CASIA OpenIR  > 毕业生  > 硕士学位论文
航空发动机气路建模及其典型部件故障诊断研究
黄强
学位类型工学硕士
导师王健
2017-05-27
学位授予单位中国科学院研究生院
学位授予地点北京
关键词航空发动机气路建模 故障特征约简 粗糙集与主元分析 故障诊断 支持向量机 遗传算法
摘要航空发动机作为飞机的心脏,其故障诊断问题研究关系到飞机的飞行安全,意义重大。将智能算法应用于航空发动机故障诊断,有利于进一步提高故障诊断的准确性,快速进行故障定位,降低维修保障成本。由于发动机结构和原理复杂,本文重点选取故障高发的发动机气路部件,开展故障建模及故障诊断研究,主要研究了航空发动机气路建模仿真方法,气路典型部件故障特征约简方法和气路典型部件故障诊断方法。主要工作与成果如下:
(1)分析研究了航空发动机气路建模及故障诊断研究的现状,对比了国内外发动机故障仿真与诊断技术的差距,并总结了各种故障诊断方法;
(2)分析研究了航空发动机气路各部件的数学模型,对比了常用建模软件的优缺点与适用性,重点选择Simulink与PROOSIS软件搭建了两种航空发动机气路模型,通过两种模型的对比分析验证了Simulink机理模型的准确性和可信性;
(3)基于建立的模型进行了发动机气路故障模拟,获取了发动机气路故障仿真数据,研究了基于粗糙集与主元分析的故障特征降维方法,进行了发动机气路故障特征约简,该方法能够在不影响故障诊断准确度的前提下有效减少故障特征数量;
(4)研究了基于支持向量机的发动机气路故障诊断方法,研究发现利用支持向量机对降维后的特征数据进行故障诊断存在支持向量机参数设置影响诊断结果的问题,提出采用自适应、模拟退火、聚合等三种遗传算法来优化支持向量机参数设置。这三种方法针对传统遗传算法可能收敛到局部极值的问题,进行了适应度函数或交叉、变异概率自适应选择上的改进,提高了故障诊断能力;
(5)对发动机气路机理模型、基于粗糙集与主元分析的故障特征降维方法和改进后的智能算法参数优化的支持向量机进行了软件集成,基于MATLAB GUI开发了航空发动机气路故障诊断软件,为用户提供了发动机气路故障仿真和故障诊断的功能。
本文研究得到的发动机气路机理模型和故障诊断方法,为发动机气路故障仿真和故障诊断提供了有效、可信的技术途径,开发的航空发动机气路故障诊断软件能够支持发动机用户开展相关研究,对推动发动机故障诊断发展具有重要的意义。
其他摘要Aero-engine is the heart of the aircraft. The fault diagnosis of aero-engine plays an important role in the flight safety. Artificial intelligence can improve the diagnosis of aero-engine in the fields of fault diagnosis accuracy, quick fault locating, and maintenance cost reduction. Because the structures and principles of the engine are complex, it is hard to study the whole engine system. Therefore, this thesis aims at fault modeling and fault diagnosis mainly for the engine gas path that has a high failure rate.
The research focuses on the aero-engine gas path modeling and simulation method, the fault feature reduction method for typical gas path components, and the fault diagnosis method for typical gas path components. The main effort and contributions of the thesis consist of the following parts:
(1) This work has studied the state-of-the-art technologies of aero-engine gas path modeling and fault diagnosis, identified the gap between domestic and foreign technologies for engine gas path fault simulation and diagnosis, and summarized various fault diagnosis methods;
(2) This work has investigated the mathematical models of the aero-engine gas path components and compared the commonly used modeling tools. This work has used Simulink and PROOSIS to build two types of aero-engine gas path models. We verified the accuracy and credibility of the Simulink mechanism model by the comparative analysis of the two models;
(3) Based on the model built above, this work has carried out the failure simulation of engine gas path and obtained the simulation data. This work has studied the fault feature reduction method based on rough set and principal component analysis, and reduced the characteristics of engine gas path fault without reducing the accuracy of fault diagnosis;
(4) This work has studied the method of aero-engine gas path fault diagnosis based on support vector machine. The work showed that the support vector machine-based fault diagnosis of the dimension reduced feature data has a problem: the parameter setting of support vector machine affects the results of diagnosis. The work has proposed three types of genetic algorithm: adaptive method, simulated annealing, and aggregation method for optimizing the parameter setting of the support vector machine. These three methods improved the fitness function or adaptive selection of crossover and mutation probability for addressing the local extremum issue of the traditional genetic algorithms to enhance the fault diagnosis ability;
 (5) This work has integrated aero-engine gas path mechanism model, dimension method based on rough set and principal component analysis, and support vector machine improved by intelligent algorithm for parameter optimization. The work has developed the aero-engine gas path fault diagnosis software on MATLAB GUI to provide the functions of gas path fault simulation and fault diagnosis.
The engine gas path mechanism model and fault diagnosis method obtained in the thesis provide effective and credible approach for the engine gas path fault simulation and fault diagnosis. The fault diagnosis software above mentioned can support the aero-engine related research. It has important significance in promoting the development of engine fault diagnosis.
文献类型学位论文
条目标识符http://ir.ia.ac.cn/handle/173211/14639
专题毕业生_硕士学位论文
作者单位中国科学院自动化研究所
推荐引用方式
GB/T 7714
黄强. 航空发动机气路建模及其典型部件故障诊断研究[D]. 北京. 中国科学院研究生院,2017.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
硕士学位论文-黄强-终稿 - 改10.p(2900KB)学位论文 暂不开放CC BY-NC-SA请求全文
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[黄强]的文章
百度学术
百度学术中相似的文章
[黄强]的文章
必应学术
必应学术中相似的文章
[黄强]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。