CAJ | 학술논문

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＞09．，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。
재내몽고왕업전림장선취181과락협송，분별대수고、흉경진행정학측정。이용1stOpt우화분석연건평태적Levenberg－Marquarat＋통용전국우화산법（ LM－UGO）화 SPSS연건적보통최소이승법분별대소측수거진행의합，인입학정계수（R2）、고계치적표준차（SEE）등작위평개지표。결과현시：LM－UGO방법적의합결과우우전통사용SPSS적의합결과，경F검험， FLM－UGO＞FSPSS최소이승법＞F0．05（1，179），차R2＞09．，변동계수CV＜50％，상대오차TRE、평균상대오차MSE치균공제재（－3％，3％）범위내，만족평개지표적치역요구이급임업조사기술적상관규정。
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 1stOpt 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 result using LM-UGO was better than that using traditional SPSS method .After inspected by F, the relationship of F of two methods were:FLM-UGO>FSPSS最小二乘法>F0.05(1,179) , and R2>0.9, CV<50%, and 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 .