周道成, 段忠东, 欧进萍. 短期观测资料的海洋极值环境要素概率模型估计[J]. 工程力学, 2009, 26(3): 176-181.
引用本文: 周道成, 段忠东, 欧进萍. 短期观测资料的海洋极值环境要素概率模型估计[J]. 工程力学, 2009, 26(3): 176-181.
ZHOU Dao-cheng, DUAN Zhong-dong, OU Jin-ping. EXTREME VALUE DISTRIBUTION OF OFFSHORE ENVIRONMENTAL FACTOR BASED ON SHORT-TERM OBSERVATION DATA[J]. Engineering Mechanics, 2009, 26(3): 176-181.
Citation: ZHOU Dao-cheng, DUAN Zhong-dong, OU Jin-ping. EXTREME VALUE DISTRIBUTION OF OFFSHORE ENVIRONMENTAL FACTOR BASED ON SHORT-TERM OBSERVATION DATA[J]. Engineering Mechanics, 2009, 26(3): 176-181.

短期观测资料的海洋极值环境要素概率模型估计

EXTREME VALUE DISTRIBUTION OF OFFSHORE ENVIRONMENTAL FACTOR BASED ON SHORT-TERM OBSERVATION DATA

  • 摘要: 首先将最大熵分布应用于极值环境要素;其次根据其参数估计的特点,利用LS-SVM良好的泛化能力和小样本学习能力,采用 Bootstrap方法得到的“理想”极值数据样本对LS-SVM函数进行估计,建立根据现场短期观测资料估计其极值环境要素矩的方法;结合最大熵分布的参数估计,建立了由现场短期观测资料估计其极值概率模型的新方法;最后通过模拟试验和实际数据验证了该方法的有效性和合理性。

     

    Abstract: The probabilistic properties of the extreme value of environmental factor are described by the maximum entropy distribution for its excellent adaptability. Because LS-SVM has some important properties such as good generalization and excellent learning, it is employed to calculate the moment of the environmental factor according to short-term observation data and based on the extreme value samples obtained by Bootstrap method. The parameters of maximum entropy distribution can be calculated by the moments, so a new method for the distribution of the environmental factor based on short-term observation data is obtained. The method is verified by simulation experiment and actual data.

     

/

返回文章
返回