多算法融合管道泄漏检测预警系统试验研究

管道泄漏不仅会造成环境污染,还可能引发火灾、人员伤亡等严重后果和巨大的经济损失。传统的管道泄漏在线检测与定位系统均使用单一算法进行信号处理,漏判误判多,应用效果较差。针对目前集输管道泄漏检测准确性和灵敏度较低的现状,研究一种多算法融合的方法,通过采集管道压力、流量等多种特征信号,采用最优加权融合算法,将专家经验分析法、向量机(SVM)模式识别法、序贯概率比检测法等多种信号处理方法进行综合应用,扩展了时间和空间上的检测范围,提高了泄漏检测系统的灵敏度和可靠性。利用正常运行集输管道进行了现场泄漏应用试验,试验结果验证了多算法融合检漏方法比单一检漏方法具有明显的优越性。

Experimental Research on Multi-algorithm Fusion Pipeline Leakage Detection and Early Warning System

Oil pipeline Leakage causes environmental pollution along with other serious consequences such as fire, casualties, and huge economic losses. The traditional pipeline leakage on-line detection and location system processes signal by a single algorithm which brings lots of omissions and misjudgments and results in poor effect. Aiming at the current situation of low accuracy and sensitivity of leakage detection in gathering and transportation pipelines, a multi-algorithm fusion method was studied. By collecting multiple characteristic signals such as pipeline pressure and flow rate, the optimal weighted fusion algorithm was adopted, and the expert experience analysis method, Vector machine(SVM)pattern recognition method, sequential probability ratio detection method, and other signal processing methods were comprehensively applied, which expanded the detection range in time and space, improved the sensitivity and reliability of the leak detection system. The field leakage tests on a working pipeline were carried out and verified the multi-algorithm fusion leakage detection method has obvious advantages over the single leakage detection method.