基于神经网络的马尾松人工林密度指数模型  

Density index model of the Pinus massoniana plantation based on artifical neural network(ANN)

作　　者：刘光武 陈晨 王柯力 LIU Guangwu;CHEN Chen;WANG Keli(Henan Forestry Vocationan College,Luoyang 471002,Henan,China;Forestry Bureau of Nanzhao County,Nanzhao 474650,Henan,China)

机构地区：[1]河南林业职业学院,河南洛阳471002 [2]河南省南召县林业局,河南南召474650

出　　处：《浙江农林大学学报》2020年第1期100-104,共5页Journal of Zhejiang A&F University

基　　金：河南省科技攻关计划项目(172102110239)

摘　　要：[目的] 通过对马尾松Pinus massoniana人工林密度指数模型的研究,为制定木材产量及质量的提升决策提供参考。 [方法] 以河南省薄山林场马尾松人工林为研究对象,采用147块标准地数据,以林分平均胸径为输入向量,以林分密度为输出向量,建立了林分密度指数人工神经网络(ANN)模型,并与Reineke的林分密度指数模型进行比较。 [结果] ① 薄山林场马尾松人工林最大密度线斜率b为-1.516 3,马尾松标准平均胸径为14 cm,Reineke的林分密度指数模型精度为92.11%,t检验结果显著;②构建了网络结构为1:2:1的林分密度指数ANN模型,模型拟合精度为92.57%,均方误差为0.001 469 7。③无论采用Reineke林分密度指数还是人工神经网络技术,在拟合株数密度随林分平均胸径的变化趋势时,幼龄林组拟合效果都不理想,这与幼龄林组数据数量偏少有关。 [结论] 所建模型可为薄山林场马尾松抚育经营决策提供依据。Objective To provide reference for wood production and quality improvement management measures, a study was conducted of the density index model of the Pinus massoniana plantation. Method Taking the P. massoniana plantation in the Boshan Forest Form in Henan Province as the research object with data collected of 147 standard plots, artifical neural network (ANN) model of the stand density index was established taking stand DBH as input vector and the plant number density as output vector and compared with Reineke's stand density index model. Result (1) The maximum density slope b of the P. massoniana plantation in the Boshan Forest Form was -1.516 3, the standardized mean DBH was 14 cm, the fitting accuracy of Reineke's stand density index model was 92.11%, and the effect of t test was significant. (2) The fitting accuracy of ANN model was 92.57% and the mean square error (MSE) was 0.001 469 7. (3)Either with Reineke density index or ANN technique employed, the young forest group demonstrate lower accuracy in fitting the variation trends of number density against that of stand DBH, which attributes to the smaller data of the young forest group. Conclusion The above established model was expected to provide reference for the operational decisions of the P. massoniana plantation in the Boshan Forest Form in Henan Province.

关 键 词：院 森林经理学曰 马尾松曰 人工林曰 林分密度指数曰 人工神经网络 

分 类 号：S757.9[农业科学—森林经理学]