Research of Turbofan Engine Complex Faults Based on Manifold Learning
Aiming at the problem that complex fault of engine is nonlinear and overlapping,which makes the accuracy of fault diagnosis too low,this paper introduces the feature extraction method of supervised local tangent space transform(SLTSA),combines with SVM classifier fault diagnosis al-gorithm.Moreover,the dimensionality reduction,the number of neighbors and the selection principle of supervisory parameters that affect the performance of SLTSA are provided,and parameters are op-timized and selected by artificial intelligence.The simulation results show that the SLTSA can effec-tively extract the nonlinear features of the fault from the spectrum data,and improve the accuracy of fault diagnosis,reaching 95.46％.Therefore,the SLTSA algorithm designed in this paper can effec-tively extract the feature of complex fault and provide basis for accurate fault diagnosis.
万方数据
维普
