欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
针对目前预测农作物产量只利用年产量或其变形，而没有考虑气象因素对产量预测的影响导致误差偏大的问题，在基于商空间粒度理论框架下的农作物产量预测中，考虑气象因素如光照、平均气温、降水量对农作物产量的影响，提出利用支持向量机方法构造模型时气象时间序列进行数据挖掘（产量预测）。粒度分析和实验结果表明：混合粒度预测模型不仅降低了问题求解的复杂性，而且误差较低，其预测值平均绝对百分误差为0．8849，均方根误差37．3，希尔不等系数为0．0044，与其他预测模型相比误差最小。基于商空间理论的支持向量机产量预测模型可较好地应用于产量预测中。
英文摘要：
The current way to predict the yield of crops is based on annual output or the variation of annual output. However, it neglects the influence of meteorological factors on yield prediction, which leads to big errors. Based on Support Vector Machine （SVM） within the framework of quotient space theory, this paper takes the influences of such meteorological factors as light, average temperature and rainfall into consideration to predict the yield of crops. It is an initial attempt to adopt SVM for conducting data mining on meteorological time series （yield prediction）. The granularity analysis and experimental results show that the mixed-granularity-model can reduce the complexity of problem solving and yield smalter errors with mean absolute percent error of 0. 884 9, root mean squared error of 37.3, and the value of Theil Inequality Coefficient of 0.004 4. Among all the prediction models, the errors are the smallest. The yield prediction model based on SVM within the framework of quotient space theory has a good performance in yield prediction.
同期刊论文项目
同项目期刊论文
期刊信息
位置：