Abstract: Aiming at the problems in the reliability assessment of liquid rocket engine， such as insufficient utilization of test data， strong subjectivity， and insufficient correlation with design improvements， a two-line assessment method on reliability was presented to improve the accuracy of reliability assessment results under limited test information. In addition， the weak links of engine could be identified from the perspective of risk control. This method used two main lines for reliability assessment. Firstly， taking the engine component test data as input， a reliability model was established based on the engine working sequence and engine failure mode， and the probability of component failure was brought into the reliability model to obtain the prior information of engine reliability using uncertainty propagation method. Then， the engine test data was used to obtain the sample information of engine whose life times was considered to follow a Weibull distribution. Subsequently， the Bayesian method was used for information fusion to calculate the posterior distribution of engine reliability. This method expands the amount of data for reliability assessment， identifies the key component failures of the engine from a risk control perspective， and improves the engineering application value of reliability assessment.

Key words: Liquid rocket engine;Fault diagnosis;Reliability model;Reliability assessment;Data fusion;Risk control

摘要： 针对液体火箭发动机可靠性评估过程中试验数据利用不充分、主观性强、与设计改进关联不足等现状，给出了一种可靠性双线评估方法，在有限试车条件下提升可靠性评估结果的准确性，并从风险控制角度识别出发动机的薄弱环节。该方法采用两条主线开展可靠性评估，第一条线以发动机部组件试验数据为输入，以发动机工作时序为线索，识别发动机故障模式，建立可靠性模型，将部组件故障模式的发生概率带入可靠性模型中，采用不确定性传播算法，得到发动机可靠性先验分布。第二条线以发动机整机试验数据为输入，采用威布尔可靠性评估方法，得到发动机可靠性追加信息；随后采用贝叶斯方法进行双线融合，计算发动机可靠性的后验分布。通过可靠性双线评估方法，扩充了可靠性评估的数据来源，并从风险控制角度辨识了发动机的关键部组件及其故障模式，提升了可靠性评估的工程应用价值。

关键词: 液体火箭发动机;故障诊断;可靠性模型;可靠性评估;数据融合;风险控制