基于相位敏感光时域反射计(Φ-OTDR)的分布式光纤振动探测是海缆锚害预警的重要防护手段。针对探测海缆锚害的入侵事件类型判别需求, 提出将反向传播(BP)神经网络分类器应用于基于Φ-OTDR的海缆防锚害系统中, 介绍了基于Φ-OTDR+BP神经网络分类器的海缆防锚害系统原理, 采用信号时域特征和时频域特征作为特征向量, 构建基于BP神经网络的分类器, 实现了对入侵事件类型的判别。试验结果表明, 分类器的模式识别准确率达到100%。

Distributed optical fiber vibration detection based on phase sensitive optical time domain reflectometer(OTDR) is an important protection method for early warning of submarine cable anchor damage. Aiming at the requirements of distinguish intrusion event types for detecting anchor damage of submarine cable, the backpropagation(BP) neural network classifier is applied to the submarine cable anchor damage prevention system based on Φ-OTDR. The principle of the submarine cable anchor damage prevention system based on Φ-OTDR+BP neural network classifier is introduced. The characteristics of signal in time domain and time frequency domain are used as feature vectors. A classifier based on BP neural network is constructed to recognize the types of intrusion events. The experimental results show that the pattern recognition accuracy of the classifier reaches 100%.