Академический Документы
Профессиональный Документы
Культура Документы
To cite this article: Anargyros T. Baklezos, Christos D. Nikolopoulos & Christos N. Capsalis
(2017): An equivalent dipole method with novel measurement positioning for modeling electric
emissions in space missions, Electromagnetics, DOI: 10.1080/02726343.2017.1376902
Article views: 17
ABSTRACT Keywords
Prediction of the electric emissions in space missions is critical due to Electromagnetic
the sensitivity of their payload. A reliable method to predict such compatibility; electric fields;
emissions is the accurate electric source identification. In this work, measurement; modeling;
stochastic algorithm
every space component is modeled by a small number of electric
Downloaded by [Tulane University] at 02:32 02 October 2017
Introduction
Various space missions like Cluster, Rosetta, Swarm, Pathfinder, GOCE, LISA, and
BepiColombo or Solar Orbiter carry payload, which consists of various scientific instruments
in order to measure the electromagnetic field and particle populations in space plasma
(Antonucci et al. 2011; Benkhoff et al. 2010; Drinkwater et al. 2003; Danzman K., 2008;
Escoubet, Schmidt, and Goldstein 1997; Friis-Christensen et al. 2008; Müller et al. 2013;
Scheeres et al. 1998). Such missions have instruments that are necessarily sensitive to magnetic
and electric fields and set strict requirements on electro-magnetic cleanliness and compat-
ibility. Focusing on electrostatic cleanliness and LF electric cleanliness, the instruments are
meant to measure slow time-variant electric fields corresponding to the frequencies below 200
KHz. In order to meet the stringent requirements, both correct characterization of the electric
field sources and the accurate prediction of emissions are mandatory. In the case of the slow
varying AC electric sources, the frequency variations are low and can be well approximated by
quasi-static modeling in each distinct frequency like the electrostatic approach. Quasi-static
modeling employing electric dipoles as the core elements of the emissions scheme, can then
result through Fast Fourier Transform (FFT) and Real-Time Spectrum Analysis to a method
of fully describing the emissions of a component or an Equipment Under Test (EUT).
Much work (Li et al. 2013; Mikki S. M. & Y. M. M. Antar, 2012) has been put at providing
a model for electric field emissions of an EUT. However these models reflect the mode of
operation in high frequencies or used only to predict emissions in order to conform to EMC
standards (European Cooperation for Space Standardization 2003, 2012).
The worst case approach, where all the electric moments orientation are directed towards the
observation point, results to the maximum possible value of the electric field. This ensures the
prevention of the electric field’s underestimation (when multiple EUTs are present); whereas the
overestimation of the field is considered “safe”. Nevertheless, in many cases the accurate
prediction of the electric field is required; regardless of under/over estimation. Due to difficulty
to acquire measurements for the spacecraft and for all the possible test cases (EUTs possible
positions and orientation), the development of a modeling tool, in which the emissions of
various EUTs are considered in order to provide electromagnetic cleanliness at system (space-
craft) level, is imperative.
The case of searching and modeling electric sources from a measurement setup constitutes an
inverse scattering electromagnetic problem and covers many different applications (Bojarski, N.,
1982; Li et al. 2013; Mikki and Antar 2012; Tsitsas N., 2013; Wang Y. M. Chew W.C., 1989;
Zaiping, N. & Yerong, Z., 1998).
In this paper a multiple electric dipole model (EDM) is developed and validated. The
proposed model aims at identifying the electric source and thus accurately predicting the
Downloaded by [Tulane University] at 02:32 02 October 2017
extrapolated field of an EUT from near field measurements. Correct EUT electric field
calculation at an arbitrary position provides the capability of prediction at spacecraft level.
