Weibull分布参数辨识改进及对浙江毛竹林胸径年龄分布的测度

The algorithm update of Weibull Distribution parametric identification and its application on measuring the distribution of diameter and age of Moso bamboo forests in Zhejiang Province,China

非线性模型迭代参数初始值的选取及拟合的结果能否可用非常重要且一直是一个难题。传统的方法只是凭感觉偿试不进行定量评估,对于用阻尼最小二乘进行拟合只注重精度高低,没有从理论上研究和分析评价得出的结果能否可用。为了克服这一不足,使参数选取较好,阻尼最小二乘得出结果可用。利用非线性回归原理在这两方面做了探讨,提出了二个判别指标:βT<1,βN<1,分析了每一指标在参数不同状态的控制作用,并在Weibull分布函数参数辨识中做了应用。针对目前毛竹林林分结构规律研究的不足,利用浙江省230个毛竹林样地资料研究了省域尺度的毛竹林胸径分布规律,经SPSS软件中的P-P图检验与柯尔莫哥洛夫检验,表明毛竹林胸径结构服从Weibull分布;再经对Weibull分布模型拟合求解与作图对比,结果显示Weibull分布能很好地描述毛竹林胸径分布规律。利用Weibull分布三参数与林分特征因子间的相关关系,经分析研究,导出了一个改进的二元分布模型,用该二元分布模型很好地测度了浙江省毛竹林胸径和年龄分布规律,最后得到了浙江省毛竹林株数按胸径年龄分布的理论频数。

It is very important and difficult to select initial parametric value and to evaluate the result which is applicable or not of nonlinear model. To solve nonlinear equation with Levenberg-marqurt iteration method, the traditional approach to select initial parametric value is based on the sense and multiple testing, and only focuses on the accuracy of the fitted value while ignores the applicability of the result theoretically. These shortcomings were overcome in this study through. In order to conquer these shortcomings, this paper makes the Nonlinear Regression Statistics and two key estimation indices β^T〈1,β^N〈1. The control function of β^T〈1,β^N〈1 in the various parametric states were analyzed and used to solve parameters of Weibull Distribution function as well. About 230 samples＇ datum of Moso bamboo （ Phyllostachys pubescens） in Zhejiang Province were applied to study the distribution pattern of diameter of Moso bamboo forests. The results show that the diameter distribution pattern of Moso bamboo forest is conformed to Weibull Distribution through the P-P chart and Kolmogorov test. Using of the relationship with Weibull＇s three parameters and characteristic factors of forests, two-dimensional distribution function is induced. In the presented report, we also demonstrated that two-dimensional distribution function had high accuracy in the measurement on the distribution patterns of diameter and age of Moso bamboo forests. Finally, the theoretical probability value of diameter and age distribution of Zhejiang Province were presented.