基于趋势修正的灰预测模型优化研究

针对序列增长趋势不完全满足准指数规律时的灰色预测建模问题,提出基于GM(1,1)模型与序列增长趋势之间偏差修正的建模方法,将GM(1,1)模型还原式中的常数项作为灰变量处理,加入调整系数以缩小拟合值与实际值之间的增长趋势差异,利用灰色离散模型拟合调整系数的变化过程,将得到的调整系数拟合值带入原时间响应函数,进而得到趋势修正的原始序列拟合值;运用新的建模方法对南京市第三产业用电量进行拟合和预测,证明了方法有效提升了GM(1,1)建模精度,并且拟合序列和实际序列的灰色绝对关联度得到提高.

Optimization of Grey Model Based on Trend Correction

For the problem of modeling the sequence not meeting completely the quasi exponential law,this paper puts forwards a new method of correcting the difference of changing trends between the original sequence and the model by revising coefficients;the constant in the restoring formation is transformed to a grey variable,and an adjust coefficient is used to reduce the difference between the fitting sequence and the original sequence,then use the discrete grey model to simulate the coefficient sequence,with which the fitting value is acquired from the revised time response function;the application predicts the electricity power consumption of Nanjing's tertiary industry,and the result shows that the new modeling method has improved the accuracy of GM(1,1) model,as well as the absolute degree of grey incidence of fitting sequence and the original sequence has been improved.