从理论上介绍一阶线性化算法和一阶条件期望线性化算法求解非线性混合效应模型参数,并利用这2种算法分别对杉木优势木平均高进行拟合(选用常用的Logistic模型作为基础模型,把区组作为随机效应因子).结果表明:2种算法对杉木优势木平均高进行拟合时精度都很高.通过对2种线性化算法进一步比较可得,在分析单木水平非线性混合效应优势木平均高模型时,2种算法拟合效果非常接近,因此在实际应用中可以选择其中任意一种算法对杉木优势木平均高进行拟合.%First-order linearization algorithm and first-order conditional expectation linearization algorithm were introduced theoretically to estimate the parameters in nonlinear mixed effects model. Through using the two algorithms to analysis the dominant height for Cunninghamia lancelata (In this paper the common Logistic model is chosen as a basic model) , it shows that both of the algorithms have high accuracy to fit the dominant height of Chinese Fir. Compared with the two linearization algorithms, we can see that the two algorithms have a close fitting effects for analyzing the single level nonlinear mixed effects dominate height model. Therefore, we can chose orbitrarily each of these algorithm to analyze the dominant height model for Cunninghamia lanceolata in practice.
展开▼
