Converter transformer internal discharge fault monitoring system and monitoring method

Abstract

The invention discloses a converter transformer internal discharge fault monitoring system and monitoring method, and relates to the field of discharge fault monitoring, the converter transformer internal discharge fault monitoring system comprises a broadband current sensor, the broadband current sensor is internally provided with a synchronous high-speed acquisition processing unit, a sub-band data processing compensation unit, a multi-band data characteristic analysis unit, a synchronization module and a communication module, the broadband current sensor comprises a power source, a converter transformer iron core and a clamp grounding wire part, the arc current sensor couples current signals with different frequency bands from the converter transformer iron core and the clamp grounding wire part, internal discharge faults of a converter transformer, especially later arc type discharge, are researched, and the broadband current sensor capable of effectively monitoring arc current is researched and developed. According to the method, high-frequency current, arc current and harmonic current are monitored respectively, a discharge fault development stage identification model based on different frequency band current relations is established by using energy spectrum offset characteristics in a discharge fault development process, and whole-process monitoring of the discharge fault in the converter transformer is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of discharge fault monitoring, and in particular to a system and method for monitoring internal discharge faults of a converter transformer. Background Art

[0002] With the rapid development of UHV DC transmission technology in my country, UHV DC transmission projects are increasing. As an important equipment in the UHV DC transmission system, the converter transformer is a key hub equipment connecting DC and AC in the converter station. Once a converter transformer fails, it will cause the converter valve to lock, causing local or even large-scale power outages. At present, State Grid Corporation has experienced a number of serious explosions caused by the breakdown discharge of converter transformers, which poses a huge threat to the personal safety of on-site operation and maintenance personnel, equipment safety, and cross-regional DC safety.

[0003] In the existing technology, there is still no relatively scientific and mature evaluation criteria for the monitoring of internal discharge defects of converter transformers. Everyone focuses on the field of local discharge signal extraction and intelligent signal identification inside converter transformers. However, there is a lack of in-depth research on how to defend the last line of defense of converter transformers in the event of destructive discharge inside converter transformers, especially when arcing occurs after the later fault development and deterioration, especially in the research and application of full-process monitoring technology based on arc current. Therefore, it is of great significance to carry out research based on arc current technology and multi-physical quantity early warning monitoring of discharge signals, discharge defect development criteria and alarm strategies for evaluating the operating status of converter transformers and preventing the occurrence of serious accidents such as explosions. Therefore, a converter transformer internal discharge fault monitoring system and monitoring method are needed to meet people's needs. Summary of the invention

[0004] The object of the present invention is to provide a system and method for monitoring internal discharge faults of a converter transformer, so as to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solution: a system and method for monitoring internal discharge faults of a converter transformer, comprising:

[0006] Broadband current sensor, which is equipped with a synchronous high-speed acquisition processing unit, a frequency band data processing and compensation unit, a multi-band data feature analysis unit, a synchronization module, a communication module, and a power supply;

[0007] The arc current sensor couples current signals of different frequency bands from the commutation transformer core and the clamp grounding wire.

[0008] Synchronous high-speed acquisition processing unit converts analog signals conditioned and amplified by multiple channels into digital signals;

[0009] The frequency band data processing and compensation unit fuses, compensates and synthesizes the converted digital signals in the frequency domain to obtain the discharge current signal;

[0010] A multi-band data feature analysis unit extracts feature information of different frequency bands for analysis;

[0011] Communication module, transmits data and analysis and diagnosis results to the background;

[0012] The power supply provides working power for the frequency-band signal conditioning unit, the synchronous variable time base high-speed acquisition processing unit, the frequency-band data processing and compensation unit, the multi-band data feature analysis unit, the synchronization module, and the communication module.

[0013] Preferably, by using high-frequency pulse current and ultrasonic waves for comprehensive monitoring, different physical quantities radiated by the same discharge source are collected synchronously to obtain fault sample data when the equipment has a partial discharge fault;

[0014] The transmission ratio model is used to determine whether the signal comes from inside the converter transformer. The above signal analysis can basically exclude the influence of external interference signals including bushing end signals and external discharge signals, thus ensuring the accuracy of the monitoring signal.

