Risk function based fuzzy judge SVM algorithm and its application

The concept of risk is introduced and a fuzzy judge based SVM (FJ-SVM) is put forward. Fuzzy theory is applied to improve the algorithm of SVM, the judge probability for fault is increased and diagnosis is more sensitive to fault. So the probability of mistaking fault for nature is decreased and the slight fault of facility can be identified correctly and timely. It is very important for improving the security of facility in the process. In this paper, the diagnosis example of working status in turbine axletree and the classification results are given to prove the feasibility of this algorithm. The algorithm of fuzzy samples based SVM (FS-SVM) is compared with FJ-SVM to indicate the rationality of FJ-SVM.