基于KPCA-BLSTM的航空发动机多信息融合剩余寿命预测

Remaining Useful Life Prediction of Aero-engine Multi-information Fusion Based on KPCA-BLSTM

复杂航空发动机在运行过程中易出现多退化信息而导致寿命预测不精确的问题,为此提出基于核主成分分析(KPCA)和双向长短时记忆(BLSTM)神经网络的多信息融合寿命预测模型。首先采用KPCA对多维退化数据集进行降维处理和信息融合,得到能够表征设备退化的低维特征数据集;然后利用BLSTM神经网络对带有多维退化信息的航空发动机剩余寿命进行预测,得到监测数据与剩余寿命的映射关系;最后采用CMAPSS航空发动机退化数据集对提出的多信息融合寿命预测模型进行仿真验证,并与其他三种模型结果进行对比。结果表明:KPCA-BLSTM神经网络能够对多维退化信息下的剩余寿命进行精准预测,本文提出的预测模型的误差与得分优于其他三种模型,而且预测精度更高。

The inaccurate life prediction caused by multiple degradation information can be easily appeared during the operation of complex aero-engine,a multi-information fusion life prediction model based on kernel principal component analysis(KPCA)and bidirectional long short-term memory(BLSTM)neural network is proposed.Firstly,the kernel principal component method is used to perform dimensionality reduction and information fusion on the multi-dimensional degraded data set to obtain a low-dimensional feature data set that can characterize equipment degradation. Then,the BLSTM neural network is used to predict the remaining useful life(RUL)of aero-engine with multi-dimensional degradation information to obtain the mapping relationship between the monitoring data and the remaining life. Finally,the C-MAPSS aero-engine degradation data set is used to simulate and verify the proposed multi-information fusion life prediction model,and the results are compared with other three models.The results show that the proposed KPCA-BLSTM neural network can predict RUL under multi-dimensional degradation information accurately,the error and score of the proposed model are better than the other three models,and the model has higher prediction accuracy.