【摘要】 利用多项式拟合我国春小麦产量,依据实际值散点图与拟合曲线偏离程度的大小,从拟合残差数据中给定一个阈值(异常临界值),将原始数据进行筛选。将残差绝对值大于异常临界值的时刻点及相应的原始数据选出,形成新的数据序列构建灰色灾变GM(1,1)模型,对未来春小麦产量可能出现的灾变点进行预测。去除灾变点数值的其余数据序列用新的多项式拟合曲线预测,提高预测精度,同时也减少单一预测模型可能出现的长期预测误差。更多还原

【Abstract】 China’s spring wheat yield is fitted by polynomials.Then,according to the magnitude of the deviation of the actual value of the scatter plot and the fitting curve,a threshold(abnormal threshold)is given from the fitting residual data to filter the original data.When the residual absolute value is larger than the abnormal threshold,the corresponding time point and the corresponding raw data are selected.Thus,a new data sequence is obtained.The new data sequence is used to construct the gray catastrophe GM(1,1)model to predict the possible catastrophe of future spring wheat production.The remaining data sequences to remove the disaster point values are predicted by the new polynomial fitting curve to improve the prediction accuracy.At the same time,it also reduces the long-term prediction error that may occur in a single prediction model.更多还原