| 摘要: |
| 为解决新疆加工番茄病虫害预测问题中样本数据的非线性和高维性等问题,采用投影寻踪回归模型对加工番茄病虫害预测进行研究。根据新疆某种植基地的样本数据,将投影寻踪回归模型与改进状态转移算法结合,建立了改进状态转移算法优化的基于Hermite多项式的投影寻踪病虫害预测模型。投影寻踪病虫害预测模型将高维的数据投影到低维空间,利用加入正交变换的状态转移算法优化得到投影方向和多项式系数。试验结果表明,利用该模型对新疆某种植基地2003-2008年的样本数据训练效果误差<0.2,等级预测达到完全正确;对2009-2011年的病虫害等级预测准确率>95%。基于改进状态转移算法的Hermite投影寻踪回归模型可靠性及预测精度很高,能有效的解决病虫害预测中存在的数据非线性、高维性等实际难题。该模型应用于加工番茄病虫害的预测具有一定的可行性和实用性。 |
| 关键词: 投影寻踪回归模型 状态转移算法 正交变换 病虫害预测 Hermite多项式 |
| DOI：10.11841/j.issn.1007-4333.2015.05.31 |
| 投稿时间：2014-12-05 |
| 基金项目:国家自然科学基金项目(61064005) |
|
| Projection pursuit regression model for pest prediction of processing-tomoto optimized by state transition algorithm |
|
WANG Cong, ZHANG Hong-li
|
| (College of Electrical Engineering, Xinjiang University, Urumqi 830047, China) |
| Abstract: |
| In order to solve the practical problem of nonlinear and high dimensions of sample data,which appeared on forecasting processing-tomato diseases and pests in Xinjiang,the projection pursuit regression model was used in this research.To provide theoretical basis for pest prevention and control,according to sample data of a planting base in Xinjiang,a new method based on projection pursuit regression model of Hermite polynomial and improved state transition algorithm is proposed.The model projects the high-dimensional data into low-dimensional space and uses state transition algorithm combined with orthogonal transformation to get the projection direction and polynomial coefficients.The experimental results showed that the effect error was below 0.2 and the training effect levels predict was exactly right when the model was used to train sample data during 2003-2008;The forecasting accuracy was over 95% when this model was used to predict the data of 2009-2011.The projection pursuit regression model of Hermite polynomial optimized by improved state transition algorithm has high prediction accuracy and reliability.It can effectively solve the practical problems exist in diseases and pests forecasting such as nonlinear and high dimensional data etc. |
| Key words: projection pursuit regression model state transition algorithm orthogonal transformation diseases and pests prediction Hermite polynomial |
