Documentos de Académico
Documentos de Profesional
Documentos de Cultura
Background
In response to several mishaps in which perfectly good F-16swere flown
into the ground, the F-16System Program Office decided to integrate the Digital
Terrain System (DTS) into all F-16s. Since DTS was an off-the-shelf purchase,
this integration effort was considered a low cost way to reduce controlled flight
into terrain mishaps, especially when compared with the cost of such accidents.
Although not the primary function of DTS for reducing controlled flight into
terrain mishaps, Database Terrain Cueing (DBTC), aided pilot awareness of the
surrounding terrain by providing a cue in the head up display (HUD).
118
The F-16 Combined Test Force at Edwards Air Force Base, California
recently completed Phase 1 flight testing of the DTS. During this testing, the DTS
team identified and corrected several handling qualities problems. This paper will
not only share the results of our DBTC testing, but more importantly offer
valuable lessons for other flight test programs to use.
The heart of DTS was a portable black box called the Data Transfer
Cartridge (DTC) built by Fairchild Industries (Figure 2). The improved
DTS/DTC was a form, fit, and function replacement for the standard F-16 DTC.
The DTSDTC differed from a standard DTC in that it had a new mass memory
for storing a terrain database and a new microprocessor to drive the DTS. The
actual DTS software was contained in TERPROM (TM) software developed by
British Aerospace Systems and Equipment Company.
119
The terrain database used by the F-16 DTS came from the Level 1 Digital
Terrain Elevation Data (DTED) developed by the National Imagery and Mapping
Agency (Figure 3). The raw DTED represented the terrain by storing elevations
on posts along a grid in latitude and longitude. In the F-16 DTS, these terrain
database posts were spaced apart by about 500-600 feet in our test area. The
terrain database also contained a Vertical Obstruction Database (VOD), which
stored the location and height of manually entered man-made obstacles such as
towers.
120
121
122
Terrain Cue
Current g
25 mils = 1.0 g
Figure 7, Head Up Display DBTC Presentation
Test Methods
The DBTC function was first tested by two different ground simulations.
Lockheed-Martin Tactical Aircraft Systems (LMTAS) used an unmanned, off-line
simulation to characterize the DBTC algorithm and compare DBTC performance
to that of terrain following systems. This simulation used a pilot model and did
not address the issues involved with displaying the cue to the pilot. LMTAS
further tested DBTC on the ground by using their handling qualities simulator
(HQ Sim). This simulator was a man-in-the-loop simulation which used actual
avionics and flight control hardware and software. The aircraft was modeled by a
six degree of freedom high fidelity model that'had been used during F-16 Low
Altitude Navigation and Targeting Infrared for Night (LANTIRN) development
and flight test. This provided the first opportunity to assess the presentation of the
DBTC cue itself. Initial HQ Sim results highlighted the potential for a pilotinduced oscillation to develop while following the DBTC cue at high
groundspeeds.
Initial DBTC flight test consisted of flying low level routes over isolated
peaks, rolling terrain, and rough terrain. Test points were chosen across the F-16
speed envelope, from 340 to 590 knots groundspeed. The DBTC function was
exercised at all TCHs, from 100 to 1,000 feet. Finally, two aircraft loadings, one
123
light and one heavy, were used to characterize how changes in aircraft
performance and mass properties would affect DBTC performance (Figure 8).
8000
4-0
7000
I
+
6000
5000
4000
10
Distance (NM)
Figure 9, Initial DBTC Performance
In addition to these DBTC performance issues, pilots found the DBTC cue
to be jittery and too sensitive when they attempted to follow the cue as closely as
possible. From an operational perspective, the test team did not think that pilots
would try to fly the box as closely as the test pilots were. However, the jerky
movement of the box was a big distraction which could divert the pilots attention
away from the actual terrain. The pilot would be able to do a better job of terrain
masking by visually referencing the actual terrain. Most significantly, if a pilot
did try to follow the box closely, flight safety could be jeopardized by a divergent
pilot-induced oscillation (PIO) that developed at various airspeed and terrain
combinations. The test team determined that the potential PI0 should be further
investigated and characterized.
flight conditions using the Cooper-Harper HQ rating scale and to identify any HQ
cliffs. Additionally, the team hoped that this HQ evaluation would generate
pilot comments in order to isolate and correct the specific HQ problems.
