Professional Documents
Culture Documents
Abstract— Charge accumulation on a solid dielectric surface initiate a flashover or breakdown [1], [2]. Therefore, it is
is one of the critical concerns for the design and optimization of important to obtain the charge density on solid dielectric
the insulation system in a high-voltage power equipment, since it surface with high accuracy.
will lead to the overstress of electrical insulation. Therefore, it is
important to obtain the charge density distribution on a solid Over the past decades, tremendous progress has been made
dielectric surface with high accuracy. The acquisition of surface on this subject. The earliest method of evaluating surface
charge for insulators requires multipoint potential measurements charge distribution is the so-called dust figure technique uti-
to establish the inverse calculation for the determination of an lized since the time of Lichtenberg [3]. This technique may
unknown charge distribution. Up to now, extensive studies have give a rough estimate of surface charge density and its polarity,
been conducted on this subject; nevertheless, the methods are
either too complicated and time consuming, or only applicable but it is impossible to get the density quantitatively. In the
for specific arrangements, or with poor accuracy. In this paper, late 1960s, electrostatic probes were gradually introduced
the problem is divided into two categories, i.e., shift-variant to measure the surface charge distribution as a noncontact
system and shift-invariant system, and the basic principle of method [4], [5]. There are two types of electrostatic probes.
an improved inversion algorithm is interpreted to solve the One is the capacitive probe which measures the induced
problem. The 2-D Fourier transform and Wiener filter techniques
are employed in the algorithm for shift-invariant system thus floating potential of a sensor electrode [6]. The sensor is
the relationship between potential and charge density can be usually grounded through a known capacitance, while the
processed in spatial frequency domain, which tremendously guard electrode is directly grounded in most cases. The other
simplifies the conventional procedure. The accuracy and reso- one is the Kelvin probe [7], with a feedback loop that allows to
lution of the algorithm are discussed in detail with the aid of adjust the probe potential until it equals the value at the point
numerical examples. In the end, experiments are conducted and
the effectiveness of the algorithm is verified. being measured. No matter what type of probe was used, it was
initially thought that a probe output corresponds uniquely to
Index Terms— 2-D fourier transform, inversion algorithm, the accumulated charge (density) facing the probe [4], [5].
Kelvin probe, surface charge, surface potential, Wiener filter.
Until 1976, the idea of a matrix relationship between probe
I. I NTRODUCTION outputs and charge densities was reported for the first time
by Specht [8]. Since then, extensive debates have been held
S OLID dielectrics are used as supports of stressed conduc-
tors in any high-voltage insulation system. In compressed
SF6 gas or vacuum insulation systems where high electric
in the 1980s about whether evaluation of charge densities on
a solid dielectric surface from surface potential measurement
was possible or not [9]–[12], finally reaching to a consen-
field is involved, charge accumulation on solid dielectrics may
sus that accumulated charge could be in general determined
seriously influence surface discharge and insulation character-
by multipoint measurements [6]. In other words, the output
istics, especially for dc equipment. These deposited charges
potential of the probe is determined by all the deposited
will, if of significant magnitude, lead to the overstress of
charges on the measured object. Meanwhile, different inverse
insulation due to local field distortion, and may eventually
calculation methods have also been developed to solve the
Manuscript received February 9, 2017; revised May 8, 2017; accepted matrix relationship, including integral equation method [13],
July 3, 2017. Date of publication October 16, 2017; date of current version charge simulation method [14], λ-function method [15], [16],
November 8, 2017. This work was supported by the National Basic Research
Program of China under Grant 2014CB239502. The Associate Editor coordi- -function method [17], and some methods in [18] and [19].
