Parameter Identification Method for Interferometric Vibration Detection of Optical Fiber in Cable Trench

This paper focuses on the problems of cable external damage and proposes a parameter identification method for vibration detection in cable trench, based on the principle of Mach-Zehnder fiber interference. The different vibration signals are collected by the vibration test system, and the original signal data are preprocessed, which including time domain characteristics, wavelet packet energy feature, and IDT features. The Relief algorithm is used to calculate the weight of all kinds of features within the set. The vibration signals are classified and identified by support vector machines and BP neural networks. The experimental results show that the system is designed reasonably, and the recognition accuracy rate of vibration signals is high to over 80% averagely.