logoGlobal Energy Interconnection

Contents

Figure(0

    Tables(0

      Global Energy Interconnection

      Volume 5, Issue 1, Feb 2022, Pages 118-130
      Ref.

      Research on characteristics of acoustic signal of typical partial discharge models

      Hang Ji1 ,Xing Lei1 ,Qiang Xu1 ,Chengjun Huang2 ,Ting Ye1 ,Shangqing Yuan1
      ( 1.State Grid Shanghai Municipal Electric Power Company, Shanghai 200122, P.R.China , 2.Power Monitoring and Diagnostic Technology Ltd., San Jose CA 95119, USA )

      Abstract

      The pulse current method, acoustic and ultrasonic partial discharge (PD) detection, and voiceprint PD detection are commonly used detection methods for the PD detection of power equipment.To study the characteristics of PD signals of typical discharge models based on the principles of the above three detection methods, an acoustic detection experimental system consisting of a needle-tip model and a surface model was built.Acoustic tests were carried out on needle-tip models with different curvature radii and surface discharge models with different lengths of conductive paste.The experimental results showed that acoustic and ultrasonic PD detection and voiceprint PD detection exhibited different sensitivities to the needle-tip discharge models, and the combination of acoustic and ultrasonic PD and voiceprint PD detection was more beneficial for the comprehensive detection of cable PD signals.Based on voiceprint recognition technology, this study drew FFT (Fast Fourier Transformation) diagrams of different types of PD acoustic signals and analyzed the differences in the ultrasonic signal frequency distribution.The frequency band of the voiceprint PD signal of the needle-tip discharge models was concentrated in the range 17-27 kHz, and the frequency band of the voiceprint PD signal of the conductive paste discharge models was concentrated in the range 20-25 kHz.The measurement of voiceprint PD signals in these frequency bands were strengthened when the PD of a cable was detected on-site, which provides the basis for the use of the cable model for on-site PD detection.

      0 Introduction

      Partial discharge (PD) has a significant influence on the insulation condition of power equipment.PD changes indicate a deterioration of the electrical insulation performance of equipment and may lead to power system accidents.At present, the on-line detection method is an effective method to solve the time-consuming and laborious problem of cable fault repair and is of great significance in improving the reliability of cable operations [1-3].In reference [4], the problem of cable defect diagnosis and positioning is introduced.The positioning method is mainly for centralized defects, and the positioning effect of local defects in actual situation is not good.PD detection mainly solves two problems: PD location and PD-type identification.

      In terms of PD location, PD detection methods mainly include pulse current, ultra-high frequency (UHF) detection,acoustic detection, and optical detection [5].Among these, the pulse current method is presently the standard method because of its high sensitivity, easy calibration,and ability to measure discharge quantity.However, it can only be tested off-line.The UHF detection method has a mature technology and high sensitivity.However, its anti-electromagnetic interference and diffraction abilities are poor.Therefore, it is widely used in gas-insulated switchgear (GIS) and other equipment with less interference[6].The acoustic measurement method has strong antielectromagnetic interference ability and is suitable for the on-site PD detection in complex electromagnetic environments.However, it is vulnerable to environmental acoustic noise, has poor sensitivity, can hardly reflect the amount of discharge, and is not conducive to estimating the degree of insulation deterioration.To overcome these technical limitations, different methods are used to improve.The high sensitivity of an ultrasonic phased array sensor was used to improve positioning accuracy [7]; although the effectiveness of the new positioning method was verified,its experimental environment was ideal, and only a single noise source was considered.In reference [8], a control method for a fiber-optic ultrasonic sensor was proposed that improved the possibility of fiber-optic ultrasonic sensors being interfered with by low-frequency vibration and noise in the environment during field applications.In reference [9],acoustic imaging technology was used to study the abnormal sound of typical corona discharge equipment at the highvoltage end of GIS, and the rapidity and intuitiveness of acoustic imaging technology was applied to the PD location.In reference [10], a propagation-time estimation algorithm was proposed to calculate the propagation time and improve the accuracy of the PD location of a transformer.In addition, PD detection research has also been carried out on new sensor designs [11], multi-sensor fusion, and algorithm improvement [12-14].

