关键词: 最小二乘支持向量机, 可变惩罚因子, 遗传算法, 二叉树结构, 故障诊断

Abstract: Abstract: On how to minimize the interference of man-made factors on fault diagnosis, BTGAVPF- LSSVM fault diagnosis method was proposed. First, in view of the penalty factor to reduce support vector machine (SVM) selection is influenced by the researchers experience, the characteristics of variable penalty factor to established (VPF-SVCM), and proved that the algorithm of the dual problem. Secondly, because of the support vector machine (SVM) is not easy to solve the problem, using the least squares solution of VPF-SVCM, VPF-LSSVCM algorithm was put forward. Then, by using the genetic algorithm search VPF–LSSVCM, GAVPF-LSSVCM algorithm was put forward. Finally, according to the practical problems of fault diagnosis, build GAVPF-LSSVM algorithm of binary tree structure. Numerical simulation experiment results showed that compared with the original SVM algorithm, the proposed BTGAVPF - LSSVCM algorithm diagnosis is increased by nearly 14.3%. Therefore, in the field of the fault diagnosis accuracy requirements higher, BTGAVPF LSSVCM model is applicable.

Key words: Least Squares Support Vector Machine(LSSVM), Variable penalty factor, Genetic algorithm(GA), Binary tree(BT) structure, Fault diagnosis