nating the review process was Dr. Sasan Bakhtiari. (Corresponding author: However, the above methods are either too complicated and
Guixin Zhang.) time consuming or with poor accuracy. Because, on the one
B. Zhang is with the Department of Electrical Engineering, Tsinghua
University, Beijing 100084, China, and also with the Institute of Materials hand, these methods try to solve the matrix relationship by
Science, University of Connecticut, Storrs, CT 06269 USA. directly taking the inversion of the matrix, whose dimension
W. Gao, Z. Qi, and G. Zhang are with the Department of Elec- (N × N) corresponds to the measured points N. When the
trical Engineering, Tsinghua University, Beijing 100084, China (e-mail:
guixin@mail.tsinghua.edu.cn). measured object is large and the sampling density is high,
Q. Wang is with the School of Mechatronic Engineering, Beijing Institute N can be such a huge number (for example, N = 65536
of Technology, Beijing 100081, China. in [20]) that it becomes unpractical to use these methods, due
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org. to the unacceptable computational time. On the other hand,
Digital Object Identifier 10.1109/TIM.2017.2730981 practically, noise is usually superimposed on the measured
0018-9456 © 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
ZHANG et al.: INVERSION ALGORITHM 3317
signal. The more data used in the calculation, the more The current signal I is amplified and demodulated by a
possibility the matrix will become ill posed, in which case, phase-sensitive demodulator to produce a voltage proportional
the solution often makes no sense, including extremely large to the amplitude of I . The output of the integrator drives
errors. This paper, therefore, interprets the basic principle a high-voltage amplifier circuit to replicate the voltage on
of a latest inversion algorithm to solve the problem, which the tested surface. The amplified voltage is applied to the
borrows the concept of image restoration from the area of probe thus nullifying the electric field between the tested
imaging processing. The problem is divided into two cat- surface and the sensor. The current I vanishes if the probe
egories, i.e., shift-variant system and shift-invariant system. reaches the same potential as the tested surface potential,
2-D Fourier transform (2D-FT) and Wiener filter techniques which serves as the voltmeter’s output. This voltage-following
are employed in the algorithm for shift-invariant system, which technique makes the measurement independent of the distance
was first proposed by Kumada and Okabe [21], thus the matrix D0 , at least within a certain range of D0 .
relationship can be processed in spatial frequency domain.
The amplitude-frequency characteristics of Wiener filter is B. Multipoint Measurement Method
studied, which reveals that the Wiener filtering executes an
optimal tradeoff between inverse calculation and noise smooth- As mentioned in Section I, to obtain surface charge distri-
ing. It also has a better compromise between computational bution requires multipoint probe outputs to establish inverse
efficiency and quality of the restoration with respect to some calculation. The probe output, therefore, corresponds to a lin-
other inversion schemes. The accuracy and resolution of the ear superimposition of the effects of all the deposited charges
algorithm are discussed in detail with the aid of numerical on the measured object. Suppose the entire measuring surface
examples. In the end, experiments are conducted and the is subdivided into N elements and each element j acquires a
effectiveness of the algorithm is verified. charge density σ j . Then, the probe output ϕi obtained above
the center of the element i can be given by
II. BASIC P RINCIPLE OF M EASUREMENT
N
control system drives the Kelvin probe to scan along the slope
surface of the insulator within an exact separation of 2 mm.
The detail of this platform can be found in [25] and [30].
Fig. 5. Estimated surface charge along the radial direction by inverse Fig. 6. Fourier transformation of the estimated surface charge distribution
calculation for a unit surface charge on r = 28 mm. and the cut-off frequency of the measurement system.
Fig. 12. Charge density and surface potential distribution of a unit charge in spatial frequency domain.
Fig. 13. Amplitude-frequency characteristics of the transfer function H, inverse filter 1/H, and Wiener filter W .
The results are shown in Fig. 12. Note that since the measuring Then, the Wiener filter can be built as
pitch r = 0.5 mm, σ 0 , and ϕ 0 in spatial frequency domain H ∗ (μ, ν)
holds values up to a spatial frequency of 1 cycle/mm equivalent W(μ, ν) = . (25)
|H(μ, ν)|2 + c2
to the sampling frequency of f s = 1/(2r ). From Fig. 12,
we can see that the charge distribution σ 0 has high-frequency Fig. 13 illustrates the amplitude-frequency characteris-
components while the spectrum of potential distribution ϕ 0 tics of the transfer function H, inverse filter 1/H, and
only concentrates on the low-frequency region. Wiener filter W. The transfer function H of the measuring
According to the flowchart in Fig. 10, the transfer function system converges to a very small value in the high-frequency
of the system can be obtained by region, as shown in Fig. 13(a). While the inverse filter 1/H,
which consists the reciprocal values of H, diverges in the high-
frequency region as shown in Fig. 13(b). It can enlarge the
Φ 0 (μ, ν) system noise which is mainly in the high-frequency region.
