针对某石化公司的常减压装置在炼制中东进口原油后冲刷腐蚀明显加重的问题，在管线弯头等部位进行定点测厚，对其腐蚀速率进行监控。监测的结果是准光滑数据序列，一阶累加生成后具有准指数规律，符合建立灰色微分方程条件。针对此灰色系统建立GM(1,1)模型，利用该模型对弯头部位的腐蚀速率进行模拟预测。经过检验发现，预测结果序列与原始数据数列之间的平均误差小于15%，预测精度达到85%以上，表明GM(1,1)模型用于常减压装置管线的腐蚀速率的预测是可行和有效的。

Serious erosion-corrosion was found in the crude distillation unit for proeessing middle east crude oil in a petrochemical company. The thickness of some positions like elbow was measured every month, and the thickness change was monitored. The monitored data series is quasi smooth sequence. The series after accumulated generating has quasi exponent law. The series meets the conditions for establishing gray differential equations. GM(1,1) model was established by grey system theory according to this series. The model was used for the simulated prediction of the elbow corrosion rate. The examined statistical data show that the average relative error was less than 15% and the prediction accuracy was above 85%. So, using GM(1,1) to predict the corrosion rate of crude oil distillation unit's pipelines is feasible and effective.