      PD-type identification mainly focuses on a pattern recognition algorithm for the data obtained by the above methods.With the rise of artificial intelligence technology,various identification methods using deep learning have also emerged.Detecting the characteristics of PD signals not only quickly diagnoses fault types but also helps to predict the state of power equipment [15].At present, the traditional mode feature extraction method mostly adopts mathematical statistics in the time-frequency domain as a feature parameter [16,17].However, in the case of the coexistence of noise interference and multi-source discharge, there are problems such as unclear features and signal overlap [18,19].Because an artificial intelligence algorithm can effectively separate different types of similar signals in the time and frequency domains, it has been widely used in the field of PD pattern recognition and state evaluation [20,21].However, similar to the application of most artificial intelligence algorithms, most studies have only provided good results with laboratory models.In addition, the amount of data in the database has been small,the type was different, and a standard was lacking [22].

      In practical applications, the acoustic measurement method is easily affected by environmental noise.At present,it is mostly used for GIS, electrical cabinets, open overhead lines, and other equipment with less interference.There are few studies on the effectiveness of acoustic and ultrasonic PD detection and voiceprint PD detection for electrical equipment under actual working conditions, especially for cable PD detection.To improve the sensitivity of acoustic PD detection, the set measurement frequency band should cover the main components of the measured frequency to eliminate or reduce various interferences.However, for the application of acoustic imaging detection in PD detection,voiceprint PD detection will be disturbed by the ultrasonic frequency band of a leakage point above 20 kHz and will be affected by the noise frequency band below 20 kHz.The accurate setting of the frequency band is key to ensuring the correctness of voiceprint PD detection results.During an on-site voiceprint PD test, the best detection effect of the PD signal can be determined by adjusting the voiceprint atlas of different detection bands, which increases the difficulty of voiceprint PD detection.The effectiveness of a PD detection technology depends on its sensitivity to the signal characteristics emitted by a PD and the anti-interference of noise.Therefore, some researchers have selected to study the PD signal characteristics of various cables as their primary work [23,24].

      Local defect diagnosis and location require signal characteristics, so it is necessary to analyze models under different conditions [25], including cable insulation materials [26,27], ambient temperature [28], and defect types [29,30].In reference [31], the finite element method(FEM) is used to analyze the electric field variation of high voltage cable intermediate joint under different defect models, and the PD of the cable is tested.In reference[32], the metal particle defects were set on the silicone rubber / XLPE interface of 35 kV cable intermediate joint to explore the development process of partial discharge of insulation defects, and two characteristic quantities that can characterize the statistical law of partial discharge at each stage were extracted.

      To solve the problem of the effectiveness of voiceprint PD detection methods for cable PD detection, a PD measurement system for different curvature-radius needle models and corona models in cables was built, and the characteristics of acoustic signals and voiceprint signals generated by the PD of the experimental models were analyzed.Compared with the advantages and disadvantages of different detection methods, the optimal cable PD detection method and the optimal detection frequency band of voiceprint PD were obtained.

      1 PD detection methods

      1.1 Pulse current method

      The basic principle of the pulse current method is shown in Fig.1.When sample Cx produces a PD, the pulse current generates an instantaneous voltage change at both ends of the detection impedance through the coupling capacitor Ck, that is, the pulse voltage ΔU.The pulse voltage can be used to measure the basic parameters of the PD after transmission, amplification, and display.

      Fig.1 Basic principle of the pulse current method

      The pulse current method measures the lower-frequency components in the PD spectrum to avoid radio interference.Traditional measuring instruments are generally equipped with a pulse peak meter to indicate the pulse peak and an oscilloscope tube to display the pulse size, number,and phase.The amplifier gain is very large, and its test sensitivity is quite high, and it can use the pulse injection of known charge to correct the quantity and measure the discharge q.When PD occurs, the pulse current flows through the sample, which is the apparent charge.

      The Fourier integral of the above formula can be obtained as

      where I() is the complex frequency spectrum of the pulse current.If ω→0, then

      Therefore, the DC component of the pulse current spectrum corresponds to the apparent charge.

      Another detection parameter of the pulse current method is the phase distribution diagram of PD; that is, the phase and amplitude of the discharge time are recorded by the detection instrument in a voltage cycle, and each discharge event is expressed in the form of a point or line.For example, for a needle-tip model with a curvature radius of 0.13 mm, the amplitude spectrum measured by the pulse current method is shown in Fig.2 when 7 kV is applied and the partial discharge is 26.23 pC.The pulse current detection method can determine the strength of the PD by observing the number of PD pulses in a cycle.The detection method can calibrate the PD magnitude.The pulse current method has the advantages of high sensitivity, ease of calibration, and monitoring safety, but can only detect the power equipment PD offline.

