%0 Journal Article
%T 基于TCN-BiLSTM的航空发动机寿命预测<br>Aero-Engine Life Prediction Based on TCN-BiLSTM
%A 李振乾
%A 裴莹莹
%A 陆莹
%A 王俊兴
%A 王晓彤
%A 韩金颖
%J Computer Science and Application
%P 163-176
%@ 2161-881X
%D 2025
%I Hans Publishing
%R 10.12677/csa.2025.151017
%X 航空发动机是飞机的重要组成部分&#65292;其性能和可靠性直接影响飞机的安全性和经济效益。针对航空发动机寿命预测精度低和数据复杂度高的问题&#65292;提出基于一种基于TCN和BiLSTM融合模型用于航空发动机寿命预测。该模型通过TCN捕捉长期依赖关系和处理长序列数据&#65292;通过BiLSTM处理上下文信息和提取高级特征&#65292;将筛选出来的数据特征输入TCN-BiLSTM模型中预测航空发动机的剩余寿命。本文采用NASA开发的C-MAPSS数据进行实验&#65292;通过数据仿真验证&#65292;模型可以较为准确地预测出航空发动机的剩余寿命&#65292;并与SVG、MLP、CNN、LSTM和CNN-LSTM的预测结果相比较&#65292;对比结果证明TCN-BiLSTM模型的RSME和Score均低于上述模型方法&#65292;从而证明本文提出方法预测效果更好。<br>Aero-engine is an important part of aircraft, and its performance and reliability directly affect the safety and economic benefits of aircraft. Based on the problem of low accuracy and high data complexity of aero-engine life prediction, a fusion model based on TCN and BiLSTM is proposed. The model captures long-term dependencies and processes long sequence data through TCN, processes context information and extracts advanced features through BiLSTM, and inputs the screened data features into the TCN-BiLSTM model to predict the remaining life of the aero-engine. In this paper, the C-MAPSS data developed by NASA is used to experiment. Through data simulation, the model can accurately predict the remaining life of the aero-engine, and compared with the prediction results of SVG, MLP, CNN, LSTM and CNN-LSTM, and the comparison results prove that the RSME and Score of TCN-BiLSTM model are lower than the above model method, thus proving that the prediction effect is better.
%K 航空发动机&#65292
%K 剩余寿命预测&#65292
%K 融合&#65292
%K TCN&#65292
%K BiLSTM<br>Aero-Engine
%K Residual Life Prediction
%K Fusion
%K TCN
%K BiLSTM
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=106331