LEAKAGE LOCALIZATION FOR PIPELINES BASED ON BP NEURAL NETWORK AND EXPERIMENTAL VERIFICATION

Research work based on pressure gradient method shows that the location of leakage depends on the friction factor. And the traditional methods using the friction coefficient are not suitable for determining the leakage location of running pipelines. The method to locate leak point is put forward based on BP neural network forecasting friction coefficient. The BP neural network forecasting friction factor is determined from the average flow rate before and after leakage occurrence. Experimental results confirm that this method can effectively forecast the friction coefficient and reasonably locate the leak point of long-distance pipeline.