[0015] According to the fault sample data, a two-dimensional spectrum corresponding to the fault sample data is generated, and a partial discharge characteristic parameter of the fault sample data is extracted from the two-dimensional spectrum;

[0016] A multi-source parameter fusion analysis model is established by utilizing their homology and phase relationship to achieve accurate identification of internal discharge and external interference signals.

[0017] Preferably, after using the frequency-division arc current sensor to obtain the original data of multiple signals, the high-speed acquisition and processing unit completes the synchronous high-speed acquisition and processing of the pulse signal, low-frequency cluster signal, and power frequency and harmonic signal, and uses data processing fusion technology to perform compensation and data reorganization. At the same time, the multi-band data feature analysis unit can obtain the signal of a certain frequency band separately according to specific requirements, and perform multi-band waveform data feature analysis through edge computing to determine the severity of the internal discharge fault of the converter transformer.

[0018] Preferably, (1) determining a filter function Hd(ejW);

[0019] (2) Determine the amplitude function A from Hd(ejW) d (Ω):

[0020]

[0021] (3) Determine the filter amplitude-frequency characteristic function by formula 1:

[0022]

[0023] (4) Add window to get the impulse response of the actual filter:

[0024] h[k]=h d [k]*W N [k] (3)

[0025] The window function W N [k] is:

[0026]

[0027] (5) Using Formula 3 and Formula 4, the sampled signal x[k] is bandpass filtered to obtain the processed data y[k]:

[0028] y[k]=x[k]*h[k] (5)

[0029] (6) Use Formula 5 to separate and calculate y[k] in the frequency domain:

[0030]

[0031] In the frequency domain data synthesis implementation, F1S, F1E are the low-end cutoff frequency and high-end cutoff frequency of frequency band 1; F2S, F2E are the low-end cutoff frequency and high-end cutoff frequency of frequency band 2; F3S, F3E are the low-end cutoff frequency and high-end cutoff frequency of frequency band 3;

[0032] Then perform frequency domain synthesis:

[0033] Yout(n)=Y1(n)*S1(n)+Y2(n)*S2(n)+Y3(n)*S3(n) (7)

[0034] Where: Yout(n) is the frequency domain output of the arc current sensor;

[0035] Y1,2,3(n) are the outputs of the sensor units in the 1st, 2nd, and 3rd frequency bands respectively;

[0036] S1,2,3(n) are the correction coefficients of the sensors in the 1st, 2nd and 3rd frequency bands respectively;

[0037] The arc current signal is constructed and implemented by using Formula 7 to transform the data after frequency domain synthesis. The transformed data ensures the sensitivity and dynamic range of each detection frequency band. The arc current signal obtained by frequency domain fusion can be continuously analyzed in time domain and frequency domain, and then feature extraction is realized.

[0038] Preferably, a multi-layer progressive logic alarm strategy is used: primary judgment is made using waveform recognition, secondary judgment is made using discharge energy change and amplitude change rate, tertiary judgment is made using the correlation between high-frequency partial discharge current, arc current and power frequency current changes combined with discharge source location information, and final alarm information is output in combination with other physical quantities, which can ensure accurate alarm when destructive discharge or arc occurs inside the converter transformer.

[0039] The beneficial effects of the present invention are:

[0040] 1. In the present invention, the internal discharge faults of the converter transformer, especially the late arc-type discharge, are studied, and a broadband current sensor capable of effectively monitoring the arc current is developed to monitor the high-frequency current, arc current and harmonic current respectively. The energy spectrum shift characteristics in the development process of the discharge fault are utilized to establish a discharge fault development stage identification model based on the relationship between currents in different frequency bands, thereby realizing the full process monitoring of the internal discharge faults of the converter transformer.

[0041] 2. In the present invention, the partial discharge monitoring means of the converter transformer is utilized. The partial discharge monitoring means of the converter transformer mainly include two methods: high frequency and ultrasonic. These two monitoring modes use different physical quantities as characteristic information, and have different effectiveness and sensitivity for different types of discharges inside the converter transformer. The homology and phase relationship are utilized to establish a multi-source parameter fusion analysis model to achieve accurate identification of internal discharge and external interference signals.

