Study On The Ship Pipeline Leakage Detection Method Based On Fuzzy Neural Network

Ship pipeline system is used to transport liquid and gaseous material, now it is animportant part of ship transportation. However, ship pipeline leakage and breakageaccidents which will cause serious injuires，economic loss and environmental pollutionotfen happen due to some factors, for example，less solid and wear of pipe joint,corrosion damage of pipeline itself, vibration duirng ship operation as well asman-made improper operation. Therefore, the technology of ship pipeline leakagedetection has become an important research topic. At present a lot of methods have beenapplied in research of pipeline leakage detection, which can be divided into threecategoires, namely，the method based on information, the method based on model andmethod based on knowledge. Especially the method based on knowledge is generallyconcerned by industry delegates. In this paper fuzzy neural network is put forward,which is applied to ship pipeline leakage detection. Research contents mainly includethe following aspects:1? By analyzing the mechanism of ship pipeline leakage，pressure and flow signalis used for the research basis of leakage detection. By using previous knowledge andexperimental data of pipeline leakage detection for reference and analyzing the workingcondition of pipeline leakage, the changes of flow and pressure are obtained whenleakage happening, fuzzy neural network is built for ship pipeline leakage detection，atthe same time, fuzzy diagnosis rule and learning samples and test data are built.2.The network parameters are trained by fuzzy BP neural network. The reliabilityand accuracy of network diagnostics results are veirfied by the simulation expeirment.3.According to the shortcomings that slow convergence speed and easy to fall intolocal optimum for ofzzy BP neural network, fuzzy RBF neural network is built. Thereliability and accuracy of this method are veirfied by the simulation expeirment of realdata on the spot‘4，Based on theoretical research for ifizzy neural network of ship pipeline leakagedetection, ship pipeline leakage detection system based on this method is developed,5. Consider from the basic composition of ship pipeline leakage detection sysetm,the sources of random error and system error are analyzed in this system. Finally，themethod is put forward to amend and restrain random error, sensor error, data acquisitionerror and date processing error.