以太行山区栓皮栎天然次生林为研究对象,以65块标准地及40株解析木为数据源,构建了优势木平均高神经网络模型和传统的函数曲线模型。结果表明:太行山栓皮栎天然次生林适宜的人工神经网络模型结构为1∶2∶1。与传统函数曲线模型相比,人工神经网络模型具有不依赖现存数学函数,拟合精度高等优点,更适合用来建立林分生长模型。

We investigated sixty-five plots and forty analytic dominant individuals in the natural secondary Quercus variabilis forest in Taihang Mountains to establish the growth model of the dominant tree average height based on ANN and the existing function.The result showed that the optimum model structure of natural secondary Quercus variabilis forest in Taihang Mountains was 1∶2∶1.ANN didn’t depend on existing function,which was better than existing function in the fitting effect and was more suitable for building the growth model.