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
Model, described in Chapter 6 of this paper. The weighted average change in collision
risk over a family of representative scenarios would approximate the relative change in
risk due to a blunder under a postulated decrease in separation standards. How one would
construct these scenarios and changes in probabilities is a serious challenge.
SEPARATION SAFETY MODELING
5-18
5.4 FAST-TIME (MONTE CARLO) AND REAL-TIME SIMULATION
MODELS
A real-time simulation is a carefully controlled experiment combining the power and
graphical display capabilities of today's computers with the knowledge and skill of human
subjects, such as pilots and air traffic controllers, to operate equipment, either simulated or
real, such as aircraft and radar facilities. Since time cannot be compressed because of the
human participants, the simulations are referred to as real-time simulations. A real-time
simulation is typically an extremely complex endeavor involving several participants, (e.g.,
controllers, engineers, analysts, and ergonomists) requiring elaborate facilities for its
performance, and possibly requiring a year or longer to complete.
Real-time simulations of aircraft and air traffic control operations have been used
increasingly by the FAA and EUROCONTROL in the evaluation of new procedures and
airport configurations. In real-time simulations, controller and aircrew/aircraft
performance can be measured individually and combined in the system performance
measures. One benefit of conducting real-time simulations with human operators is that
unexpected effects of the procedure on system performance can be identified. A limitation
of real-time simulations is that only a relatively small number of conditions can be tested
economically. Thus, the data collected is usually a very small subset of all the possible
conditions.
Real-time simulations are used to provide the Air Traffic Management (ATM) community
the most accurate information possible on the impact of change and new developments in
its ATM facilities and procedures. They provide the capability to study modifications in
airspace sectorization, aircraft routing, control procedures, the state-of-the-art in
controller interfaces and new tools. One can assess the effects of delays, fuel expenditure,
safety aspects of new procedures, and controller workload. To run a real-time simulation,
the client provides the controllers (typically 20-25). The simulation facility provides a
representation of the normal or proposed ATC environment. The controllers come to the
FAA or EEC (Eurocontrol Experimental Centre) facilities for an agreed period arranged
to address the objectives of the simulation. During this time, data and a variety of
measures are collected to allow the client to obtain objective and subjective assessments of
the environment being studied. On the basis of the researchers knowledge and past
experience, recommendations based on the results may also be offered.
The value of real-time simulation data must be weighed against the “simulation effect.”
Simulation, by definition, is not “the real thing.” This effect is visible as controllers doing
things they would not do in the normal practice of ATC and considerably higher simulated
traffic levels being considered “acceptable” than would be the case in reality. This can be
seen as desirable or undesirable. In some cases, it is also evident that participants tolerate
poor interaction with the simulated system because it is a simulation.
APPROACHES TO COLLISION RISK ANALYSIS
5-19
The purpose of a real-time simulation is to produce estimates of parameters that are used
in the decision process. Because of the small set of data collected in real-time simulations,
there is the possibility of a relatively large margin of error in the observed decision
parameters. Only a relatively small number of combinations of human factors, such as
pilot and controller reaction times, and aircraft dynamic parameters have been used to
produce the estimate of the decision parameters. Therefore, confidence intervals are
typically established to estimate the range of the decision parameters. Since the sample
sizes are relatively small, the confidence intervals are correspondingly large. It is possible
that a decision to accept or reject a procedure could be in error because of the size of the
confidence interval used to estimate the decision parameter. In addition, because of
economic or time considerations, it may have been necessary to omit some possible
configurations of interest from the real-time simulation.
One way to refine and/or expand the results of a real-time simulation is to conduct a fasttime
computer simulation, commonly called a Monte Carlo simulation. This technique has
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(29)
