Partial Discharge Detection Method of Electrical Equipment Based on UHF Method

Partial discharge is an early symptom of insulation defects. In the field application, the detection method can not accurately identify those noise signals mixed in the partial discharge signal, which limits the detection to a certain extent. Based on UHF method, a partial discharge detection method for electrical equipment is designed. When partial discharge occurs, the exchange between charges, the radiation of electromagnetic waves and the consumption of energy will occur at the discharge position, causing the change of electrode potential. For the phase distribution spectrum, the signal characteristics produced by partial discharge are extracted from three aspects: statistical information, time domain information and frequency domain information. During detection, UHF sensor is placed on the basin insulator, and then UHF signal is received. Insulation defects are identified according to various differential discharge patterns formed by the discharge characteristics of different defects. Test results show that the signal spectrum detected by UHF partial discharge roughly matches the typical discharge spectrum, and it is determined that there is partial discharge in this area. This method has high positioning accuracy and meets the detection requirements of electrical equipment.


Introduction
Electrical equipment mainly includes transformers, high-voltage circuit breakers and other electrical appliances, power supply lines and power lines used for connection and other components.They are sealed and installed in a metal shell filled with insulating gas.In order to ensure the stable operation of the power system, it is necessary to inspect the electrical equipment frequently.Whether there is abnormal noise in the equipment, whether the instrument parameters are abnormal, and whether the equipment will be damaged if the temperature is too low or too high, all these require the participation of inspectors' audio-visual and olfactory senses.Correctness largely depends on the experience, professional level and working attitude of inspectors.Insulation failure is usually accompanied by partial discharge, and the long-term accumulation of partial discharge will cause varying degrees of damage to the performance of electrical equipment [1] .Partial discharge is an early symptom of insulation defects.Therefore, the insulation defects can be found as soon as possible by carrying out partial discharge detection, and the maintenance personnel can take corresponding measures to prevent the partial discharge from further deterioration.Partial discharge is an important technical index used relatively more to determine the insulation condition of electrical equipment at this stage.It has been unanimously recognized by more and more equipment manufacturers, operators and enterprises.Various on-site testing equipment and data acquisition methods may not have unified standards, resulting in the error of testing results and the delay of troubleshooting.Some tiny air gaps and debris residues in the manufacturing process provide congenital conditions for the formation of partial discharge electric field.In the environment of high voltage and strong magnetic field, weak partial discharge is very easy to occur.With the continuous development of industrial technology and economic level, people have higher requirements for the security and reliability of power supply system.In addition, the state pays more attention to the safety performance of substations, so that the insulation detection of electrical equipment has received more attention.The electromagnetic wave generated by partial discharge will continue to transmit along the cavity.When it is transmitted to the discontinuity of wave impedance, it will leak out, and the gas in the equipment cavity will be decomposed with partial discharge, resulting in uneven pressure in the cavity, resulting in ultrasonic waves [2] .When the electrical equipment fails due to internal defects, it takes longer to complete the maintenance work, which brings greater harm than other equipment.Long term partial discharge will cause serious damage to the insulation performance of the switchgear.Partial discharge will cause the exchange impact of charged particles, causing the temperature of the discharge part to rise and overheat, and then carbonizing the insulator [3] .The application of conventional pulse current method to test the partial discharge phenomenon of equipment, in the field application, often because they can not accurately identify those noise signals mixed in the partial discharge signal, the detection is limited to a certain extent.When partial discharge occurs, the pulse signal generated by it cannot exist for a long time, and the pulse signal with a frequency of several megahertz gathers and climbs in a short time.In addition to internal UHF and ultrasonic, light effect is also produced during partial discharge, and light radiation effect is brought about when charged particles are combined.Similarly, SF6 gas will also decompose during partial discharge.This paper presents a partial discharge detection method of electrical equipment based on UHF method, which improves the detection sensitivity and provides a reference for equipment detection.

