A Partial Discharge Detection Method Based on Parameter Optimization of Current Transformers

Abstract

The present invention discloses a partial discharge detection method based on the parameter optimization of a current transformer. The partial discharge detection method is as follows: First, construct a transfer function for the current transformer to detect partial discharge. Second, set the combination of parameters to be optimized for the current transformer. Third, perform frequency domain analysis on the transfer function corresponding to each group of parameters to be optimized respectively. Fourth, select the best combination of parameters to be optimized according to the open-loop gain and phase shift magnitude obtained from the frequency domain analysis. Fifth, place the current transformer on the power supply input cable of the device under test; determine whether partial discharge occurs in the device under test according to the output signal of the current transformer. The present invention constructs a transfer function for the current transformer to identify partial discharge signals, and optimizes the load resistance, the number of turns of the secondary coil, and the relative permeability of the current transformer during the detection process according to the transfer function, thereby improving the signal-to-noise ratio and gain of the current transformer when detecting partial discharge signals.

Description

Technical Field

[0001] The present invention belongs to the technical field of partial discharge detection, and particularly relates to a partial discharge detection method based on the optimization of current transformer parameters. Background Art

[0002] In an electrical system, the degradation of dielectric materials is usually related to partial discharges released in voids and cracks (bubbles) at the conductor-dielectric interface in solid insulation systems. If it is a liquid dielectric or corona, it is related to gases. Under the operating stress conditions of an electrical insulation system, bubbles, cracks, and voids may exceed their dielectric strength, resulting in discharges in the dielectric, reducing stiffness, and ultimately leading to the complete or partial failure of the insulation. Therefore, quality and compliance tests are carried out through the analysis of partial discharges. In addition, considering the cyber-physical system framework, when any component is about to collapse due to partial discharges, the partial discharge analysis can be extrapolated as a component of the power system to issue an alarm in advance. Partial discharge tests are classified according to measurement techniques, and electrical methods are widely used. Electrical methods can be used for high-voltage electrical equipment such as power transformers, instrument transformers, medium- and high-voltage and extra-high-voltage cables, high-voltage bushings, and rotating machinery. In addition, electrical methods are divided into conventional methods and unconventional methods. Conventional partial discharge methods are carried out in accordance with IEC60270 - High-voltage test techniques: partial discharge - charge measurement techniques, using coupling capacitors with a central measurement frequency of up to 1 MHz. On the other hand, unconventional partial discharge methods (or electromagnetic methods), based on IEC62478 - High-voltage test techniques - Measuring partial discharges by electromagnetic and acoustic methods, the working frequency range of the sensors used in this method includes: high-frequency - HF (3 - 30 MHz), very high-frequency - VHF (30 - 300 MHz), and ultra-high-frequency - UHF (300 MHz - 3 GHz). Various forms of noise may appear during the PD test, including white noise, regular pulses, oscillations, etc. White noise is similar to the partial discharge signal and is the most difficult noise to eliminate.

[0003] The time measurement efficiency of partial discharges depends on the signal-to-noise ratio and gain of the current transformer. However, the time measurement efficiency of partial discharges decreases with the noise coupled to the current transformer during on-site measurements. To overcome this problem, this patent aims to optimize the signal response of the current transformer through transfer function optimization, redefine the physical and geometric parameters of the current transformer, and apply it to the time measurement of partial discharges strongly affected by background noise to achieve a reasonably high signal-to-noise ratio. Summary of the Invention

[0004] The purpose of the present invention is to provide a partial discharge detection method based on the optimization of current transformer parameters.

[0005] A partial discharge detection method based on the optimization of current transformer parameters includes the following steps:

[0006] Step 1. Construct the transfer function F(s) for detecting partial discharge of the current transformer as follows:

[0007]

[0008] where s is the independent variable of the transfer function; M c is the mutual inductance of the current transformer; N p takes the value of 1; N s is the number of turns of the current transformer coil, R L is the load resistance, C s is the parasitic capacitance of the current transformer, L s is the inductance of the current transformer, R s is the coil resistance of the current transformer.

[0009] The expression of the mutual inductance M c of the current transformer is as follows:

[0010]

[0011] where u is the permeability of the current transformer, and its value is u r ·u 0 , u r is the relative permeability of the coil of the current transformer, u 0 is the air permeability. A c is the cross-sectional area of the iron core of the current transformer, l w is the total length of the coil of the current transformer.

