Research on quantitative method of GIS partial discharge

Optical method is a commonly used method for detecting partial discharge in electrical equipment. How to use photometric quantitative detection is one of the most urgent problems to be solved. Based on the in-depth analysis of the physical meaning of the integral value of the PD optical signal, it is concluded that the linear integral value of the signal is linear with the apparent discharge. The experimental results show that the integral value of the PD photometric signal at different polarities and different distances is linear with the apparent discharge. The larger the gap distance between the needle and the plate electrode is, the larger the slope is. The linear relationship between the average primary integrated value of the PD photometric signal and the average apparent discharge is different. With the increase of the applied voltage, the PD discharge intensity and the pulse repetition rate increase, and the phase interval of the PD becomes wider. The actual air gap volume of the PD becomes larger, and the variation interval becomes wider, that is, the apparent discharge amount distribution interval becomes wider, and the maximum value and the minimum value of the distribution interval become larger. Therefore, it is possible to obtain the PD apparent discharge amount by using the photometric signal, and it is effective and feasible to quantitatively detect the internal PD of the GIS by optical method.


Introduction
Since partial discharge is one of the important parameters for characterizing the insulation state of electrical equipment, accurate detection of electrical equipment PD is the key to realize on-line monitoring and state assessment of insulation status of electrical equipment [1]. At present, methods for detecting PD of electrical equipment mainly include pulse current method, ultra-high frequency method, chemical detection method, and optical method. The PD pulse current method is currently the only quantitative detection method recommended by the academic community. However, due to its narrow detection frequency range and weak anti-interference ability, it has low detection accuracy in large-scale field and is usually only used for quantitative detection in low-interference environments [2]. On the contrary, UHF detection has high sensitivity and strong immunity, but its quantitative problem still plagues many scholars. The chemical detection method can reflect the overall degree of PD, but the detection period is long and does not meet the online monitoring requirements. At present, most PD detection methods acquire PD signals from outside the electrical equipment. Due to the complexity of some electrical equipment, it is difficult to accurately determine the location of the PD.
In recent years, the optical measurement method [3] is used to detect PD as the mainstream. The PD is measured by detecting the optical signal generated by the PD light effect. It is not affected by electromagnetic interference, and has high sensitivity. The biggest advantage is that it can accurately EMCEME 2019 IOP Conf. Series: Materials Science and Engineering 782 (2020) 032018 IOP Publishing doi: 10.1088/1757-899X/782/3/032018 2 detect and judge the PD position inside the complex structure. The optical method has been successfully used for PD detection in GIS, but this method is only for qualitative detection of PD, and cannot be used for quantitative judgment of PD discharge level, which is the main bottleneck encountered by this method [4][5]. In literature [6][7][8], the relationship between the amplitude of the photometric method and the apparent discharge under corona discharge and creeping discharge is studied respectively. It is concluded that the two have a linear relationship, but they are not analysed and explained from the theory. Therefore, based on the above basics, based on the theory, the relevant experimental platform is established, and the comparison between PD photometry and pulse current method under the needleplate electrode model is carried out. The relationship between the integral value and the apparent discharge is studied. The reader is provided with a reference method for quantitative detection of PD using photometry.

Theoretical Analysis of the Relationship between Signal and Discharge Quantity in optical method
The photometric method uses a photoelectric converter to convert a photo sensor to a weak optical signal generated by a PD and convert it into an electrical signal output. In this paper, a photomultiplier tube is used as a photoelectric converter, and its working principle is shown in Figure 1. When the working voltage of the photomultiplier tube is constant, the anode output current can be expressed by the following formula: Where i(t) is the anode output current, G is the current gain, S d is the photocathode sensitivity, and P s (t) is the optical signal power.
Since the optical signal generated by the PD is a pulse signal, it is known from the equation (1) that the anode output current is also a pulse signal, wherein the influence of the photomultiplier tube anodeto-ground capacitance on the pulse signal is not negligible. The output circuit and equivalent circuit diagram are shown in Figure 1. Available from equivalent circuits: Where u (t) is the output voltage of the photomultiplier tube, 1 The simultaneous (2) and (3) versions are available: The simultaneous (1) and (4) are available: According to the above derivation, the light energy E absorbed by the PD photo sensor can be obtained as follows: The signal of a single PD optical method is indicated by A, and the simultaneous (6) can be obtained: It can be seen from the attenuation characteristics of the optical signal in the medium that the relationship between the light energy E light released by the primary PD and the received light energy E can be expressed by the following formula [9]: Where μ is the absorption coefficient of the photo sensor, k ξ is the gas absorption coefficient, L is the distance between the sensor and the PD source, and ρ is the gas density. The simultaneous (3.47) and (3.48) versions are available:  There are two main reasons why PD produces a light effect. One is charged particle composite luminescence, PD generates a large number of charged particles. When charged particles with opposite charge signs meet, it is possible to recombine and release energy in the form of light radiation. Second, charged particles cause luminescence, PD the generated free charged particles are accelerated by a strong electric field. When the freely charged particles collide with the neutral particles, the kinetic energy of the freely charged particles is transferred to the neutral particles for excitation, and when the excited neutral particles return to the ground state, they are absorbed. The energy is released in the form of radiant light. The reason why the light effect is generated by the PD is that the light energy released by the PD is essentially converted from the PD discharge energy [10][11][12]. Therefore, the relationship between the light energy Elight released by the PD and the PD discharge energy Ef can be referred to as the luminous efficiency η. The luminous efficiency η is a physical quantity describing the efficiency of conversion of discharge energy into light energy in the principle of light source, which is defined as: The luminous efficiency η is related to the characteristics of the luminescent medium, the electric field strength E, the electrode spacing l, and the environmental factors. For the pulse type discharge, it is also related to the duration t c of the discharge current pulse, and the discharge energy has little influence on the luminous efficiency η, which can be approximated in The luminous efficiency η at different discharge energies is a constant under certain other influencing factors.
The PD discharge energy E f can be calculated by the equation (11): Where u i is the peak of the initial discharge voltage of PD and q is the amount of PD apparent discharge. The relationship between the light energy E light released by the PD and the PD apparent discharge amount q can be expressed by the following formula: The simultaneous (9) and (12) are available: It can be seen that the primary integral value of the PD photometric signal is linear with the apparent discharge amount, wherein the slope k and the working parameters of the photomultiplier tube, the relative position of the photo sensor and the PD source, the distance between the electrodes, and the initial discharge. Voltage peaks and environmental factors are related.

