中文摘要：

采用大兴安岭东部落叶松Larch的材料，分别运用人工神经网络和回归分析方法编制了落叶松的一元材积表，人工神经网络在拟合过程中以落叶松胸径作为输入向量，以单株材积作为输出向量，并选出既符合林木材积曲线分布规律，又具有较高拟合准确度的网络模型，其网络结构为1：2：1，网络对象名为Enet。用该模型拟合的材积表其拟合准确度迭95．69％，拟合误差为7．2997；而用回归分析法拟合的材积表其拟合准确度为91．56％，拟合误差为14．2930。由此表明，用人工神经网络编制的材积表精度明显高于回归分析法，误差明显小于回归分析法，拟合的材积更接近实际材积。

英文摘要：

Based on the data of Larch in eastern region of Daxinganling, the single entry volume table of Larch is established by using artificial neural networks and regression analysis. By using breast height diameter of Larch as input variables and single timber as output variables, a network model is selected and it orresponds to the discipline of forest volume curve distribution and has a high fitting accuracy. The network structure is 1 ： 2：1 and named Enet. The fitting precision of the model is 95.69% and the fitting error is 7. 299 7 ; while the fitting precision by using regression analysis is 91.56% and the fitting error is 14.293 0. The results showed that the precision of the single entry volume table fitted by artificial neural networks is obviously higher than that of regression analysis and the error is obviously tess than that of regression analysis. The volume fitted by artificial neural networks is closer to the actual volume.