[0012] Step 2. Preset the number of turns N s of the current transformer coil, the load resistance R L , the parasitic capacitance C s of the current transformer, the inductance L s of the current transformer, the coil resistance R s of the current transformer, the cross-sectional area A c of the iron core of the current transformer, and the total length l w of the coil of the current transformer; Set several combinations of parameters to be optimized. The combinations of parameters to be optimized include the load resistance R L , the number of turns N s of the secondary coil, and the relative permeability μ r .

[0013] Step 3. For the transfer function F(s) corresponding to each combination of parameters to be optimized, perform frequency domain analysis respectively to obtain the Bode diagram and frequency response curve of the current transformer corresponding to different combinations of parameters to be optimized.

[0014] Step 3: Use the frequency range corresponding to the identified partial discharge type as the target frequency range; extract the parts of the Bode plot and the frequency response curve within the target frequency range, and extract the open-loop gain and the phase shift magnitude of this part.

[0015] Step 4: Taking the larger the open-loop gain and the smaller the phase shift magnitude as the criteria, select the optimal combination of parameters to be optimized; use the combination of parameters to be optimized as the parameters of the current transformer.

[0016] Step 5: Place the current transformer determined in Step 4 on the power supply input cable of the device under test; judge whether partial discharge occurs in the device under test according to the output signal of the current transformer.

[0017] Preferably, in Step 4, when the combination of parameters to be optimized with the largest open-loop gain is different from the combination of parameters to be optimized with the smallest phase shift magnitude, if stability is preferred, take the combination of parameters to be optimized with the largest open-loop gain as the parameters of the current transformer. If hysteresis time is preferred, take the combination of parameters to be optimized with the smallest phase shift magnitude as the parameters of the current transformer.

[0018] Preferably, the identified partial discharge type in Step 3 is the partial discharge type most likely to occur in the target device.

[0019] Preferably, if the identified partial discharge type is corona discharge, the target frequency range is 15 kHz to 24 kHz.

[0020] Preferably, if the identified partial discharge type is arc discharge, the target frequency range is below 30 kHz and above 1 MHz.

[0021] Preferably, if the identified partial discharge type is surface discharge, the target frequency range is 3 MHz to 10 MHz.

[0022] Preferably, if the identified partial discharge type is air gap discharge and the monitoring target is a cable, the target frequency range is 3 MHz to 100 MHz.

[0023] Preferably, if the identified partial discharge type is air gap discharge and the monitoring target is a GIS joint, the target frequency range is 10 Hz to 100 kHz.

[0024] Preferably, in Step 5, after denoising the output signal of the current transformer, then perform partial discharge judgment.

[0025] Preferably, the process of denoising is as follows:

[0026] (1) Use the output signal of the current transformer as the signal x to be processed; decompose the signal x to be processed step by step through empirical mode decomposition, and obtain the first residual r 1 .

[0027] (2) Calculate the first mode: Take the initial value of k as 1.

[0028] (3) Decompose through empirical mode decomposition to obtain the (k + 1)-th residual r k+1 ; calculate the (k + 1)-th mode

[0029] (5) If the (k + 1)-th residual r k+1 is less than the preset value, the decomposition is completed, and the updated signal to be processed Otherwise, increase k by 1 and repeat step (3).

[0030] The specific beneficial effects of the present invention are as follows:

[0031] 1. The present invention constructs a transfer function for the current transformer to identify partial discharge signals, and optimizes the load resistance R L , the number of turns N s of the secondary coil, and the relative permeability μ r in the detection process of the current transformer, thereby improving the signal-to-noise ratio and gain when the current transformer detects partial discharge signals, and improving the time measurement efficiency of partial discharge detection.

[0032] 2. The present invention uses an improved empirical mode decomposition algorithm to denoise partial discharge signals to improve the denoising efficiency and improve the detection of partial discharge. Description of the Drawings

[0033] Figure 1 is a flow chart of the present invention.

[0034] Figure 2 is a schematic diagram of using a current transformer to detect partial discharge in the present invention.

[0035] Figure 3 is a flow chart of the denoising process in the second step of the present invention.

[0036] Figure 4 is a schematic diagram of the structure of the current transformer used in the present invention.

[0037] Figure 5 is an equivalent circuit diagram when the present invention performs partial discharge detection.

