Health monitoring of marine diesel engine based on graph network
In order to realize the intelligent condition monitoring of marine diesel engines,the graph network al-gorithm is introduced.In this paper,a dual graph sampling graph network inductive learning algorithm(Dual-GraphSAINT)is proposed for diesel engine health monitoring.Firstly,in order to fully mine the potential infor-mation in the diesel engine vibration signal,the adjacency graph is constructed for the vibration signal and fault state based on the local and global consistency assumptions respectively.Then,the node state is learned based on the connection relationship between the data based on the double adjacency graph,so as to accurately monitor the potential health state during the operation of the diesel engine.At the same time,Dual-GraphSAINT is an in-ductive learning algorithm,which breaks through the disadvantage that the traditional graph network cannot gen-erate an effective embedded representation of unknown nodes,so it can analyze the health state of diesel engine in real time.The proposed health monitoring scheme obtained the best performance compared with the traditional data-driven and deep learning schemes.The health monitoring performance of the marine diesel engine is signifi-cantly improved.
万方数据
维普