Analysis of partial discharge mechanism of electrical equipment
There are some local insulation weaknesses in each insulation structure for various reasons.Under the action of electric field, these weaknesses will start discharging earlier than other insulation structures.However, partial discharge will not immediately form insulation penetration breakdown at these weak points.This insulation between conductors only produces discharge phenomenon at local places, which is called partial discharge phenomenon for short.Partial discharge is a kind of pulse discharge, which produces corresponding physical and chemical changes in the interior and edge of power equipment, such as light, sound, electricity and mechanical vibration [4].On the one hand, it involves the movement of charges, the consumption of energy and other situations, on the other hand, it will emit electromagnetic waves, ultrasonic waves and so on, which is an extremely complex process.At the beginning of partial discharge, the amount of partial discharge is very small, and the discharge range is also very limited, which will hardly cause damage to the power cable.When partial discharge occurs, the exchange between charges, electromagnetic wave radiation and energy consumption will occur at the discharge position.Among them, the obvious point is that there is a small pulse voltage signal, which appears on the terminal of the voltage applied by the test object.The partial discharge phenomenon is not detected, and if it is allowed to continue, the damage to the insulation layer will continue to deepen, and the discharge range will gradually expand [5].Physical and chemical changes related to partial discharge can monitor the internal insulation of power equipment and provide detection signals.The insulation strength and breakdown field strength of power equipment are quite high, and the breakdown process is very fast when partial discharge occurs in a very small range.The electric field near the metal protrusions inside the equipment will change greatly.Generally, there are two situations: one is that the field strength area around the metal protrusion will form a high field strength due to gas, and partial discharge will occur when the field strength reaches a certain degree.However, the breakdown field strength in the whole cavity is still higher than other positions, so it will not penetrate the whole cavity, and only partial discharge will occur.The types of partial discharge produced by different defects are also different, and the harm degree is not nearly the same.Common insulation defects of electrical equipment are usually divided into internal discharge, surface discharge, corona discharge and suspension discharge [6].The simplest model of discharge is that there is an air gap in the insulating medium.Mechanical vibration and poor contact affect each other, and eventually lead to the change of electrode potential.For many electrodes with potential change, the relevant equivalent capacitance will have partial discharge during charging and discharging, and strong electromagnetic wave signals and ultrasonic signals will be generated at the same time [7].Air gap discharge usually occurs in the air gap between the medium and the electrode.There are many reasons for the generation of air gaps, which may be the existence of air gaps in the production of insulating components, or the cracks in insulating materials caused by physical reasons such as thermal expansion and contraction.At this time, when voltage is applied to the circuit, the voltage of the internal air gap without spark discharge is expressed as: In formula (1), and represent the voltage and capacitance of the air gap inside the insulating medium; indicates the applied voltage of the circuit; is the capacitance of the insulating medium in series with .
There may be some bubbles, cracks or foreign matters on the surface of insulating materials, resulting in that the insulating materials are not very pure.When the applied voltage is relatively high, the air gap will be punctured first, but the surrounding medium can maintain its insulation performance.
Because the field strength of the breakdown gas medium is much smaller than that of the solid, and the electric field strength inside the gas is higher than that of the solid [8] .Therefore, partial discharge often occurs in the air gap of insulating materials when a penetrating channel has not been formed in the medium.With partial discharge, some decomposition products will be generated, which are corrosive to a certain extent, which will pollute the surrounding insulator surface.

Experimental Method
In the partial discharge detection of electrical equipment, how to accurately and effectively extract the characteristic parameters contained in the signal is a key step.The selection of characteristic parameters directly determines the final effect of partial discharge fault detection of electrical equipment.In this paper, the signal generated by partial discharge is extracted from statistical information, time domain information and frequency domain information.The initial amplitude time waveform received by the upper computer, which is composed of dense columnar data.These columnar data correspond to sampling points at different times.It can be found that it is difficult to intuitively obtain effective feature information from the graph, so it is necessary to statistically process the graph.Intercept the waveform data generated when partial discharge fault occurs, and obtain the information characteristic parameters of the signal, reflecting the overall energy and fluctuation degree of the signal [9] .The abscissa of the amplitude time diagram is divided into multiple sinusoidal cycles, and multiple cycles are superimposed in one cycle.According to the amplitude, count the discharge times, average discharge and maximum discharge at different phases of a cycle.The specific process of feature extraction for this kind of phase distribution map is shown in Figure 1.Here, it is defined as the ratio of the number of positive (or negative) half cycle pulses of the characteristic spectrum to the number of phase intervals [10] .In order to obtain the stable performance of the overall signal energy, the root mean square and root mean square amplitude are extracted.Based on this, two parameters of skewness and steepness are introduced to reflect the symmetry and prominence of the signal.The skewness represents the symmetry of the probability distribution of partial discharge in the power frequency cycle, and the calculation formula is as follows: In formula (2),  represents the inclination; a represents discrete value; i and m represent the serial number and total number of data; ,  , b and  represent the probability, mean and standard deviation of the distribution.Steepness indicates the flatness of probability distribution of partial discharge at power frequency cycle.The calculation formula is as follows: In formula (3), 2  represents steepness.Time domain signal is the most direct expression of fault signal.In order to realize accurate analysis and rapid processing of data, on the basis of retaining the original partial discharge signal, the relevant waveforms before the fault is triggered and after the fault is stabilized are eliminated, which are not included in the scope of time-domain feature analysis, and only the signal data collected during the fault is retained.The average discharge diagram is a reflection of the average discharge of partial discharge signals over a period of time.A large number of discharge amplitudes are concentrated in one cycle, and the average values of all amplitudes at different phases in a cycle are counted [11] .The discharge times map is to count the discharge times at different phases in a superimposed cycle.The extracted primary envelope is then extracted, and the resulting envelope is the secondary envelope.According to the observed figures and analysis and calculation, the secondary envelope and its shape are more conducive to extract more feature quantities in the time domain information.It can eliminate the redundant information generated by the interference of other factors in the field work to a certain extent, making the feature extraction of timedomain waveform more intuitive and convenient.Discharge asymmetry is the ratio of the average level of positive (or negative) half cycle discharge parameters of the characteristic map, and the expression is: In formula (4),  represents the discharge asymmetry; 1 W and 2 W are the sum of positive and negative half cycle discharge parameters respectively; 1 M and 2 M are the number of non-zero values of positive and negative half cycles of voltage respectively.Using fast Fourier transform, the time domain is converted to frequency domain, and the number of spectral peaks and corresponding frequency values are read.The average value, spectral variance and other relevant data in the frequency domain signal can be calculated directly.