      Fig.2 Pulse current test spectrum data

      1.2 Acoustic and ultrasonic PD detection method

      The generation of PD is accompanied by a series of physical phenomena, including the excitation of ultrasonic signals that radiate outward from the interior of power equipment and are received by ultrasonic sensors in the discharge space area.The method of monitoring the size and position of a PD by measuring the sound wave generated by the PD is called acoustic measurement.The acoustic information generated by a PD is relatively small,and the frequency spectrum is widely distributed, ranging from 101 to 107 Hz.The acoustic frequency spectrum detected in different power detection equipment is different owing to different discharge states, propagation media,and environmental conditions.In an acoustic PD detector,the sound pressure generated by a PD is measured and converted into an electrical signal by a sensor and then amplified.The sound pressure level is calculated as

      where P0 is the reference sound pressure, and P is the sound pressure amplitude.

      When the acoustic impedance and the distance between the sound source and the probe for measuring the positive pressure are constant, the sound energy WA is proportional to the square of the sound pressure

      and

      where K1 and K2 are constants, and QA is the reading of the acoustic PD instrument.

      The principle of ultrasonic PD detection is illustrated in Fig.3.The detection method has no connection with the electrical circuit, is not subject to electrical interference,and can realize live detection.The PD of the power equipment can be analyzed by observing and analyzing the characteristics of the collected ultrasonic signal.Acoustic and ultrasonic PD detection uses a portable partial inspection instrument for detection.In this study, we adopted the PDS-T95 acoustic PD detection instrument to obtain the effective value, cycle maximum, 50 Hz frequency component, and 100 Hz frequency component in the AE amplitude and waveform spectra of PD signals to judge the strength of a PD.For the needle-tip model with a curvature radius of 0.35 mm, an applied voltage of 18 kV, and a test distance of 2.8 m between the PDS-T95 PD monitor and the tip, the AE amplitude spectrum and AE waveform spectrum of PDS-T95 are shown in Fig.4.

      Fig.3 Basic principle of space ultrasonic PD

      Fig.4 Acoustic and ultrasonic PD patterns

      1.3 PD detection method for voiceprint

      Voiceprint imaging detection methods based on microphone array measurement technology adopt visual acoustic imaging technology to measure the sound pressure on the surface of a holographic.A reconstruction algorithm is used to reconstruct the sound field on the surface of the tested equipment, measure the amplitude of the sound source, and display the spatial distribution of the sound source in the form of an image.In this manner, a cloud image of the spatial sound field distribution-sound image is obtained, in which the color and brightness of the image represent the strength of the sound.The monitoring technology can effectively measure the distribution of the sound field and is a combined application of sound images and visible videos.

      Visual voiceprint imaging technology integrates multidimensional information of space, time, and frequency and has the advantages of non-contact testing and provision of intuitive results.It can effectively realize fault analysis and defect location of power equipment and has incomparable advantages for cable PD detection.Ultrasonic radiation occurs when a PD defect occurs in a cable.The microphone array of the voiceprint PD instrument obtains the ultrasonic sound pressure signal of the cable PD model and performs a Fourier transform on the sound signal to obtain the maximum frequency band of the sound pressure level fmax.The sound pressure signal of the maximum frequency band was calculated based on a beamforming algorithm to obtain the position of the sound pressure signal in each sound source area on the cable surface as

      Where B(τ,θ) is the position of the sound pressure signal corresponding to the maximum frequency band of the sound pressure level fmax; Pn is the sound pressure signal corresponding to the maximum frequency band of the sound pressure level fmax, θ is the focusing direction of the sound source, kn is the weight vector of the sound pressure signal corresponding to the maximum frequency band, n = 1,2,...,N,N is the number of microphones, τ=ln sinθ /c0, ln is the distance between the microphone with serial number n and the reference microphone, and c0 is the propagation speed of the sound wave in the propagation medium.

