| 摘要: |
| 对表征发动机故障的原始特征，以提高故障类可分性为目的，利用信息论中互信息的理论，采用贪婪算法，去除原始特征中不利于所设计分类器的特征，得到最有利于分类的子特征集。采用子特征集对故障进行分类，得到了较理想的结果。 |
| 关键词: 液体推进剂火箭发动机 发动机故障 故障诊断 特征选择 |
| DOI： |
| 分类号:V430 |
| 基金项目: |
|
| FEATURE SELECTION OF LIQUID ROCKET ENGINE FAULT DIAGNOSIS |
|
Liu Bing, Zhang Yulin
|
|
Dept.of Aerospace Technology,National Univ.of Defence Technology,Changsha,410073
|
| Abstract: |
| The original features of the fault of liquid rocket engine were selected by greed algorithm according to the theory of mutual information.The features selected here increased the divisible ability of the fault pattern.Good results were obtained by dividing the faults using the subspace of feature |
| Key words: Liquid propellant rocket engine Engine fault Fault diagnosis Feature selection |