Partial discharge detection based on UHF method
UHF detection is a detection technology to obtain partial discharge information by detecting highfrequency electromagnetic wave signals generated in the process of partial discharge through sensors.
Sensors are divided into external and built-in types due to their placement mode.In the field actual measurement environment, the operating equipment will produce corona interference.However, due to the high frequency band of UHF method, corona interference is effectively reduced, so it has strong anti-interference ability.UHF sensor is mainly responsible for receiving the electromagnetic wave signal, and then converting the electromagnetic wave signal into voltage signal.After corresponding amplification processing, it is transmitted to the detector host to complete A/D conversion and data processing.Finally, use the network cable to transmit the preprocessed data to the notebook computer.During detection, UHF sensor is placed on the pot insulator, and then receives UHF signal.Because UHF electromagnetic wave generated by partial discharge has little attenuation when propagating in electrical equipment, this method has high sensitivity.The propagation speed of the electromagnetic wave generated by partial discharge in the cavity is close to the speed of light.The position of the discharge source can be calculated according to the placement position of the sensor and the time difference of the electromagnetic wave reaching different sensors [12] .Usually, the time difference of the signal obtained by the sensor is analyzed through the oscilloscope.If the resolution is close to 1ns, the positioning accuracy can be as high as 30cm, so the positioning is more reliable and effective.UHF signals generated by partial discharge of different defect types often have different characteristics in time domain and frequency domain.Therefore, the abnormal signal of partial discharge can be judged and the defect type can be diagnosed by combining the distribution characteristics of UHF signal in time domain and frequency domain.And insulation defects can be identified according to various differential discharge patterns formed by the discharge characteristics of different defects.
If there is obvious difference between the detection map and the background map, and there is a typical partial discharge detection map, the judgment result is defect [13] .The basic principle of this technology is shown in Figure 2. Detection principle of UHF method Using the principle that the time domain signal obtained from the nearest point to the discharge source is ahead of the point far away from the discharge source, the partial discharge position is detected and identified.Two or more UHF Sensors and a multi-channel digital oscilloscope are used for testing.The time difference is the time interval between different sensors successively acquiring the discharge signal of the same waveform [14].The sensor is bound on the adjacent position, and the distance is about 0.5m.According to the time difference of the detected discharge signal, the location of the partial discharge source is obtained by using formula (5): In formula (5), d represents the distance between the discharge power supply and the sensor placed on the left; S is the sensor spacing; v is the propagation speed of electromagnetic wave; t  is the time difference between the time domain signal wavefront captured by the sensor.The analysis and diagnosis software on the computer processes the pre data into peak value retention map, PRPS pulse sequence phase distribution map and PRPD phase distribution map, which are displayed and stored in real time.Then the typical characteristic atlas library is used to analyze and diagnose the collected partial discharge signals, and the diagnosis results of the types of partial discharge defects are given.
Then the high-speed oscilloscope is used to accurately locate the discharge source.The partial discharge characteristics of different insulation defects in electrical equipment will be different.Different defects have different degrees of damage to the equipment.Through the amplitude, phase and other characteristics of partial discharge signal, the type, nature and characteristics of insulation defects can be effectively analyzed [15] .In the detection process, the frequency of the external power supply must be guaranteed to be 50Hz to ensure the accuracy of the phase mode.There is no need to add an amplifier immediately before starting the test.If a weak abnormal signal is found, connect the amplifier to amplify the signal for convenient judgment.When setting the detection parameters, it specifically includes the naming of the substation and the specific location of each channel sensor.In addition, in order to make the signal better displayed on the spectrum of each channel, it is necessary to complete the threshold setting of detection signals of different channels in combination with the real noise situation on site.

