Fault diagnosis of a cooling package for AECS based on Principal Component Analysis

In this paper a method of Principal Component Analysis (PCA) is presented in doing the fault detection in the cooling package of Aircraft Environmental Control System (AECS). The cooling package is constituted of the heat exchanger and turbine. Due to the complexity of the fault pattern, the multitude of the data, and the relationships among so many data, the system fault is difficult to detect. In order to solve these problems, a PCA-based fault detection model is established to do the simulation research; by collecting the process data represented the fault characters, and using the PCA to analyze these process data. The simulation results show that the new algorithm is proved to be efficient; PCA has a good fault detection performance and a high sensitivity to predict the early fault. Besides, the fault can be located correctly by the PCA statistics and the contribution process diagram. Finally, some necessary sensors need to be added in the PCA and the AECS according to the schematic of aircraft air circulation system for fault detection and isolation. For example, temperature sensor and flow sensor can be added in the outlet of heat exchanger, and a rotation speed sensor can be added in the turbine cooler.