欢迎访问《中国农学通报》,

中国农学通报 ›› 2010, Vol. 26 ›› Issue (20): 131-135.

所属专题: 园艺

• 林学 园艺 园林 • 上一篇    下一篇

基于人工神经网络的天然栎树生长建模

段群迷 张广亮 高光芹 黄家荣 张作明   

  • 收稿日期:2010-06-21 修回日期:2010-08-27 出版日期:2010-10-20 发布日期:2010-10-20
  • 基金资助:

    基于

Growth Modeling of Natural Oak (Quercus acutissima Carr.) Based on Artificial Neural Network

  • Received:2010-06-21 Revised:2010-08-27 Online:2010-10-20 Published:2010-10-20

摘要:

以薄山林场天然栎树为研究对象,用人工神经网络方法研建其生长模型,对提高森林资源信息化管理水平具有重要的应用价值。以20株栎树解析木数据为训练样本,对所建模型进行训练和分析。结果表明:最佳网络结构为1:2:3,其总体拟合准确度为98.37%;模型的龄阶准确度最大99.29%,最小93.18%,平均96.52%;胸径、树高、材积的平均准确度分别为97.17%、99.30%、92.30%。总之,所建模型能够满足林业上森林资源调查与预测的精度要求,是一个网络结构简单直观、数学表达形式简捷、使用方便的单输入多输出神经网络模型。

关键词: 苏云金杆菌, 苏云金杆菌, 肠毒素基因, 定位, 序列分析

Abstract:

Taking natural oak (Quercus acutissima Carr.) from Boshan forest farm as research subject, a growth model of natural oak was created with artificial neural network modeling technology, which has an important application for improving informatization management of the forest resources. With the data of 20 oak analytic trees as training samples, the built model was trained and analyzed. The results showed that: The best network structure was 1:2:3 and the total fitting accuracy of the model was 98.37%; Concretely, the age gradation accuracy was 93.18% to 99.29% and the mean value was 96.52 %; The mean accuracy of breast height diameter, height and volume was 97.17%, 99.30%, 92.30% respectively. In a word, the built model would satisfy the accuracy requirement for investigating and forecasting the forest resources in forestry, and it was a single-input and multiple-output neural network model, the network structure of which was simple and intuitive, mathematical expression of which was short-cut, and it would be easy to use.