Partial discharge detection process
UHF partial discharge detector is used to detect the electrical equipment of a substation.Record the time when sensor 1 and sensor 2 detect the partial discharge signal respectively.Use the time difference method to locate and analyze the signal.The partial discharge signal is processed to suppress the signal interference outside the frequency band of the discharge signal of the detection object.Because the response is a finite length sequence, it has prominent linear phase characteristics.Filter the collected signal, and the FIR filter description equation is as follows: In formula (6),   F u represents the output sequence; u indicates acquisition signal;  represents the sampling sequence; j is signal phase; z is the filter coefficient length;   j  is the filter coefficient.
Observe the amplitude frequency response until the phase and amplitude frequency characteristics consistent with the specific requirements are finally obtained.UHF detection sensor probe shall be installed on the basin insulator of the air chamber where the busbar, transformer, various switches and other equipment are located in the interval.Using UHF detection equipment to obtain the PRPD map, it is found that the map shown in Figure 3 is detected in the area where the transformer and circuit breaker are located.The signal spectrum detected by UHF partial discharge roughly matches with the typical discharge spectrum to determine that there is partial discharge in this area, and further locate the location of the discharge source.

Detection and location results and analysis
Based on the detection of partial discharge of electrical equipment in this area, the coordinate system is established with the positioning base station as the coordinate origin, and the coordinates of the current detection point are calculated by rotating the antenna array for one week.Calculate the positioning error between the coordinate value detected by UHF method and the real coordinate of the discharge source.In order to test the effectiveness of the method proposed in this paper, the methods based on pulse current and ultrasonic are selected as the comparison methods.The positioning error obtained by each detection method is shown in Table 1.It can be seen from Table 1 that the positioning error of partial discharge detection method of electrical equipment based on UHF method is 0.183m, which is 0.158m and 0.224m lower than that based on pulse current method and ultrasonic method.Therefore, this design method successfully locates a fault discharge location, and the positioning accuracy is high, which meets the requirements of partial discharge detection of electrical equipment.Through UHF partial discharge detection, the advantages of condition based maintenance are effectively brought into play, defects are found early and eliminated in time, which provides a guarantee for the safe and stable operation of the power grid, and has strong practicality.

Conclusions
Partial discharge is an early symptom of insulation defects.Therefore, the insulation defects can be found as soon as possible by carrying out partial discharge detection, and the maintenance personnel can take corresponding measures to prevent the partial discharge from further deterioration.The electromagnetic wave generated by partial discharge will continue to transmit along the cavity, and will leak out when it is transmitted to the discontinuity of wave impedance.The gas in the equipment cavity is decomposed with partial discharge, resulting in uneven pressure in the cavity, resulting in ultrasonic waves.In the field application, the detection method can not accurately identify those noise signals mixed in the partial discharge signal, which limits the detection to a certain extent.Based on UHF method, a partial discharge detection method for electrical equipment is proposed in this paper.UHF sensor is mainly responsible for receiving the electromagnetic wave signal, and then converting the electromagnetic wave signal into voltage signal.After corresponding amplification processing, it is transmitted to the detector host to complete a/d conversion and data processing.Finally, use the network cable to transmit the preprocessed data to the notebook computer.Insulation defects are identified according to various differential discharge patterns formed by the discharge characteristics of different defects.This method successfully locates a fault discharge location, and the location accuracy is high, which meets the requirements of partial discharge detection of electrical equipment.The subsequent research analyzes the UHF signal data of the composite defects of multiple discharge sources, obtains its characteristics and laws, and summarizes the applicability of various discharge source location methods.

Figure 1 .
Figure 1.Feature extraction process Extract common statistical parameters such as mean, absolute mean and variance from statistical features.The maximum and minimum values and local peaks are extracted from the perspective of dynamic change of signal energy.The number of partial peaks is used to characterize the distribution of partial peaks during partial discharge.Here, it is defined as the ratio of the number of positive (or negative) half cycle pulses of the characteristic spectrum to the number of phase intervals[10] .In order to obtain the stable performance of the overall signal energy, the root mean square and root mean square amplitude are extracted.Based on this, two parameters of skewness and steepness are introduced to reflect the symmetry and prominence of the signal.The skewness represents the symmetry of the probability distribution of partial discharge in the power frequency cycle, and the calculation formula is as follows:

Figure 2 .
Figure 2. Detection principle of UHF methodUsing the principle that the time domain signal obtained from the nearest point to the discharge source is ahead of the point far away from the discharge source, the partial discharge position is detected and identified.Two or more UHF Sensors and a multi-channel digital oscilloscope are used for testing.The time difference is the time interval between different sensors successively acquiring the discharge signal of the same waveform[14].The sensor is bound on the adjacent position, and the distance is

Table 1 .
Partial discharge detection positioning error (m)