基于1D-CNN和Bi-LSTM的航空发动机剩余寿命预测

Residual Life Prediction of Aeroengine Based on 1D-CNN and Bi-LSTM

剩余寿命预测对航空发动机的预防性维修有重要指导作用,是保障飞机安全运行,提高维修保障效率的重要手段。一维卷积神经网络(1-dimensional convolutional neural network,1D-CNN)和双向长短时记忆神经网络(Bidirectional long short memory, Bi-LSTM)被应用于航空发动机剩余寿命预测模型。首先,根据工程经验在多状态参数的主成分分析的基础上对退化过程进行随机分布拟合,得到综合性能退化量;然后将多变量时间序列样本和对应的性能退化量代入1D-CNN模型进行回归分析,从而得到性能退化分析模型;再通过Bi-LSTM对性能退化量进行时间序列预测,得到性能退化的未来趋势;最后通过设定性能退化阈值,得到剩余寿命预测结果,从而得到从多状态参数-性能退化分析-性能退化预测-剩余寿命预测的实时动态感知模型。实例分析结果表明,提出的混合模型与其他单一深度学习和传统模型相比,有更低的回归分析误差和退化预测误差,能够得到更准确可靠的剩余寿命预测结果。

Residual life prediction plays an important role in the preventive maintenance of aeroengine, and it is an important means to ensure the safe operation of aircraft and improve the efficiency of maintenance support. A residual life prediction model of aeroengine based on 1-dimensional convolution neural network(1 D-CNN) and bidirectional long short memory neural network(Bi-LSTM) is proposed. Firstly, according to the engineering experience, on the basis of the principal component analysis of multi-state parameters, the degradation process is randomly distributed and fitted to obtain the comprehensive performance degradation amount;then, the multi variable time series samples and the corresponding performance degradation amount are brought into the 1 D-CNN model for regression analysis to obtain the performance degradation analysis model;then, the performance degradation amount is predicted by the Bi-LSTM time series, The future trend of performance degradation is obtained. Finally, by setting the performance degradation threshold, the residual life prediction results are obtained, and the real-time dynamic perception model from multi state parameters performance degradation analysis performance degradation prediction residual life prediction is obtained. The results show that the proposed hybrid model has lower regression analysis error and degradation prediction error compared with other single deep learning and traditional models, and can get more accurate and reliable residual life prediction results.