SVR的树木生长过程建模及其参数优化研究  

Growth Process Modeling for Tree Based on SVR and Its Parameter Optimization

作　　者：王珊[1] 燕飞[1] 

机构地区：[1]北京林业大学,北京100083

出　　处：《湖南农业科学》2010年第2期103-106,共4页Hunan Agricultural Sciences

基　　金：国家"948"项目(2008-4-62)

摘　　要：计算机树木模型可以广泛地应用于理解、预测和调控树木的生长和产量,在树木生长及生态系统等领域的研究中已经成为一种很重要的研究方法。采用支持向量回归对树木高度进行建模与预测,减小了样本数据量小造成的误差,克服了树木生长方程选择的困难。对支持向量机的参数采用遗传算法进行筛选,提高数据拟合精度。最后将预测结果与采用各类生长方程模型的预测结果进行比较,结果表明用支持向量机预测方法有较高的精度。Tree model established on computer can be widely used to understanding, predicting and controlling the growth and production of tree, which had become one of the most important methods in the domain of tree growth and ecology system. Using support vector regression machine to model and predict the height of trees, which could reduce errors caused by little size of data and solve the problem that it＇s diffcuh to choose growth function of tree. Parameters of support vector regression machine were selected by genetic algorithm to improve the precision of data fitting. In the end, the predict results gotten from support vector regression machine were compared with the predict results gotten from each growth function models, which indicated that the predict method of support vector regression machine has the higher accuracy.

关 键 词：树木 生长过程模拟 支持向量回归机 遗传算法 

分 类 号：S111[农业科学—农业基础科学]