The Application of NN-DS Theory in Natural Gas Pipeline Network Leakage Diagnosis

For the problems of low precision of conventional leakage diagnosis methods, the paper presents a novel natural gas pipeline network leakage diagnosis approach based on neural network and DS evidence theory.The principle is that different premonition information is dealt with by neural network and the result preprocessed is taken to the two levels information fusion by evidence theory for final diagnosis result.The approach makes full use of redundant and complementary leakage information.Computer simulation validates the approach is feasible.Detection results show that the approach improves the precision of leakage diagnosis and decreases the recognition uncertainty.