Abstract: Based on vibration signal of turbopump, this paper presents a fault diagnosis method using negative selection mechanism of the biology immune system, researching on a abnormality online monitoring method which suits for turbopump of liquid rocket engine. This method uses biology clone and learning mechanism technique to improve negative selection algorithm to generate detectors having different monitoring radius, and covers abnormality space effectively, at the same time, avoids problems such as the low efficiency of generating detectors etc. The results of monitoring example show that this method not only can solve the difficulty of obtaining fault sample preferably and extract the turbopump of liquid rocket engine state character effectively, but also can detect abnormality by caused various fault of the turbopump accurately. The exact diagnosis precision indicates this method is feasible, and has better on-line quality, accuracy and robustness. This method explores a new route for turbopump of liquid rocket engine fault monitoring.

Key words: Liquid rocket engine; Turbopump; Fault diagnosis; Artificial immune

摘要： 针对传统的液体火箭发动机涡轮泵故障诊断方法只能在有样本数据并且样本数据充足的情况下才能进行准确诊断以及诊断时难以提取状态特征的缺点,提出一种适用于涡轮泵在线监测及诊断方法,该方法利用生物免疫系统的反面选择机理,利用生物克隆和学习机理使改进型反面选择算法产生的检测器具有不同的检测半径,使其能更有效地覆盖异常空间,能有效地提取涡轮泵的状态特征,避免了检测器产生效率低等问题。实例诊断结果表明:该方法较好地解决了故障样本难以获取及有效地提取涡轮泵的状态特征的问题,能准确监测出涡轮泵各种常见故障所引起的异常并能准确诊断,较高诊断精度表明该方法是可行的,并且具有较好的在线性、准确性及鲁棒性,为液体火箭发动机涡轮泵故障异常检测探索了一条新路。

关键词: 液体火箭发动机;涡轮泵;故障诊断;人工免疫