| 侯姗姗,王鹏新,田苗.基于相空间重构与RBF神经网络的干旱预测模型[J].干旱地区农业研究,2011,29(1):224~230 |
| 基于相空间重构与RBF神经网络的干旱预测模型 |
| Application of phase space reconstruction and RBF neural network model in drought forecasting |
| |
| DOI：10.7606/j.issn.1000-7601.2011.01.41 |
| 中文关键词: 预测 条件植被温度指数 相空间重构 径向基函数神经网络 |
| 英文关键词:forecasting vegetation temperature condition index phase reconstruction RBF neural network |
| 基金项目:国家自然科学基金项目（40871159，40571111,40371083）; 国家高技术研究发展计划课题（2007AA12Z139）; 欧盟FP7项目(Call FP7-ENV-2007-1 212921) |
|
| 摘要点击次数: 184 |
| 全文下载次数: 141 |
| 中文摘要: |
| 通过引入混沌和相空间重构理论，将一维时间序列重构为多维序列，并与径向基函数神经网络模型相结合，建立了基于条件植被温度指数(VTCI)的干旱预测模型，并对其进行了验证。结果表明：干旱预测结果与实际干旱监测结果中的VTCI最大值、最小值、平均值及标准差十分接近，所有样本点的预测值的相对误差的绝对值均低于9%。经过α=0.05的显著性水平检验，模型预测值与实测监测值的相关系数达0.99，表明模型预测精度较高，预测结果在实际工作中具有一定的应用价值。 |
| 英文摘要: |
| The chaos theories and phase reconstruction are introduced into drought forecasting, and the vegetation temperature condition index (VTCI) time series are expended to multivariate time series. Based on multi-dimension VTCI time series, phase reconstruction and neural network model are combined to establish the drought forecasting model. The results show that the minimum, maximum, average and standard deviation between predicted value and measured value are very close, all the sample sites predictive values of the relative error are less than 10%, which indicates that the predictive method has high accuracy. And after α=0.05 significance level test, the correlation coefficients between the predicted value and measured value are all around 0.99. The method also demonstrates the utility and efficiency for drought forecasting. |
| 查看全文 查看/发表评论 下载PDF阅读器 |