Research of Fiber-Optical Fault Diagnosis Based on Support Vector Machine (SVM) Mining

Abstract

The distribution range of fiber-optical is wide, and the alarm information is scattered, the managers cannot monitor on the line fault in real time. The fiber-optical fault diagnosis is difficult, and the traditional diagnosis cannot form an effective fault diagnosis mechanism, the fault diagnosis effect is not satisfying. Large number of fiber-optical data can be used for the fiber-optical fault nodes localization, the fiber-optical fault nodes can be predicted based on the sampled data. For improving the fault diagnosis efficient and effectiveness, an improved fiber-optical fault diagnosis method is proposed based on the support vector machine (SVM) fault feature extraction and data mining. The fault feature of fiber-optical is extracted, and the fault feature matrix is established. With the SVM filter, the redundant data of features are deleted, the computing cost and complexity of fiber-optical fault diagnosis is reduced. According to the relevant principle of SVM, the fiber-optical fault feature classification problem is transformed into a nonlinear programming problem, and the solution is obtained. Finally, the fiber optical fault diagnosis is realized. Simulation result shows that, the improved algorithm is applied in the fiber-optical diagnosis, the fault diagnosis accuracy is improved greatly, and it has strong stability and practicability in application.