Research on Fault Recognition Method Based on Variable-Risk SVM

Because the tradition methods of fault pattern recognition can not distinguish the different loss by different misjudgments, the variable-risk support vector machines (SVM) is proposed in this paper. Then, the optimal classification face is redesigned and expert's experience is integrated when using an actual data to recognize the fault, which makes the result more reliable. Finally, this method has already applied in the diesel engine fault diagnosis successfully.