曝光台 注意防骗
网曝天猫店富美金盛家居专营店坑蒙拐骗欺诈消费者
delays, cancellations), airport attributes (forecasts, fleet mix, taxi times), and general
reference information (weather, airport capacities, national groundhold programs).
Key databases that are integrated and filtered in PMAC include:
- Airline Service Quality Performance (ASQP),
- Enhanced Traffic Management System (ETMS),
- Terminal Area Forecast (TAF),
- Official Airline Guide (OAG), and
- National Climatic Data Center (NCDC) weather data.
PMAC, Version 2.0, resides in stand-alone mode on two Pentium PC workstations in the
FAA’s ASD-400 computer laboratory. The front-end graphical user interface (GUI) runs
SEPARATION SAFETY MODELING
8-6
in Windows 3.1 with Visual Basic 4.0 and works in conjunction with supporting add-on
tools such as First Dimension, True Grid 4.0, and Crystal Reports 5.0. Its relational
database, FOXPRO 2.6, operates on the back-end and maintains all data files. PMAC
presently contains complete 1995, 1996, 1997, and partial (2-month lag) 1998 data sets
of delay, demand, weather, cancellation, and diversion data. System Query Language
(SQL) features are also available in the PMAC group through ACCESS, FOXPRO, and
VISICALC. PMAC is scheduled to be operational in a Windows NT platform by May
1998. Information on PMAC may be accessed on the FAA ASD-400 Web Site at
http://www.faa.gov/asd/: First click on “Operations Research” and then on “NAS
Performance Models and Tools.”
8.5 MODEL VALIDATION
In addition to providing input parameters to a risk model, data are also important for
validating the risk estimates produced. Validation is a particularly difficult aspect for risk
models for air traffic systems because the required safety performance is extremely high.
This means that it is not possible to validate a model’s predictions against actual accident
occurrence data. To partially overcome this problem, risk models may be used to predict
more frequent events that are precursors to accidents. For instance, data on serious losses
of separation could be collected and compared with the results predicted by the model.
Data on such events can be obtained from existing systems and used to partially validate
the performance of the model. Another validation technique is to compare results from
two, independently developed models that use different approaches to estimate risk.
APPENDIX A
AN ANNOTATED OUTLINE OF
FACTORS POTENTIALLY AFFECTING SEPARATION SAFETY
(Factors that may need to be included in the model)
SEPARATION SAFETY MODELING
A-2
[This page intentionally blank]
APPENDIX A
FACTORS POTENTIALLY AFFECTING SEPARATION SAFETY
A-3
This outline was presented in Section 7 of this report. In this version, many of the
individual factors are annotated with additional explanatory material and information.
(The additional material is presented in italic text bounded by square brackets.) As was
stated before, the subject headings of the outline were chosen for convenience¾other
headings might serve as well. The placement of factors under a heading is often
arbitrary¾often a factor might well fit under a number of headings. Thus, the headings do
not imply mutual exclusivity. A factor listed under one heading should perhaps be listed
under other headings as well, but to save space, we have tried (but not always succeeded!)
to minimize this. However, some factors appear under more than one category because
they may take on a slightly different meaning under each. The order does not imply
expected importance.
Although not shown here, interactions among factors in different parts of the outline will
occur¾some of these are the unfortunate ones that cause accidents. This outline is
certainly not a complete list, but it contains more factors than could be considered
explicitly in any feasible risk model. Part of the modeling task is to determine which of the
factors and interactions are the critical ones.
One final remark with respect to the independence of factors. Often, it appears that two
or three factors are independent of each other¾that the occurrence of one has no effect
on the (probability of) occurrence of others. There may, however, be another, supposedly
exogenous factor, that can influence all of the others. Consider the case of an airline with
poor management, one that spends as little as possible on training and maintenance. This
airline may hire a maintenance company with similar predilections, may postpone
maintenance and repair, and may not adequately train its pilots or ground crews. In such a
case, a ground crew error, for example, might suggest greater than usual probabilities of
maintenance problems and pilot errors. In other words, interactions (dependencies)
中国航空网 www.aero.cn
航空翻译 www.aviation.cn
本文链接地址：
a concept paper for separation safety modeling(57)
