A comparative study of genetic algorithm parameters for the inverse problem-based fault diagnosis of liquid rocket propulsion systems
A comparative study of genetic algorithm parameters for the inverse problem-based fault diagnosis of liquid rocket propulsion systems
复制标题
DOI：
10.1007/s11633-007-0255-5
复制
发表时间：
2007-07
影响因子：
4.3
通讯作者：
Erfu Yang;Hongjun Xiang;Dongbing Gu;Zhenpeng Zhang
中科院分区：
文献类型：
--
作者：
Erfu Yang;Hongjun Xiang;Dongbing Gu;Zhenpeng Zhang
液体火箭推进系统的故障诊断是航天发射活动中一个非常重要的问题，特别是在伴随载人航天任务时，高效的故障诊断系统可以显着提高安全性和可靠性。目前，基于逆问题的故障诊断方法已成为故障诊断领域的研究热点。该方法为基于模型的推进系统故障诊断提供了一种新的策略。为了解决LRPS故障诊断中的反问题，近年来，遗传算法已被采用作为可用的数值优化工具的首选和有效的选择。然而，遗传算法有许多控制参数要事先选择，仍然缺乏健全的理论工具来分析这些参数对诊断性能的影响。本文通过一系列的数值试验，比较研究了遗传算法参数对诊断结果的影响。研究了基于反问题的LRPS故障诊断中各个算法参数对最终诊断结果的影响，为遗传算法参数的选择提供了合理的估计。最后提出了一些建设性的意见，为今后将遗传算法应用于LRPS的故障诊断实践提供了参考。
Fault diagnosis of liquid rocket propulsion systems (LRPSs) is a very important issue in space launch activities particularly when manned space missions are accompanied, since the safety and reliability can be significantly enhanced by exploiting an efficient fault diagnosis system. Currently, inverse problem-based diagnosis has attracted a great deal of research attention in fault diagnosis domain. This methodology provides a new strategy to model-based fault diagnosis for monitoring the health of propulsion systems. To solve the inverse problems arising from the fault diagnosis of LRPSs, GAs have been adopted in recent years as the first and effective choice of available numerical optimization tools. However, the GA has many control parameters to be chosen in advance and there still lack sound theoretical tools to analyze the effects of these parameters on diagnostic performance analytically. In this paper a comparative study of the influence of GA parameters on diagnostic results is conducted by performing a series of numerical experiments. The objective of this study is to investigate the contribution of individual algorithm parameter to final diagnostic result and provide reasonable estimates for choosing GA parameters in the inverse problem-based fault diagnosis of LRPSs. Some constructive remarks are made in conclusion and will be helpful for the implementation of GA to the fault diagnosis practice of LRPSs in the future.