噪声较大时该算法比传统SMC-PHDF算法的滤波效果有所提升。展开更多
In tomographic statics seismic data processing, it 1s crucial to cletermme an optimum base for a near-surface model. In this paper, we consider near-surface model base determination as a global optimum problem. Given ...In tomographic statics seismic data processing, it 1s crucial to cletermme an optimum base for a near-surface model. In this paper, we consider near-surface model base determination as a global optimum problem. Given information from uphole shooting and the first-arrival times from a surface seismic survey, we present a near-surface velocity model construction method based on a Monte-Carlo sampling scheme using a layered equivalent medium assumption. Compared with traditional least-squares first-arrival tomography, this scheme can delineate a clearer, weathering-layer base, resulting in a better implementation of damming correction. Examples using synthetic and field data are used to demonstrate the effectiveness of the proposed scheme.展开更多
文摘SMC-PHDF(Sequential Monte Carlo-Probability Hypothesis Density Filter)算法由于不受高斯和线性的限制,在目标跟踪领域有着广泛的应用;然而当系统量测噪声较大,很多样本的归一化权重很小而成为无效样本,最终导致SMC-PHDF算法滤波精度较低;针对这一问题提出似然分布自适应调整的SMC-PHDF算法,通过在更新步骤中自适应调整粒子权值,增加先验密度和似然的重叠区,从而达到提高滤波性能的目的;仿真结果表明:在系统量测噪声较大时该算法比传统SMC-PHDF算法的滤波效果有所提升。
基金funded by the National Science VIP specialized project of China(Grant No.2011ZX05025-001-03)by the National Science Foundation of China(Grant No.41274117)
文摘In tomographic statics seismic data processing, it 1s crucial to cletermme an optimum base for a near-surface model. In this paper, we consider near-surface model base determination as a global optimum problem. Given information from uphole shooting and the first-arrival times from a surface seismic survey, we present a near-surface velocity model construction method based on a Monte-Carlo sampling scheme using a layered equivalent medium assumption. Compared with traditional least-squares first-arrival tomography, this scheme can delineate a clearer, weathering-layer base, resulting in a better implementation of damming correction. Examples using synthetic and field data are used to demonstrate the effectiveness of the proposed scheme.
