Abstract:  A fuzzy support vector machine (FSVM) is proposed to improve the accuracy of equipment fault diagnosis. FSVM is good at dealing with an equipment fault system or data with ambiguous characters. When equipment fault data have heterogeneous or irregular and uneven distribution characteristics, satisfactory results cannot be obtained by using a traditional FSVM based on single kernel. To overcome this drawback, a FSVM algorithm based on multikernel function is proposed to deal with equipment fault diagnosis. By applying this method to the rolling bearing fault diagnosis, the test result shows that this method is superior to an original FSVM and can identify rolling bearing fault patterns more effectively.