Online phase-by-phase monitoring and evaluation method for local insulation aging defects of converter transformers

Abstract

The present invention proposes an online phase-by-phase monitoring and evaluation method for local insulation aging defects of converter transformers, monitoring the voltage between the beginning of the winding and the ground and the voltage between the neutral point and the ground after the differential mode-common mode (DM-CM) switch timing switching and the common mode-differential mode (CM-DM) switch timing switching of the converter transformer. After the DM-CM switch timing switching, the phase difference of the local insulation aging defect of the converter transformer is determined by observing whether the damped oscillation of the transient time domain waveform of the neutral point-to-ground voltage is significantly changed compared to the inconspicuous change in the healthy state; on the basis of determining the phase difference, the transient frequency domain characteristics of the neutral point-to-ground voltage after the CM-DM switch timing switching are used to evaluate the location and aging degree of the defect. The monitoring method proposed by the present invention is non-invasive, has no additional hardware requirements, and the interruption and insertion timing are flexible and controllable, does not affect the normal operation of the system, and is suitable for converters with different topologies and different switching modulation strategies.

Description

Technical Field

[0001] The invention belongs to the field of electronic device defect detection, and in particular relates to an online phase-by-phase monitoring and evaluation method for local insulation aging defects of a converter transformer. Background Art

[0002] Since operating factors such as electrical stress, thermal fatigue, and water vapor intrusion have different effects on the aging of insulation materials, the aging conditions of various insulation components in the power transformer unit are different, resulting in non-uniform aging. Power transformer through-the-circuit faults often develop from local insulation defects. The key feature of local insulation aging defects is that defects begin to develop in a very small range, and are likely to cause insulation breakdown during long-term operation or after a sudden huge impact. Therefore, it is necessary to take preventive protection against local aging defects of power transformers as soon as possible.

[0003] In modern power systems, power electronic converters are increasingly used. Compared with power transformers in traditional power frequency systems, the local aging of this type of converter transformer connected to the converter is more serious. The main reasons include: 1) Due to the electrical stress of the periodic switching action of the converter, taking into account the working conditions such as polarity reversal, the local discharge phenomenon inside the converter transformer is serious; 2) The current harmonics in the power electronic conversion system are rich, which increases the eddy current and stray losses of the converter transformer core, which may cause abnormal temperature rise or local overheating of the internal conductor and insulation, making the aging distribution more uneven. Therefore, the local insulation aging defects of this type of converter transformer are more likely to evolve into through-going faults. In some special application scenarios, such as high-speed rail traction systems, due to frequent load switching, the temperature fluctuations in the unit make the thermal aging distribution of the insulation system extremely uncertain. Therefore, higher requirements are needed for the local insulation status monitoring of converter transformers.

[0004] At present, the insulation condition monitoring method of converter transformers still follows the long-term standard implementation. It is challenging to locate and evaluate the degree of local insulation aging defects. The key technical difficulty lies in that it does not rely on intrusive sensing units or other external hardware devices. Using conventionally configured measurement units, it can directly and reliably provide detailed information on the internal insulation aging defects of converter transformers, including aging phase, distribution position along the winding, and severity during the development process.

[0005] Traditional methods for online monitoring of transformer insulation status mostly detect additional products of insulation degradation and make judgments on the degree of aging based on signal concentration or intensity. These online test items respectively utilize physical signals (temperature, vibration), chemical signals (dissolved gas in oil), partial discharge signals (sound, light, electromagnetic waves) and electrical signals (current, insulation electrical parameters). Among them, physical signals do not directly reflect the characteristics of insulation aging, chemical signals and partial discharge signals are mostly suitable for oil-immersed transformers, and electrical signals can usually cover all attribute transformer units, but generally belong to the category of relay protection or offline testing, and the accuracy generally cannot meet the needs of online monitoring of insulation defects. At present, the most widely discussed are two types of online monitoring methods based on dissolved gas in oil and partial discharge, as well as fault analysis methods based on electrical quantities, but they also have certain limitations.

