无确定度逆向云模型新算法

Novel algorithm of backward cloud model without certainty

针对现有结果未揭示决定云模型雾化特性的本质因素以及无确定度逆向云模型算法误差较大这两个问题进行研究。通过对正向云模型的数学分析,指出云滴定量数据的标准差决定云模型雾化特性,提出用熵和超熵的比值度量云滴离散程度,称为雾化因子。分析和实验表明,云分布对应雾化因子取值在3～18之间,当其大于18时,云分布退化为正态分布,无确定度逆向云模型算法不再适用。在此基础上,利用云分布四阶原点矩来估计云模型的数字特征,提出一种新的无确定度逆向云模型算法,不同雾化因子和云滴数量的对比实验结果表明,所提算法在对云模型数字特征估计的准确性与稳定性方面优于现有算法。

The essential influence on the atomized feature of cloud model was indefinite and the error of backward cloud model without certainty was larger. To solve these problems, it demonstrated that the standard deviation of cloud drop quantitative da- ta determined the atomized property of cloud model by the mathematical analysis of forward cloud model. The ratio of entropy to hyper entropy, named atomized factor, can measure the disperse degree of cloud drops. The analysis and experimental re- sults reveal that the range of atomized factor corresponding to cloud distribution is 3 ~ 18, cloud distribution becomes normal distribution and backward cloud models without certainty are ineffective when the atomized factor is larger than 18. On this ba- sis, it proved a novel algorithm of backward cloud model without certainty by the property of the fourth moment of cloud distri- bution. The comparative experiments with the different atomized factors and cloud drop numbers indicate that the proposed al- gorithm is superior to other algorithms in accuracy and stability.