论文部分内容阅读
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。
Firstly, the maximum entropy distribution is applied to extreme environmental elements. Secondly, based on the characteristics of parameter estimation, LS-SVM good generalization ability and small sample learning ability are used to compare the “ideal” extreme value data samples obtained by Bootstrap method LS-SVM function to establish the method of estimating the moment of environmental elements from the short-term observation data. Based on the parameter estimation of the maximum entropy distribution, a new method of estimating the extreme probability model from the field observation data is established. Finally, Simulation and actual data verify the effectiveness and rationality of this method.