基于BP信号识别的光纤油气管道监测系统

介绍了一种基于BP神经网络信号识别算法的分布式光纤管道安全预警系统,利用提取的现场振动信号时域、频域短时和长时特征,对基于BP神经网络分类器模型进行训练,模型实现了对人工挖掘和机械挖掘的智能分辨。其中BP分类器模型的最大误报率为3.3 %,平均误报率为1 %,最大漏报率为3.2 %,平均漏报率为1 %。将该BP模型应用在不同时长的现场信号识别测试中,实现了最低为5 %的漏报率,因此BP信号识别算法能够实现对管线入侵信号的有效识别及分类,提升传感系统可靠性。

Fiber sensor pipeline monitoring system based on BP signal distinguish

This paper reports a distributed optical fiber pipeline safety early warning system which based on the BP neural network signal recognition algorithm.Using the time-domain,frequency-domain features of short-time and long-time of the vibration signals,the BP neural network model is trained.The model realizes the intelligent distinguish of artificial dig or mechanical dig.The maximum false alarm rate of BP classifier model is 3.3 %,the average false alarm rate is 1 %,the maximum missing alarm rate is 3.2 %,and the average missing alarm rate is 1 %.The BP model is applied to the signal distinguish test with different time lengths,and the lowest rate of missing alarm rate is 5 %.Therefore,the BP signal distinguish algorithm can effectively identify and classify the pipeline intrusion signal,and improve the reliability of the sensor system.