Field Results of a GPS/INS-Based Approach to Measuring Ship Flexure Onboard an Aircraft Carrier Sep 2005
M.G. Petovello, K. O’Keefe, G. Lachapelle and M.E. Cannon
Position, Location And Navigation (PLAN) Group | Department of Geomatics Engineering
Schulich School of Engineering | University of Calgary
"ABSTRACT
The Shipboard Relative GPS (SRGPS) component of the US Department of Defense Joint Precision Approach and Landing System (JPALS) aims to deliver automatic landing capabilities to inbound aircraft aboard aircraft carriers. To accomplish this, GPS data collected on the yardarm needs to be translated to the desired touchdown point (TDP) on the flight deck. This in turn requires that all relative motion between the yardarm and the TDP (ship flexure) be properly accounted for, either via direct compensation or in the accuracy and integrity allocations. Since the magnitude of ship flexure at sea has not yet been quantified a data collection campaign was organized to, in part, allow for a direct assessment thereof. The results of this data collection campaign are presented herein and indicates that ship flexure is on the order of a few centimeters in horizontal and vertical components.
INTRODUCTION
The Joint Precision Approach and Landing System (JPALS) being developed by the United States Department of Defense is intended to provide accurate and reliable guidance information to military aircraft landing on land and aircraft carriers using GPS augmented with other sensors. For land-based operations, Local Differential GPS (LDGPS) will be used, whereas Shipboard Relative GPS (SRGPS) techniques will be employed for aircraft carrier landings.
For the LDGPS case, the situation is similar to the Local Area Augmentation System (LAAS) implemented for civil aviation, with fixed reference stations generating differential GPS data to be sent to the incoming aircraft. The difference between LDGPS and LAAS is that the latter does not attempt to resolve the carrier phase ambiguities.
Although SRGPS is conceptually similar to LDGPS, the major practical difference is that the reference receivers are constantly in motion, as they are now mounted directly on the aircraft carrier. Since the reference GPS antennas cannot be mounted at the aircraft’s desired touchdown point (TDP) on the flight deck (due to operational limitations)), they are typically mounted on the ship’s yardarm. However, since the inbound aircraft requires knowledge of its position relative to the TDP, the GPS measurements need to be geometrically translated to this point. Moreover, this translation must account for all ship motion, most notably the ship’s attitude variations.
Practically, the translation of the GPS reference station data to the TDP is done using the assumed known baseline between the two points (obtained for example from a survey), knowledge of the ship’s attitude and the assumption that the ship is a rigid body. However, in the current context, the latter assumption may not be fully justified. In particular, during high dynamics such as during a turn or while in rough seas, the hull and/or mast of the ship will likely deform (ship flexure). Ultimately, this means that the baseline between the TDP and the GPS receiver is a dynamic quantity.
Since the magnitude of ship flexure has not yet been assessed, a data collection campaign onboard an aircraft carrier was conducted, in part, to address this problem. To this end, GPS and INS data was collected from multiple locations on the ship over several days. This paper presents ship flexure results obtained from data collected onboard an aircraft carrier under operational conditions. The objectives of the paper are as follows:
- Present methods of measuring the relative motion of two points (which, in the current context, is a direct indication of ship flexure);
- Identify the limitations of the above approaches in a practical sense and, where appropriate, propose solutions;
- Provide an initial quantification of aircraft carrier ship flexure at sea; and
- Determine if the estimated ship flexure correlates with ship motion.
The paper begins with an overview of the relevant methodology, including a brief review of methods used to estimate relative motion of two points. Next, the ship trip is described in detail, including the type and location of the various GPS and INS sensors. Finally, the results of the tests are presented and limitations are identified....
...Correlating Flexure and Ship Motion
Finally, an attempt was made to correlate the flexure effects with ship dynamics. The reason for this is that if a strong correlation exists, the flexure effects could, in principle, be compensated in real-time. To this end, an attempt was made to correlate flexure with roll, pitch, roll rate, pitch rate, azimuth rate, lateral acceleration and longitudinal acceleration. Interestingly, the only strong correlation was found to be between the y-axis flexure and roll or lateral acceleration of the ship (the roll and lateral acceleration are closely related). Figure 12 shows the correlation between the y-axis flexure and the lateral acceleration of the ship. As can be seen, the correlation is quite strong, with a correlation coefficient of -0.45 (45% correlation). For the May 6 data (with a higher sea state) the correlation increases to (negative) 78%, indicating that as the acceleration increases, the flexure increases accordingly. Interestingly however, the correlation with ship’s roll is rather small suggesting that perhaps the amount of ship’s roll may also be an important parameter. More investigation is required in this regard.
CONCLUSIONS
This paper set out to investigate the amount an aircraft carrier flexes under operational conditions. The paper began with a brief overview of the basic methodology for accomplishing this task, with focus given to the relevant error equations. A discussion of how to separate the various error sources from each other was also presented.
Using data collected aboard the USS Dwight D. Eisenhower (CVN 69), it was determined that the ship dynamics were too low to reliably estimate azimuth using an EGI. As such, a method of computing the residual EGI attitude errors was developed. Results indicated that this method was able to estimate the relative attitude errors between two EGI units to a few arc seconds. With the EGI attitude errors compensated, the ship flexure errors were then computed. The following conclusions can be drawn from this analysis:
1. Multiple antennas located on the yardarm of the ship exhibit similar short- and long-term flexure estimates.
2. The flexure in the starboard direction is larger than in the forward direction, and this is attributed to the larger variation in roll versus pitch.
3. The standard deviations of the flexure estimates are on the order of 1-2 cm and are approximately constant for all data sets, except when the sea state increases.
4. When the sea state increases, the “horizontal” (x- and y-axis) flexure estimates increase by approximately a factor of 1.6. The “down” (z-axis) flexure appears to be insensitive to sea state, suggesting that the GPS-only errors are dominant.
5. The only correlation between flexure and ship dynamics was found to be between the starboard flexure and the lateral acceleration of the ship. This correlation was found to be 45% for mild sea states and 78% for a sea state of three."
http://plan.geomatics.ucalgary.ca/paper ... S%2005.pdf (0.8Mb)
A4G Skyhawk: www.faaaa.asn.au/spazsinbad-a4g/ & www.youtube.com/channel/UCwqC_s6gcCVvG7NOge3qfAQ/videos?view_as=subscriber