摘要
为了提高现有烤烟醇化过程中化学成分预测模型的精度,提出了组合预测模型。选取灰色GM(1,1)模型、三次指数平滑、一元非线性回归3种单一预测模型,采用标准差法分配权重,对单一预测结果进行集成,以还原糖为例,建立了化学成分的组合预测模型。结果表明,组合预测模型能够较好的提高预测精度,较GM(1,1)、三次指数平滑和一元非线性回归模型预测精度分别提高了0.0184、0.0283和0.0082。
Combination forecasting model of chemical constituents during the aging process was put forward for improving pre- cision of existing models. Using standard difference method to assignment weight, GM （ 1, 1 ）, triple exponential smoothing and one unknown non linear regression were used to integrate combination forecasting model. Precision of the model was su- perior to GM （ 1, 1 ）, triple exponential smoothing and one unknown non linear regression respectively increase by 0.0184, 0. 0283 and 0. 0082.
出处
《安徽农学通报》
2012年第21期67-69,共3页
Anhui Agricultural Science Bulletin
基金
浙江中烟工业有限责任公司资助项目"打叶复烤加料醇化剂筛选与应用研究"
关键词
组合预测模型
标准差法
烤烟
醇化过程
Combination forecasting model
Standard difference method
Flue -cured tobacco
Aging process
