【摘要】 基于振动特征量研究提高灰色模型 (GM)建模精度的方法。建立了基于机械系统振动特征量的时间序列趋势预测GM(1,1)并进行了后验差精度检验 ;针对模型精度低问题对模型修正进行了重点研究。在基于原始时间序列的一次累加生成运算 (1 AGO)序列与它的预测序列间的 1 AGO残差序列对基本GM (1,1)修正失效的基础上 ,提出了设置阈值考虑相对误差、调节量系数考虑原始数据幅度分布的GM(E ,1,1)。实例表明 ,该模型用于机械系统振动特征量预测时具有动态自适应补偿误差影响的功能更多还原

【Abstract】 The grey model (GM) is used to predict the performance of a system without sufficient information or with highly nonlinear property. The methods to enhance the modeling precision for the universal GM(1,1) applied to predicting the development of a mechanical pump based on vibration features were studied. The precisions of the models were evaluated with posterior error indicators. Because the precision was very low, a method for modifing its model were investigated specifically. Since it modified the GM(1,1) model on the basis of 1-AGO series of residual errors between original data and predicted data,and it was invalid, therefore, a new GM(E,1,1) based on the universal GM(1,1) is proposed corresponding to stochastic factors. First, it considers relative errors between original data and GM(1,1) predicted data and sets up threshold to control the action modified. Secondly, the adjusted amount coefficient is used to control the valid amount based on original data distributed magnitudes. The case shows that the GM(E,1,1) has "Good" indicator for predicting the development of the pump′s vibration features. It is able to adapt to the dynamic variety of external factors.更多还原