基于ARIMA的往复式压缩机气缸振动故障预测

Prediction of Cylinder Vibration Fault of Reciprocating Compressor Based on ARIMA

以中国石油青岛石化厂的P301A往复式压缩机气缸的振动速度数据为研究对象,使用自回归积分滑动平均模型(ARIMA),对其振动故障进行了预测。首先,使用单位根检验法,对设备的振动数据进行序列平稳性判定;然后,对通过平稳性判定检验的数据进行白噪声检验,绘制其自相关函数图及偏自相关函数图,并通过模型拟合确定最佳p、q参数值;最后,对序列进行预测,得到30组预测数据。对比实际数据与预测数据,两者相差很小,说明该方法在往复式压缩机气缸振动故障上预测效果良好。该研究将大数据分析与生产设备的故障预测相结合,相比于传统的设备参数在线监测来判断故障,在保证较高准确率的前提下可以提前预判设备故障,极大地提高了技术人员的工作效率,减少了设备因故障所带来的损失。

Taking the vibration velocity data of P301A reciprocating compressor cylinder in Qingdao Petrochemical Plant of PetroChina as the research object,the vibration failure of P301A reciprocating compressor cylinder is predicted by using ARIMA.Firstly,the unit root test method is used to determine the sequence stability of the vibration data of the equipment,and then the white noise test is carried out for the data passing the stability test.Then the autoregressive function diagram and partial autoregressive function diagram are drawn,and the best p and q parameters are determined by model fitting.Finally,the sequence is predicted,and 30 groups of prediction data are obtained.Compared with the actual data and the predicted data,the difference between them is very small,which shows that the method has a good prediction effect on the vibration fault of the reciprocating compressor cylinder.This research combines big data analysis with fault prediction of production equipment.Compared with the traditional online monitoring of equipment parameters to determine the fault,it can predict the equipment fault in advance under the premise of ensuring high accuracy,which greatly improves the work efficiency of technicians and reduces the loss of equipment caused by the fault.