      The waveform diagram of voiceprint PD detection is shown in Fig.5, and the voiceprint PD detection schematic is shown in Fig.6.After voiceprint PD detection, the voiceprint PD spectrum on the voiceprint PD detector can be uploaded to the local diagnostic software, and the spectrum can be analyzed in the time domain, frequency domain, spectrogram, and cestrum.

      Fig.5 PD pattern of voiceprint

      Fig.6 Schematic diagram of PD detection

      Through the introduction of the principles of the above three PD monitoring technologies, a comparison of the three PD monitoring technologies is shown in Table 1.

      Table 1 Comparison of three detection methods

      PD method Pulse current method Space ultrasonic PD detection method PD detection method for voiceprint Technical indicators PD magnitude, number of discharge pulses, discharge pulse amplitude Effective value, periodic maximum, 50 Hz and 100 Hz frequency components Time domain waveform, frequency domain waveform, spectrum, cepstrum Test method Contact test Non-contact test Non-contact test Advantage High sensitivity More PD information enhances the ability to judge the strength and type of PD signal Intuitive results

      2 PD models

      2.1 Needle-tip model

      The corona discharge caused by the needle-tip model is directly related to the maximum electric field intensity at the top of the needle tip.When the electric field intensity exceeds the electric field intensity threshold, a discharge occurs near the needle tip.The discharge voltage of a needle-tip gas chamber depends on the distance between the needle tip and the ground and the curvature radius of the tip electrode.The amount of discharge depends only on the radius of curvature of the tip electrode.When the radius of curvature of an electrode tip increases, the voltage of the discharge chamber increases slightly, and the discharge capacity increases.To reduce the error caused by the electric field calculation, the needle-tip discharge model of the coaxial electrode was studied, and its schematic diagram is shown in Fig.7.

      Fig.7 Needle-tip discharge model

      The discharge waveform of the needle-tip model is shown in Fig.8.Needle-tip discharge occurred when the gas was located around the conductor.Taking a needle-plate electrode system as an example, when the applied voltage increases to the breakdown field strength of the gas, the field strength near the tip is the highest; thus, the first discharge always occurs when the needle-tip charge is negative.The discharge pulse appears near the 270° phase of the applied voltage.As shown in Fig.8 (a), the discharge pulse was almost equal in amplitude and interval.With an increase in voltage, the number of discharge pulses increased, and at higher voltages, discharge pulses appeared in the positive half cycle.With an increase in voltage, the magnitude of the negative half-cycle discharge remained almost unchanged,but the frequency of discharge pulse increased.When the voltage was high enough, a small amount of discharge with a large amplitude appeared in the positive half-cycle, and the positive and negative half-cycles showed extremely asymmetric characteristics as shown in Fig.8 (b).

      Fig.8 Needle-tip model discharge waveform

      At the dielectric interface, the continuity of the normal components with different dielectric induction strengths can be obtained as

      where r is the radius of curvature of the tip, V is the external applied voltage of the tip model, Vi is the applied voltage outside the air gap, Δ is the air gap between the tip and the insulating medium, R is the distance between the model and the ground, ε1 is the air dielectric constant, and ε2 is the dielectric constant.

      The relationship between the electric potential and electric field intensity satisfies E=-∇φ, and the electric field intensity near the needle-tip metal conductor surface can be obtained according to the electric potential distribution as

      2.2 Surface PD model

      The PD process along an insulator surface is similar to that of the internal discharge process.When an electric field component is parallel to a dielectric surface, surface discharge is triggered.Taking a simple insulating sleeve as an example, it is a typical surface discharge model with a very uniform electric field and a strong vertical component.The surface discharge model is shown in Fig.9, and the series breakdown voltage between the air gap and solid medium was calculated as

      Fig.9 Surface discharge model

      where the breakdown voltage of the air gap is U i =Δ·Ei,Ei is the breakdown field strength of the air gap, Δ is the airgap length, d is the thickness of the medium, and ε is the relative dielectric constant of the solid medium.

      For the surface discharge model, if the electrode system is asymmetric and the discharge only occurs at the edge of one electrode, then the discharge pattern is asymmetric.When the discharge electrode is connected to a high voltage source and the non-discharge electrode is grounded, the discharge pattern that appears in the negative half cycle has a small discharge magnitude and more discharge occurrences, while the positive half cycle has a large discharge magnitude and fewer occurrences of discharge shown in Fig.10 (a).In contrast, if the discharge electrode is grounded and the nondischarged electrode is connected to a high voltage source,the discharge pattern is also reversed as shown in Fig.10 (b).