[0038] Figure 6Bode plots and frequency response curves of five models in the embodiments of the present invention (the upper part is the Bode plot and the lower part is the frequency response curve). Detailed implementation manners

[0039] The present invention will be further described below with reference to the accompanying drawings.

[0040] As Figure 1 shown, a partial discharge detection method based on parameter optimization of a current transformer includes the following steps:

[0041] Step 1: Generation of partial discharge noise containing Gaussian white noise: By setting the pulse duration, pulse rise time, and pulse fall time, a partial discharge pulse signal s(i) is formed, and Gaussian white noise r(i) is added to generate a partial discharge signal g(i) = s(i) + r(i) for simulating a partial discharge pulse under the strong influence of noise.

[0042] Specific process of generation and transmission of partial discharge signals:

[0043] As Figure 2 shown, the device under test generates a partial discharge signal, which enters the current transformer with the addition of a noise signal. After the current transformer detects the signal interfered by noise, it transmits the data to the measurement system.

[0044] Generation of partial discharge signal model:

[0045] The characteristics of partial discharge current pulses are analyzed using the parameters rise time (T r ), pulse width (T w ), and fall time (T f ). All local current pulses originate from the starting time t 0 of the voltage step, where t 0 refers to the time when the increased voltage reaches 10% of the initial voltage (V 0 ). That is to say, T r represents the time interval between 10% and 90% of the pulse amplitude, T f represents the interval between 90% and 10% of the pulse amplitude, and T w represents the time interval between 50% of the rising signal and 50% of the falling signal, the pulse height, and the maximum pulse amplitude (100% of the pulse amplitude). Therefore, a conceptual model of the partial discharge current pulse is established:

[0046] V(t) = V 0 (10 -αt - 10 -βt )

[0047] α and β are time constant parameters related to the waveform.

[0048] In the present invention, its pulse parameters are set to a peak value of 300 mA, T r = 0.48853 ns, T f = 4.6785 ns, α = 7.47 8 and β = 3.85 9 .

[0049] Addition of Gaussian white noise:

[0050] In the analysis of partial discharge detection, background noise is the signal detected during on-site measurement. However, considering that partial discharge is a pulse with random and non-stationary properties, therefore, the noise obtained in the on-site measurement signal has random characteristics. Thus, noise addition processing is performed on the above-generated signal pulses. Gaussian white noise is the most similar to partial discharge and is also the most difficult noise to eliminate. Therefore, Gaussian white noise is added to the generated partial discharge signal.

[0051] Assume that the Gaussian process is represented by X(t) in the interval [0, t], with the weight on a certain function g(t) being X(t), and the product of g(t)×t is integrated within this interval, which is expressed as:

[0052]

[0053] where the mean square value of Y is finite for a certain weighting function g(t), and for each g(t) in this type of function, Y is called a Gaussian distributed random variable. That is to say, if each Y(t) is a Gaussian random variable, then X(t) is a Gaussian process. Then, if the probability density function (PDF) of the Y random variable, the Y random variable has a Gaussian distribution:

[0054]

[0055] μ Y is the mean, and σ Y is the standard deviation. When the mean μ Y of the Gaussian random quantity Y = 0 and the variance σ Y 2 = 1, the normalized Gaussian formula distribution is usually written as: N(0,1), and the value of the signal will be searched at the time interval of ±3σ.

[0056] White noise is a Gaussian process, and its power spectral density is independent of frequency. Therefore, Gaussian noise is used as the additive Gaussian white noise of the pulse generator for partial discharge to simulate on-site characteristics.

[0057] Its signal-to-noise ratio (SNR) is 20log 10 Av s / Av n .

[0058] Av s represents the partial discharge pulse signal of the input, and Av n represents the noise.

[0059] Step 2: As Figure 3 shown, noise reduction: Use a current transformer with a Faraday cage on the outside to sleeve the outside of the power supply line of the device under test. Use the current transformer to collect the partial discharge signal g(i), and use the improved empirical mode decomposition algorithm to perform noise reduction processing on the partial discharge signal g(i). The specific process is as follows:

[0060] 2-1: Take the partial discharge signal g(i) as the signal x to be processed; through the expression of empirical mode decomposition x (i) = x + β 0 ·E 1 (w (i) ), decompose the signal x to be processed step by step; obtain the set I = (x (1) , x (2) ,..., x (n) ), and the first residual r 1 = M 1 w (1) ; where β 0 is the magnitude of the additional noise, and its value is greater than 0. E 1 (w (i) ) is the value of the white noise processing set through the EMD operator. w (i) is white noise. M 1 is the local mean operator.