[0042] 3. In the present invention, according to the multi-source parameter information of partial discharge and the frequency-divided current signal relationship based on the arc current, judgment criteria are formed in the stages of partial discharge signal recognition, defect trend development and destructive discharge failure, and early warning and alarm strategies that meet on-site needs are established. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A schematic diagram of the structure of a converter transformer internal discharge fault monitoring system and monitoring method proposed by the present invention;

[0044] Figure 2 This is a schematic diagram of the data fusion processing flow proposed by the present invention;

[0045] Figure 3 This is a schematic diagram of frequency domain synthesis proposed by the present invention;

[0046] Figure 4 A schematic diagram illustrating the signals proposed in the present invention;

[0047] Figure 5 This is a schematic diagram of the data of the initial stage of explosion proposed by the present invention;

[0048] Figure 6 This is a schematic diagram of the explosion development stage data proposed by the present invention;

[0049] Figure 7 This is a schematic diagram of the signal spectrum analysis proposed by the present invention. DETAILED DESCRIPTION

[0050] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0051] Reference Figure 1-Figure 7 , a converter transformer internal discharge fault monitoring system and monitoring method, comprising:

[0052] Broadband current sensor, which is equipped with a synchronous high-speed acquisition processing unit, a frequency band data processing and compensation unit, a multi-band data feature analysis unit, a synchronization module, a communication module, and a power supply;

[0053] The arc current sensor couples current signals of different frequency bands from the commutation transformer core and the clamp grounding wire.

[0054] Synchronous high-speed acquisition processing unit converts analog signals conditioned and amplified by multiple channels into digital signals;

[0055] The frequency band data processing and compensation unit fuses, compensates and synthesizes the converted digital signals in the frequency domain to obtain the discharge current signal;

[0056] A multi-band data feature analysis unit extracts feature information of different frequency bands for analysis;

[0057] Communication module, transmits data and analysis and diagnosis results to the background;

[0058] The power supply provides working power for the frequency-band signal conditioning unit, the synchronous variable time base high-speed acquisition processing unit, the frequency-band data processing and compensation unit, the multi-band data feature analysis unit, the synchronization module, and the communication module.

[0059] Furthermore, by using high-frequency pulse current and ultrasonic waves for comprehensive monitoring, different physical quantities radiated by the same discharge source are collected synchronously to obtain fault sample data when the equipment has a partial discharge fault;

[0060] The transmission ratio model is used to determine whether the signal comes from inside the converter transformer. The above signal analysis can basically exclude the influence of external interference signals including bushing end signals and external discharge signals, thus ensuring the accuracy of the monitoring signal.

[0061] According to the fault sample data, a two-dimensional spectrum corresponding to the fault sample data is generated, and a partial discharge characteristic parameter of the fault sample data is extracted from the two-dimensional spectrum;

[0062] A multi-source parameter fusion analysis model is established by utilizing their homology and phase relationship to achieve accurate identification of internal discharge and external interference signals.

[0063] Furthermore, after using the frequency-divided arc current sensor to obtain the original data of multiple signals, the high-speed acquisition and processing unit completes the synchronous high-speed acquisition and processing of pulse signals, low-frequency cluster signals, power frequency and harmonic signals, and uses data processing fusion technology for compensation and data reorganization. At the same time, the multi-band data feature analysis unit can obtain the signal of a certain frequency band separately according to specific requirements, and perform multi-band waveform data feature analysis through edge computing to determine the severity of the internal discharge fault of the converter transformer.