      Fig.10 Discharge waveforms of surface discharge model

      3 PD experiments

      The circuit used for the PD experiments was the same as that of the experiments with the needle-tip and surface models.

      3.1 Experimental method

      The PD test method measured the number of pulses caused by a PD, and the experiment adopted the step-bystep boosting method.Because the pulse current PD detector is the most sensitive detector, when it detected the first PD signal, the voltage was calibrated as the initial discharge voltage.To test the sensitivity of different acoustic detection methods, T95 and conventional voiceprint cameras were used to collect the acoustic and ultrasonic PD and voiceprint PD signals of each model under three groups of voltages.After the initial discharge of a defect, the defect was boosted again, and the boost amplitude was 1-3 kV each time to ensure that the acoustic signal of a stable discharge was collected.

      3.2 Acoustic PD experiment with needle-tip model

      Considering the process requirements of cable manufacturers are different, the cable joint have difference size tips or burr.A total of five needle-tip models were set up for PD acoustic testing.Using the matching digital microscopic image processing system of an optical microscope, the measured curvature radii of the five needletip models were 0.1, 0.13, 0.3, 0.4, and 0.5 mm.The No.2 needle tip was made of aluminum, while the other four needle tips were made of stainless steel.

      Fig.11 Experimental circuit for PD detection of model defects

      Fig.12 Needle-tip models with different radii of curvature

      Fig.13 PD voiceprint test of needle-tip models

      According to the above experimental method, five needle-tip models with different curvature radii were tested.The test results are presented in Table 2.

      Table 2 Test results of different needle-tip models

      Model Applied voltage/kV Pulse count AE Amplitude/dB Voiceprint/dB 1 8≤5 Nan N 10 5 < X ≤ 20 Nan N 11 ≥20 -1 Y, 3.5 2 8≤5 Nan N 9 5 < X ≤ 20 -1 N 10 ≥20 -1 Y, -1.9 3 17 ≤5 4 3.8 19 5 < X ≤ 20 8 7.1 22 ≥20 11 9.7

      continue

      Model Applied voltage/kV Pulse count AE Amplitude/dB Voiceprint/dB 4 20 ≤5 8 6 23 5<X≤20 12 11.4 25 ≥20 13 12.3 5 23 ≤5 11 8.4 26 5<X≤20 15 13.8 29 ≥20 17 15.7

      It can be concluded from Fig.14 that the larger the curvature radius of the needle tip, the greater the initial discharge voltage.Because the No.2 needle tip was made of aluminum, the initial discharge voltage was lower than that of the No.1 needle tip with a radius of curvature of 0.1 mm.

      Fig.14 Relationship between applied voltage and discharge pulse count of needle-tip models with different radii of curvature

      When the number of discharge pulses of the No.2 needle tip was between 5 and 20, the acoustic and ultrasonic PD detection and voiceprint PD detection exhibited different sensitivities.Fig.15 shows the results of the acoustic and ultrasonic PD detection.The AE waveform shows two obvious pulse signals in two cycles with obvious tip discharge characteristics.At this point, voiceprint PD detection revealed no PD point.

      Fig.15 AE diagram of No.2 needle tip with application of 9 kV

      FFT was used to transform the voiceprint PD signal when the maximum voltage was applied to the needle tips of models with different curvature radii, and the PD spectrum analysis diagrams shown in Fig.16 were obtained.

      Fig.16 Voiceprint PD spectra of needle-tip models with different needle tip radii of curvature

      The following conclusions can be drawn from the experimental results of the needle-tip models.

      (1) In the same experimental environment, the greater the needle-tip radius of curvature, the greater the voltage level required to achieve the same number of PD pulses.

      (2) When the radius of the needle tip was 0.13 mm, the sensitivity of voiceprint PD detection was lower than that of the pulse current detector and T95.Using only an acoustic camera for PD needle-tip model monitoring will miss discharge defects with less than 20 discharge pulses.

      (3) The voiceprint PD signal frequency of the needletip discharge models with different curvature radii is concentrated in the range 17-27 kHz, and a trough existed at approximately 30 kHz.

