Once the system is verified, validation is required to assess the degree to which the behavior of the simulated system reflects the behavior of the real-world system it has been designed to represent.  An essential element of validation is comparison of numerical values of selected system performance measures from the simulation with values recorded from observed behavior of the real world, where such data is available. 

 With respect to the CAT sensor technology assessment itself, it would be useful to extend the experimental design to consider other factors in addition to lookahead, such as false alarm rate, detection rate, traffic loading, and pilot responses to continuous (rather than one-time) alarms, and to allow route changes as well as altitude changes for CAT avoidance. 

 Important recommended developments to the RFS architecture and agent models include: 

. Enabling RFS agents and other models (such as MIDAS) interacting with RFS agents to be continually synchronized in simulated time, e.g., controlled by a single master clock.

. Enabling the networking agent to access information directly from agents other than vehicles

. Developing a library of automatic measurement and statistical reporting agents

. Allowing for dynamic updating of environmental agents (e.g., weather and CAT)

. Developing a user interface to enable analysts other than RFS authors and programmers to construct and execute application system models out of existing component agent models.

 Important recommended developments to MIDAS include: 

. Enabling MIDAS and other models (such as RFS agents) interacting with MIDAS to be continually synchronized in simulated time, e.g., controlled by a single master clock

. Enabling the MIDAS updateable world representation to be initialized dynamically as the simulation progresses for human operators that enter the simulation conditionally or otherwise at times later than the start (e.g., pilots of aircraft entering a sector)

. Removing the current constraint on the number of human operators that can be practically simulated

. Generalizing the code so that models of new human operators and of operator tasks and procedures can be created through data entry rather than code changes

. Developing a user interface to enable analysts other than MIDAS authors and programmers to set up and execute human performance simulations.

Undertaking these recommended tests and developments will build credibility for and confidence in the application of human-centered, agent-based, fast-time simulation for aviation safety analysis.
 
 

I. INTRODUCTION

This report constitutes the final report deliverable under Contract Number NAS2-99072 for the final year of a three-year research project entitled “Development of Fast-Time Simulation Techniques to Model Safety Issues in the National Airspace System.”  This project was sponsored by the NASA Aviation Safety Program and was conducted by a research team comprised of ATAC Corporation as the prime contractor, and the Massachusetts Institute of Technology (MIT) and the University of California at Berkeley (UCB) as subcontractors.  San Jose State University (SJSU) and Georgia Institute of Technology (Georgia Tech) also collaborated and participated fully in the final phases of this project under separate funding vehicles with NASA.
Background
Analysis of safety issues in the National Airspace System (NAS) has traditionally been largely dependent on the analysis of incident and accident reports.  In addition to such databases as the National Transportation Safety Board Accident and Incident Database and the FAA Operational Error and Pilot Deviation System databases, a large number of voluntary reports on incidents are filed every year with the NASA Aviation Safety Reporting System (ASRS).  Recently, there has been growing interest in the use of operational data to identify potential safety problems or to better understand known problems.  Airlines are beginning to collect and analyze flight data from quick access recorders, in programs that have come to be termed Flight Operations Quality Assurance (FOQA) (Flight Safety Foundation, 1992).  In 1993, the FAA and NASA commenced a collaborative program to develop a set of analytical tools and methodologies termed the Aviation Performance Measuring System (APMS) to allow the very large quantities of flight-recorder data to be processed in an automated way and to support statistical analysis, data reduction and exploration, and causal modeling (NASA, 1998).  The FAA has begun to address how it might also use operational data to develop system safety performance measures (Gosling, 1998).
　
中國航空網(wǎng) www.k6050.com
航空翻譯 www.aviation.cn
本文鏈接地址：DEVELOPMENT OF FAST-TIME SIMULATION TECHNIQUES(8)