The rest of the paper is organized as follows: in Section II, the mathematical back-
ground and formulation is briefly discussed. In Section III, the proposed algorithm is
presented, while section IV consists of several simulation results concerning the electric
behavior and the EDM accuracy of the EUT.
point ðxm ; ym ; zm Þ, when the near field approximation is considered, is (Jackson J.D., 1999;
Paul C.R., 1992):
~ 1 1
E¼ ½3^rð^r ~
pÞ ~
p 3 (1)
4πε0 r
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
and r ¼ ðxm xi Þ2 þ ðym yi Þ2 þ ðzm zi Þ2 :
Downloaded by [Tulane University] at 02:32 02 October 2017
Assuming N dipoles are contributing, consisting the EUT, the total electric field components
at the measurement point m, are:
XN
1 3ðsm si Þ C psi
Es Total ¼ 3 ; s : x; y; z (3)
i¼1
4πεo ri 5 r
The proposed methodology exploits the dipole analysis presented in this section to match
the set of field measurements in m points, identifying the dipole sources parameters
(position, electric moment vector); hence the EDM.
In EMC/EMI tests, the common measurement practice is to measure the magnitude of the
electric field at a standard position (European Cooperation for Space Standardization
2003,
2012). However, measurement of the magnitude of the componentjEx j; Ey ; jEz js, is also
possible. Optimal modeling would require vector measurements, which provide detailed field
information, but are not available to the majority of the test facilities. The proposed methodology
of this work can be adjusted to use any type of the aforementioned measurement with minor
changes.
Where Nm is the number of the measurement points and E is the field measuredðEmeas Þ
and modeledðEmodel Þ. In case of field components, the RMS error is given by:
Downloaded by [Tulane University] at 02:32 02 October 2017
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 ffi
P3 PNm
j¼1 ð i¼1 Ej meas ðiÞ Ej model ðiÞ Þ
ðRMSÞ ¼ (6)
3Nm
Generally a potential solution for PSO is called a particle. In this implementation every
particle is represented by the 6 variables (coordinates in the problem space) that
characterize a source dipole. The particles traverse the problem space following the
optimum particles. Optimum particle coordinates in the problem space are associated
with the best solution that is determined by a fitness function. The PSO algorithm
follows the flowchart of the Figure 2 to best-fit the measured electric field at observa-
tion points m according to Eq. (5), which consists the fitness function in case of field
magnitude measurements.
Assuming that one electric dipole is located atðxi ; yi ; zi Þ, and having electric
momentðpxi ; pyi ; pzi Þ, the electric field at the observation point Nm ðxm ; ym ; zm Þ can be
calculated from Eq. (2) and the total electric field values can be represented (for all the set
of Nm measurements) with a ð1xNmÞ matrix Emodel . The PSO algorithm takes as input the
positions of the measurement points Nm , and the electric field measurement values at
these points (1xNm) matrix Emeas . Optimum particles in the problem space produce a
Emodel matrix that minimizes the RMS per iteration. The algorithm termination criterion is
fulfilled when fitness function/RMS error values are lower than 1e-6. In this manner a
solution is achieved, that is, a dipole that produces a Emodel matrix that best fits the Emeas .
The coordinates of the solution particle represent a dipole that best emulates the electric
behaviour of the EUT under study at the points of measurement. In case of electric field
component measurements the algorithm remains exactly the same except Emodel and Emeas
become ð3xNmÞ matrices and the fitness function is expressed by Eq. (6).
When N dipoles are considered for the EDM, Eq. (3) describes the direct electric field
components. In this case the PSO’s solution space expands to 6 Ndimensions.
ELECTROMAGNETICS 5
Downloaded by [Tulane University] at 02:32 02 October 2017
Simulation results
The first task in order to feed the algorithm with the necessary data is to populate the Nm
set of measurement points, that is, to determine the number and location of necessary
measurement points, with regard to origin.
To solve the inverse problem with the minimum number of measurements (e.g. 6
independent measurement points should be enough to predict the 6 parameters of one
dipole source), the measurement points should ideally be located at independent positions.