[0061] 2-2: Calculate the first mode in the first stage (k = 1):

[0062] 2-3: Decompose by the method of empirical mode decomposition, continue to decompose it into the steps shown in 2-1, and obtain r 1 + β 1 E 2 (w (i) ), and obtain the second residual r 2 = M(r 1 + β 1 E 2 (w (i) )); and define the second mode Take the initial value of k as 3.

[0063] 2-4: Decompose the (k - 1)th mode by the method of empirical mode decomposition to obtain the kth residual r k = M(r k-1 + β k-1E k (w (i) )). Calculate the k-th mode Proceed to Step 2-5.

[0064] 2-5. If the k-th residual r k is less than the preset value, the decomposition is complete. Take the sum of the 1st mode to the k-th mode plus the k-th residual r k , as the denoised partial discharge signal g(i); otherwise, increase k by 1 and repeat Step 2-4.

[0065] After decomposing the original signal into K intrinsic mode functions using the improved empirical mode decomposition algorithm, these intrinsic mode function segments contain significant and insignificant information. The intrinsic mode functions containing less important information will be considered noise and should be deleted. To determine which information needs to be deleted and which needs to be retained, a statistical significance test is introduced. Therefore, first, the partial discharge signal g(i) is decomposed into intrinsic mode functions. Then, the upper and lower limits of the confidence level are selected. Finally, the energy density of the intrinsic mode functions is compared with the confidence level, and the part of the energy density that lies above the upper bound and below the lower bound of the confidence level is regarded as white noise and removed. After applying the statistical significance test, the intrinsic mode functions with non-significant information are removed, and the intrinsic mode functions with significant information are combined to reconstruct the signal, thus achieving denoising of the signal.

[0066] Step Three: Identify the type of partial discharge corresponding to the partial discharge signal g(i). Using the principle of detecting partial discharge signals with a current transformer, establish a physical model and obtain the transfer function. Use the transfer function to perform sensor modeling by setting the geometric core material and winding parameters of the current transformer to obtain the frequency response curve and determine the transfer function. Use the transfer function to convert the partial discharge signal g(i) into a signal function Z(i) for analysis.

[0067] Establishment of the current transformer model:

[0068] The physical and geometric parameters of the current transformer are related to the toroidal coil wound on a high relative permeability core. The electrical response of the current transformer is based on structural and electrical aspects and is determined by the geometric structure, number of turns, ferromagnetic core material, and load resistance (terminals). The electrical and structural parameters of the current transformer are as Figure 3 shown. Place an FC Faraday cage outside the current transformer, whose function is to isolate external interference and prevent the filtered signal from being disturbed by external noise. Figure 3 In, R L is the load resistance, r i is the inner radius, r o is the outer radius, r is the sensor radius, r cis the iron core radius, N is the number of turns of the sensor coil (secondary circuit), is the cross-sectional area, l m is the magnetic flux path of, I(t) is the current in the secondary circuit, V o (t) is the output voltage of the circuit, i.e., the output signal of the current transformer. The radius of the current transformer is r = (r i + r o ) / 2, l m = 2πr, r c = (r o - r i ) / 2. The cross-sectional area of the iron core is A c = πr c 2 , the radius of the coil p w = r c . The length of a single coil is l c = 2πr c , and the length of the entire coil is l w = l c ·N.

[0069] The Figure 4 parameter analysis current transformer in is simplified into an equivalent circuit as shown in Figure 5 .

[0070] The following electrical parameters are provided: The resistivity of the current transformer winding ρ = 1.6800 * 10 -8 (Ω·m), relative permeability u r = 2000, permeability in vacuum u o = 4π * 10 -7 (H / m), absolute magnetic permeability u = 0.0025 (H / m), dielectric constant in vacuum e o = 8.8540 * 10 -12 (F / m), relative dielectric constant e r = 1, e = 8.85240 * 10 -12 (F / m), resistivity of copper ρ o = 1.68 * 10 -8 (Ω·m).

