Fault Diagnosis of Aero-engine Based on Support Vector Machines

The aero-engine has a complex structure, with little and nonlinear fault data, in this paper, it shows a type of fault diagnosis approach based on Support Vector Machine (SVM) combining the characteristic of aero-engine fault data. In addition, during the Directed Acyclic Graph([1]) multiclass classify algorithm, firstly computing the Class Mean Value([2]) of example data, then get the classify priority level of fault type, construct the multiclass classify machine with the type of binary tree according to priority level. Under test, the approach is reasonable, and has better diagnosis speed and accuracy comparing with other classifying methods.