However, according to Eq. (2), the field component magnitude is not linear with respect to
6 A. T. BAKLEZOS ET AL.
ði 1Þ2π
r0 ¼ 30cm; δr ¼ 1cm; φ ¼
Nm
.
In the simulation results presented in the following subsections two different
approaches of the modeled EUT are studied. First the case of one dipole source is
simulated both with field magnitude measurements and electric field compo-
electric
nents magnitude jEx j; Ey ; jEz j measurements. The proposed algorithm is able to
achieve accurate source identification in both measurement setups and the comparative
results are presented.
Afterwards the case of two dipole source is studied. The necessary changes in the
proposed algorithm in order to adapt to this case are discussed and the simulation results
are presented.
Table 1. Source identification results for various distortion levels (electric field magnitude case, with
Nm = 10).
Prediction Prediction
EUT Source Model Prediction With 2% distortion With 5% distortion Prediction with 10% distortion
x (cm) 10.7 10.70237 10.72342 10.7982 10.83255
y (cm) −10.5 −10.5122 −10.4183 −10.8289 −10.9188
z (cm) −5 −5.0646 −5.05243 −6.0661 −5.70974
pxi ðpCmÞ 50 50.38421 50.06876 55.31514 46.23526
pyi ðpCmÞ 30 30.02363 29.82999 28.00792 36.96043
pzi ðpCmÞ −20 −19.2896 −19.2844 −7.42932 −12.9829
Downloaded by [Tulane University] at 02:32 02 October 2017
Figure 5. Single dipole EUT vs Model with zero distortion electric field Emissions at measurement
points (Number of measurement points Nm = 10).
It is clear that the algorithm achieved accurate source identification, via almost
perfect matching of the measured electric field. In order to further explore the robust-
ness of the algorithm and its tolerance to distortion levels, three additional cases
including various distortion levels of the measured field (2%, 5% and 10%) were
simulated. The corresponding results are also tabulated in Table 1 and the fields
produced at measurement points Nm are depicted at Figure 6. It is worth mentioning
that even with distortion level as high as 10%, the algorithm correctly identifies the
source with an error in position below 5mm, while the error in moment value is lower
than 15%. These error values are considered adequately low for the model to accurately
represent the EUT in terms of electric field production. Additional confirmation of the
model’s validity can be acquired using appropriate metrics, such as RMS or through
standardized methods like FSV (Duffy et al. 2006; IEEE Standard P1597 2008; Orlandi
et al. 2006) for the fitting of the electric field, produced by the model at a random
different to the Nm -set of points Next (extrapolated field). Furthermore, in order to
ensure the validity of the model it is proposed the Next set to have four times the points
of the Nm set (Next = 40). Due to the fact that EUT and model source are in good
agreement, it is expected that the extrapolated fields will also be near identical. For this
reason a simple RMS error metric is deemed sufficient to evaluate the extrapolated
field’s fitness. Results of the extrapolated fields are depicted in Figure 7 and the RMS
errors for the three distortion cases are tabulated in Table 2.
ELECTROMAGNETICS 9
Figure 6. Single dipole EUT vs Model with various distortion values electric field Emissions at measure-
Downloaded by [Tulane University] at 02:32 02 October 2017
Figure 7. Single dipole EUT vs Model with various distortion values electric field Emissions at extra-
polation points (Nm = 10).
Additionally, the RMS error of the field values for the various source models at the
extrapolation points are tabulated in Table 2 indicating the potency of the proposed
methodology to accurately identify the EUT source.
The proposed
methodology
can be easily adapted to work with electric field components
magnitude jEx j; Ey ; jEz j measurements instead of electric field magnitude ðjEjÞ measure-
ments. Under this mode of operation the necessary minimum number of measurement points is
reduced. It should be notedthat, for every point of measurement in this operation mode, three
variables jExmeas j; Ey meas ; jEzmeas j are obtained. In contrast to the electric field magnitude
measurement case, where ten measurement points are considered adequate, in this mode only
two measurement points are enough to achieve equivalent results for source identification. The
10 A. T. BAKLEZOS ET AL.
Figure 8. Single dipole EUT vs Model with various distortion values electric field Emissions at extra-
Downloaded by [Tulane University] at 02:32 02 October 2017
Figure 9. Single dipole EUT vs Model with various distortion values electric field Emissions at extra-
polation points (Nm = 6, Next = 24).
results tabulated in Table 3 indicate that the algorithm is capable to predict accurately the
aforementioned electric source.