[0071] For the specific description of the model, assume the parameters of the current transformer: outer diameter d 0 = 0.1m, outer radius r 0 = 0.05m, inner diameter d i = 0.065m, inner radius r i = 0.0325m, number of turns of the primary coil N P = 1, number of turns of the secondary coil N s = 10, coil diameter d w= 0.0005 m, coil radius r w = 0.00025 m. Core radius r c = 0.0088 m, core area A c = 2.4053 * 10 -4 (m 2 ),core diameter dr c = 0.0175 m. Sensor radius r = 0.0575 m. Magnetic flux path l m = 0.3613 m. Coil cross - sectional area A w = 2.4053 * 10 -4 (m 2 ),length of a single coil l c = 0.0550 m, coil pitch p w = 0.0088 m.

[0072] Winding resistance value of the current transformer:

[0073]

[0074] The voltage of the current transformer is expressed as:

[0075]

[0076] Mutual inductance M c is expressed as:

[0077]

[0078] where, u = u r ·u 0 , u r is the relative permeability (determined according to the value of the material), u 0 is the air permeability, with a value of 4π * 10 -7 (H / m). N s is the number of turns of the secondary coil, i.e., the winding of the sensor. i p (t) is the current signal output by the current transformer.

[0079] Leakage inductance (self - inductance) L of the current transformer loop is expressed as:

[0080]

[0081] Inductance L of the current transformer s is expressed as:

[0082] L s = M c 2 ·L loop

[0083] The parasitic capacitance C of the current transformer s is expressed as:

[0084]

[0085] where the absolute permittivity ε = ε 0 ε r , the relative permittivity ε 0 = 8.854 * 10 -12 , and the relative permittivity of air ε r = 1.

[0086] That is, the transfer function F(s) of the current transformer is

[0087]

[0088] where s is the independent variable of the transfer function.

[0089] The frequency resonance ω of the transfer function F(s) 0 is:

[0090]

[0091] The damping coefficient ξ of the transfer function F(s) is:

[0092]

[0093] To improve the signal response of the current transformer, the present invention optimizes its frequency response through the transfer function and adjusts the parameters affecting the transfer function to optimize the geometric parameters of the current transformer. The present invention provides 5 models, namely S1, S2, S3, S4, and S5. The S1 model is the basic model, and the parameters are given as R L = 50, N s = 10, μ r = 2000. Based on S1, the parameters are adjusted to introduce S2, S3, S4, and S5. The parameters of S2 are given as R L = 50, N s = 30, μ r = 2000; the parameters of S3 are given as R L = 250, N s = 10, μ r = 2000; the parameters of S4 are given as R L = 250, N s = 10, μ r = 2300; the parameters of S5 are given as R L = 150, N s = 20, μ r= 2100. Optimize the transfer function using the above five groups of parameter models. Through the above parameters, the transfer functions of 5 models can be obtained.

[0094] In S1

[0095]

[0096] In S2

[0097]

[0098] In S3

[0099]

[0100] In S4

[0101]

[0102] In S5

[0103]

[0104] Step 4: Optimization verification: Based on the changes in structural parameters and electrical parameters, analyze its Bode plot and impulse response curve to evaluate the result of signal optimization.

[0105] Through MATLAB simulation, the Bode plots and frequency response curves of 5 models can be obtained, as Figure 6 shown.

[0106] Through the MATLAB simulation software, the physical parameter performance obtained by each model can be calculated. In S1, its poles are s 1 = -2.92*10 5 , s 2 = -4.65*10 9 , damping ratio ξ = 6.3104*10 1 , and the bandwidth is 5.03*10 9 Hz. In S2, its poles are s 1 = -3.24*10 4 , s 2 = -1.55*10 9 , damping ratio ξ = 1.0930*10 2 , and the bandwidth is 4.89*10 8 Hz. In S3, its poles are s 1 = -1.46*10 4 , s 2 = -9.28*10 9 , damping ratio ξ = 1.2621*10 1 , and the bandwidth is 4.95*10 9Hz. In S5, its poles are s 1 = -2.09*10 5 , s 2 = -7.55*10 8 , damping ratio ξ = 3.0465*10 1 , and the bandwidth is 1.321*10 9 Hz.