[0064] Further, (1) determine the filter function Hd(ejW);

[0065] (2) Determine the amplitude function A from Hd(ejW) d (Ω):

[0066]

[0067] (3) Determine the filter amplitude-frequency characteristic function by formula 1:

[0068]

[0069] (4) Add window to get the impulse response of the actual filter:

[0070] h[k]=h d [k]*W N [k] (3)

[0071] The window function W N [k] is:

[0072]

[0073] (5) Using Formula 3 and Formula 4, the sampled signal x[k] is bandpass filtered to obtain the processed data y[k]:

[0074] y[k]=x[k]*h[k] (5)

[0075] (6) Use Formula 5 to separate and calculate y[k] in the frequency domain:

[0076]

[0077] As attached Figure 2 The frequency domain data synthesis is shown in the attached Figure 3As shown, F1S, F1E are the low-end cutoff frequency and high-end cutoff frequency of frequency band 1, F2S, F2E are the low-end cutoff frequency and high-end cutoff frequency of frequency band 2, and F3S, F3E are the low-end cutoff frequency and high-end cutoff frequency of frequency band 3;

[0078] Use Formula 7 for frequency domain synthesis:

[0079] Yout(n)=Y1(n)*S1(n)+Y2(n)*S2(n)+Y3(n)*S3(n) (7)

[0080] Where:

[0081] Yout(n) is the frequency domain output of the arc current sensor;

[0082] Y1,2,3(n) are the outputs of the sensor units in the 1st, 2nd, and 3rd frequency bands respectively;

[0083] S1,2,3(n) are the correction coefficients of the sensors in the 1st, 2nd and 3rd frequency bands respectively;

[0084] As attached Figure 2 As shown, the arc current signal is constructed and implemented. Formula 7 is used to transform the data after frequency domain synthesis. The transformed data ensures the sensitivity and dynamic range of each detection frequency band. The arc current signal obtained by frequency domain fusion can be continuously analyzed in time domain and frequency domain, and then feature extraction is realized.

[0085] Furthermore, it includes a multi-layer progressive logic alarm strategy: using waveform recognition for primary judgment, using discharge energy change and amplitude change rate for secondary judgment, using the correlation between high-frequency partial discharge current, arc current and power frequency current changes combined with discharge source location information for tertiary judgment, and combining other physical quantities to output the final alarm information, which can ensure accurate alarm when destructive discharge or arc occurs inside the converter transformer.

[0086] Working principle of the present invention:

[0087] Arc discharge is the discharge form in the later stage of the internal discharge fault of the converter transformer. From the perspective of the development stage of the converter transformer fault, it has exceeded the scope of partial discharge. The energy released when arc discharge occurs is much greater than that of partial discharge, which is more harmful to the insulation system of the converter transformer and is the main factor causing explosion accidents. The existing partial discharge monitoring sensors mainly monitor partial discharge pulse current. When the internal discharge fault develops to a certain extent, the sensitivity of the existing sensors will decrease and cannot effectively characterize the development degree of the discharge fault, especially for arc discharge.

[0088] Based on the original high-frequency pulse current sensor, a sensor for arc current monitoring inside the converter transformer is developed. After using the frequency-divided arc current sensor to obtain the original data of multiple signals, the high-speed acquisition and processing unit completes the synchronous high-speed acquisition and processing of pulse signals, low-frequency cluster signals, and power frequency and harmonic signals, and uses data processing fusion technology for compensation and data reorganization. At the same time, the multi-band data feature analysis unit can obtain the signal of a certain frequency band separately according to specific requirements, and perform multi-band waveform data feature analysis through edge computing to determine the severity of the internal discharge fault of the converter transformer.

[0089] When a discharge defect occurs inside the converter transformer, it will simultaneously radiate high-frequency pulse current, ultra-high frequency electromagnetic waves and ultrasonic signals. These signals have different physical properties, but they are homologous and phase-related. The homology of these three signals can cover different discharge types and discharge locations inside the converter transformer, and their homology can be used to complete the confirmation of internal discharge signals.