Two different measurement cases, one for the minimum set of measurements points
(Nm = 2) and another case with Nm = 6 are simulated and presented in Tables 3 and 4
respectively.
The second set with Nm = 6 is studied in order to provide more accurate results for
distortion level of 10%. The field emissions produced by the two models at the
ELECTROMAGNETICS 11
Table 3. Source identification results for various distortion levels (electric field components case, with
Nm = 2).
Prediction Prediction
Source Prediction With 2% distortion With 5% distortion Prediction with 10% distortion
x (cm) 10.7 10.7 10.73036 10.81352 10.5098
y (cm) −10.5 −10.5 −10.5077 −10.6406 −10.4986
z (cm) −5 −4.99999 −4.9776 −5.02306 −4.76776
pxi ðpCmÞ 50 49.9999 49.84541 50.77059 55.07816
pyi ðpCmÞ 30 30.00014 29.78419 31.72995 29.62491
pzi ðpCmÞ −20 −19.9999 −18.7206 −22.6274 −21.9452
Table 4. Source identification results for various distortion levels (electric field components case, with
Nm = 6).
Prediction Prediction
Source Prediction With 2% distortion With 5% distortion Prediction with 10% distortion
x (cm) 10.7 10.69996 10.6914 10.81524 10.69945
y (cm) −10.5 −10.4999 −10.5674 −10.4206 −10.6728
Downloaded by [Tulane University] at 02:32 02 October 2017
extrapolation points are depicted in Figure 8 and Figure 9 against the EUT emissions for
the different measurement cases.
The RMS errors of the field values for the various source models at the extrapolation
points in case of electric field component measurements for Nm = 2 and Nm = 6 setups are
tabulated in Table 5.
Table 6. Two dipole source identification results for various distortion levels (electric field components
case, with Nm = 8).
Prediction With zero Prediction With 2% Prediction With 5% Prediction with 10%
Source distortion distortion distortion distortion
1st Dipole
x (cm) 10.7 10.69999 10.67278 10.75787 10.96121
y (cm) −10.5 −10.5 −10.4762 −10.5216 −10.5209
z (cm) −5 −5.00001 −5.09841 −5.09905 −4.64557
pxi ðpCmÞ 50 50.00007 50.21402 50.03554 49.25645
pyi ðpCmÞ 30 30.00016 30.78329 29.93391 26.11154
pzi ðpCmÞ −20 −19.9999 −20.0885 −18.856 −17.0341
2nd Dipole
x (cm) −10.5 −10.4999 −10.4227 −10.5449 −10.9261
y (cm) 0.5 0.499953 0.474221 0.367288 1.096822
z (cm) −0.3 −0.30007 −0.24016 −0.4463 0.175741
pxi ðpCmÞ −15 −15.0004 −15.1029 −15.2695 −11.666
pyi ðpCmÞ −20 −20.0002 −20.2578 −19.5858 −19.3933
pzi ðpCmÞ −16 −16.0001 −16.0375 −15.6492 −15.6953
Downloaded by [Tulane University] at 02:32 02 October 2017
Figure 10. Two dipole EUT vs Model with various distortion values electric field Emissions at measure-
ment points (Component Case with Nm = 8).
respectively. Table 6 clearly states that the identification of the complex dipole source is
achieved for all distortion levels with satisfactory accuracy.
As a secondary metric, the RMS error of the field values for the various source models
at the extrapolation points are tabulated in Table 7 indicating the potency of the proposed
methodology to accurately identify even more complex EUT sources.