[0107] As Figure 6 shown, the gain of S1 can reach 13.8 dB. At the resonance frequency, its phase has a 180° shift. Under these 5 models, by changing the physical parameters (R L , N s , μ r ) of the current transformer, different performances can be obtained for the current transformer. The gain of S2 reaches 4.23 dB. Due to the increase of C s and L s relative to the increase of the S1 model, the curve of its phase charge center is flatter. In S4, μ r = 2300, and its gain reaches 13.9 dB. The characteristic response of S4 to S1 is almost the same, and μ r does not change its parameter characteristics. Therefore, this results in the overlap of the S1 curves. The gain of S5 reaches 17.3 dB, which is the intermediate response between the S1, S4, and S3 models, reflecting its parameter characteristics.

[0108] Partial discharge is mainly divided into corona discharge (such as near the tip of high-voltage wires and live conductors), arc discharge (such as partial discharge occurring at the opening circuit of switchgear), surface discharge (such as partial discharge between the solid edge of a transformer and transformer oil), and air gap (internal) discharge. Based on the optimized design of the transfer function, its R L , N s , μ r values can be controlled to enhance the detection of partial discharge.

[0109] When the main discharge form of the equipment is corona discharge, its frequency is concentrated in 15 - 24 kHz, and the statistical order of the open-loop gain effect from large to small is S5 > S2 > S4 > S1 = S3. The order of the phase shift magnitude from small to large is S3 < S1 < S4 < S5 < S2. It can be seen that although S5 and S2 have strong stability, their phase shifts are relatively large. Although S1 and S3 have relatively large offsets, their stability is relatively poor. Therefore, S4 is selected.

[0110] When the main discharge form of the device is arc discharge, the frequency of the high-frequency arc is concentrated in the range below 30 kHz or above 1 MHz. Below 30 kHz, the statistical results of the open-loop gain effect are sorted from large to small as S5, S2, S4, S1 = S3. The phase shift is sorted from small to large as S3, S1, S4, S5, S2. Although S5 and S2 have strong stability, their phase shifts are relatively large. Although S1 and S3 have relatively large offsets, their stability is relatively poor. Therefore, the selection order is S4, S1, S3, S5, S2. Above 1 MHz, generally speaking, the statistical results of the open-loop gain effect are sorted from large to small as S3, S5, S4, S1, S2. The phase shift is sorted from small to large as S5, S3, S1, S2, S4. S5 and S3 perform better, and S1, S2, and S4 all perform poorly in some aspects. Combining these two working frequencies, there are two preferences. Taking the stability of the system as the standard, S5 can be selected; taking the phase shift of the system as the standard, S3 can be selected.

[0111] When the main discharge form of the device is surface discharge, its frequency range is 3 - 10 MHz. Generally speaking, the statistical results of the open-loop gain effect are sorted from large to small as S3, S5, S4, S1, S2. The phase shift is sorted from small to large as S5, S3, S1, S2, S4. S5 and S3 perform better, so S3 and S5 can be selected.

[0112] When the main discharge form of the device is air gap (internal) discharge, it is usually caused by defects in cables, bushings, GIS joints, etc. The frequency range of the cable is 3 MHz - 100 MHz, and the statistical results of the open-loop gain effect are sorted from large to small as S3, S5, S4, S1, S2. The phase shift is sorted from small to large as S5, S3, S1, S2, S4. S5 and S3 perform better, so S3 and S5 can be selected. The frequency of the GIS joint is 10 - 100 kHz, and the statistical results of the open-loop gain effect are sorted from large to small as S5, S3, S4, S1, S2, and the phase shift is sorted from small to large as S2, S5, S4, S1, S3, so S3 can be selected.

[0113] For a clearer and more accurate detection of partial discharge, multiple current transformers can be set to detect at a certain interval (2 m), and the interactive data of multiple current transformers can be used to ensure the data accuracy of partial discharge.

[0114] Therefore, the present invention uses a preprocessing method to optimize the signal of the transfer function, and by changing the circuit and coil structure characteristics of the current transformer, the signal response of the current transformer is optimized, thereby increasing the signal-to-noise ratio and gain of the current transformer, and thus improving the time measurement efficiency of partial discharge detection.