[0090] The detection principle of high-frequency pulse current mode is that the winding of the transformer (reactor) is made of oil-paper insulation material between the core, clamps and shell, and there is distributed capacitance. The high-frequency signal of 3MHz-30MHz in the spectrum content of the discharge signal is transmitted from the winding to the core through the distributed capacitance, and connected to the ground through the core, clamps and shell grounding wires. The high-frequency HFCT sensor installed on the grounding wire of the equipment can detect the partial discharge pulse signal. The main features of high-frequency pulse partial discharge monitoring technology: using high-frequency sensors and high-speed broadband sampling units to obtain enough discharge information and provide effective diagnostic basis; can fully preserve various discharge spectra, such as: Q-Φ spectrum, N-Φ spectrum, PRPD spectrum, PRPS spectrum; by comparing the differences in waveform characteristics between different discharge pulse signals and between discharge and interference, it can effectively separate and cluster different discharge pulses and distinguish different discharge types.

[0091] The detection principle of ultrasonic mode is that once partial discharge occurs inside the transformer, ultrasonic waves will be generated, which will propagate around in the transformer oil in the form of spherical waves at a speed of about 1400m / s. As long as an ultrasonic detector is placed in the oil or on the outside of the transformer box wall, the ultrasonic pulse signal generated by the discharge can be received to confirm the internal partial discharge defects. The time difference of the electroacoustic signal is approximately equal to the propagation time of the ultrasonic wave. According to the propagation speed and direction of the ultrasonic wave, the location of the discharge source can be determined. The main features of ultrasonic monitoring technology: Ultrasonic waves are extremely sensitive to internal surface discharge defects; the discharge type is analyzed by recording the ultrasonic wave time domain waveform information and frequency domain characteristics; ultrasonic detection technology has great advantages in defect location of large oil-immersed transformers. The ultrasonic signal has a certain coverage range. According to the amplitude and waveform changes, the approximate location of the discharge source can be determined, and the time difference between the ultrasonic wave and the electrical signal can be used to accurately locate the discharge source.

[0092] The fusion monitoring technology of the converter transformer discharge fault can utilize the correlation and phase synchronization between multiple physical quantities to achieve complementary advantages and mutual verification, and realize the suppression of external interference signals and the confirmation of internal discharge signals.

[0093] According to the electrical aging and breakdown characteristics of oil-paper insulation, six key parameters in the partial discharge process were extracted to establish a discharge fault development stage identification model. The selected parameters are Qmax (maximum discharge amplitude), Qave (average discharge amplitude), Feqave (discharge pulse spectrum), Fnc (number of discharge pulses), dmax (pulse distance), and Teqave (pulse duration). The visualization relationship of multiple feature quantities is established through the characterization spectrum, and a method for identifying the development stage of discharge defects based on the fault development characterization spectrum is proposed. In order to make the spectrum highly comparable, the six types of partial discharge features should be normalized first. Due to the randomness and dispersion of partial discharge, the average values ​​of partial discharge feature quantities in the same development stage are not equal. Therefore, the coordinate range of the fault development spectrum should not only be unified after normalization, but also be slightly larger than the corresponding feature quantity range to increase the universality of the spectrum.

[0094] The spectra of discharge faults at different stages show that the shapes of the spectra of discharge faults at different development stages are different, while the shapes of radar spectra of different discharge types at the same development stage are similar. Further research found that no matter how the partial discharge feature quantities are arranged, there are obvious differences in the spectra shapes of different electrical development stages, and they are mainly reflected in the center of gravity and aspect ratio of the spectra. The electrical aging stage can be identified according to the shape characteristics of the spectra. On the one hand, the shape characteristics of the spectra are quantitatively and effectively reflected; on the other hand, the original partial discharge feature space is reduced in dimension (to less than three dimensions) to establish a visual identifier, and the center of gravity coordinate ratio and aspect ratio of the spectra can also be used as new identification features. The sub-band current information obtained by the arc current sensor is combined with the above-established development stage recognition model to establish an early warning strategy for internal discharge faults in the converter transformer.

[0095] Through the study of the internal discharge fault of the converter transformer, it is found that when the insulation defect occurs inside the converter transformer, it is initially manifested in the form of partial discharge. At this time, through the high-frequency pulse current sensor, it can be seen that the discharge form appears in the form of pulses, and the spectrum analysis shows that its frequency domain range can cover 10kHz-30MHz, which can be monitored by the high-frequency partial discharge sensor. At the same time, it will also radiate 300MHz-2GHz ultra-high frequency signals and 20kHz-200kHz ultrasonic signals. With the development of the discharge fault, the discharge degree is intense and arcs are generated. At this time, the pulse characteristics of the discharge form are no longer obvious, and it appears clustered or columnar. Its spectrum range drops below 1MHz, and its main energy range is concentrated in the range of tens of kilohertz.

