GRU-BP在数字化车间关键部件寿命预测中的研究

Research on GRU-BP for Life Prediction of Key Components in Digital Workshop

准确的剩余使用寿命预测在数字化车间的预测和健康管理中起着至关重要的作用.如数字化车间安全保证中最重要的环节是故障的排查和维修,但是目前故障的排查和维修都是需要在停机的状态下完成,这不仅可能会影响生产加工的进度,还可能会增加人工干预的风险,剩余使用寿命预测则可以很好的解决这个问题.本文针对传统的剩余使用寿命预测方法缺乏输入时序的考虑,提出了一种基于GRU-BP的多对一双GRU层神经网络剩余使用寿命预测方法.并使用数控车间的铣削刀和轴承数据进行实验,验证了该方法在剩余使用寿命预测中可行,且精度优于输入时序无关的梯度增强方法 XGBoost和传统的BP神经网络.

Accurate remaining useful life prediction plays a crucial role in the prediction and health management of digital workshops.For example,the most important part of the digital workshop safety assurance is the troubleshooting and repair of the fault,but the current troubleshooting and repair of the fault needs to be completed in the state of shutdown,which may not only affect the progress of production and processing,but also may increase manual intervention. The remaining useful life prediction can solve this problem. In view of the lack of input timing in the traditional residual useful life prediction method,this paper proposes a GRU-BP based multi-pair and GRU layer neural network remaining useful life prediction method. The experimental results show that the proposed method is feasible in the prediction of remaining useful life,and the accuracy is better than the gradient boosting method XGBoost and the traditional BP neural network without considering the input timing factor.