H(μ, ν) = . (24)
σ 0 (μ, ν) Fig. 13(d) shows that the amplitude of the spectrum of 1/H
3324 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 66, NO. 12, DECEMBER 2017
C. Spatial Resolution
As discussed in Section III-C, the “imaging” relationship
between the real charge density σ and the estimated charge
density σ̂ is governed by the output transfer function GH,
as shown in Fig. 4. Note that, here, G = W. At a spatial fre-
quency of |H(μc , 0)| = c, the amplitude of the output transfer
function GH becomes 0.5 and consequently a value equal to
only half the original signal can be restored. According to the
definition of spatial resolution, the resolution of the overall
measuring system is therefore can be obtained as the reciprocal
of the cut-off frequency at which |G(μc , 0)H(μc , 0)| = 0.5
or |H(μc , 0)| = c. Fig. 14 shows the output transfer
function GH when c equals to 1%, 0.5%, and 0.25%
of H(0, 0). The predicted spatial resolutions are 4.3, 2.3, Fig. 16. Surface charge accumulation on a cone-type insulator after
and 1.2 mm, respectively. application of −20 kV dc voltage for 6 h in air.
D. Accuracy Analysis Based on (16), the SNR of the calculation results are 28 and
Suppose a surface charge pattern, resembling a “THU” logo, 22 dB, respectively.
is set on the measured subject. The charge density of each
letter is set as 0.5, 1, and 0.5 pC/mm2 , as shown in Fig. 15(a).
V. E XPERIMENTAL R ESULTS
Fig. 15(b) shows the surface potential distribution calculated
by electrostatic field computation. Next, Gaussian noise with By using the measuring system described in Section III-A,
noise level of 0.25% and 0.5% is added to it, respectively. the accumulated charge on the cone-type insulator is mea-
Then, the estimated surface charge density distribution can sured after the application of −20 kV dc voltage for 6 h
be obtained by using the inversion algorithm, as shown in atmospheric air. Fig. 16(a) shows the measured surface
in Fig. 15(c) and (d), where c equals to 0.5%|H(0, 0)|. We can potential distribution which is the output of the Kelvin probe.
see that the estimated surface charge distribution agrees Fig. 16(b) shows the estimated surface density distribution by
very well with the originally assumed charge distribution. using (12). Obviously, charge density distribution gives more
ZHANG et al.: INVERSION ALGORITHM 3325
R EFERENCES
[1] S. Tenbohlen and G. Schroder, “The influence of surface charge on
lightning impulse breakdown of spacers in SF6 ,” IEEE Trans. Dielectr.
Electr. Insul., vol. 7, no. 2, pp. 241–246, Apr. 2000.
[2] S. Kumara, S. Alam, I. R. Houqe, Y. V. Serdyuk, and S. M. Gubanski,
“DC flashover characteristics of a polymeric insulator in presence of
surface charges,” IEEE Trans. Dielectr. Electr. Insul., vol. 19, no. 3,
pp. 1084–1090, Jun. 2012.
[3] G. C. Lichtenberg, “De nova methodo naturam ac motum fluidi electrici
investigandi,” Akademie der Wissernschaften, Göttingen. Phys. Math.,
K1. Novi Commentarii, vol. 8, pp. 168–180, 1777.
Fig. 17. Surface charge measurement on a flat insulator after negative dc [4] D. K. Davies, “The examination of the electrical properties of insulators
corona charging. by surface charge measurement,” J. Sci. Instrum., vol. 44, no. 7,
pp. 521–524, 1967.
[5] T. R. Foord, “Measurement of the distribution of surface electric charge
details than potential distribution. Therefore, more accurate by use of a capacitive probe,” J. Phys. E, Sci. Instrum., vol. 2, no. 5,
behaviors of surface charge are likely to be obtained. pp. 411–413, 1969.
[6] T. Takuma, M. Yashima, and T. Kawamoto, “Principle of surface charge
Surface charge measurements on a flat insulator made of measurement for thick insulating specimens,” IEEE Trans. Dielectr.
epoxy resin are also carried out. As shown in Fig. 17(a), Electr. Insul., vol. 5, no. 4, pp. 497–504, Aug. 1998.
a “needle-to-plane” electrode is used for charging the insulator. [7] W. A. Zismann, “A new method of measuring contact potential differ-
ences in metals,” Rev. Sci. Instrum., vol. 3, no. 7, pp. 367–368, 1932.
