航空学报 > 2023, Vol. 44 Issue (22): 128522-128522   doi: 10.7527/S1000-6893.2023.28522

基于故障分辨能力的火箭发动机测量参数选择方法

张效溥(), 任枫, 徐鹏里, 李志敏, 宿彩虹   

  1. 上海宇航系统工程研究所,上海 201109
  • 收稿日期:2023-02-02 修回日期:2023-03-13 接受日期:2023-04-21 出版日期:2023-06-29 发布日期:2023-06-27
  • 通讯作者: 张效溥 E-mail:earoke@163.com

Selection method of measuring parameters for rocket engine based on fault recognition

Xiaopu ZHANG(), Feng REN, Pengli XU, Zhimin LI, Caihong SU   

  1. Shanghai Aerospace System Engineering Institute,Shanghai 201109,China
  • Received:2023-02-02 Revised:2023-03-13 Accepted:2023-04-21 Online:2023-06-29 Published:2023-06-27
  • Contact: Xiaopu ZHANG E-mail:earoke@163.com

摘要:

针对液体火箭发动机测量参数选择这一难题,提出一种基于模型的、提高故障分辨能力和系统可靠性的液体火箭发动机测点选取方法。基于发动机系统非线性静态特性数学模型,建立常见发动机故障下的故障特征库,并采用飞行数据验证其准确性;分别基于凝聚层次聚类算法、蒙特卡洛方法和失效模式影响分析(FMEA)构建了发动机测量特征子集的故障分辨种类数、鲁棒性和系统可靠性3种评价指标,并基于改进的多目标二进制粒子群算法(MOBPSO)开展优化设计。优化后的测点排布,可分辨故障从9种提高到13种,鲁棒性与原排布相当,风险指数略有上升;进一步探究了副系统混合比在故障分辨中的重要作用并分析其机理。本文提出的方法对其他复杂、闭环动力系统测量特征的选择具有较好的应用价值。

关键词: 液体火箭发动机, 基于模型方法, 测点选择, 粒子群优化, 凝聚层次聚类, 主成分分析

Abstract:

To address the problem of measuring parameters selection of liquid rocket engines, we propose a model-based method to improve fault recognition and system reliability. The nonlinear and static mathematical model of the engine system is built, the fault feature list metrics of fault recognition, robustness and system reliability of the engine measurement feature subset are then established respectively, and the optimization design is proposed based on the improved multi-objective binary particle swarm optimization (MOBPSO). With the optimized measuring parameters, the number of distinguishable faults has increased from 9 to 13, the robustness is equivalent to the original layout, and the risk criterion has increased slightly. The important role of the subsystem mixture ratio in fault recognition is further explored and its mechanism analyzed. The method proposed in this paper has a good application value for the selection of measurement characteristics of other complex, closed-loop dynamic systems.

Key words: liquid rocket engine, model-based method, measuring parameters selection, particle swarm optimization, agglomerative hierarchical clustering, principal component analysis

中图分类号: