STATISTICS BASED VIRTUAL PILOT FOR FLIGHT DATA ANALYSIS AND PILOT TRAINING

摘要

This paper describes a statics based virtual pilot that provides a proper baseline for flight maneuvers in test and pilot training. Flight parameters, such as the warning signal, pilot response time, time to level wings, altitude loss during recovery, were estimated for thousands of flight maneuvers over the course of several years with several different pilots. Each maneuver was then characterized by its associated entry parameters, and like maneuvers were then compared to the statistical baselines. These baselines were then compared to individual flights. The statistical results were then presented in a 2D and 3D visualization environment to compare one or more datasets. Pilots were able to view their flights when compared to previous flights and learn what improvements could be made. Additionally, when a sufficient number of tests have been performed and data have been collected, particular high-acceleration maneuvers may no longer need to be performed in subsequent tests in order to validate bus signals and software features; once the proper entry is made in flight, the virtual pilot can perform the recovery, leaving less wear and tear on the aircraft and the pilot. The proposed virtual pilot can save the Air Force millions of dollars per year by eliminating expensive re-flights, unnecessary maintenance and fuel, and creates a safer set of tests for the pilot.