PSYCH OpenIR  > 社会与工程心理学研究室
民机着陆异常事件人因操作特征与绩效评价研究
Alternative TitleHuman Operation Charcteristics and Performance Evaluation of Civil Aircraft Landing Incidents
汪磊
Subtype博士
Thesis Advisor吴昌旭
2014-10
Degree Grantor中国科学院研究生院
Place of Conferral北京
Degree Discipline心理学
Keyword着陆安全 人因 操作特征 绩效评价
Abstract  尽管飞机着陆的时间仅占总飞行时间的 1%,但着陆却是民机飞行事故发生最多的阶段,事故统计表明过去十年全球有23%的飞行事故发生在该阶段,其中驾驶舱人因差错一直是导致这些事故最主要的诱因,占到事故致因的 60%以上,因此解决着陆阶段的人因问题是改善飞行安全最有效、直接的途径。本研究以飞行快速存取记录器(Quick Access Recorder,QAR)数据为基础,综合应用数据挖掘、统计建模、模拟飞行实验、问卷调研等方法对着陆异常事件的人因操作特征和绩效评价问题展开研究。
研究一采集了293个航班的飞行QAR数据样本,从中挖掘出远接地和重着陆两类着陆异常事件的飞行参数特征,然后对远接地、重着陆、正常着陆的拉平操作特征进行了差异性分析,最后应用多元线性回归和Logistic回归模型分析了拉平操作与着陆绩效的相关性,研究结论表明:拉平是最终着陆阶段最重要的操作,拉平操作开始的时机、完成速率与着陆绩效(接地距离与垂直载荷)紧密相关,拉平操作时间和拉平操作起始高度两个因素对着陆接地距离和接地载荷产生直接影响,飞行员均匀、柔和、一致地拉杆和收油门对应着良好的着陆绩效。
研究二应用眼动追踪设备在模拟飞行环境中对 10 名被试开展了目视着陆操作眼动实验,结论表明:注视时间百分比、注视转移次数、平均注视时间组合而成的注视模式特征与着陆绩效之间存在较强相关性,在着陆过程中绩效优组对舱外景区的注视百分比显著高于绩效差组、绩效优组总的注视转移次数显著多于绩效差组、绩效优组对舱内仪表的平均注视时间显著短于绩效差组。
研究三将着陆过程中影响目视判断和操作的因素设计进一套着陆情境问卷中,通过对专家飞行员实施情境问卷调研,并应用熵权法处理了 32 份有效问卷数据,结论表明:着陆过程中影响飞行员目视判断最重要的五个因素分别是能见度、助航设施、飞行经验、飞行能力、地面参照,影响飞行员手动操作最重要的五个因素分别是飞行技能、飞行经验、风速、着陆重量、着陆构型。
研究四基于 QAR 数据和风险评价原理提出了一个着陆操作绩效评价模型,即通过计算每次着陆后远接地、重着陆、擦机尾三类着陆异常事件的发生风险来评价飞行员的操作绩效,并依据该模型开发了一个飞行着陆操作绩效评价系统, 结论表明该系统可以实现着陆操作绩效评价、飞行数据处理、事件查询统计等功能,QAR数据的进一步应用验证了模型和系统的可用性。
总之,本研究应用真实飞行数据对着陆关键人因操作的特征、机理、影响因素和绩效评价做了积极的探索,成果将为航空公司优化着陆操作训练、开展操作绩效评价、控制着陆异常事件、预防着陆近地事故提供直接依据和参考,对提高飞行安全水平具有重要现实意义。
Other AbstractLanding  only  accounts  for  1%  of  the  whole  flight  time,  but  holds  the  most amount of  flight accidents. The global accident statistics  indicated that 23% of  flight accidents happened in landing phase in the past ten years. Meanwhile, human error in flight deck is the most important inducing factor and contributed to 60% of accidents. Resolving human  factors  issues  in  landing will  be  the most effctive way  to promote current  flight  safety  level. Based  on  flight Quick Access Recorder  (QAR)  data,  this research  aims  to  find  the  human  operation  characteristics  of  landing  incidents  and evaluate  pilot  operation  performance  by  jointly  using  methods  of  data  mining, statistical modeling, flight simulating experiment and pilot survey.  
In  Study  I,  293  cases  of  flight  QAR  data  were  collected,  and  the  analysis  of variance  was  carried  out  for  finding  flight  parameter  characteristics  of  abnormal landing  at  a  flare  initial  point  and  in  the whole  flare  process. Then  two  regression  models were developed to analyze the potential correlations between  flare operations and  landing performance. Finally,  it was concluded that flare would greatly  influence touchdown distance and vertical acceleration through the two key factors of flare time and  height.  The  control  column  and  throttle  operation  would  affect  landing performance  conjointly.  Pilots’  quick  and  steady  pulling  up  columns  and  softer throttle  reduction  are  helpful  for  a  better  flare  operation  and  better  landing performance.
In Study II, an experiment with eye tracking in flight simulator was conducted to find  the human  fixation pattern  in  landing phase. The results of experiment  indicated that  strong  correlation  existed  between  landing  performance  and  fixation  pattern consisting  of  fixation  percentage,  number  of  fixation  transfer  and  fixation  average time.  In  visual  landing  process,  the  group  with  excellent  operation  performance allocated  more  time  and  fixation  points  on  outside  view  and  the  average  time  on instrument was significantly less than the group with poor performance.
In  Study  III,  these  factors  influencing  visual  perception  and  operation  were considered and designed into a landing scenario questionnaire. Then the questionnaire survey was  implemented  among  experienced pilots  and  entropy weight method was used  to deal with obtained 32  sample data. The  results  indicated  that  five  important factors affecting visual perception and judgment were visibility, navigation aids, flight experience, flight ability and ground visual references. These top five factors affecting manual  operation  in  landing  were  flight  skill,  flight  experience,  wind  speed,  gross weight and configuration.  
In Study  IV, a  landing operation performance evaluation model based on  flight QAR  data  and  risk  theory.  The  pilot  operation  performance  could  be  evaluated according to the occurrence risk of long landing, hard landing and tail strike incidents after  each  landing. Meanwhile  based on  this model, an applicable  landing operation performance evaluation system was developed with function of flight data processing, performance  evaluation  and  incident  statistics.  Finally  the  QAR  data  was  used  to examine the usability of the evaluation model and system.
To  sum  up,  the  whole  research  examined  human  operation  and  visual characteristics,  influencing  factors  and  performance  evaluation  of  landing  incidents, these  results would be  the basis of  further safety  landing  research. These  findings  in this research can also be applied  into practice  for airlines  to optimize  flight  training, evaluate  flight  operation  performance,  prevent  landing  incidents  and  improve  flight safety in further.
Subject Area应用心理学
Language中文
Document Type学位论文
Identifierhttp://ir.psych.ac.cn/handle/311026/19665
Collection社会与工程心理学研究室
Affiliation中国科学院心理研究所
Recommended Citation
GB/T 7714
汪磊. 民机着陆异常事件人因操作特征与绩效评价研究[D]. 北京. 中国科学院研究生院,2014.
Files in This Item:
File Name/Size DocType Version Access License
汪磊-博士学位论文.pdf(4723KB)学位论文 限制开放CC BY-NC-SAView Application Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[汪磊]'s Articles
Baidu academic
Similar articles in Baidu academic
[汪磊]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[汪磊]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: 汪磊-博士学位论文.pdf
Format: Adobe PDF
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.