Abstract | ||
---|---|---|
A method based on the neural network to predict the strains of the gas generator in a liquid rocket engine is presented for the fault analysis of the gas generator. A modified back-propagation algorithm is proposed to train the neural network. The training and testing samples are generated with an experiment. In the experiment, four strains in the risk domain of the gas generator and three forced displacements of the flange are employed to generate the sample patterns. To reduce the number of training samples while maintaining the sample completeness, the variation of samples is arranged using an orthogonal array. Results indicate that the method is helpful for the fault analysis of the gas generator and evaluation of the strain level caused by assembling errors. |
Year | DOI | Venue |
---|---|---|
2008 | 10.1109/PACIIA.2008.22 | PACIIA (1) |
Keywords | Field | DocType |
neural network,training sample pattern,training sample,orthogonal array,three forced flange displacement,strain level,aerospace components,forced displacement,gas generator,backpropagation,liquid rocket engine,modified back-propagation algorithm,strain prediction,failure analysis,fault diagnosis,flanges,sample pattern,oa technique,sample completeness,back-propagation algorithm,rocket engines,fault analysis,neural nets,bp neural network,bp neural networks,testing,strain,generators,artificial neural networks,force | Orthogonal array,Fault analysis,Pattern recognition,Computer science,Gas generator,Artificial intelligence,Liquid-propellant rocket,Flange,Artificial neural network,Backpropagation,Completeness (statistics) | Conference |
Volume | ISBN | Citations |
1 | 978-0-7695-3490-9 | 0 |
PageRank | References | Authors |
0.34 | 3 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Feng Li | 1 | 0 | 0.34 |
Changhua Deng | 2 | 0 | 0.68 |
Shaowei Song | 3 | 0 | 0.34 |
Jie Duan | 4 | 0 | 0.34 |