      3.3 Acoustic PD experiment with surface model

      The surface discharge model was simulated by coating different lengths of conductive paste on the postinsulator surface, which can be used to simulate the partial discharge phenomenon caused by different degrees of dust accumulation on the surface of the cable sleeve.The DDG-A conductive paste was an electrical contact coating with good electrical properties.The main materials were conductive powder and grease having the characteristics of oxidation resistance and high-temperature resistance.In the experiment, to study the surface discharge models with different degrees of conductive paste, the lengths of conductive paste were set to 2, 4, 6, 8, 10, and 12 cm from the high-voltage end of the insulator.The surface discharge model is shown in Fig.17.

      Fig.17 Surface discharge models

      The conductive paste length of the actual surface PD test model was 2 cm.Surface PD models with conductive paste lengths of 4, 6, 8, 10, and 12 cm were set up in the same manner.

      According to the experimental method, the conventional PD pulse test and different surface PD models were tested using a T95 voiceprint camera.The test process is shown in Fig.18.The experimental results are presented in Table 3, and the spectra of surface discharge model PD signal voiceprint were shown in Fig.19.

      Table 3 Acoustic experiment results of surface discharge models

      Model number Conductive paste length/cm Applied voltage/kV Pulse count AE Amplitude/dB Voiceprint/dB 1 2 14 5 22 Y, 5 2 4 12 6 13 Y, 5.8 3 6 10 5 17 Y, 6 4 8 7 4 21 Y, 7.8 5 10 6 4 8 Y, 4.8 6 12 5 5 123 Y, 11

      Fig.18 Acoustic experiment setup for surface discharge model

      Fig.19 Surface discharge model PD signal voiceprint spectra

      The following conclusions can be drawn from the experimental results of the surface PD models.

      (1) The greater the length of conductive paste on the surface of the insulator, the shorter the creepage distance of the insulator and the smaller the initial discharge voltage,and it was easier to create a PD.

      (2) Acoustic and ultrasonic PD detection and voiceprint PD detection have the same sensitivity for surface PD models with different lengths of conductive paste.

      (3) The voiceprint PD signal of the insulator surface discharge model was concentrated in the range 20-25 kHz,and a trough existed at approximately 30 kHz.

      4 Conclusions

      Based on the principles of pulse PD detection, acoustic and ultrasonic PD detection, and voiceprint PD detection,the discharge principles of needle-tip PD and surface PD models were studied.The acoustic test system of five needle-tip models with different radii of curvature and six surface PD models with different lengths of conductive paste were built in the laboratory.Through the analysis of the acoustic test results, the following conclusions were drawn.

      (1) Compared with the insulator surface discharge model, the acoustic and ultrasonic PD detection and voiceprint PD detection exhibited different sensitivities to PD signals of some needle-tip discharge models.The PD defects of some needle-tip models will not be detected if only voiceprint PD detection is used for PD detection.The combination of acoustic and ultrasonic PD detection and voiceprint PD is more conducive to the detection of PD defects in the needle-tip discharge model.

      (2) After the frequency analysis of the voiceprint PD signal, the voiceprint PD signal frequency bands of the needle-tip discharge models were concentrated in the range 17-27 kHz, and the voiceprint PD signal of the conductive paste discharge models with different paste lengths was concentrated in the range 20-25 kHz.When voiceprint PD detection of a cable was performed on-site,the test of the voiceprint PD signal in this frequency band was strengthened, and the anti-interference design of the voiceprint PD test in this frequency band was enhanced.

      Acknowledgements

      This work was supported by the science and technology project of State Grid Shanghai Municipal Electric Power Company (No.52090020007F) and National Key R&D Program of China (2017YFB0902800).

      Declaration of Competing Interest

      We declare that we have no conflict of interest.

      Reference

      References

      1. [1]

        IEEE Std 1234-2019 (2019) IEEE guide for fault-locating techniques on shielded power cable systems - redline.IEEE, NY [百度学术]

      2. [2]

        Chen Z, Zhao P, Zhao J, et al.(2018) Review and prospect of DC cable transmission in the world.Journal of Global Energy Interconnection, 1(4): 487-495 [百度学术]

      3. [3]

        Zheng X, Lu L, Fu W (2016) On-line monitoring technique for partial discharge of high voltage explosion-proof motor.High Voltage Engineering, 42(5): 1651-1658 [百度学术]

      4. [4]

