Leakage Diagnosis Method for Pipelines Based on Multi-Weight Neural Network
For reasons of low accuracy of traditional leakage, a pipeline leakage diagnosis method based on multi-weight neural networks is presented to recognized leak signal in city gas pipelines. By using analysis and modeling a multi-weight neural networks are established at normal node to simplify network structure. The Information entropy of leakage characteristic parameters of negative pressure wave was used as input eigenvector respectively for primary diagnosis. It has been applied for leakage diagnosis in city gas pipelines with the whole computational process done by a computer. Results of simulation and tests show that this method has its advantage in dealing with multi-coupled fault situations and is featured by a high probability of accuracy, which not only proves the method to be effective, but also provides a theoretical basis and a new way for leak diagnosis of other pipelines.