基于BP神经网络的杉木标准树高曲线

Generalized Height-diameter Model for Chinese Fir Based on BP Neural Network

以福建省将乐县国有林场29块杉木人工林实测数据为例,运用BP神经网络建模技术建立树高预测模型。分别确定输入量和隐层节点数,再经训练和优选,得到的最优模型结构为2∶5∶1,决定系数为0.902 3,均方误差为1.7842。结合传统的两个标准树高曲线方程,利用检验数据分别对模型进行验证。结果表明:BP神经网络模型不管是拟合效果还是预测效果都明显优于传统方程,可以作为有效的树高预测技术。

We used the data of 29 plots of Chinese fir located in national forest farm of Jiangle in Fujian Province to build height prediction model by BP neural network. First, the input variable and the hidden nodes were determined, then, by training and optimization, an optimum modal was developed, with a model structure of 2 ： 5 ： 1, a determinate coefficient of 0.902 3 and error of mean square of 1.784 2. And then, it was compared with two traditional generalized height-diameter equations, the validation datasets were used to test the models, respectively. The fitting effect and prediction effect of BP neural network model are better than those of traditional equations, and BP neural network model can be used as effective tree height prediction technology.