Turbine Engine Health Assessment Based on Graph Neural Network

Aiming at the research field of aero-engine health state assessment, for the problem of missing or unobtainable part of the monitoring data which leads to a big challenge in health assessment, this paper proposes a health state assessment method for complex systems based on Graph Neural Network(GNN), aiming to realize the health assessment of the system through partial inspection data. Firstly, the GNN node graph is drawn based on the structural mechanism among the components in the complex system, secondly, the health state of some components is obtained based on the monitoring data, and the node attributes and edge attributes are set to establish the Graph Convolution Network(GCN) health assessment model; finally, the effectiveness of the proposed method in the complex system is verified by the performance simulation data of an aviation turbofan engine.