Remaining useful life prediction of FCV for aircraft air-conditioning system based on CAReCNN and Sequence-to-Sequence
Lists
Only you see this
Abstract
The aircraft's air conditioning system, a critical subsystem, ensures cabin stability across varying flight conditions. However, it is essential to forecast its remaining useful life (RUL) due to the elevated failure rate resulting from prolonged operation in extreme environments, which impacts both safety and economic advantages. To solve the problem of strong coupling in multi-scale heterogeneous data and the difficulties of feature extraction in aircraft systems, a method for estimating how much longer something will work is suggested. It combines an attentional residual convolutional feature extractor with a sequence-to-sequence architecture. Initially, the data undergoes preprocessing to provide a sample dataset, which is subsequently processed by the attention residual convolution feature extractor to produce a feature map. Second, the feature map that was collected is used as an encoder input for the sequence-to-sequence architecture. This makes it easier to guess how long the related system will still work by using a decoder that is improved by the attention mechanism. The suggested method is validated using flow control valve (FCV) Quick Access Recorder (QAR) data from the air conditioning system of the aircraft, and the experimental findings demonstrate that the method effectively predicts remaining useful life.
BibTeX
@misc{huang2025remaining,
author = {Jiayi Huang and Shaojie Zhang and Pengfei Huang and Mengjie Zeng and Tianyu Tu},
title = {{Remaining useful life prediction of FCV for aircraft air-conditioning system based on CAReCNN and Sequence-to-Sequence}},
year = {2025},
doi = {10.1109/safeprocess67117.2025.11268216},
}Loading related content...