[0096] Taking the monitoring signal when a poor contact discharge defect occurs inside an actual converter transformer as an example, high-frequency pulse current, ultra-high frequency and ultrasonic monitoring are carried out, and obvious discharge signals can be detected externally, and the high-frequency pulse current signal and the ultra-high frequency signal are consistent in the spectrum morphology, and are basically the same in phase. The high-frequency signal and the ultra-high frequency signal are ahead of the ultrasonic signal in phase, which can explain that the monitoring data during the test is divided into two stages. In the initial stage of the explosion test, the high-frequency pulse current monitoring data is 23mV, the arc current monitoring data is 110mV, the power frequency current monitoring data is 1570mV, and the ultrasonic monitoring data is 87mV. In the stage of violent development of internal arcing, the high-frequency pulse current monitoring data is 29mV, the arc current monitoring data is 524mV, the power frequency current monitoring data is 2047mV, and the ultrasonic monitoring data is 686mV. Through the analysis of the above data and spectrum, it can be clearly seen that there are obvious changes in both the electrical signal and the acoustic signal, and the ultrasonic amplitude has increased dramatically, indicating that ultrasonic waves can effectively monitor when discharge defects occur inside the converter transformer. In addition, it is found that the pulse signal amplitude of high-frequency partial discharge changes relatively little during the development stage of internal faults, and the pulse density increases; the arc current amplitude changes greatly, and the pulse density also increases; the power frequency current amplitude also changes greatly. This shows that the existing high-frequency pulse partial discharge sensor is highly sensitive to the pulse current signal occurring in the initial stage of the discharge fault of the converter transformer, which has been widely recognized in engineering applications. When the internal changes drastically and the pulse discharge develops into a large-energy current discharge, the amplitude change is relatively not very obvious, and this result cannot be used as the basis for hazard assessment for on-site monitoring. From the above spectrum analysis, it can be seen that the energy distribution is relatively balanced during pulse current discharge, while the spectrum energy of the arc current has a concentration point. The arc current monitoring sensitivity is relatively low compared to the high-frequency partial discharge signal at the initial stage of the fault, but it is more sensitive to the defect change process. From the analysis of the actual monitoring effect of the converter transformer, the existing partial discharge detection means are highly sensitive to the pulse discharge signal, and the current discharge and large-energy arc discharge cannot be accurately expressed through data changes, and there are problems in describing the development degree of the discharge fault. It is feasible to use arc current to monitor the hazard degree of the internal discharge fault of the converter transformer through experiments and on-site verification.

[0097] The early warning monitoring technology for converter transformers requires the study of a three-layer progressive logic alarm mechanism, the establishment of a complete alarm strategy based on fault tree technology, the use of a variety of monitoring parameters fusion analysis for primary judgment to determine whether the signal comes from inside the converter transformer, the use of pulse energy changes and amplitude change rates for secondary judgment, the use of partial discharge signals, arc current change correlation combined with discharge source location information for tertiary judgment, and ultimately output of alarm information, which can ensure accurate alarm when strong discharge occurs inside the converter transformer and that the signal comes from inside the converter transformer.

[0098] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A system and method for monitoring internal discharge faults of a converter transformer, characterized in that: include: Broadband current sensor, which is equipped with a synchronous high-speed acquisition processing unit, a frequency band data processing and compensation unit, a multi-band data feature analysis unit, a synchronization module, a communication module, and a power supply; The arc current sensor couples current signals of different frequency bands from the commutation transformer core and the clamp grounding wire. Synchronous high-speed acquisition processing unit converts analog signals conditioned and amplified by multiple channels into digital signals; The frequency band data processing and compensation unit fuses, compensates and synthesizes the converted digital signals in the frequency domain to obtain the discharge current signal; A multi-band data feature analysis unit extracts feature information of different frequency bands for analysis; Communication module, transmits data and analysis and diagnosis results to the background; The power supply provides working power for the frequency-band signal conditioning unit, the synchronous variable time base high-speed acquisition processing unit, the frequency-band data processing and compensation unit, the multi-band data feature analysis unit, the synchronization module, and the communication module.

