基于残差修正的非等时距GM(1，1)模型及其应用

Unequal Interval Forecast Grey Model GM(1,1) Based on Residual Error Correction and Application

针对很多预测案例中历史数据序列非等时距的特点,构建了非等时距的GM(1,1)预测模型,将序列的时间间隔作为乘子嵌入模型中,同时通过动态采用最新分量作为初始值、动态优化背景值和累积残差修正等方法,解决了非等时距预测模型长期预测精度不易控制的难题.将该模型应用于某发动机油液监测数据预测中,预测效果较好.

In order to solve the forecast problem of unequal interval monitor data, an unequal interval forecast grey model（UIFGM1,1） is established, the unequal interval is set into the model as an multiply factor; in order to improve the accuracy of UIFGM when it is used in long time forecast, an dynamic initialization value metabolism method , an background value optimized method and the residual error correction method are also introduced into the model, an example on oil monitor of engine is applied to prove the model＇s forecast accuracy.