Experimental system and experimental method
In order to study the relationship between PD apparent discharge and photometric signal, PD signal was measured by pulse current method and optical method. The experimental platform consists of a doubleshielded sealed box, a PD source artificial model, a PD pulse current and photoelectric signal measurement system, and a non-smooth power test power supply. The experimental wiring is shown in Figure 2.

Figure 2. Correction circuit
The PD pulse current signal measurement uses a series impedance of 50Ω to input the pulse signal into the digital storage oscilloscope through the high frequency cable. The PD photometric method uses a fluorescent fiber optic sensor to receive the optical effect signal generated by the PD, and then couples and transmits the optical signal induced by the fluorescent fiber to the photodetector by using the transmission type ordinary optical fiber, and the photodetector converts the optical signal into a voltage signal. Finally, the pulse current signal is input to the digital storage oscilloscope via the high frequency cable. The photodetector uses a H9656-02 photomultiplier tube (PMT) with a spectral response range of 300-880 nm and a peak sensitivity wavelength of 500 nm. The acquisition and storage signals are recorded by a high-speed digital oscilloscope (Tektronix DPO7104 oscilloscope, analog bandwidth 1GHz, maximum sampling rate 20GS/s, storage depth 40M), which can meet the requirements of PD optical signal and pulse current signal acquisition.
According to the IEC60270 standard pulse current method, the pulse voltage amplitude U and the apparent discharge amount q are proportional to each other, and the measurement circuit is corrected to obtain the pulse voltage amplitude and the apparent discharge amount proportional coefficient K, thereby measuring the pulse according to the pulse current method. The voltage amplitude U calculates the apparent discharge amount q. The discharge quantity correction circuit is shown in Figure 2. A partial discharge calibrator is connected in parallel with the sample to generate a pulse signal with a known discharge amount on the sample. The peak value of the pulse voltage across the 50Ω sense resistor can be measured by a digital oscilloscope. Adjusting the output of the partial discharge calibrator can obtain the relationship between U and q, and obtain the calibration curve shown in Figure 3 by fitting.  (2) The high-speed oscilloscope is used to collect the PD signal at the same time measured by the photometric method and the pulse current method.
According to the above steps, the PD test under four insulation defects [13] was performed separately.