2. A converter transformer internal discharge fault monitoring system and monitoring method according to claim 1, characterized in that: By using high-frequency pulse current and ultrasonic waves for comprehensive monitoring, different physical quantities radiated by the same discharge source are collected synchronously to obtain fault sample data when the equipment has partial discharge faults; The transmission ratio model is used to determine whether the signal comes from inside the converter transformer. The above signal analysis can basically exclude the influence of external interference signals including bushing end signals and external discharge signals, thus ensuring the accuracy of the monitoring signal. According to the fault sample data, a two-dimensional spectrum corresponding to the fault sample data is generated, and a partial discharge characteristic parameter of the fault sample data is extracted from the two-dimensional spectrum; A multi-source parameter fusion analysis model is established by utilizing their homology and phase relationship to achieve accurate identification of internal discharge and external interference signals.

3. A converter transformer internal discharge fault monitoring system and monitoring method according to claim 1, characterized in that: After using the frequency-division arc current sensor to obtain the original data of multiple signals, the high-speed acquisition and processing unit completes the synchronous high-speed acquisition and processing of pulse signals, low-frequency cluster signals, power frequency and harmonic signals, and uses data processing fusion technology for compensation and data reorganization. At the same time, the multi-band data feature analysis unit can obtain the signal of a certain frequency band separately according to specific requirements, and perform multi-band waveform data feature analysis through edge computing to determine the severity of the internal discharge fault of the converter transformer.

4. A converter transformer internal discharge fault monitoring system and monitoring method according to claim 1, characterized in that: (1) Determine the filter function Hd(ejW); (2) Determine the amplitude function A from Hd(ejW) d (Ω): (3) Determine the filter amplitude-frequency characteristic function by formula 1; (4) Add window to obtain the impulse response of the actual filter; h[k]=h d [k]*W N [k] (3) The window function W N [k] is: (5) Using Formula 3 and Formula 4, the sampled signal x[k] is bandpass filtered to obtain the processed data y[k]: y[k]=x[k]*h[k] (5) (6) Use Formula 5 to separate and calculate y[k] in the frequency domain: In the frequency domain data synthesis implementation, F1S and F1E are the low-end cutoff frequency and high-end cutoff frequency of frequency band 1; F2S, F2E are the low-end cutoff frequency and high-end cutoff frequency of band 2; F3S, F3E are the low-end cutoff frequency and high-end cutoff frequency of band 3; Then perform frequency domain synthesis: Yout(n)=Y1(n)*S1(n)+Y2(n)*S2(n)+Y3(n)*S3(n) (7) Where: Yout(n) is the frequency domain output of the arc current sensor; Y1,2,3(n) are the outputs of the sensor units in the 1st, 2nd, and 3rd frequency bands respectively; S1,2,3(n) are the correction coefficients of the sensors in the 1st, 2nd and 3rd frequency bands respectively; The arc current signal is constructed and implemented by using Formula 7 to transform the data after frequency domain synthesis. The transformed data ensures the sensitivity and dynamic range of each detection frequency band. The arc current signal obtained by frequency domain fusion can be continuously analyzed in time domain and frequency domain, and then feature extraction is realized.

5. A converter transformer internal discharge fault monitoring system and monitoring method according to claim 1, characterized in that: It includes a multi-layer progressive logic alarm strategy: primary judgment is made using waveform recognition, secondary judgment is made using discharge energy changes and amplitude change rate, tertiary judgment is made using the correlation between high-frequency partial discharge current, arc current and power frequency current changes combined with discharge source location information, and final alarm information is output in combination with other physical quantities, which can ensure accurate alarm when destructive discharge or arc occurs inside the converter transformer.

Images