In response to the issue of unclear early fault characteristics of diesel engine valves and an insufficient number of fault samples leading to low accuracy in fault diagnosis and recognition

this paper proposes a diesel engine valve clearance fault diagnosis method based on markov transition field (MTF) and convolutional attention mechanism module fusion with improved gaussian prototype network (CBAM-IGPN) embedded in the network under small samples. Firstly

based on the characteristics of the vibration signal of the diesel engine cylinder head

the one-dimensional vibration signal of the cylinder head is encoded into a two-dimensional feature map through MTF. Secondly

by improving the embedding network in GPN

the initial convolutional neural networks (CNN) are improved into deep convolutional neural networks (DCNN) to enhance the model′s mining of deep information in feature maps. CBAM is added to the convolutional layer of DCNN to enhance the models attention to important regions. Finally

the encoded feature map is input into CBAM-IGPN for training and testing to obtain classification results. The results indicate that the method proposed in this article has higher accuracy in diagnosing diesel engine valve faults under small sample conditions.