基于BP人工神经网络的兴安落叶松天然林全林分生长模型的研究

以大兴安岭地区兴安落叶松天然林为研究对象，基于688块固定标准地数据，采用MATLAB中log-sigmoid型函数(logsig)和线性函数(purelin)为神经元的作用函数，依据全林分生长模型的概念，以年龄（A）、地位级指数（SCI）和林分密度指数（SDI）作为输入变量，以林分每公顷蓄积量（M）作为输出变量，构建和训练了全林分生长的BP人工神经网络模型，并与常规建模方法进行了对比研究。结果表明，BP人工神经网络模型的拟合精度高达99.6%，检验精度为98.9%，说明与其它建模方法相比人工神经网络建模具有较高的拟合精度和适应性，对林分生长具有更好的预测能力。

Whole Stand Growth Model for Natural Dahurian Larch Forests Based on BP ANN

Based on 688 permanent sample plots，the whole stand growth model of BP ANN was developed for natural dahurian larch(Larix gemelinii Rupr.) forests in area of the Daxing’an Mountains, using log-sigmoid function (logsig) and linearity function (purelin) of MATLAB as the neural functions. According to the concept of whole stand growth model，the Age(A)，Site Class Index (SCI) and Stand Density Index (SDI) were considered as input variables and the Stand Volume(M) was as the output variable in the model. A comparative study between BP ANN model and the common method was also conducted. The result of the model performance analysis was showed that the BP ANN model in this paper had high fitting accuracy (99.6%) and precision (98.9%), it is better than the common methods in fitting and adaptability and it was suitable to predicting stand growth.