1、中图分类号：TP391论文编号：1028704 11-B001学科分类号：080202 硕士学位论文基于任务段的航空发动机载荷谱聚类方法研究学科、专业航空工程研究方向结构强度与振动指导教师南京航空航天大学研究生院 能源与动力学院二一七年十二月NanjingUniversity of Aeronautics and AstronauticsThe GraduateSchoolCollege of X X X XResearch on Kernel-Based Associative Memories,Clustering Algorithms and their ApplicationsA Th
2、esis in学科专业英文名称byAdvised byProf. Submitted in Partial Fulfillmentof the Requirementsfor the Degree ofDoctor of Philosophy (Master of Engineering)February, 2022承诺书本人声明所呈交的博/硕士学位论文是本人在导师指导下进行的研究工作及取得的研究成果。除了文中特别加以标注和致谢的地方外，论文中不包含其他人已经发表或撰写过的研究成果，也不包含为获得南京航空航天大学或其他教育机构的学位或证书而使用过的材料。（保密的学位论文在解密后适用本承诺书）作
3、者签名：日 期：航空发动机载荷谱能够为航空发动机及其零部件进行应力分析、结构设计、整机的可靠性、零部件耐久性试验以及寿命评定提供准确参考。其中，航空发动机载荷谱所对应的飞行任务剖面包含了发动机的飞行信息，是飞机在实际飞行时所产生的具体数据，反映了发动机的实际使用状况。飞行任务剖面聚类后，易于确定典型剖面,可以对飞行任务剖面的编制和预测提供很大的帮助，在实际工程中具有很大意义。然而，飞行任务剖面由于数量众多，杂乱无序，难以找到相似剖面的规律性，分类问题迄今为止并未能得到很好的解决。针对此问题，就需要提及聚类的问题。本文开展了基于二代机任务段的航空发动机载荷谱聚类方法研究，提出了利用聚类散点图对航
4、空发动机飞行任务剖面进行分类的方法，并对某战斗机发动机31个飞行任务剖面进行了聚类分析。通过比对分析飞行任务剖面中的飞行高度、飞行马赫数、法向过载、转速四个重要参数，选取了飞行高度和飞行马赫数作为划分飞行任务剖面的原始参数；依据其对应的飞行任务段均值生成聚类散点图；按照散点分布区域的差异性对其进行初步划分，并根据滤波后剖面特征对密集点区域进一步划分；最终将剖面类型划分为五大类。本文开展了针对三代机飞行任务剖面的聚类分析研究。首先，通过飞行高度与飞行马赫数的均值生成了聚类散点图。之后，依据雨流计数法统计的各剖面对应的飞行高度、飞行马赫数、转速、法向过载的循环数，划分出了巡航、地慢状态。随后，依据
5、飞行高度参数划分出了空对地与空对空状态。最后，依据油门杆角度划分出了油门杆状态。将三代机剖面类型划分为五大类。本文确定了三代机的各类剖面所对应的典型飞行任务剖面。之后，将典型飞行任务剖面按照任务混频、起飞降落段均值处理的方式组成了综合任务谱。最后，通过穿级计数法与雨流计数法对各类典型飞行任务剖面与综合任务谱分别进行了计数比照。结果说明，所编制的综合任务谱合理有效。关键词：航空发动机载荷谱，聚类分析，飞行任务剖面ABSTRACTAeroengine load spectrum can provide an accurate reference for the stress analysis, s
6、tructural design, machine reliability, durability testing of components and life assessment of aeroengines and their components. Among them, the aero-engine load spectrum corresponding to the flight mission profile contains the flight information of the engine is the aircraft in the actual flight ge
7、nerated by the specific data, reflecting the actual use of the engine. After the flight mission profile clustering, it is easy to determine the typical profile, which can greatly help the compilation and forecast of the mission profile, which is of great significance in practical engineering. Howeve
8、r, the classification of flight mission sections has so far failed to be well solved because of the large number of clutter and disordered navigation systems that make it difficult to find the regularity of similar sections. To solve this problem, we need to mention the clustering problem.In this pa
9、per, the study of aeroengine load spectrum clustering method based on the second-generation machine mission segment is carried out. A method to classify the aeroengine mission profile by clustering scatterplot is proposed. The profile of 31 missions of a fighter engine Cluster analysis. By comparing
10、 and analyzing four important parameters of flight mission, such as flight altitude, flight Mach number, normal overload and speed, flight altitude and flight Mach number are selected as the original parameters of flight mission profile. Based on the corresponding flight mission segment According to
11、 the difference of scatter distribution area, the scatter points are divided according to the characteristics of the scatter points, and then the dense point areas are further divided according to the filtered profile features. Finally, the section types are divided into five categories.This article
12、 has carried on the cluster analysis research to the third generation aircraft flight mission section. First, a clustered scattergram is generated from the mean of flight altitude and flight Mach number. Then, according to the rain flow count method of the corresponding sections of the flight height
13、, flight Mach number, speed, the number of normal overload cycle, divided cruise, slow to the state. Subsequently, the air-to-ground and air-to-air conditions were classified according to the flight altitude parameters. Finally, according to the throttle lever angle out of the throttle lever state.
14、Three generations of machine section type is divided into five categories.This paper identifies the typical flight profiles corresponding to the various sections of the three generations of machines. After that, the typical flight mission profile is composed of the task map, the average of take-off
15、and landing segments. Finally, the comparison between the cross sections of various typical missions and the spectrum of comprehensive tasks is carried out through the method of wear-leveling counting and rain-flow counting. The results show that the comprehensive task spectrum is reasonable and effective.Keywords:Aeroengine load spectrum, cluster analysis, mission profileIABSTRACTII第一章绪论- 1 -选题背景- 1 -国内外研究现状- 2 -飞行任务剖面研究现状- 2 -1.2.2 聚类法研究现状- 5 -1.2.3 载荷谱编制- 9 -本文的主要研究工作- 11 -第二章二代机聚类分析- 12 -引言- 12 -载荷谱预处理- 12 -聚类方法比较- 13 -2.3 划分参数的选取- 20 -飞行任务剖面聚类232.4.1 高空高速飞行剖面- 24
