895
1 INTRODUCTION
The satellite system era started in the mid1980s,
whenthe American NAVSTAR GPS system became
fully operative. Currently, there are 11 satellite
systems in different parts of the world, with more
under construction, which have had an increasing
impact on the global economy [Czaplewski 2015].
Satellite systems
have become a widely recognised
andusedtoolinmanytypesofnavigation.Maritime
navigation mainly employs systems for
determination of vessel coordinates, but also for
ensuringsailingsafetyateverystageofaseajourney
[Urbanskietal.2008].Asinsailing,themainjobthat
satellitesystemsdo
inairnavigationistopositionan
aircraft, especially during its takeoff and landing
[Bialyetal.2011].Satellitesystemsareusedinland
navigationwithpopularSatNavreceivers,whichare
sopopularthatitisnotoftenrealisedthatnavigation
signalscomefromsatellitesystems.Railvehiclesare
another
areainwhichlandnavigationisused.Here,
satellitesystemsareusedinstocktaking,diagnostics
anddesignworkinrailways.Thegreatestadvantage
of mobile satellite measurements is their ability to
perform measurements in a unified and coherent
system of spatial coordinates [Specht, Koc 2016;
Spechtetal.2019].
Satellitesystemsarealsousedin
otherareasofresearchandpracticalapplications,e.g.
in geodesy. Apart from conventional methods of
determining coordinates for a point on the globe,
Use of a Least Squares with Conditional Equations
Method in Positioning a Tramway Track in the Gdansk
Agglomeration
K.Czaplewski,C.Specht,P.Dąbrowski&M.Specht
GdyniaMaritimeUniversity,Gdynia,Poland
Z.Wiśniewski
UniversityofWarmiaandMazury,Olsztyn,Poland
W.Koc,A.Wilk,K.Karwowski,P.Chrostowski&J.Szmagliński
GdańskUniversityofTechnology,Gdańsk,Poland
ABSTRACT: Satellite measurement techniques have been used for many years in different types of human
activity,includingworkrelatedtostakingoutandmakinguseofrailinfra structure.Firstandforemost,satellite
techniquesareappliedtodeterminethetramwaytrackcourseandto
analysethechangesofitspositionduring
itsoperation.
This paper proposes using the least squares with conditional equations method, known in geodesy (LSce).
Whenapplied,thismethodwillallowforimprovementofthefinaldeterminationaccuracy.Thispaperpresents
asimplifiedsolutiontotheLScealignmentproblem.Thesimplification
involvesreplacementoftheparameter
binding equations with equivalent observational equations with properly selected weights.The results
obtained with such a solution were demonstrated with a randomly selected section of a tramway track in
Gdańsk.
Thearticlepresentsthetheoreticalfoundationsofthetestmethod,theexperimentorganisationandthe
results
obtained with MathCad Prime 3.0 software. It also presents the outcome of a study associated with the
executionoftheprojectNoPOIR.04.01.01000017/17entitled“Developinganinnovativemethodofprecision
determinationofarailvehicletrajectory”executedbyaconsortiumoftheGdańskUniversityofTechnology
andGdyniaMaritimeUniversity.
http://www.transnav.eu
the International Journal
on Marine Navigation
and Safety of Sea Transportation
Volume 13
Number 4
December 2019
DOI:10.12716/1001.13.04.25
896
they are also used‐combined with mathematical
alignmentmethodstoincreasetheaccuracyoffinal
determinations, e.g. [Bakula, Kazmierczak 2017,
Czaplewski, Waz, 2017, Swierczynski, Czaplewski
2015].
For several years, coauthors of this paper have
been presenting opportunities of using satellite
techniquesinmeasurementsofrailwayinfrastructure
[Koc 2012,
Specht et al. 2011, Specht et al. 2014].
Researchinthisareaisalsoconductedinotherparts
oftheworld,e.g.[Gikasetal.2008,Chenetal.2015,
2018]. The European Commission has financed
severalprojectsaimedatapplyingsatellitesolutions
for positioning multiple units and the results were
published,amongothers,in[Filipetal.,2001,Urech
etal.,2002,MertensP.,FranckartJ.P.,2003].Thereis
also the Positive Train Control (PTC) system in the
USA, which has been used on most American
railways since 2015. It employs the NAVSTAR GPS
systeminboth its basic form and inthe differential
version[Bettsetal.2014].Currently,the Automated
Train Management System (ATMS) is under
construction in Australia; starting in 2020, it will
enable train positioning with high reliability and
accuracy[ACIL2013].However,eachsolutionbased
only on satellite systems faces the problem of
positioning accuracy because of GNSS system
accessibility, disrupted not only for natural reasons
[Czaplewski2018]. Therefore,theauthorsattempted
toadaptmethodsofresultalignment,whicharewell
knowningeodesy.Thispaperproposestheuseofthe
least squares with conditional equations method to
increasetheaccuracyofmobilemeasurements,which
canbesupportedbyadditionalprecisionpositioning
of GNSS receiver antennas on measurement
platforms.Theproposedadaptationofthealignment
methodwasverifiedusingameasurementcampaign
carriedoutinGdanskinautumn2018.
2 LEASTSQUARESWITHCONDITIONAL
EQUATIONSMETHODWITHASIMPLIFIED
SOLUTION
There are alignment problems in geodetic
measurements in which the parameters of an
observational equations system must meet extra
conditions (Wisniewski 1985, 2013, 2016). These
conditions in a geodetic network can apply to
coordinates of some points being part of the given
network.Thisisasituationdealtwithinthetaskof
monitoring the position of a tramway track. It is
possible when the distance between measurement
points (antennas of GNSS receivers) are measured
with accuracy sufficient to treat the quantities as
relativelyerrorfree(comparedtootherobservations
in the network). Coordinates of the antennas on a
measurement platform, contained in vector
1
[,, ]
T
r
XXX
, must be a solution of the
followingsystemofconditionalequations
(1)
On the other hand, according to the general
principles of determination of geodetic network
coordinates, vector X determination is based on
satellite observations which enable formulating the
followingsystemofobservationalequations
(2)
where
1
[, , ]
T
n
y
yy
is a vector of observation.
1
[, , ]
T
n
vvv
denotes a vector of random
observation errors with the covariance matrix
21
0
v
CP
(P matrix of weights,
0
2
coefficient of variance). With known approximate
coordinates
00 0
1
[,,]
T
r
XXX
of receiver
antennas, the equation

