基于灰色预测和线性回归的烟叶产量预测模型

Forecast model of tobacco production based on grey dynamic model and multivariate linear regression

为解决作物产量预测样本数据偏少的问题,同时体现出产量与气候因素的关联性,采用灰色预测模型与线性回归相结合方法构建烟叶产量预测模型(TPFM),符合实际需求也便于应用。模型基于灰色预测GM(1,1)模型,利用历年烟叶产量计算产量趋势曲线,根据实际需求划分不同产量状态。使用线性回归方法确定一个多种气候因素与烟叶产量之间的关联公式,通过该公式提取每个状态的状态特征值,然后与预测年份的状态特征值进行比对,与其偏移最小的一个状态就是该年份的产量状态。经过数据验证,模型准确性较高,能够满足烟草预测的要求。

In this paper, grey dynamic model and multivariate linear regression model were used to build Tobacco Production Forecast Model （TPFM）, which solved the problem of too few crop forecast data, and reflected the correlation of yield and climatic factors. This model meets the practical needs and facilitates the practical application. According to the outputs of tobacco over the years, TPFM calculated the trend curve based on grey dynamic model, and got different production states according to the actual needs. An incidence formula between climatic factors and tobacco production could be determined using the linear regression method, and then the eigenvalue of each production state could be extracted using this formula. Comparing the eigenvalue of each state with the eigenvalue of the year which is also calculated with the incidence formula, the state with minimum offset is the production state of that year. TPFM has been verified through actual data, and it has high accuracy in forecasting tobacco production.