基于1stOpt软件的树高测量模型  

Tree Height Measure Model Based on 1stOpt Software

作　　者：郭欣雨[1] 冯仲科[1] 曹忠[1] 范永祥[1] 

机构地区：[1]精准林业北京市重点实验室(北京林业大学),北京100083

出　　处：《东北林业大学学报》2015年第5期38-41,83,共5页Journal of Northeast Forestry University

基　　金：国家自然科学基金项目(41371001)

摘　　要：在内蒙古旺业甸林场选取181棵落叶松,分别对树高、胸径进行精确测定。利用1stOpt优化分析软件平台的Levenberg-Marquarat+通用全局优化算法(LM-UGO)和SPSS软件的普通最小二乘法分别对所测数据进行拟合,引入确定系数(R2)、估计值的标准差(SEE)等作为评价指标。结果显示:LM-UGO方法的拟合结果优于传统使用SPSS的拟合结果,经F检验,FLM-UGO>FSPSS最小二乘法>F0.05(1,179),且R2>0.9,变动系数CV<50%,相对误差TRE、平均相对误差MSE值均控制在(-3%,3%)范围内,满足评价指标的值域要求以及林业调查技术的相关规定。We selected 181 deciduous pine tree precious data of height and DBH in Wangyedian Forest in Inner Mongolia, and used Levenberg-Marquarat＋universal global optimization algorithm （LM-UGO） of lstOpt optimization analyzing and com- puting software platform and ordinary least square method in SPSS software to fit. We also evaluated two different models by using new evaluation method. The fitting resuh using LM-UGO was better than that using traditional SPSS method. After inspected by F, the relationship of F of two methods were： FLM_UG0 〉FsPss最小二乘法 〉F0.05（1,179） , and R2 〉0.9, Cv〈50%, arid the value of TRE and MsE were both less than ＋3%. The results can meet the requirements of assessment index range and the relevant provisions of the Forestry Survey Technology.

关 键 词：1stOpt软件 树高模型 通用全局优化算法 最小二乘法 

分 类 号：S711[农业科学—林学]