基于数字孪生及机器学习的设备故障预警研究

结合数字孪生和机器学习技术,可以实现对设备预警的更加及时和准确的监测,不仅能够帮助提高生产效率,减少设备故障损失,还能够降低人为干预的需求,提高设备管理的智能化水平。通过本文设计的孤立森林模型,在设备故障预警中具有较好的性能,相比高斯离群监测,在AUC方面提高了0.078,相比支持向量机在AUC方面提高了0.0977,相比随机森林模型在AUC方面提高了0.1023,相比One Class SVM模型在AUC方面提高了0.0685,并且通过本文设计的数字孪生系统,可以将预警结果进行实时监测,提高对设备故障的预警能力。

By combining digital twin and machine learning technology,more timely and accurate monitoring of equipment warning can be achieved.This not only helps to improve production efficiency and reduce equipment failure losses,but also reduces the need for human intervention and improves the intelligence level of equipment management.The isolated forest model designed in this article has good performance in equipment fault warning.Compared with Gaussian outlier monitoring,it has improved AUC by 0.078,support vector machine by 0.0977,random forest model by 0.1023,and OneClassSVM model by 0.0685.Moreover,through the digital twin system designed in this article,the warning results can be monitored in real time,improving the warning ability for equipment faults.