A Fault Diagnosis Method of Rotating Machinery Based on One-Dimensional, Self-Normalizing Convolutional Neural Networks.

As shown in Table 3 , the fault diagnosis accuracy shows a decreasing trend as the number of network layers increases; i.e., the model possesses the best generalization capability when it contains only a single layer.

also in 47,180 other papers

In the actual applications of rotating machinery, the working conditions are complex and variable, which makes it impossible to obtain adequate data samples under each load level; thus, the diagnosis capability under cross-load level conditions is also an extremely significant performance indicator.

also in 7,787 other papers