        Shan B, Li S, Yang X, et al.(2021) Key problems faced by defect diagnosis and location technologies for XLPE distribution cables.Transactions of China Electrotechnical Society, 36(22):4809-4819 [百度学术]

      5. [5]

        Zhang C, Cao M, He J, et al.(2021) Detection approach of abnormal discharges based on dual-spectral radiation characteristics.Insulating Materials, 54(9): 79-85 [百度学术]

      6. [6]

        Du J, Chen W, Zhang Z, et al.(2020) Propagation characteristics of utral high frequency signals in an actual power transformer.High Voltage Engineering, 46(6): 2185-2191 [百度学术]

      7. [7]

        Xie Q, Li Y, Lü F, et al.(2009) Method for PD location in oil combining ultrasonic phased array with wideband array signal processing.Proceedings of the CSEE, 29(28): 13-17 [百度学术]

      8. [8]

        Zhang M, Zhou H, Ma G, et al.(2020) Phase feedback control of optical fiber GIS partial discharge ultrasonic testing system.High Voltage Engineering, 46(12): 4138-4145 [百度学术]

      9. [9]

        Shao Y, Wang X, Peng P, et al.(2021) Research on defect detection method of power equipment based on acoustic imaging technology.China Measurement &Test, 47(7): 42-48 [百度学术]

      10. [10]

        Wang Y B, Chang D G, Fan Y H, et al.(2017) Acoustic localization of partial discharge sources in power transformers using a particle-swarm-optimization-route-searching algorithm.IEEE Transactions on Dielectrics and Electrical Insulation, 24(6):3647-3656 [百度学术]

      11. [11]

        Yadam Y R, Ramanujam S, Arunachalam K (2021) An ultrawideband conical monopole with radome for detection of partial discharges.IEEE Sensors Journal, 21(17): 18764-18772 [百度学术]

      12. [12]

        Zhou L, Cai J, Hu J, et al.(2021) A correction-iteration method for partial discharge localization in transformer based on acoustic measurement.IEEE Transactions on Power Delivery, 36(3):1571-1581 [百度学术]

      13. [13]

        Samaitis V, Mažeika L, Jankauskas A, et al.(2021) Detection and localization of partial discharge in connectors of air power lines by means of ultrasonic measurements and artificial intelligence models.Sensors, 21, 20 [百度学术]

      14. [14]

        Lu Y, Fan L, Tao F, et al.(2021) Application of artificial intelligence in partial discharge detection part I: denoising and fault location.Insulating Materials, 54(5): 10-20 [百度学术]

      15. [15]

        Song H, Dai J, Li Z, et al.(2019) An assessment method of partial discharge severity for GIS in service.Proceedings of the CSEE, 39(4): 1231-1240 [百度学术]

      16. [16]

        Chen J, Xu C, Li P, et al.(2021) Feature extraction method for partial discharge pattern in GIS based on time-frequency analysis and fractal theory.High Voltage Engineering, 47(1): 287-295 [百度学术]

      17. [17]

        Hou Z, Guo R, Li J (2020) Partial discharge characteristics of the surface discharge in SF6/N2 of the mixed gas under DC voltage.Transactions of China Electrotechnical Society, 35(14): 3087-3096 [百度学术]

      18. [18]

        Ma X, Zhang D, Zhu H, et al.(2020) Research on noise reduction of a partial discharge signal of a high voltage cable based on EWT.Power System Protection and Control, 48(23): 108-114 [百度学术]

      19. [19]

        Wu T, Sun K, Shi W (2020) Research on cable partial discharge denoising based on an AVMD-adaptive wavelet packet method.Power System Protection and Control, 48(14): 95-103 [百度学术]

      20. [20]

        Seri P, Ghosh R, Montanari G C (2021) An unsupervised approach to partial discharge monitoring in rotating machines:detection to diagnosis with reduced need of expert support.IEEE Transactions on Energy Conversion, 36(3): 2485-2492 [百度学术]

      21. [21]

        Zhu Y, Zhang Y, Cai W, et al.(2021) Data augmentation and pattern recognition for multi-sources partial discharge based on boundary equilibrium generative adversarial network with auxiliary classifier.Proceedings of the CSEE, 41(14): 5044-5053 [百度学术]

      22. [22]