Conclusion
In the present work and in order to contribute to a credible pre-verification toolbox for space
applications, a theoretical analysis of a stochastic based approach is proposed in order to
characterize an EUT with a number of electric dipoles based on a set of measurements.
Simulation results showcase the accurate identification of a complex electric source/EUT.
ELECTROMAGNETICS 13
Figure 11. Two dipole EUT vs Model with various distortion values electric field Emissions at extra-
Downloaded by [Tulane University] at 02:32 02 October 2017
In the proposed scheme an Electrical Dipole Model was developed for accurate complex
source identification even in cases where the simulated measurements were distorted by noise.
The accuracy of the proposed methodology was evaluated by two approaches, direct source
comparison and RMS error of the extrapolated electric fields. The validity of the proposed
methodology is confirmed from both approaches. The algorithm has been designed in a way
that extra dipoles can be iteratively added until the field in measurement points is matched, as
evaluated by the fitness function. Although in the frequency range of interest an adequate
number of dipoles is one or two, the proposed algorithm can be expanded to a much higher
number of dipoles. It should be noted; however, in this case, source identification should not be
the target of the methodology and the objective of the algorithm should be to match the field at
the measurement points. In order to accurately predict more than two dipoles, a set of much
higher number of measurement points would be needed, since the inverse problem (electric field
equations) is not linear to the variables.
Future work could include sets of actual measurements and the expansion of the
proposed methodology in order to include low frequency electric sources. Moreover
Future work of the authors would include the investigation of the correlation of the
number of needed measurement points to the number of dipoles.
ORCID
Anargyros T. Baklezos http://orcid.org/0000-0002-0532-6216
Christos D. Nikolopoulos http://orcid.org/0000-0003-1344-4666
14 A. T. BAKLEZOS ET AL.
References
Antonucci, F., M. Armano, H. Audley, G. Auger, M. Benedetti, P. Binetruy, C. Boatella, et al. 2011.
LISA pathfinder: Mission and status. Classical and Quantum Gravity 28:9. doi:10.1088/0264-
9381/28/9/094001.
Benkhoff, J., J. Casteren, H. Hayakawa, M. Fujimoto, H. Laakso, M. Novara, P. Ferri, H. R.
Middleton, and R. Ziethe. 2010. BepiColombo—comprehensive exploration of mercury:
Mission overview and science goals. Planetary and Space Science 58 (1):2–20. doi:10.1016/j.
pss.2009.09.020.
Bojarski, N. 1982. Low frequency inverse scattering. IEEE Transactions on Antennas and
Propagation 30 (4):775–78. doi:10.1109/TAP.1982.1142846.
Carrubba, E., A. Junge, F. Marliani, and Α. Monorchio. 2012. Particle Swarm Optimization to solve
Multiple Dipole Modelling problems in space applications. Aerospace EMC, 2012 Proceedings
ESA Workshop On 1:21–23.
Danzmann, K. 2008. LISA mission overview. Advances in Space Research 25 (6):1129–36.
doi:10.1016/S0273-1177(99)00973-4.
Drinkwater, M. R., R. Floberghagen, R. Haagmans, D. Muzi, and A. Popescu. 2003. GOCE: ESA’s
first earth explorer core mission. In Earth gravity field from space—From sensors to earth
Downloaded by [Tulane University] at 02:32 02 October 2017
sciences. In the Space Sciences Series of ISSI, Vol. 18, ed. G.B. Beutler, M.R. Drinkwater, R.
Rummel and R. von Steiger, 419–432., Dordrecht, Netherlands, Kluwer Academic Publishers.
ISBN: 1-4020-1408-2
Duffy, A. P., A. J. M. Martin, A. Orlandi, G. Antonini, T. M. Benson, and M. S. Woolfson. 2006. Feature
Selective Validation (FSV) for validation of computational electromagnetics (CEM). Part I – The FSV
method. IEEE Transactions on Electromagnetics Compatibility 48:449–59. doi:10.1109/
TEMC.2006.879358.