A negative dc corona (Udc = −5 kV) occurs at the tip of [8] H. Specht, “Oberflächcnladungen bei rotatians-symmetrischen isolier-
the needle with a separation of 10 mm above the sample stoffkörpern,” ETZ-A, vol. 97, pp. 474–476, 1976.
center. The sample is first corona charged for one minute [9] L. G. Christophrou, Gaseous Dielectrics III. New York, NY, USA:
Pergamon, 1982, pp. 566–567.
and then transferred to the measurement platform, as shown [10] A. Pedersen, “On the Electrostatic field near the charged surface of an
in Fig. 17(b). The measured surface potential and estimated insulator with special reference to surface charge probe measurement,”
charge density distribution is shown in Fig. 17(c) and (d), in Gaseous Dielectrics IV. L. G. Christophorou and M. O. Pace, Eds.
New York, NY, USA: Pergamon, 1984, pp. 414–421.
respectively. We can see that the surface potential distribution [11] M. Yashima, H. Fujinami, and T. Takuma, “Measurement of accumulated
after negative corona discharging has a bell-like shape and charge on dielectric surfaces with an electrostatic probe,” Gaseous
the potential values are all negative. As for charge density Dielectrics, vol. 5, pp. 242–248, 1987.
[12] M. N. Horenstein, “Measuring surface charge with a noncontacting
distribution, both positive and negative charges are detected. voltmeter,” in Proc. Conf. Rec. IEEE Ind. Appl. Soc. Annu. Meet., vol. 3.
The existence of hetero charges may be due to the attractive Toronto, ON, Canada, Oct. 1993, pp. 1811–1816.
Coulombic force toward the sample center. And it has been [13] C. E. Sudhakar and K. D. Srivastava, “Electric field computation from
probe measurements of charge on spacers subjected to impulse voltages,”
verified by dust figure, as shown in Fig. 17(e). in Proc. 5th Int. Symp. High Voltage Eng., Braunschweig, Germany,
1987, pp. 1–25, paper 33.14.
[14] H. Ootera and K. Nakanishi, “Analytical method for evaluating surface
VI. C ONCLUSION charge distribution on a dielectric from capacitive probe measurement
This paper presents an improved inversion algorithm to application to a cone-type spacer in + or −500 kV DC-GIS,” IEEE
Trans. Power Del., vol. 3, no. 1, pp. 165–172, Jan. 1988.
calculate charge density on solid dielectric surface based [15] A. Pedersen, “On the electrostatics of probe measurements of surface
on potential measurement. After the interpretation of basic charge densities,” in Proc. Gaseous Dielectr. V, 1987, pp. 235–240.
principle of the multipoint measurement, the algorithms for [16] T. O. Rerup, G. C. Crichton, and I. W. McAllister, “Using the λ function
to evaluate probe measurements of charged dielectric surfaces,” IEEE
shift-variant system and shift-invariant system are discussed Trans. Dielectr. Electr. Insul., vol. 3, no. 6, pp. 770–777, Dec. 1996.
in detail. [17] D. C. Faircloth and N. L. Allen, “High resolution measurements of
1) In a shift-variant system, Wiener filter based on the surface charge densities on insulator surfaces,” IEEE Trans. Dielectr.
Electr. Insul., vol. 10, no. 2, pp. 285–290, Apr. 2003.
Tikhonov’s regularization is introduced to solve the ill- [18] A. Tatematsu, S. Hamada, and T. Takuma, “Estimation of charge
posed problem of the transfer function matrix H and distribution on a bulky solid dielectric using regularization technique,”
the restoration characteristics are explainable in terms Electr. Eng. Jpn., vol. 163, no. 2, pp. 1–13, 2008.
[19] Q. Wang, G. Zhang, and X. Wang, “Characteristics and mechanisms
of eigenvalues and eigenvectors of the output transfer of surface charge accumulation on a cone-type insulator under DC
function GH. voltage,” IEEE Trans. Dielectr. Electr. Insul., vol. 19, no. 1, pp. 150–155,
2) In a shift-invariant system, the restoration can be Feb. 2012.
[20] H. Mu and G. Zhang, “Calibration algorithm of surface charge density
processed strictly in the spatial frequency domain by on insulating materials measured by Pockels technique,” Plasma Sci.
