北京山区油松人工林单木材积生长量BP神经网络模型  

BP Neural Network Model for Volume Growth of Single Trees in Pinus tabulaeformis Plantations in Beijing Mountain Area

作　　者：史宇[1] 余新晓[1] 张佳音[1] 高志亮[1] 张振明[1] 朱建刚[1] 

机构地区：[1]北京林业大学,北京100083

出　　处：《东北林业大学学报》2010年第2期20-22,共3页Journal of Northeast Forestry University

基　　金：国家"十一五"科技支撑计划项目(2006BAD03A0201);北京市科技计划重大项目(D0706001000091)

摘　　要：采用综合分析法和相关分析法选定每公顷蓄积量、标准木年龄、标准木树高、标准木胸径为输入变量,以标准木的上个生长季的材积连年生长量为输出变量,建立三层结构的油松(Pinus tabulaeformisCarr.)人工林单木材积生长量BP神经网络模型。利用北京山区78块各类型油松人工林样地调查数据对模型进行训练和仿真模拟。通过综合比较得出网络模型结构为4∶11∶1的网络模型具有最好的模拟效果,训练精度可达94.024%。此模型可以准确、简单、快速的对北京山区油松人工林的单木材积生长量进行有关的分析、计算、模拟和预测。Three-layer BP neural network models for volume growth of single trees in Pinus tabulaeformis plantations in Beijing mountain area were established using current annual increment of standard tree in the previous growing season as the output variable,and volume per hectare,the age,height and diameter at breast height of the standard tree as the input variables which were chosen by the methods of comprehensive analysis and correlation analysis.The neural network models were trained and simulated by the data of 78 sample plots in different types of P.tabulaeformis plantations in Beijing mountain area.Comprehensive comparison of the networks shows that the network model with the structure of 4∶11∶1 has the best simulation results,with a training accuracy of 94.024%.This model can be applied to the analysis,calculation,simulation and prediction of volume growth of single trees in P.tabulaeformis plantations in Beijing mountain area,and it is proved to be an accurate,simple,and rapid model.

关 键 词：北京山区 油松 人工林 单木材积生长量 BP神经网络 

分 类 号：S791.254[农业科学—林木遗传育种] TP183[农业科学—林学]