The total length of the global oil and gas pipelines has increased rapidly during the last decades. Subsequently, prewarning systems for pipeline intrusion damage incidents have become significantly important. The traditional remote optical fiber prewarning systems however often have high false-alarm and missing alarm rates due to the complex environment that reduces signal recognition accuracy. To address this issue, in this paper we analyze the signal characteristics of various dangerous events and then use dimensional signals in the space-time domain to develop a novel signal recognition method based on object detection. Our objective is to improve recognition accuracy and reduce the false alarm rate. Our experiments and test scenarios confirm that the recognition accuracy of this method is greater than 98. We also examined the proposed method on Pipechina MaFang to MiYun with a total length of 41 km, and show that it significantly improves signal recognition and further provides good adaptability under various interferences.
展开▼