2D-FT, which avoids matrix calculation and significantly Technol., vol. 13, no. 6, pp. 645–650, 2011.
simplifies the conventional procedure. [21] A. Kumada, S. Okabe, and K. Hidaka, “Resolution and signal processing
technique of surface charge density measurement with electrostatic
3) The spatial resolution of the algorithm is evaluated based probe,” IEEE Trans. Dielectr. Electr. Insul., vol. 11, no. 1, pp. 122–129,
on the PSF of the restoration system. And the accuracy Feb. 2004.
3326 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 66, NO. 12, DECEMBER 2017
[22] A. Kumada and S. Okabe, “Charge distribution measurement on a Boya Zhang was born in 1990. He received the
truncated cone spacer under DC voltage,” IEEE Trans. Dielectr. Electr. B.S. degree in electrical engineering from Tsinghua
Insul., vol. 11, no. 6, pp. 929–938, Dec. 2004. University, Beijing, China, in 2013. He is currently
[23] H. Iwabuchi et al., “Influence of surface-conductivity nonuniformity on pursuing the Ph.D. degree with the Department of
charge accumulation of GIS downsized model spacer under DC field Electrical Engineering, Tsinghua University.
application,” Electr. Eng. Jpn., vol. 181, no. 2, pp. 29–36, 2012. Since 2016, he has been a Visiting Scholar with
[24] H. Iwabuchi et al., “Influence of tiny metal particles on charge accumu- the Institute of Materials Science, University of
lation phenomena of GIS model spacer in high-pressure SF6 gas,” IEEE Connecticut, Storrs, CT, USA. His current research
Trans. Dielectr. Electr. Insul., vol. 22, no. 5, pp. 1895–1901, Oct. 2013. interests include HVDC gas-insulated system and
[25] B. Zhang, G. Zhang, Q. Wang, C. Li, J. He, and Z. An, “Suppression charge accumulation phenomena under dc stress and
of surface charge accumulation on Al2 O3 -filled epoxy resin insulator dc insulation materials.
under DC voltage by direct fluorination,” AIP Adv., vol. 5, no. 12,
pp. 127207–127212, 2015.
[26] B. Zhang, Z. Qi, and G. Zhang, “Thermal gradient effects on surface
charge of HVDC spacer in gas insulated system,” in Proc. IEEE Conf.
Elect. Insul. Dielectr. Phenomena, Toronto, ON, Canada, Oct. 2016,
pp. 703–706. Wenqiang Gao was born in 1991. He received the
[27] B. Zhang, Q. Wang, and G. Zhang, “Measurement and modeling of B.S. degree in electrical engineering from Sichuan
surface charge accumulation on insulators in HVDC GIL,” Cigre Sci. University, Chengdu, China, in 2014. He is currently
Eng., vol. 3, pp. 81–88, Oct. 2015. pursuing the M.S. degree with the Gas Discharge
[28] B. Qi, C. Gao, C. Li, L. Zhao, X. Sun, and Y. Lei, “The accumulation and Plasma Laboratory, Department of Electrical
characteristic of surface charges on GIS Insulator under switching Engineering, Tsinghua University, Beijing, China.
impulse voltages in SF6,” in Proc. IEEE Conf. Elect. Insul. Dielectr. His current research interests include surface
Phenomena, Ann Arbor, MI, USA, Oct. 2015, pp. 130–133. charge measurement techniques.
[29] T. Vu-Cong, L. Zavattoni, P. Vinson, and A. Girodet, “Surface charge
measurements on epoxy spacer in HVDC GIS/GIL in SF6,” in Proc.
IEEE Conf. Elect. Insul. Dielectr. Phenomena, Toronto, ON, Canada,
Oct. 2016, pp. 93–96.
[30] B. Zhang, Z. Qi, and G. Zhang, “Charge accumulation patterns on spacer
surface in HVDC gas-insulated system: Dominant uniform charging and
random charge speckles,” IEEE Trans. Dielectr. Elect. Insul., vol. 24,
Zhe Qi was born in 1992. He received the B.S.
no. 2, pp. 1229–1238, Apr. 2017.
degree in electrical engineering from Tsinghua Uni-
[31] C. L. Lawson and R. J. Hanson Solving Least Squares Problems.
versity, Beijing, China, in 2015.