Escoubet, C. P., R. Schmidt, and M. L. Goldstein. 1997. Cluster-Science and mission overview. In
The cluster and phoenix missions, ed. C.P. Escoubet, R. Schmidt, R., C. Russell, 11–32.
Dordrecht, Netherlands, Springer Science+Business Media B.V.
European Cooperation for Space Standardization. 2003. Standard ECSS-E-40-12A “Space engineer-
ing, SpaceWire - Links, nodes, routers and networks”. Noordwijk, The Netherlands, ESA
Publications Division.
European Cooperation for Space Standardization. 2012. Standard ECSS E ST 20 07, C Rev. 1, “Space
engineering, Electromagnetic compatibility”. Noordwijk,The Netherlands, ESA Publications
Division.
Friis-Christensen, E., H. Lühr, D. Knudsen, and R. Haagmans. 2008. Swarm–an Earth observation
mission investigating geospace. Advances in Space Research 41 (1):210–16. doi:10.1016/j.
asr.2006.10.008.
IEEE Standard P1597. 2008. Standard for validation of computational electromagnetics computer
Q9 modeling and simulation – parts 1&2, New York, IEEE
Jackson, J. D. Classical electrodynamics, 1999 New York, USA, John Wiley.
Kennedy, J., and R. Eberhart. 1995. Particle swarm optimization. Proceedings of IEEE International
Conference on Neural Networks 4:1942–48.
Li, P., Y. Li, L. Jiang, and J. Hu. 2013. A wide band equivalent source reconstruction method
exploiting the Stoer-Bulirsch algorithm with the adaptive frequency sampling. IEEE Transactions
Antennas Propagat 61:5338–43. doi:10.1109/TAP.2013.2274032.
Mikki, S. M., and Y. M. M. Antar. 2012. Near-field analysis of electromagnetic interactions in
antenna arrays through equivalent dipole models. IEEE Transactions on Antennas and
Propagation 60:1381–89. doi:10.1109/TAP.2011.2180318.
Müller, D., R. G. Marsden, O. S. Cyr, and H. R. Gilbert. 2013. Solar orbiter. Solar Physics 285:25–70.
doi:10.1007/s11207-012-0085-7.
Orlandi, A., A. P. Duffy, B. Archambeault, G. Antonini, D. E. Coleby, and S. Connor. 2006. Feature
Selective Validation (FSV) for validation of computational electromagnetics (CEM). Part II –
Assessment of FSV performance. IEEE Transactions on Electromagnetics Compatibility 48:460–
67. doi:10.1109/TEMC.2006.879360.
ELECTROMAGNETICS 15
Paul, C. R. Introduction to electromagnetic compatibility, 2006. Hoboken, New Jersey, USA, John Wiley &
Sons, Inc.
Scheeres, D. J., F. Marzari, L. Tomasella, and V. Vanzani. 1998. ROSETTA mission: Satellite orbits around
a cometary nucleus. Planetary and Space Science 46:649–71. doi:10.1016/S0032-0633(97)00200-6.
Tsitsas, N. L. 2013. A low-frequency electromagnetic near-field inverse problem for a spherical
scatterer. Journal of Computational Mathematics 31:5. doi:10.4208/jcm.1304-m4388.
Wang, Y. M., and W. C. Chew. 1989. An iterative solution of the two-dimensional electromagnetic
inverse scattering problem. International journal of imaging systems and technology. International
Journal of Imaging Systems and Technology 1 (1):100–08. doi:10.1002/(ISSN)1098-1098.
Zaiping, N., and Z. Yerong. 1998. Hybrid Born iterative method in low-frequency inverse scattering
problem. IEEE Transactions on Geoscience and Remote Sensing 36 (3):749–53. doi:10.1109/
36.673668.
Downloaded by [Tulane University] at 02:32 02 October 2017