涡轮泵实时故障检测的多特征参量自适应阈值综合决策算法

Multi-feature Adaptive Threshold Compositive Decision-making Algorithm for Real-time Fault Detection of Turbopump

为了实时监控液体火箭发动机涡轮泵的状态,提高安全性,降低故障带来的破坏程度,提出了一种多特征参量自适应阈值综合决策算法(MATA)。研究了该算法的特征参量选取、阈值区间的确定、阈值的自适应计算(包括特征参量均值与标准方差的自适应计算)、故障综合决策逻辑、故障数据对阈值贡献的剔除等方法,利用某型火箭发动机地面试车涡轮泵振动测量数据和某型转子试验平台实时测量数据对该算法进行离线和实时在线故障检测试验验证,结果表明MATA没有发生误检测情况,具有实时故障检测的能力。因此,MATA适合于液体火箭发动机涡轮泵的实时故障检测。

In order to monitor the conditions of Liquid Rocket Engine (LRE) turbopump in real time, enhance its safety and minimize the loss of its faults, a Multi-feature Adaptive Threshold compositive decision-making Algorithm (MATA) was presented and realized in this paper. Some methods, such as the selection of signal features, confirmation of threshold ranges, adaptive estimation of thresholds (the adaptive iterative computing of features means and standard deviations), compositive decision-making logics of faults, and refreshment of thresholds without fault data, were researched. Then, MATA was validated with the historical data from LRE test as well as real-time data from rotor test platform. It is shown that MATA can detect faults in real time and give no false alarm in this case. As a conclusion, MATA is suitable for the real time fault detection of LRE turbopump.