Englewood Cliffs, NJ, USA: Prentice-Hall, 1974.
He was a Technician with the Gas Discharge and
[32] Liu J. The Regularization Method and Application of the Ill-Posed
Plasma Laboratory, Department of Electrical Engi-
Problems. Beijing, China: Science Press, 2005.
neering, Tsinghua University. His current research
[33] W. K. Pratt, Digital Image Processing, New York, NY, USA: Wiley,
interests include high-voltage testing and signal
1987.
processing and programming.
[34] A. N. Tikhonov and V. Y. Arsenin, Solution of Ill-Posed Problems.
Washington, DC, USA: V. H. Winston & Sons, 1977.
[35] A. Y. Ng, “Feature selection, L1 vs. L2 regularization, and rotational
invariance,” in Proc. 21st Int. Conf. Mach. Learn., Banff, AB, Canada,
2004, p. 78.
[36] P. C. Hansen, “Analysis of discrete ill-posed problems by means of the
L-curve,” SIAM Rev., vol. 34, no. 4, pp. 561–580, Dec. 1992.
[37] S. Okabe and A. Kumada, “Measurement methods of accumulated Qiang Wang was born in 1984. He received the
electric charges on spacer in gas insulated switchgear,” IEEE Trans. B.S. degree in electrical engineering from Xi’an
Power Del., vol. 22, no. 3, pp. 1547–1556, Jul. 2007. Jiaotong University, Xian, China, in 2006, and the
[38] R. M. Schaffert, Electrophotography. London, U.K.: Focal Press, 1965. Ph.D. degree in electrical engineering from Tsinghua
[39] M. Schueller, U. Straumann, and C. M. Franck, “Role of ion sources University, Beijing, China, in 2012.
for spacer charging in SF6 gas insulated HVDC systems,” IEEE Trans. In 2015, he joined the Department of Mechachon-
Dielectr. Electr. Insul., vol. 21, no. 1, pp. 352–359, Feb. 2014. ics, Beijing Institute of Technology, Beijing, as a
[40] A. Winter and J. Kindersberger, “Stationary resistive field distribution Researcher involved in gaseous discharge, plasma
along epoxy resin insulators in air under DC voltage,” IEEE Trans. and intelligent control technology.
Dielectr. Electr. Insul., vol. 19, no. 5, pp. 1732–1739, Oct. 2012.
[41] S. Kumara, Y. V. Serdyuk, and S. M. Gubanski, “Surface charge decay
on polymeric materials under different neutralization modes in air,” IEEE
Trans. Dielectr. Electr. Insul., vol. 18, no. 5, pp. 1779–1788, Oct. 2011.
[42] Electrostatic Voltmeter Solutions Brochure, accessed on Jan. 2017.
[Online]. Available: http://www.trekinc.com/pdf/Trek_ESVM.pdf
[43] R. Bracewell, “Fourier analysis and imaging,” in The Two-Dimensional Guixin Zhang was born in 1963. He received the
Convolution Theorem. New York, NY, USA: Springer, 2003, ch. 6. B.S. and M.S. degrees from the Department of
[44] J. Kindersberger and C. Lederle, “Surface charge decay on insulators in Electrical Engineering, Tsinghua University, Beijing,
air and sulfurhexafluorid—Part II: Measurements,” IEEE Trans. Dielectr. China, in 1987 and 1993, respectively, and the Ph.D.
Electr. Insul., vol. 15, no. 4, pp. 949–957, Aug. 2008. degree from the Department of Physics, Nanyang
[45] B. Zhang and G. Zhang, “Interpretation of the surface charge decay Technological University, Singapore, in 1999.
kinetics on insulators with different neutralization mechanisms,” J. Appl. He is currently a Full Professor with the
Phys., vol. 121, no. 10, pp. 105105–105110, 2017. Department of Electrical Engineering, Tsinghua
[46] J. Deng, A. Kumada, K. Hidaka, G. Zhang, and H. Mu, “Resid- University. His current research interests include
ual charge density distribution measurement of surface leader with plasma, high-voltage insulation, optical fiber sensor,
feedback electrostatic probe,” Appl. Phys. Lett., vol. 100, no. 19, signal processing, and electronic current/voltage
pp. 192904–192906, 2012. transformer.