The defined HQ tracking task was to keep the flight path marker inside the
DBTC cue box while flying wings level at a particular airspeed. Each test point
specified an airspeed and a terrain route. Terrain routes were selected to sample
isolated peak, rolling terrain, and rough terrain types. To limit the scope of this
HQ evaluation, all test points were flown at 500 feet TCH. The desired criteria
for this HQ evaluation was defined as following the cue with at least half of the
flight path marker (FPM) in the DBTC box for the entire route with no more than
five excursions per minute. This represented the pilot following the DBTC cue to
within +I-0.2 g. The adequate criteria was defined as following the cue with the
flight path marker at least touching the DBTC box for the entire route with no
more than five excursions per minute. This represented the pilot following the
cue to within +/- 0.4 g. Since PIOs were anticipated, pilots were told that
overshoots while correcting back to the box were not to be counted as excursions.
A thorough postflight video review was used to determine task performance.
Following each HQ test point, pilots were asked to give a quantitative
Cooper-Harper rating. These ratings were based on the tracking performance that
was attained and the amount of pilot compensation required to meet this level of
tracking performance. Since some of the HQ deficiencies seemed related to
turbulence, pilots were also asked to give turbulence ratings following each run
based on the definitions from the Department of Defense Flight Information
Handbook. In general, pilots rated the severity of turbulence, as light, moderate,
or severe, based upon. its effect on aircraft attitude and altitude. Turbulence was
also rated as either intermittent or continuous based on its frequency. Perhaps
even more important than the quantitative HQ ratings, the pilot comments that
these HQ evaluations inspired helped considerably in isolating specific HQ
problems.
First, the TRN map error state was added to the height of terrain within the worst
case profile used by DBTC. Additionally, the TRN map error uncertainty was
added into the calculation of the aircraft worst case height. During a single
iteration of the DBTC function, the worst case profile may jump in altitude as
indicated by the dashed terrain profile in Figure 1 1. Similarly, the TRN estimate
of worst case height may jump as indicated by the white aircraft. After these
calculations are made, the DBTC function would generate a new g-command. In
the example in Figure 1 I , DBTC commanded the white aircraft back to the preset
TCH over the dashed terrain by generating a large positive discrete jump in the
DBTC box. By removing these two calculations, Lockheed-Martin engineers
hoped that more consistent worst case profiles and worst case aircraft heights
would reduce the magnitude and occurrence of discrete jumps in the DBTC box.
MaphrUncertaintyJ
,.
.......
-- .....-._._.
..._._.
._..
_.....
Map Error State Jlrmp
\
Another step in improving the DBTC handling qualities was to reduce the
jitter in the DBTC cue. Figure 12 shows a pilotized block diagram of the DBTC
control loop. The very center of this and all things in the world is the pilot
himself. The pilot directly responded to the HUD cue and produced an input to
the aircraft. The top box represents the DBTC algorithm generating the required g
commands. The bottom row represents the feedback of aircraft g. This signal
was smoothed out by filtering high frequency noise. The difference between
commanded and actual g was then used to generate the displacement of the DBTC
box in the HUD.
128
129
...........
Time (sec)
130
The left graph in Figure 15 shows how parameters in the low pass filter
were changed to reduce the magnitude of high frequency noise. The parameters
of interest were damping ratio (6) and natural frequency (0,).
The two curves
show the change from 0.7 to 1.1 damping ratio. The right graph in Figure 15
shows the velocity-dependent lag filter. At low airspeeds, the time lag (T) was
minimal, but at higher airspeeds, this time lag increased to slow down the
response of the DBTC box to changes in aircraft g. Thus, the inherent pitch
sensitivity of the F-16 at higher airspeeds was compensated by the lag in the
feedback loop.