Claims

1. A partial discharge detection method based on the parameter optimization of a current transformer, characterized in that: It includes the following steps: Step 1: Construct the transfer function for detecting partial discharge of current transformers As follows: Among them, is the independent variable of the transfer function; is the mutual inductance of the current transformer; takes the value of 1; is the number of turns of the coil of the current transformer, is the load resistance, is the parasitic capacitance of the current transformer, is the inductance of the current transformer, is the coil resistance of the current transformer; Mutual inductance of current transformer The expression is as follows: Among them, is the permeability of the current transformer, and its value is , is the relative permeability of the coil of the current transformer, is the air permeability; is the cross-sectional area of the iron core of the current transformer, is the total length of the coil of the current transformer; Step 2: Preset the number of turns of the current transformer coil , load resistance , parasitic capacitance of the current transformer , inductance of the current transformer , coil resistance of the current transformer , cross-sectional area of the iron core of the current transformer , total length of the coil of the current transformer ; Set several combinations of parameters to be optimized; The combinations of parameters to be optimized include load resistance , number of turns of the secondary coil and relative permeability ; Step 3: For the transfer function corresponding to each group of optimized parameter combinations , perform frequency domain analysis respectively to obtain the Bode diagram and frequency response curve of the current transformer corresponding to different optimized parameter combinations; Step 3: Take the frequency range corresponding to the identified partial discharge type as the target frequency range; extract the parts of the Bode plot and the frequency response curve within the target frequency range, and extract the open-loop gain and the phase shift magnitude of this part; Step 4: Take the larger the open-loop gain and the smaller the phase shift magnitude as the criteria, and select the best combination of parameters to be optimized; use the combination of parameters to be optimized as the parameters of the current transformer; When the combination of parameters to be optimized with the largest open-loop gain is different from the combination of parameters to be optimized with the smallest phase shift magnitude, if stability is preferred, take the combination of parameters to be optimized with the largest open-loop gain as the parameters of the current transformer; if the lag time is preferred, take the combination of parameters to be optimized with the smallest phase shift magnitude as the parameters of the current transformer; Step 5: Place the current transformer determined in Step 4 on the power supply input cable of the device under test; judge whether the device under test has partial discharge according to the output signal of the current transformer.

2. A partial discharge detection method based on the parameter optimization of a current transformer according to Claim 1, characterized in that: The identified partial discharge type in Step 3 is the partial discharge type most likely to occur in the target device.

3. A partial discharge detection method based on the parameter optimization of a current transformer according to Claim 1, characterized in that: If the identified partial discharge type is corona discharge, the target frequency range is 15 kHz to 24 kHz.

4. A partial discharge detection method based on the parameter optimization of a current transformer according to Claim 1, characterized in that: If the identified partial discharge type is arc discharge, the target frequency range is below 30 kHz and above 1 MHz.

5. A partial discharge detection method based on the parameter optimization of a current transformer according to Claim 1, characterized in that: If the identified partial discharge type is surface discharge, the target frequency range is 3 MHz to 10 MHz.

6. A partial discharge detection method based on the parameter optimization of a current transformer according to Claim 1, characterized in that: If the identified partial discharge type is air gap discharge and the monitoring target is a cable, the target frequency range is 3 MHz to 100 MHz.

7. A partial discharge detection method based on the parameter optimization of a current transformer according to Claim 1, characterized in that: If the identified partial discharge type is air gap discharge and the monitoring target is a GIS joint, the target frequency range is 10 Hz to 100 kHz.

8. A partial discharge detection method based on the parameter optimization of a current transformer according to Claim 1, characterized in that: In Step 5, after the output signal of the current transformer is denoised, the partial discharge judgment is carried out.

9. A partial discharge detection method based on the parameter optimization of a current transformer according to Claim 8, characterized in that: The process of denoising is specifically as follows: (1) Take the partial discharge signal g(i) as the signal x to be processed; through the expression of empirical mode decomposition =x+ , decompose the signal x to be processed step by step; obtain the set , and the first residual = ; where is the magnitude of the added noise, and its value is greater than 0, is the value of the white noise processing set by the EMD operator; is white noise; is the local mean operator; (2) Calculate the first mode in the first stage: = x - ; (3) By means of empirical mode decomposition, decompose to obtain + , and obtain the second residual =M( + ); Define the second mode = - = -M( + ); Take the initial value of k as 3; (4) Decompose the (k - 1)-th mode by empirical mode decomposition to obtain the k-th residue ; Calculate the k-th mode = - ; Proceed to step (5); (5) If the k-th residual is less than the preset value, the decomposition is completed, and the sum of the first mode to the k-th mode plus the k-th residual is taken as the denoised partial discharge signal g(i); otherwise, k is incremented by 1 and step (4) is repeated.

Images