The Study Of Oil Pipeline Weak Leak Detection And Location Based On Pressure And Flow Signals

The pipeline leak detection technology is the key of oil pipeline safetytransportation. One of the most economic and popular method of pipeline leakdetection is the negative pressure wave method. With proper signal processing toolsand techniques, the detection and location accuracy could be relatively high. But mostof the traditional methods are hard to detection the weak leak of pipeline due to thecomplexity and low signal noise ratio of weak leak signals. This paper proposed somenovel signal processing and detection approaches to the detection and location ofweak pipeline leak.A method of signal characters extraction based on statistical model changedetection theory was proposed. A signal test model was established based on thechange pattern of leak signals. The General likelihood ratio(r) based on GLRT testwas calculated as a character to represent the pressure and flow signals’ abruptnessand extent of change. The degree of change(d) was also calculated as a character torepresent the change degree of signals. After analyzed different kinds of pressure andflow signals the characters of r and d was proved effectively to distinguish the patternof change of the pressure and flow signals. Beside the signal Ap-proximateEntropy(ApEn) was calculated as a character to show the complexity of pressuresignals. The results of different kinds of signals proved the effectiveness of selectedcharacteristic.The small leak detection method was researched. An online threshold-baseddiscriminate algorithm was developed to be used in simple pipelines. This algorithmwas proved effective in field experiments. To complex pipelines BP neural networkswas utilized to discriminate the types of signals include pipe leak signals, pipeoperations signals and normal pipe signals. Use previous classified groups of differenttype pipe pressure and flow signals to train and test the network. The result accuracyof BP network is96%The modified MLE of change time in the pressure signals was considered aeffectively inflection estimation in the pipeline leak location method with negativepressure wave. The calculate method based on Dynamic programming was utilized toimprove the calculate speed. Use simulated signals to validate the precision and time consuming of the DP program. The modified estimation of change time based on DPwas proved satisfied with the requirement of weak leak detection. Field leak locationexperiments shown that algorithm designed in this paper could achieve a locationprecision less than500m.