一种用于航空发动机RUL预测的推理网络模型

Inference Network Model for RUL Prediction of Aero-engine

剩余使用寿命预测对于航空发动机的故障预测和健康管理至关重要.为解决传统长短期记忆网络只利用最后一步学习到的特征进行回归的问题,本文提出了一种基于注意力机制的航空发动机剩余使用寿命预测模型.利用长短期记忆网络处理航空发动机的时序数据,自动提取与时间有关的特征,采用注意力机制为不同传感器特征和不同的时间步进行了加权.此外,本文还考虑到了不同操作条件对发动机剩余使用寿命的影响,将自动提取的特征与操作条件进行了特征融合.实验结果表明,本文提出的模型能有效预测航空发动机的剩余使用寿命,为基于状态的维护提供了可靠的支持.

Remaining useful life prediction is very important for the failure prediction and health management of aero-engines.In order to solve the problem of traditional long-short term memory network using only the characteristics learned in the last step for regression,this paper proposes an attention mechanism-based aero-engine remaining useful life prediction model.The long-short term memory network is used to process the time series data of aero-engines,and the time-related features are automatically extracted.The attention mechanism is used to weight different sensor features and different time steps.In addition,this paper also considers the influence of different operating conditions on the remaining useful life of the engine,and combines the automatically extracted features with the operating conditions.Experimental results show that the model proposed in this paper can effectively predict the remaining useful life of aero-engines and provide reliable support for condition-based maintenance.