Research On Pipeline Leak Detection Method Based On Fuzzy Neural Network Optimized By PSO

'With the development of pipeline transportation, pipeline is getting more and more important in national economy status,but the pipeline leakage accident also sometimes occurs.The pipeline leakage will not only cause the wastage of power resource and the huge economic losses, but also pollute the environment, threat the safety of people.Therefore, long pipeline's safe operation receives more and more attention, the pipeline leakage detection and localization have become the current important research subject. There are many methods in pipeline leakage detection and localization. The method based on knowledge does not need to establish the complex model of pipeline system, and its diagnosis result is not affected by the uncertainty factors,and it has the qualitative and the quantitative analysis dual effects, so it receives widespread attention in the field in recent years. Based on this, the methods of pipeline leakage detection and estimation have been further studied, under the background of the actual data in the pipeline leakage network, in the earlier period based on the fuzzy BP neural network's leak detection method study's foundation, in view of existence some questions,with the union of some new technologies and new achievements in recent years,in this work.The main contents as the following aspects:1.In view of the fuzzy BP neural network exists convergence rate slow and easy to fall into partially smallest and so on questions,this article has constructed the fuzzy RBF neural network pipeline leakage detection method optimized by PSO, the real data's simulation research confirmed that the fuzzy RBF neural network has a better reliability and the accuracy than the fuzzy BP network.2.In view of general fuzzy neural network operating function to fuzzy logic fusion insufficiency, and the weight optimization easy to fall into partially most excellent and so on. In the article, the fuzzy neural network that the fuzzy operator of the general probability, the probability and generalized substitutes its operating function is used in pipeline leakage detection, using the divergent-convergent PSO(DCPSO) algorithm to optimize the weights of this fuzzy neural network, established the DCPSO-FNN neural network; based on the fixed weights, using basic PSO and DCPSO separately to optimize the pessimism, happy parameter of the generalized fuzzy operator, the simulation result indicated that this method has the higher detection and estimation precision. 3.Considered the structure of the network is quite huge, and the partial weights' influence are vary weak, in order to enhance the detection efficiency of the network, this article has carried on cutting out this DCPSO-FNN neural network based on certain cutting out algorithm,obtained the more simpler and direct fuzzy neural network.After the real data simulation, confirmed that the cutting out network has the accuracy and the reliability to pipeline leakage detection and estimation, thus has provided the convenience for the practical application.