Classifier Utility Modeling and Analysis of Hypersonic Inlet Start/unstart Considering Training Data Costs

Juntao Chang,Qinghua Hu,Daren Yu,Wen Bao
DOI: https://doi.org/10.1016/j.actaastro.2011.05.035
IF: 2.954
2011-01-01
Acta Astronautica
Abstract:Start/unstart detection is one of the most important issues of hypersonic inlets and is also the foundation of protection control of scramjet. The inlet start/unstart detection can be attributed to a standard pattern classification problem, and the training sample costs have to be considered for the classifier modeling as the CFD numerical simulations and wind tunnel experiments of hypersonic inlets both cost time and money. To solve this problem, the CFD simulation of inlet is studied at first step, and the simulation results could provide the training data for pattern classification of hypersonic inlet start/unstart. Then the classifier modeling technology and maximum classifier utility theories are introduced to analyze the effect of training data cost on classifier utility. In conclusion, it is useful to introduce support vector machine algorithms to acquire the classifier model of hypersonic inlet start/unstart, and the minimum total cost of hypersonic inlet start/unstart classifier can be obtained by the maximum classifier utility theories.
What problem does this paper attempt to address?