【摘要】 随着光纤承载业务量的逐渐增加,光纤线路产生的故障数据越来越多,传统方法对光纤故障数据的采集过程复杂,不能快速得到故障信息,导致光纤故障监测实时性能差,无法及时发现故障,因此,提出基于改进神经网络的光纤故障监测方法,通过小波变换法对光纤OTDR曲线进行分析,得到光纤信号中的所有细节,获取故障光纤信号,利用小波包对故障光纤信号进行特征提取,将提取到的光纤故障向量作为神经网络的输入,通过前向计算过程、误差计算和误差反向传播过程完成神经网络的训练。针对BP神经网络收敛速度慢和学习率、惯性系数确定方法不合理的弊端,利用自适应学习速率动量梯度下降反向传播算法对神经网络进行改进,给出利用改进方法对光纤故障进行监测的实现过程。实验结果表明,所提方法具有很高的故障监测精度。更多还原

【Abstract】 With the optical fiber carrying volume increase gradually,fiber optic lines of failure data is increasing.The traditional methods of optical fiber complex fault data collection process. can’t get the fault information quickly.Resulting in failure of fiber optic monitoring real-time performance is poor,and cannot find fault in time,therefore,in this paper,a fiber fault monitoring method based on improved neural network and through the method of wavelet transform analysis of OTDR curve is proposed. Getting all the details of fiber optic signal and obtaining fault fiber optic signal,fault optical signal to take use of wavelet packet feature extraction extracted the fiber optic fault vectors as the input of neural network by the forward calculation process,which through error calculation and error back propagation process to complete the neural network training. In view of the slow convergence speed of BP neural network and the disadvantages of vector,inertia coefficient determination method is not reasonable so we can through the adaptive learning rate momentum gradient descent back propagation algorithm to improve the neural network. Improved methods of optical fiber failure reasons are monitoring the implementation of the process. The experimental results show that the proposed method has high precision of fault monitoring.更多还原