        Fan L, Lu Y, Tao F, et al.(2021) Application of artificial intelligence in partial discharge detection part Ⅱ: pattern recognition and condition assessment.Insulating Materials, 54(7):10-24 [百度学术]

      23. [23]

        Du H, Guan H, Chen X, et al.(2021) Partial discharge characteristics of typical defects in XLPE cable under DC voltage.High Voltage Engineering, 47(2): 555-563 [百度学术]

      24. [24]

        Guo L, Cao W, Bai L, et al.(2021) Partial discharge characteristics and degradation mechanism of EPR cable insulation by thermo-oxygen aging.High Voltage Engineering,47(1): 231-240 [百度学术]

      25. [25]

        Liu B, Wang D, Ma A, et al.(2021) Study on the components of air under partial discharge condition in closed space.Power System Protection and Control, 49(9): 21-28 [百度学术]

      26. [26]

        Du B, Han C, Li J, et al.(2019) Research status of polyethylene insulation for high voltage direct current cables.Transactions of China Electrotechnical Society, 34(1): 179-191 [百度学术]

      27. [27]

        Zhou K, Wu K, Wan L, et al.(2016) Validity of inorganic nanoparticles on the suppression of partial discharge in air gap defects for medium voltage cable terminations.Transactions of China Electrotechnical Society, 31(22): 230-238 [百度学术]

      28. [28]

        Bai L, Zhou L, Xing L, et al.(2020) Effect of low temperature on interface discharge characteristics of ethylene-propylene rubber cable termination in high-cold environment.Transactions of China Electrotechnical Society, 35(3): 646-658 [百度学术]

      29. [29]

        Tang J, Pan C, Wang D, et al.(2017) Development of studies about surface charge accumulation on insulating material under HVDC.Transactions of China Electrotechnical Society, 32(8):10-21 [百度学术]

      30. [30]

        Liu Y, Liu H, Li Y, et al.(2018) Research of the electrical tree properties in XLPE under DC-AC composite voltages.Transactions of China Electrotechnical Society, 33(3): 601-608 [百度学术]

      31. [31]

        Hong J, He W, He H, et al.(2019) Research on electric field simulation of insulation defects of cable intermediate joints and detection of partial discharge in discharge chamber.Power System and Clean Energy, 35(8): 56-61 [百度学术]

      32. [32]

        Chang W, Yan C, Li C, et al.(2015) Assessment of creeping discharge initiated by metal particles on the silicone rubber/XLPE interface in cable joints.Transactions of China Electrotechnical Society, 30(24): 245-254 [百度学术]

      Fund Information

      Author

      • Hang Ji

        Hang Ji received B.S.degree at North China Electric Power University in 2006.He is working in State Grid Shanghai Municipal Electric Power Company.His research interests include power system automation, equipment management.

      • Xing Lei

        Xing Lei received M.S.degree at North China Electric Power University in 2006, and received Ph.D.degree at Shandong University in 2012.He is working in State Grid Shanghai Municipal Electric Power Company.His research interests include power system automation, equipment maintenance.

      • Qiang Xu

        Qiang Xu received bachelor’s degree at North China Electric Power University in 2002.He is working in State Grid Shanghai Municipal Electric Power Company.His research interests include power system automation, equipment maintenance.

      • Chengjun Huang

        Chengjun Huang received Ph.D.degree at Shanghai Jiaotong University in 2000.He is the chairman of Power Monitoring and Diagnostic Technology Ltd.San Jose, USA.His research interests include partial discharge detection technology, intelligent power equipment, and condition maintenance technology.

      • Ting Ye

        Ting Ye received B.S.degree at Fudan University in 2012.He is working in State Grid Shanghai Municipal Electric Power Company.His research interests include power system automation, equipment maintenance.

      • Shangqing Yuan

        Shangqing Yuan is a graduate student at Shanghai University of Electric Power.His research interests include power system automation, control engineering.

      Publish Info

      Received:2021-11-07

      Accepted:2022-01-05

      Pubulished:2022-02-25

      Reference: Hang Ji,Xing Lei,Qiang Xu,et al.(2022) Research on characteristics of acoustic signal of typical partial discharge models.Global Energy Interconnection,5(1):118-130.

      (Editor Dawei Wang)
      Share to WeChat friends or circle of friends

      Use the WeChat “Scan” function to share this article with
      your WeChat friends or circle of friends