Fault identification of aero-engine central rod-fastened rotor system based on multi-source information fusion and graph convolutional network

Accurate fault identification is the key to realizing the fault diagnosis and condition monitoring of aero-engines. Aero-engine rotor systems work in high-temperature, high-speed and high-pressure environments. It is extremely prone to failure. In this article, a rotor system fault identification method based on multi-source information fusion and graph convolutional network (MSIF-GCN) is proposed. Firstly, a simulation test bed for the center rod-fastened rotor system of aero-engines was built. Experiments with different fault states were carried out on the test bed. The corresponding acoustic emission, sound and vibration signals were collected. Secondly, the acquired signals were converted into two-dimensional time-frequency maps using continuous wavelet transform. The spatial structure between the time-frequency maps was encoded through the adjacent matrix. The waveform reverberation, dispersion characteristics and multimodal characteristics of various signals were deeply explored. The MSIF-GCN model was established to realize fault identification. The accuracy of fault identification reached 99.77%. Thirdly, the fault identification situation when using a certain type of sensor data alone was studied. The positive effect of multi-source information fusion methods on model accuracy was demonstrated. Finally, the superiority of the MSIF-GCN model was further verified by comparing it with other well-established fault identification methods. These results show that the method proposed in this article can be used as an effective means of fault identification and diagnosis for aero-engine center rod-fastened rotor systems.