[0006] For online monitoring methods of dissolved gas in oil. Since the end of the last century, the detection of dissolved gas in insulating oil has gradually been realized online, becoming the most widely used online means for diagnosing latent insulation defects of transformers. The characteristic gas method, Rogers method and IEC 60599 three-ratio method are the three most basic gas analysis and fault judgment methods. Due to the large number of gas types in the box and the complex fault types, in order to solve the problem that the coding combination and threshold setting rules are single and it is difficult to fully cover the differences in various attribute transformer units and operating environments, the oil gas analysis method combined with artificial intelligence algorithm has been widely studied. This type of method is still based on data-driven fault diagnosis, and the selection method of algorithm weights and coding combination thresholds often lacks a clear physical connection related to insulation aging. The judgment criteria of different algorithms are relatively independent and weakly correlated, and it is essentially difficult to solve the problems of transformer attribute differences and environmental sensitivity. In addition, industrial standards and academic research rarely distinguish between the coding combination and boundary threshold setting methods of gas in oil of converter transformers, and the differences caused by complex operating conditions and environmental factors of the converter system have not been discussed.

[0007] For the online monitoring method of partial discharge. According to GB / T 7354-2018, the online test items for partial discharge include current pulse detection, ultrasonic method and ultra / ultra-high frequency electromagnetic wave method. The main application difficulty of online insulation monitoring technology based on partial discharge signals is that the dimensions of various partial discharge signals are different, and the mechanisms affected by environmental factors are different. Therefore, the verification standards, detection means and calibration methods of partial discharge signals are not strict and unified, and the discharge types are difficult to unify. Usually, artificial intelligence algorithms are relied upon to complete feature recognition. In the past three decades, several scholars around the world have conducted in-depth research on partial discharge positioning technology based on multi-position sensing. By analyzing the spatial transmission characteristics and attenuation laws of different types of partial discharge signals, the detection area on the tank wall that can sensitively reflect the partial discharge position is clarified, and a multi-position sensing optimization configuration scheme is given. This type of method often has several technical defects: 1) There are individual differences, and it is difficult to cover transformer units with different structures and properties;

[0008] 2) High measurement accuracy is required; 3) The system configuration is expensive; 4) The monitoring results usually clarify the local scope of the insulation defects in the box, but cannot clarify the aging phase and component type. The online location technology of insulation defects based on partial discharge signals has not been widely used in power transformer systems. There are also few studies specifically targeting partial discharge monitoring technology for converter transformers, and few studies have discussed the complexity and particularity of online perception, feature recognition and evaluation diagnosis of partial discharge signals under complex operating conditions and environmental factors of converter systems.

[0009] For fault analysis methods based on electrical quantities. Small current faults such as transformer winding inter-turn, winding high-resistance grounding and core multi-point grounding often develop from local insulation defects. When this type of minor fault occurs in the transformer unit, the operating voltage and load current can reflect the fault characteristics. The method of diagnosing the type and degree of faults using electrical quantities at the transformer port is called fault analysis. Power transformer fault analysis methods have been widely discussed, and existing studies have proposed several methods based on different principles. Classification based on electrical quantity characteristics includes the use of power frequency amplitude, power frequency phasor and time domain transient waveform. The power frequency amplitude method usually determines the size of the electrical quantity amplitude and the threshold; the power frequency phasor method uses the symmetrical component method or Park transformation to identify asymmetrical grounding faults and inter-turn faults in the unit using negative sequence components, and has strict requirements on the accuracy and response speed of the signal acquisition unit. The time domain transient waveform method generally uses signal processing technology to extract fault characteristics. Although this type of fault analysis method has been continuously developed and improved over the past few decades, the sensitivity of minor fault detection has gradually increased to achieve more detailed fault classification and local positioning. However, they still belong to the category of fault diagnosis and relay protection of power equipment. Methods based on the power frequency components of voltage and current have difficulty distinguishing the impact of changes in system operating conditions and insulation aging in transformer units. In other words, the power frequency electrical quantities are not sensitive enough to reflect insulation aging problems. Similarly, methods based on time domain transient waveforms can usually only perform fault diagnosis based on the shape or characteristic quantity of the waveform when the internal fault reaches an observable level, and need to rely on complex data processing methods. In addition, considering the interference factors and complex operating conditions in the power electronic commutation system, the extremely small changes in electrical quantities caused by insulation aging are difficult to observe, and fault analysis methods generally do not have the function of online monitoring and preventive protection of converter transformer insulation defects.

[0010] Therefore, there is still a lack of a local defect positioning and quantitative online monitoring technology for the composite insulation system of converter transformers that takes into account both economy and flexibility. Summary of the invention

[0011] Since the switching transient pulse signal in the converter output voltage is rich in harmonics and has a continuous spectrum, it can be used to construct a wideband voltage response. Therefore, in view of the above problems, the present invention proposes an online phase-by-phase monitoring and evaluation method for local insulation aging defects of converter transformers based on active converter switch timing control. The time-frequency domain characteristics of the transient response voltage of the winding neutral point in different switching timings are used to realize online monitoring of the phase where the defect occurs, the distribution position along the winding, and the degree of aging during the development process.

[0012] An online phase-by-phase monitoring and evaluation method for local insulation aging defects of a converter transformer comprises the following steps:

[0013] Step 1: monitor the voltage between the start end of the three-phase winding and the ground and the voltage between the neutral point and the ground after the differential mode-common mode (DM-CM) switching sequence and the common mode-differential mode (CM-DM) switching sequence of the converter transformer;

[0014] Step 2: After the converter transformer differential mode-common mode (DM-CM) switching sequence is switched, the transient time domain waveform of the neutral point-to-ground voltage is analyzed to determine the phase where the local insulation aging defect of the converter transformer occurs;

[0015] Step 3: Based on the phase of the local insulation aging defect determined in step 2, a fast Fourier analysis is performed on the neutral point-to-ground voltage after the common mode-differential mode (CM-DM) switch timing is switched to obtain its amplitude-frequency characteristics in a wide frequency range, and the location and degree of aging of the local insulation aging defect of the converter transformer are evaluated accordingly.

[0016] Furthermore, in step 1, phase-by-phase monitoring of the three-phase winding is implemented within one monitoring cycle, and the voltage between the starting end of the three-phase winding and the ground and the voltage between the neutral point and the ground after the switch state is switched are obtained by executing interruption and insertion control switch timing.

[0017] Furthermore, each monitoring cycle includes six control cycles, each control cycle includes a number of modulation cycles, and the switch timing control is performed by a counter.

[0018] Furthermore, in each control cycle, when the counter steps to the specified number, an interrupt is executed after detecting that the upper bridge arm of one of the first three phases is simultaneously turned off and the lower bridge arm is simultaneously turned on, and the specified DM-CM switch timing of the phase is inserted; in the next control cycle, an interrupt is executed again after detecting that the upper bridge arm of the phase is simultaneously turned off and the lower bridge arm is simultaneously turned on, and the specified CM-DM switch timing of the phase is inserted; similarly, another four interrupts are executed in sequence, and the specified switch timings for monitoring the other two phases are inserted.

[0019] Furthermore, the switch combination states inserted into the control sequence are switched in sequence, and each time the switch state of only one phase bridge arm is changed, which does not affect the normal conduction of each bridge arm of the converter.

[0020] Furthermore, under a healthy state, the transient time-domain waveform of the neutral point-to-ground voltage has a damped oscillation with a smaller amplitude. When the damped oscillation of the transient time-domain waveform of the neutral point-to-ground voltage after the differential mode-common mode (DM-CM) switch timing is switched and changes significantly and observably compared with that under a healthy state, it is determined that the phase has a main insulation defect to the ground; when the damped oscillation of the transient time-domain waveform of the neutral point-to-ground voltage after the differential mode-common mode (DM-CM) switch timing is switched and changes little compared with that under a healthy state, the phase does not have a main insulation defect to the ground.

[0021] Furthermore, on the basis of clarifying the defect phase difference, the parallel resonance point information (f′) of the neutral point to ground voltage amplitude-frequency characteristic in the common mode-differential mode (CM-DM) control switch timing is obtained. z ,M′ z ) Use a linear function to perform linear fitting, which is f′ z =-aM′ z +b,f′ z and M′ z is the frequency and amplitude of the resonance point; the coefficient a is related to the defect location information, the larger a is, the closer to the neutral point; the coefficient b is related to the aging degree information, the smaller b is, the deeper the aging degree is.

[0022] The beneficial effects of the present invention are as follows: the comprehensive monitoring strategy proposed in the present invention is non-invasive, has no additional hardware requirements, and the interruption and insertion timings are flexible and controllable, does not affect the normal operation of the system, and is suitable for converters with different topologies and different switch modulation strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 is the common mode switch state combination; Figure 1 (a) is state (111); Figure 1 (b) is the status (000)

[0024] Figure 2 It is the switch timing control logic for local defect monitoring;

[0025] Figure 3 It is the phase monitoring switch timing; Figure 3 (a) is the A / B / C phase DM-CM switching timing; Figure 3 (b) is the CM-DM switching timing of phase A / B / C;

[0026] Figure 4 These are the implementation steps of the experimental example;

[0027] Figure 5This is the A-phase monitoring switch timing and winding start voltage waveform in the experimental example; Figure 5 (a) is the DM-CM switching timing; Figure 5 (b) is the CM-DM switching timing;

[0028] Figure 6 It is a schematic diagram of the winding structure and taps in the experimental example;

[0029] Figure 7 This is the voltage waveform in the (011→111) switching sequence in the experimental example;

[0030] Figure 8 This is the voltage waveform in the switching sequence (000→100) in the experimental example;

[0031] Fig. 9 It is the frequency domain result of the normalized voltage response in the switching sequence (000→100) in the experimental example;

[0032] Fig.10 is the normalized U in the experimental example N Parallel resonance information quantification results. DETAILED DESCRIPTION

[0033] The present invention is described in detail below in conjunction with the accompanying drawings and embodiments.

