Abstract:

In order to effectively monitor the potential performance degradation of core components in aircraft air conditioning systems, a baseline modeling method for aircraft air conditioning systems based on modified MSET was proposed in this paper. Firstly, QAR data of key components was selected as the feature parameter. Secondly, using healthy historical data as training sample, the density peak clustering method was used to screen state data and construct memory matrices for both non-low-temperature and low-temperature datasets. Then, the adaptive diagonal loading technique was applied to MSET process to reduce the abnormal fluctuations caused by the pathological state of the memory matrix. Finally, a performance baseline was established between the multivariate feature variables and the system operating state, and real flight data from Airbus A320 aircrafts was used to analysis. Results show that the proposed method can simultaneously establish performance baselines for multiple key components such as primary heat exchanger, main heat exchanger and air cycle machine. It can effectively detect flight cycles with performance degradation before component failure, and the detection results are relatively accurate, providing a reference standard for airlines' condition based on maintenance and health management.

Key words: aircraft air conditioning system, baseline modeling, multivariate state estimation technology (MSET), performance degradation, quick access recorder (QAR)