yv=FX
can then be
replaced with a linearobservational equation of the
followingform:


0
0
0
d
dd
 

XX
v
l
FX y FX X y
FX
XF AyXX
X
(3)
The quantity
dX
is a vector of unknown
incrementssuchthat
0
dXX X
.Thematrix
is a known matrix of coefficients (
()rank r
A ),
whereas
  
0
00 0
112 2
T
nn
FyFyFy


lFX y
XX X
is a vector of absolute terms. Considering
that
0
dXX X
, a system of observational
equations can likewise be reduced to a linear form
(1). Developing function
X
to the following
form
897





0
0
00
d
dd


 
XX
XXX
X
XXXBX
X
(4)
givesthefollowinglinearconditionalequation
d BX 0
(5)
where:

Taking into account the conditions binding the
parametersbeingdeterminedandapplicationofthe
goalfunctionoftheleastsquaremethodleadstoan
alignment problem of the following form
(Wiśniewski2016):
21
0
m() in
T
d
d
d


v
A
P
X
B
lv
C
v
X0
XPv
(6)
To obtain a strict solution of the problem, it is
necessary to replace the primary goal function
()
T
d
vXPv
with the following Lagrange
function
2() )(
TT
dd
XBvXPv k
(7)
wherekisavectorofunknownLagrangemultipliers.
The problem (6) can be solved practically in an
different, numerically simpler way. To this end, the
conditional equation
d BX 0
must be
replacedwithanequivalentobservationalequation
d
BX v
 (8)
Sinceitisrequiredthat
v0
,thenthevectorof
fictitious observational errors
v
must be assigned
with such a covariance matrix
21
0
v
CP
, that
fictitious residuals
ˆ
v
should meet the condition
ˆ
v0
(within thecalculationprecisionlimits).For
mutually independent observations, this results in
theadoptionofsufficiently large, diagonal elements
of the matrix of weights
P (theoretically, these
shouldbeinfinitelylargequantities).Inthismanner,
the problem (6) is replaced with a conventional
alignmentproblemwiththefollowingform
21
21
0
0
21
0
*
min
m
)
(in
(
)
T
TT
d
d
d
d
d






v
v
v
X
AX l v
Alv
CP
CP
vPv
CP
vP
BX v
X
XvvPv
(9)
where:
,,






v
Al
Av
v
l
B
Vector
v
is a combined vector of observation
errorswiththecovariancematrix
21
1
0
221
00
21
1
0
0
0
0
0










v
v
CP
P
CP
CP
P
(10)
where
*
P0
P
0P

is a combined matrix of weights (
2
0
the
coefficient of variance common to both elements of
model (10)). Problem (9) has a solution as an
estimator of increments
dX
with the following
form
1
ˆ
()
TT
d
XAPAAPl


