一种改进型神经网络的光纤预警系统适应性研究

针对传统长距离光纤预警系统识别入侵事件误报率高的问题，提出了一种基于改进型神经网络的光纤预警系统并研究了该系统在不同环境下的泛化适应性.系统的分布式传感部分应用Phi-OTDR技术，信号识别部分采用一种改进型的神经网络对入侵事件进行识别分类.最后通过3种情况下的实验，探究了系统的适应性.结果表明，该系统在光纤振动信号识别中有优秀的分类效果，并且在不同的环境条件下也具备良好的适应性. 

An Improved Neural Network Based Research on Generalization of an Optical Fiber Pre-Warning System

There is a high false alarm rate in identifying event process of traditional long-distance optical fiber pre-warning system. To solve the problem, an optical fiber pre-warning system was proposed based on an improved neural network and its adaptability to different environments was studied. Firstly, Phi-OTDR technology was applied to design the distributed sensing part of the system. And then, an improved neural network was used for the signal recognition part to identify and classify intrusion events. Finally, experiments were carried out in three cases to analyze the adaptability of the system. The results show that the system can provide an excellent classification effect in the recognition of optical fiber vibration signals, and it also has better adaptability under different environmental conditions.