基于人工神经网络的林分直径分布预测  

Forecasting stand diameter distribution based on artificial neural network

作　　者：黄家荣[1] 高光芹[1] 孟宪宇[2] 关毓秀[2] 

机构地区：[1]河南农业大学林学院 [2]北京林业大学林学院

出　　处：《北京林业大学学报》2010年第3期21-26,共6页Journal of Beijing Forestry University

基　　金：河南省科技攻关项目(0624050007);河南农业大学博士基金项目(30400242)

摘　　要：以马尾松人工林为研究对象,用人工神经网络建模技术构建了林分直径分布预测模型。经训练和优选,得到的理想模型结构为3∶6∶6∶1,训练误差指标为0.000281,总体拟合准确度为98%。模型对82块训练标准地的累积频率拟合准确度最大为100%,最小为95%,平均为98%;频率拟合准确度最大为96%,最小为75%,平均为87%。模型对18块检验标准地的累积频率预测准确度最大为99%,最小为97%,平均为98%;频率预测准确度最大为96%,最小为76%,平均为88%。所建模型具有很好的拟合效果和很强的预测能力,可用于10～30年生马尾松人工林。研究结果证明,人工神经网络技术可以作为有效的林分直径分布预测技术。A stand diameter distribution model was described by using artificial neural network modeling technology,in a Masson pine forest.By training and optimization,an ideal model was developed,with a model structure of 3：6：6：1,a training error of 0.000281 and a 98% accurate fit.For the 82 training plots,the accuracy of accumulated frequency fitting ranged from 95% to 100%,with a mean of 98%.The accuracy of frequency fitting ranged from 75% to 96%,with a mean of 87%.For the 18 test plots,the accuracy of accumulated frequency prediction ranged from 97% to 99%,with a mean of 98%.The accuracy of frequency prediction was between 76% and 96%,with a mean of 88%.Due to its high fitting accuracy and strong prediction ability,the model created by us can be used to predict stand diameter distributions in Masson pine forests,aged 10 to 30 years.The results indicate that an artificial neural network is an effective prediction technology for stand diameter distribution.

关 键 词：人工神经网络 马尾松 直径分布 预测 

分 类 号：S758.5[农业科学—森林经理学]