Study On Fault Detection And Isolation Methods For LOX/Methane Rocket Engines

For the need of fault detection and isolation to the new LOX/methane rocket engine,this thesis researches on line fault detection and isolation methods in observation of line fault detection factor,and explores the methods of engine subsystems fault isolation based on fuzzy clustering and neural network.At the same time,a test-post analysis system for engine failure is developed.Firstly,this paper carries on a thorough analysis of the working principle and structure of LOX/methane,and determines the positions where faults frequently occur,according to the specific statistical data.Secondly,the changing rules of line fault detection factors and differential values in the condition of leakage and blockage are analyzed on the basis of pipe flow equation.With this law combined with adaptive threshold algorithm,the thesis detects the testing data for instance.Then the fuzzy clustering and neural network modeling method for remaining engine subsystems is studied,making full use of the advantages of clustering algorithm and the neural network algorithm.And it makes a further improvement of accuracy in modeling.The fault isolation method based on the models is validated by test data.Finally,according to the methods described in this paper,the test-post analysis system which is based on Matlab-GUI platform is established for LOX/methane engine.The research results show that factors in fault detection for the observation of ATA fault detection method can accurately detect the related line fault s.Simultaneously,the fault isolation method of fuzzy clustering and neural network can accurately isolate the fault in engine subsystems.The two methods have completed a full range of fault detection and isolation to LO X/methane rocket engine,that has high reliability and robustness.