基于BP人工神经网络的大青山自然保护区华北落叶松人工林全林分生长模型研究

BASED ON ARTIFICIAL NEURAL NETWORK MODELING OF THE Larix principis rupprechtii Mayr PLANTATION AT DAQING MOUNTAIN IN INNER MONGOLIA

以内蒙古自治区大青山自然保护区华北落叶松人工林为研究对象,采用2006年二类调查的371个小班数据,利用Matlab软件中的"S"型对数(logsig)函数和线性(purelin)函数为神经元的作用函数,依据全林分生长模型概念的要求,选取年龄(A)、地位级(N)和郁闭度(S)作为输入变量,分别以胸径(D),树高(H)和林分每公顷蓄积(M)作为输出变量,进行全林分生长BP人工神经网络模型的构建和训练,并对模型的拟合精度和检验精度进行检验,模型的拟合精度分别为81.65%、88.53%和80.55%,检验精度分别为83.13%、79.47%和80.79%,表明人工神经网络建模具有较好的拟合精度和适应性,对林分生长具有较好的预测能力。

Make plantation Larix principis rupprechtii Mayr at Da Qingshan nature reserves in Inner Mongolia as the research object.We take 371 subcompartment data of the National Second-Class investigation data by Inner Mongolia survey and design institute of forestry in 2006.Use of Matlab software log-the sigmoid type function(logsig) and linear function(purelin) for the role of neurons.Based on the function of the concept of stand growth model,we choose age requirement(A),status level(N) and crown density(S) as input variables and DBH(D),tree height(H),,volume(M) as output variables to build and ttrain the stand growth BP artificial neural network model.And test the model fitting precision and inspection accuracy,the model fitting precision is81.65%、88.53% and 80.55%,inspection accuracy is 83.13%、79.47% and 80.79%,these show that neural network modeling has better fitting precision and adaptability for the stand growth,and has good prediction ability.