[0034] Figure 1 is a common-mode switch combination state of a two-level converter in an embodiment of the present invention. Under any switch modulation strategy, in one carrier cycle, the two-level converter has two common-mode switch combination states 111 and 000, which are recorded as

[0035] Status①(S A ,S B ,S C )=(1,1,1)

[0036] Status②(S A ,S B ,S C )=(0,0,0)

[0037] S A Indicates the switch state of the upper bridge arm of phase A of the two-level converter, S A =1 means the upper bridge arm of phase A is turned on and the lower bridge arm is turned off; S A =0 means the upper bridge arm of phase A is turned off and the lower bridge arm is turned on. A , S B , S C )=(1,1,1) means that the upper bridge arms of the three phases ABC are turned on at the same time and the lower bridge arms are turned off at the same time; (S A , S B , SC )=(0,0,0) means that the upper bridge arms of the three phases ABC are turned off at the same time and the lower bridge arms are turned on at the same time.

[0038] U Ag ,U Bg and U Cg Respectively represent the output voltage of phase A, phase B and phase C of the two-level converter (voltage from the starting end to the ground); U dc represents the DC bus voltage. t represents any time; t1 represents the start time of state ①, and Δt1 represents the duration of state ①.

[0039] In the time domain, when t∈[t1, t1+Δt1], state ① corresponds to U Ag (t) = U Bg (t) = U Cg (t) = U dc / 2. When t∈[t2, t2+Δt2], state ② corresponds to U Ag (t) = U Bg (t) = U Cg (t)=-U dc / 2 。 Corresponding to the frequency domain, there are three-phase common mode voltages with equal amplitude and phase. When the three-phase voltages are not completely consistent, there is a potential difference between any two phases of the converter output voltage, which is defined as the differential mode switch combination state.

[0040] In the two states of state ① and state ②, the three phases of the converter output voltage are consistent, which is defined as the common-mode (CM) switch combination state. Usually, the next state of the CM switch combination state is to switch the upper and lower bridge arm switch states of any phase, which is called the differential-mode (DM) switch combination state. This switch state switching is defined as the common mode-differential mode (CM-DM) switch timing. When the converter switches from the DM switch combination state to the CM switch combination state, this switch state switching is defined as the differential mode-common mode (DM-CM) switch timing.

[0041] The DM-CM switch sequence is used to determine the defective phase, and the CM-DM switch sequence is used to evaluate the location and aging degree of the defect. To achieve phase monitoring of three-phase windings within one monitoring week, the main insulation monitoring of one-phase winding to ground needs to include both the DM-CM switch sequence and the CM-DM switch sequence specified for the corresponding phase. By executing interrupts and inserting control switch sequences, the voltage measurement after the switch state is switched (after interruption) is facilitated. Figure 2As shown in the figure, each monitoring cycle contains six control cycles, each control cycle contains several modulation cycles, and the switch timing control is performed by the counter. In each control cycle, when the counter steps to the specified number, the first 000 state is detected, and interrupt ① is executed to insert the specified A-phase DM-CM switch timing; in the next control cycle, after the 000 state is detected, interrupt ② is executed to insert the specified A-phase CM-DM switch timing; similarly, the other four interrupts are executed in sequence, and the specified switch timing for monitoring the B-phase and C-phase is inserted accordingly.

[0042] Figure 3 (a) and (b) respectively give the DM-CM switching timing and CM-DM switching timing corresponding to phases A / B / C. Each interruption and timing insertion starts from the 000 state and ends at the 000 state, which does not affect the converter's execution of the normal switching timing. The switch combination states inserted into the control timing are switched in sequence. Each switch only changes the switch state of one phase of the bridge arm, which does not affect the normal conduction of each bridge arm of the converter. After each interruption is executed, the voltage between the beginning of the three-phase winding and the neutral point to the ground of the converter transformer is recorded, and the time-frequency domain characteristics of the neutral point to the ground voltage transient signal in the effective switch state in the control timing are selected for analysis.