(11)
where:
TTT
APA APA BPB
TTT
APl APl BP
Determination of the covariance matrix for the
estimatorgives
21
ˆ
0
ˆ
()
T
d
X
CAPA
(12)
where:
2
0
ˆˆ
ˆ
T
nwr
vPv
(13)
Vector
ˆ
ˆˆ
[]
TTT
vv v
is a combined vector of
residualsdeterminedfromthefunction
ˆ
ˆ
d
vAXl
(14)
898
Thefollowingvectorisacoordinateestimatorfor
GNSSreceiverantennas
0
ˆˆ
dXX X
(15)
with the covariance matrix
ˆˆ
d
XX
CC
. Let us note
thatsquaresofmeanerrorsofdeterminedestimators
are the diagonal elements of the matrix, i.e.
2
ˆˆ
[]
i
ii
X
m
X
C
3 ORGANISATIONOFTHEEXPERIMENT
The measurement campaign was conducted during
the night of 28/29 November 2018 between 11 p.m.
and 4.00 a.m. in Gdansk with a Bombardier NGT6
tramway(Fig.1).
Figure1.BombardierNGT6tramway‐source:www.gait.pl
Two mobile measurement platforms with the
measuring instruments‐GNSS receivers were
pulledbytherailwayvehicle(Fig.2)Theinstruments
weresuppliedbytwoleadingproducersofgeodetic
instruments.
a/
b/
Figure2 a‐Mobile measurement platform, b‐GNSS
receivertracks
Moreover, an inclinometer, an accelerator and a
compass were installed to conduct other
measurements not described in this paper. The
measurements were performed on a 3kilometre
route in Gda nsk (Fig. 3). They involved repeated
passageofthemeasurementunitalongasectionwith
variousnumbersofbuildingsalongit,which
affected
theaccessibilityof the satellite systems (Fig. 3).The
mean speed of the unit at which the measurements
wereperformedwas10km/h.
Figure3. Plan of the measurement unit passage‐source:
maps.google.pl
The diagram of the measurement platforms
constructed for the tests is shown in Fig. 4. The
receiversweredeployedinsuchawaythatfourwere
situatedattheverticesofasquare,andthefifthwas
situated at the intersection of its diagonals. The
construction of the mobile measurement platform
enableddesigning ageometricmeasurement
structure in the shape of a square with the sides of
155 cm to 170 cm. The precision placement of the
GNSS receivers in the track axis and above the rail
wasperformedinthelocalsystemwiththeuseofan
electronictotalstation
andaprismaticmirrorplaced
onadedicatedtripodwith anaccuracyof ca.1mm
(rms).
Figure4Adiagramofthemobilemeasurementplatform(1
rails,2pointsofforcedcentringofaGNSSreceiver,3
platformwheels,d
idistancesbetweentheGNSScentring
points)source:preparedbytheauthors
899
Thepositiondatawererecordedinrealtimewith
the 1 Hz frequency during the measurement
campaignfordifferentGNSSreceiverconfigurations:
positioningwiththeuseofGPScorrectiondata,
positioning with the use of GPS+GLONASS
correctiondata,
positioning with the use of GPS+GALILEO
correctiondata.
4 THEMETHODAPPLICATIONFORANTENNA
POSITIONALIGNMENTONA
MEASUREMENTPLATFORM.
The method described in section 2 was applied to
align the satellite measurement results, also taking
into account previous tachymetric measurements.
Thetachymetricmeasurementsallowedforprecision
positioning of the
GNSS receiver antennas on the
measurement platforms relative to one another and
dimensioningthewholemeasurementsystemonthe
measurement platform. The observation results are
showninFig.5.
Figure5.Distancesbetweenantennasonthemeasurement
platform‐source:preparedbytheauthors
The alignment process was performed for all
kindsofobservations.However,becauseofthelarge
amountofthestudymaterial,onlyitsrandompartis
presented in this paper (Fig.3). 60 seconds of
recording with one type of receivers using
adjustmentfromGPS+GALILEO+GLONASSwere
used for the
alignment. The least squares with
conditional equations method was applied with the
use of MathCad Prime 3.0 software, enabling
calculationsandexportofthealignmentresultstoa
fileinaformatcompatiblewithArcGISembeddedin
theGISlaboratoryattheDepartmentofGeodesyand
OceanographyoftheMaritime
UniversityinGdynia.
No partial results are presented because of the
largesizeofthemeasurementsample.However,the
final results are shown in Fig. 6. The graphic
presentation of the results was prepared in the
ArcGISprogram.
Figure6.Alignedrecordingresultsforaselectedsectionof
the tramway track loop in Gdansk source: prepared by
theauthors
The GNSS antenna position, aligned by the least
squares with conditional equations method, is
marked in yellow. The red colour denotes the
antenna position determined by satellite techniques
foronerecordingmoment.
Figure7showsmeanerrorsofpositioncoordinate
estimatorsobtainedfromthecovariancematrix(12).
Figure7. Errors of mean antenna position coordinate
estimatorssource:preparedbytheauthors
The errors lie within the interval

