ARMA模型在输油泵振动特征值趋势预测中的研究

Research on ARMA model in trend prediction of vibration characteristics of oil pump

输油泵机组作为管道输油系统的主要设备,其运行状态影响着油田正常的生产。传统的针对泵机组的“预防为主,计划检修”检修模式由于缺乏实时性、灵活性,已无法满足设备安全稳定运行的要求。基于这种缺陷,通过“预测性维修”的方式实时掌握输油泵组运行状态趋势能够及时察觉设备故障,为企业计划性备品备件、提高生产效率、确保生产安全提供有力保障。因此,本文提出了一种基于ARMA模型的输油泵振动特征值趋势预测方案。通过对某输油站车间输油泵进行在线监测,得到测点振动信号特征值的历史数据。使用历史时间数据进行建模,进行当前时间数据预测。通过实际数据与预测数据对比,发现ARMA模型可以较好地拟合输油泵振动信号特征值,满足当下预测趋势的需求。该方案较于其他传统趋势预测方法准确率高、计算迅速、易于理解,填补了输油泵振动信号趋势预测方面的空白。

As the main equipment of the pipeline oil transportation system,the operation of oil pump unit influences normal production of the oil field.The traditional overhaul way of“prevention first,planned overhaul”for the pump unit lacks of real-time capability and flexibility,so it cannot meet the requirements of safe and stable operation.Based on this deficiency,by grasping the operating trend of the oil pump unit through“predictive maintenance”,equipment failures can be detected in time,which provides a strong support for enterprise’s planning spare parts,improving productivity and ensuring production safety.For this reason,this paper proposed a trend prediction scheme for the vibration characteristic value of oil pump based on ARMA model.Through the on-line monitoring of the oil pump in a gas station workshop,the historical data of the characteristic values of the vibration signals of the measuring points were obtained.The historical time data was used to model the algorithm,and the current time data was predicted.By comparing the actual data with the predicted data,it is found that the ARMA model can better fit the characteristic value of the oil pump vibration signal,which meets the demand of the current prediction trend.Compared with other traditional trend prediction methods,this scheme has the advantages of high accuracy,fast calculation and easy understanding,which fills the gap in the trend prediction of oil pump vibration signals.