Abstract: Current leak detection system of pipeline only works based on an analysis method on individual pipe sections, which often cause a higherfalse alarm rate and report failure. A leak detection method is established to realize the function in mistake identifying, say abnormalpressure, through introduction of site operation information from overall situation of product oil pipeline network. Fuzzy neural networkis taken as a major classification method and the position, time and variation of working condition adjustment, and negative pressure wavetaken as an input of the fuzzy neural network. It is trained to identify causes of negative pressure waves so as to shield impacts of theworking condition adjustment on the pipeline leak detection system. Through applied experiment in Phase 1 Shandong and Anhui Pipeline ofNorth China Product Oil Pipeline Network, the feasibility of the method is verified.