ˆ
0.0004, 0.0053 mm
X
.
5 CONCLUSIONS
This paper proposes the nonstandard use of a
methodofaligningobservationresultswhichiswell
known in conventional geodesy to calculate the
resultsofsatellitemeasurementsinmobile
measurementcampaigns.Theconstantdevelopment
of satellite techniques, aided by the data processing
methods in geodesy, can bring
improved quality to
mobile measurements which require precision
determinationofpointcoordinatesinvariousstudies
and practical implementations of technical
assignments.
The justifiability of the theoretical assumptions
wasconfirmedbyrecordingdatainthemeasurement
campaign described in the paper. The use of the
proposed alignment method should be confirmed
withdatafromothermeasurementcampaigns,which
willbetheobjectofstudiesinthenextstagesofthe
900
projectexecution.Moreover,theteamwilladaptthe
solutions derived from the solutions applied in this
methodandadaptotherpopula rmethodsofaligning
observation results. The work outcome will be
presentedinfuturepublicationsshowing theresults
of studies obtained as part of the project No
POIR.04.01.01000017/17.
REFERENCES
ACILAllenConsulting,2013,Precisepositioningservicesin
the rail sector, pdf file available at:
http://www.ignss.org/LinkClick.aspx?fileticket=rpl6BIa
o%2F54%3D&tabid=56,Australia
Bakula M., Kazmierczak R. 2017. Technology of Rapid and
Ultrarapid Static GPS/GLONASS Surveying in Urban
Environments, pp. 336342. doi.
10.1109/BGC.Geomatics.2017.51.Poland
BettsK.M.,MitchellT.J.,ReedD.L.,SloatS.,Stranghoener
D.P.,
WetherbeeJ.D.,2014,Developmentand Operational
Testing of a SubMeter Positive Train Lo cation System,
Proceedings of the 2014 IEEE/ION Position, Location
andNavigationSymposium(PLANS2014),pp.452461,
USA
Bialy, J., Cwiklak, J., Grzegorzewski, M., Oszczak, S.,
Ciecko, A., Koscielniak, 2011, Aircraft Landing System
Utilizing a GPS Receiver
with Position Prediction
Functionality, [in:] Weintrit A. (ed): Methods and
Algorithms in Navigation. Marine Navigation and
Safety of Sea Transportation, Leiden, CRC Press,
BalkemaBook,2011,pp.207216.
Chen,Q.;Niu,X.;Zhang,Q.;Cheng,Y.(2015)Railwaytrack
irregularity measuring by GNSS/INS integration.
Navigation Journal ION
vol.62, issue 1, pp. 83–93.,
USA
Chen,Q.,Niu,X.,Zuo,L.,Zhang,T.,Xiao,F.,Liu,Y.,Liu,J.
(2018) A Railway Track Geometry Measuring Trolley
System Based on Aided INS. Sensors. 18(2), 538, doi:
10.3390/s18020538.,Switzerland
Czaplewski K., 2015, Global Positioning System: Political
Support, Directions of Development, and
Expectations.
TransNav, the International Journal on Marine
NavigationandSafetyofSeaTransportation,Vol.9,No.
2,doi:10.12716/1001.09.02.10,pp.229232,Poland
CzaplewskiK.2018DoesPolandNeed eLoran?.In:Mikulski
J. (eds) Management Perspective for Transport
Telematics. TST 2018. Communications in Computer
andInformationScience,vol.897,pp.525
544,Springer,
Switzerland
Czaplewski K., Waz M., 2017, Improvement in Accuracy of