The first step in optimizing the feedback filter parameters was to fly them
in the HQ simulator. This allowed engineers to bound the parameters, thus
limiting the number of .sorties required to come up with the optimum
configuration.
Figure 15, Noise Filter Bode Diagram and Velocity-Dependent Lag Parameter
On each flight test sortie, the pilot was given three DTSDTCs, one
containing the baseline configuration for the flight, and the other two varying one
or two parameters from that baseline. The pilots were not told the parameters on
any DTC until the optimization process was complete. Thus, by direct blind
comparisons of the three DTSDTCs flown by a single pilot on a single sortie, a
best configuration was chosen. The parameters that changed from the baseline
to the best configuration were evaluated by the engineers. The configurations
flown in the next sortie were chosen based on the current best configuration and
the engineers predictions of which parameters to alter further. This process was
131
repcated until a final sct of threc DTSDTCs was flown to pick the optimum
configuration.
The additional six DBTC sorties flown in follow-up flight test was money
well spent by the F-16 program office. Figure 16 dcmonstrates that the
optimization process really worked. The final best set of parameters improved
handling qualities across the board. Although the final evaluations were only
flown in light intermittent turbulence, the jitter in the box was significantly
reduced and handling qualities while flying over an isolated peak at 590 knots
moved up to Level I . Furthermore, the sensitivity of the DBTC box at high
speeds was reduced, with no significant P I 0 tendencies noted. The third plot in
Figure 16 shows that at our most demanding test point, 590 knots over rough
terrain, handling qualities improved from Level 3 to Level 2. The problem which
kept this best set of parameters from having Level 1 handling qualities was that
the DBTC box still moved too fast for a pilot to follow. This is best illustrated by
the center plot in Figure 16 which shows the correlation between Cooper-Harper
rating and terrain while flying at 590 knots. Over rough terrain, only adequate
tracking criteria could be attained.
132
8000
h
CI
7000
al
'g 6000
5000
4000
0
10
Distance (NM)
Figure 17, Improved DBTC Performance
Future Testing
Phase 2 of DTS testing is currently underway at Edwards in 1997. In this
phase, the test team hopes to find other ways to improve this system. Specifically,
the team plans to extend the DBTC HQ investigation down to lower altitudes to
see if there is any correlation between HQ and TCH. We also hope to expand the
types of turbulence encountered during our future HQ investigations.
The test team is also planning on doing a military utility evaluation of
DBTC while flying with night vision goggles (NVGs) (Figure 18). The test team
thinks that this is the area that pilots will find DBTC most useful. In high moon
illumination conditions, pilots can easily see and avoid terrain while flying at low
altitude with NVGs. However, current generation NVGs lack the depth
perception that pilots are accustomed to in the daytime. This makes it more
difficult to judge altitude above the ground. The DBTC system should
compliment NVGs well, since rather than having to look at and interpret the
digital radar altimeter readout, pilots will be able to cross-check the DBTC
133
symbol to determine where they are in relation to their desired TCH. This
increased terrain awareness should reduce pilot workload while flying at low
altitudes with NVGs.
Lessons Learned
This flight test effort not only provided an excuse to have fun flying low
and fast, but it also generated numerous lessons for the test team to pass on to our
fellow flight testers.
Our first big lesson was that even avionics flight test may call for a
handling qualities evaluation. This is especially true for new displays designed as
flight directors to help reduce pilot workload. Another excellent example is the
flight testing of helmet mounted displays, especially if used for sensor slaving.
Our test team found that the HQ evaluation using the Coopei-Harper rating scale
was an excellent tool. Not only did it allow us to correlate HQ with flight
conditions, but the pilot comments generated by our HQ evaluation allowed us to
isolate and fix specific HQ problems. Of course this method has proven valuable
for classical flying qualities evaluations over the years, but it also worked well to
wring out the HQ problems of our DBTC display.
134
135