Investigation On Fault Diagnosis And Condition Monitoring Of Seal Face On Gas Seal

Gas seal is a form of non-contact sealing，it is widely used in thepetroleum chemical industry, aviation, etc because of the well sealingperformance. With more and higher utilization rate and the usagesituation more important, the failure of the possibility of gas sealing isincreasing. In order to improve the using safety of the gas sealing, thisarticle is focus on the failure of gas seal and condition monitoring of sealface.The related parameters of gas seal failure and heir membershipfunctions are first determined by used of fuzzy neural network knowledgein this paper. Then the effective fuzzy neural network is set up throughthe establishment of the MATLAB, the using of statistical data, buildingthe learning samples and diagnosis sample, many simulationoptimizations. At last the effectiveness of the sample fuzzy neuralnetwork is checked by the diagnosis sample of gas seal. The acousticemission gas seal test bed is also designed to test the relationship betweenthe related parameters (speed, seal gas pressure, the gas pressure sealresistance) and the gas seal and acoustic emission signal, the test resultsare compared with the theory, and at the same time, some relevant method of gas seal failure analysis and the acoustic emission signal isputted forward.A new using method about fuzzy neural network structure on the gassealing failure diagnosis is put forward and the acoustic emission gassealing test bed is set up in this paper, which put forward new ideas in gasseal failure diagnosis research and condition monitoring of seal face.