Determining a Vessel’s Position with the Use of Neural
Networks Ana Robust MEstimation, Polish Maritime
research, vol. 24, issue 1, pp. 2231,
https://doi.org/10.1515/pomr20170003,Poland
FilipA.,MocekH.,BazantL.,TauferJ., MaixnerV.,
2001,
Architecture of GNSS Aided Signalling: Analysis and
Experiments, Proceedings of the World Congress on
RailwayResearch(WCRR2001),pp.15.,Germany
Gikas,V.,Daskalakis,S.(2008) DeterminingRail TrackAxis
Geometry Using Satellite and Terrestrial Geodetic Data.
SurveyReview,40(310),392405.,UnitedKingdom
Koc, W., 2012
Design of RailTrack Geometric Systems by
Satellite Measurement, Journal of Transportation
Engineering‐ASCE,vol.138,issue1,pp.114122,doi:
10.1061/(ASCE)TE.19435436.0000303,USA
MertensP.,FranckartJ.P.,2003,LowCostSignallingonLow
Density Lines, Railway Gazette International. United
Kingdom
Specht, C., Koc, W., 2016, Mobile Satellite
Measurements in
Designing and Exploitation of Rail Roads, Transportation
Research Procedia, Volume 14, pp. 625634. doi:
10.1016/j.trpro.2016.05.310
Specht C., Koc W., Chrostowski P., Szmaglinski J., 2019,
MetrologyandMeasurement SystemsAccuracy Assessment
of Mobile Satellite Measurements in Relation to the
Geometrical Layout of Rail Tracks, Metrology and
Measurement Systems,26(2),pp.309321.
DOI:10.24425/mms.2019.128359.
Specht, C., Koc, W., Smolarek, L., Grzadziela, A.,
Szmaglinski,J., Specht,M.,2014,DiagnosticsoftheTram
TrackShape withtheuse oftheGlobal Positioning Satellite
Systems (GPS/Glonass) Measurements with a 20 Hz
Frequency Sampling, Journal of Vibroengineering,
Volume 16, Issue 6, pp. 30763085, ISSN 1392
8716,
China
Specht C., Nowak A., Koc W., Jurkowska A., 2011,
Application of the Polish Active Geodetic Network for
railway track determination. [in:]: A. Weintrit & T.
Neumann (eds), Transport Systems and Processes.
Marine Navigation and Safety of Sea Transportation.
Leiden,CRCPress,BalkemaBook,2011,pp.7781.
SwierczynskiS.,
CzaplewskiK.2015MEstimationasaTool
Supporting a Vessel Traffic Controller in the VTS System,
Polish Maritime research, vol. 22, issue 3, pp. 313,
https://doi.org/10.1515/pomr20150050,Poland
Urbanski J.,MorgasW. andSpechtC., 2008,Perfecting the
Maritime Navigation information services of the European
Union, 1st
International Conference on Information
Technology,pp.14.doi:10.1109/INFTECH.2008.4621631,
Poland
UrechA.,PerezDiestroJ.,GonzalezO.,2002,GADEROS,a
Galileo Demonstrator for Railway Operation System,
ProceedingsoftheDataSystemsInAerospace(DASIA
2002),pp.19.,Ireland
Wisniewski Z., 1985, Methods for solving a system of
interdependent
conditional equations. Geodesy and
Cartography,Volume34,Issue1,pp.3952,Poland
Wisniewski Z., 2013, Zaawansowane metody opracowania
obserwacji geodezyjnych z przykładami (in Polish),
UniversityofWarmaandMazury,monography,Poland
WisniewskiZ.,2016, Rachunekwyrównawczywgeodezjiz
przykładami (in Polish), Wydawnictwo UWM,
UniversityofWarma
andMazury,academichandbook,
Poland