液体火箭发动机传感器故障检测与数据恢复算法研究  

Research of Failure Detection and Data Accommodation Algorithm of Sensor for Liquid-Propellant Rocket Engine

作　　者：黄强[1] 刘洪刚[1] 吴建军[1] 

机构地区：[1]国防科技大学航天与材料工程学院,长沙410073

出　　处：《航天控制》2012年第4期73-76,共4页Aerospace Control

基　　金：教育部"新世纪优秀人才支持计划"项目(NCET-06-09278);湖南省自然科学基金资助项目(08JJ4023)

摘　　要：以某型液体火箭发动机为研究对象,根据其传感器的故障特性,提出了基于BP神经网络的传感器故障检测与数据恢复算法。通过定义的传感器置信度来判断传感器是否发生故障,以及确定故障传感器,利用已训练好的神经网络结构对故障传感器进行数据恢复。研究内容能够实现传感器的故障检测、定位与补偿,能够有效提高发动机故障检测方法的可靠性和鲁棒性。Based on the back propagation neural network, the failure detection and data accommodation algorithm of sensor for liquid-propellant rocket engine is developed according to the failure characteristics of sensor. The confidence level is defined to detect the failure of sensor and identify the failure sensor, and the trained neural network structure is used to accommodate it. Regarding the research, the failure detection, identification, and accommodation of sensor can be realized and the reliability and robustness of fault detection method of engine are improved.

关 键 词：传感器 故障检测 数据恢复 BP神经网络 

分 类 号：V434[航空宇航科学与技术—航空宇航推进理论与工程]