Leak detection of oil transport pipelines based on prediction and modeling

For solving the problem of discriminating leaks and valve adjusting operations in leak detection of oil pipeline based on dynamic pressure transducer (DPT). In this paper, the fuzzy neural system (FNS) based chaos time series prediction method is used to detect and locate the leakage based on DPT pressure signals. Offline tests based on the real historical data demonstrate that the proposed method can not only discriminate the leak from valve adjusting operations effectively, but also achieve satisfying detection and location precision.