Experimental results and analysis
The PD apparent discharge amount q can be obtained by obtaining a calibration curve in Figure 3, as follows: Where U is the amplitude of the primary PD voltage waveform detected on the 50Ω non-inductive resistor. The integral value of the PD photometric signal can be calculated by the following formula: Where A is the primary integrated value of the photometric signal, U i is the i-th value of the primary PD waveform measured by optical method, and Δt is the sampling time interval.
At the same time, the pulse current method and the optical measurement method are used to collect the single PD signal waveform generated by the four types of insulation defect models. However, the signal waveforms obtained by the pulse current method for detecting the PDs generated by the four types of defect models are almost the same, and the optical detection method detects four types. The signal waveform obtained by the PD generated by the insulation defect model is also almost the same. Therefore, this paper only gives the single-shot PD signal waveform generated by the pulse current method and the photometric method to detect the defects of metal contaminants on the insulator surface.    Figure 4 shows the waveform of a single PD signal obtained by pulse current measurement through 50Ω detection impedance. Figure 5 shows the waveform of a single PD signal detected by optical method. The single-shot PD signal detected by the pulse current method is divided into two types: a positive polarity pulse and a negative polarity pulse, and the pulse polarity is related to the polarity of the high voltage end. The high voltage end is the positive pole. When the PD of the defect model occurs, the pulse current generated flows from the high voltage end to the low voltage end. At this time, the single PD signal waveform detected by the pulse current method is a positive polarity pulse. The high voltage end is the negative pole. When the defect model generates PD, the pulse current generated flows from the low voltage end to the high voltage end. At this time, the single PD signal waveform detected by the pulse current method is a negative polarity pulse. The photometric signal only has a positive pulse waveform, because the photometric signal reflects the change in optical energy released by the PD.  Because the distribution characteristics of the single-shot PD photometric signal and the apparent discharge amount increase with the increase of the discharge intensity under different defect models, this paper only gives the integral value of the PD photometric signal under the metal protrusion defect. The distribution interval with apparent discharge is shown in Figures 6 and 7. It can be seen from the figure that under the same experimental voltage, the integral value and the apparent discharge amount of a single PD signal at different times are not the same, but are distributed within a certain interval. As the applied voltage increases, the PD pulse repetition rate increases, and the signal increases. Both the primary integral value and the apparent discharge amount distribution interval are widened, and the maximum and minimum values of the distribution interval become large. Reason: PD only occurs in the local area near the discharge defect pole. Under AC voltage, the instantaneous voltage is greater than the initial discharge voltage value. The PD can be generated in the phase interval. According to the PD characteristics, the PD apparent discharge amount and the PD actually occur. Regarding the state of the region, the larger the air gap volume in which the PD actually occurs, the larger the apparent discharge amount.
Since the PD characteristics are related to many factors, the actual air gap volume of PD occurring in different phases is not the same, and has a certain randomness, but its volume always varies within a certain range. In general, when the applied voltage is increased and the discharge intensity and the pulse repetition rate are increased, the phase interval of the PD is widened, and the actual air gap volume of the PD is increased, and the apparent discharge amount distribution interval is widened. From equation , it can be seen that the primary integrated value of the PD photometric signal is positively correlated with the apparent discharge amount, so that the primary integrated value and the apparent discharge amount exhibit similar distribution characteristics.

The establishment of the relationship between the integral value of the signal and the discharge amount
According to Figure 6 and Figure 7, the single-shot PD and the apparent discharge of the single-shot PD photometric signal at the same discharge intensity are distributed in a certain interval, and the singleshot PD photometric signal at different discharge intensities is once. There is an overlap region between the integral value and the apparent discharge amount distribution interval, so the primary integrated value and the apparent discharge amount of the single PD photometric signal cannot be used to characterize the PD discharge intensity. Therefore, this paper defines the average integral value av Since the average primary integral value av A and the average apparent discharge amount av q of the PD photometric signal can uniquely characterize the PD discharge intensity, a one-to-one correspondence between av A and av q can be found by establishing a relationship curve between av A and av q . Realize the quantitative detection of PD in GIS by optical method.
Since this paper controls the PD discharge intensity by controlling the test voltage, 200 single-shot PD signals at the same voltage obtained by the pulse current method and the photometric method under the same conditions are used formula (15),(16),(17)and (18)   A and av q is that as the applied voltage increases, the insulation defect causes the PD to be more intense, but the properties of the PD do not change, causing the intensity of the optical effect to increase proportionally with the increase of the PD discharge energy. Comparing Figure 8(a) the relationship between the average integral value of the PD photometric signal and the average apparent discharge under different pin gap distances, it can be seen that the larger  11 the gap distance, the average signal and the average apparent discharge. The slope k of the linear relationship is larger. The reason is that the larger the gap distance is, the free travel of the freely charged particles increases, and the probability of ionization caused by the collision of the unit number of freely charged particles with the neutral particles increases, so that the luminous efficiency η of the charged particles due to the composite increases; On the other hand, the larger the gap distance, the larger the initial discharge voltage ui of the PD, and since k is proportional to the luminous efficiency η and the initial discharge voltage ui, the slope k increases as the gap distance increases.
Comparing Figures 8 (a), (b), (c) and (d), it can be seen that the linear relationship between the average primary integrated value of the PD photometric signals and the average apparent discharge is different for different insulation defect types. The reason is: the discharge properties of different insulation defect types PD are different, so that the luminous efficiency η caused by PD is different, since k is positively correlated with the luminous efficiency η, so that the average integral value and average of the PD photometric signals of different insulation defect types are averaged. The linear relationship curve corresponding to the discharge amount is different.

Validity Verification of Quantitative Detection of Partial Discharge in GIS by Photometric Method
In order to verify whether the relationship between the average primary integral value av A and the average apparent discharge amount av q of the PD photometric signal obtained in Figure 8 can be used for quantitative detection of GIS internal PD by optical method. In this paper, PD experiments of four typical defects in GIS are carried out under three different discharge intensities, and simultaneous detection by optical method and pulse current method. The experimental results are shown in Table 1

Conclusion
From the above experimental results, it is known that, based on the accurate identification of the insulation defect, the average apparent discharge amount calculated by the average integral value of the photometric method and the average apparent discharge amount corrected by the pulse current method are within 14%. At the same time, according to the research results of this paper, it is shown that the PD optical measurement signal can accurately identify the internal insulation defects of GIS. Therefore, it is effective and feasible to quantitatively detect the internal PD of GIS by optical measurement.