[0043] Based on a two-level converter and an SPWM switch modulation method, the present invention provides a switch timing control method for realizing comprehensive monitoring of local aging defects of the main insulation of the converter transformer winding to the ground. The essence of this method is to monitor the specified switch trigger state, insert the specified switch timing after the execution interruption, achieve the corresponding monitoring target, and then return to the switch trigger state to continue to execute the conventionally controlled switch timing. Therefore, the switch timing control method for realizing comprehensive monitoring of the main insulation of the converter transformer to the ground is independent of the converter topology and the switch modulation method, and has general applicability. Since insulation aging is a long and slow process, the converter switch timing is intermittently controlled. The duration of each switch state is controllable, and usually tens to hundreds of microseconds can meet the voltage spectrum analysis requirements. Therefore, the method for monitoring the main insulation to the ground based on the converter switch timing control does not affect the normal operation of the converter transformer.

[0044] The following takes the quantitative evaluation of the local aging defects of the main insulation of phase A and the defect location and aging degree as an example, and explains the determination through specific experiments. Figure 4 As shown, the steps of the converter transformer-to-ground main insulation monitoring method based on two-level converter switch timing control include: converter switch timing control, test transformer-to-ground main insulation aging simulation test, signal acquisition and data processing, and test result evaluation.

[0045] (1) Switching timing control

[0046] The switching modulation strategy of the two-level converter adopts the conventional seven-segment SVPWM (Space Vector Width Modulation, SVPWM), and the switching frequency f s =10kHz, powered by a DC power supply, U dc = 400 V. The switch timing interrupt and insertion control program is compiled in the Code Composer Studio code debugger, and the program is burned into the DSP to perform switch timing control of the specified monitoring target on the two-level converter. Figure 5 The waveform of the switch combination state of the upper bridge arm T1, T3 and T5 of the two-level converter and the voltage waveform of the A-phase winding start end to ground are shown. For easy signal capture, each state lasts for a period of T d =500μs.

[0047] (2) Testing

[0048] The two-level converter is connected to the 220V side of the test transformer. Figure 6 As shown in the figure, each phase winding has a double-layer structure, and 5 taps are used to simulate local aging defects at different locations of the main insulation to ground. An adjustable capacitor is connected in parallel between the tap and the ground to simulate different aging degrees. The local aging defect test condition of the main insulation to ground is marked as Tapi-ΔC in the following text. g -Gnd. For example, Tap3-ΔC g (2nF)-Gnd indicates that a 2nF capacitance to ground is connected in parallel at the winding tap 3 position.

[0049] Use a differential probe to test the voltage U between the beginning of the A-phase winding and the ground Ag And the neutral point to ground voltage U N , the differential probe bandwidth is 5MHz, the oscilloscope sampling window length is 10ms, and the sampling frequency is 125MS / s. The aging simulation test includes: During the A-phase DM-CM control switch timing, the A / B / C phase Tap3-ΔC g (2nF)-Gnd, complete the simulation test of local aging defects of 4 groups of phase-to-ground main insulation (including healthy conditions); within the A-phase CM-DM control switch timing, change the adjustable capacitance value three times at 5 taps respectively, and complete the simulation test of local aging defects of 16 groups of A-phase-to-ground main insulation (including healthy conditions).

[0050] (3) Data processing

[0051] Phase-by-phase monitoring of local aging defects of ground main insulation is realized based on the time-frequency characteristics of the neutral point response voltage. The effective switching transient voltage is selected in the A-phase DM-CM control switch sequence and the CM-DM control switch sequence, that is, the neutral point voltage time domain waveform after switching (011→111) and (000→100), and the transient voltage sampling window is 50μs long. The original signal is subjected to wavelet denoising (5-layer decomposition), and then the voltage transient after denoising is subjected to fast Fourier transform to obtain U N Amplitude-frequency characteristics. Considering the change of the DC bus voltage of the converter during system operation, U N The amplitude-frequency characteristic is normalized relative to the steady-state voltage in this switching state.

[0052] (4) Evaluation

[0053] The aging law of local defects in the ground main insulation is that after the defect appears at a certain position along the winding, the aging degree gradually deepens. N The amplitude-frequency characteristics determine the defect phase, and then based on the CM-DM switch timing, the N The amplitude-frequency characteristics are used to evaluate the location of local defects and the degree of aging during the development process. N The time domain waveform is used to qualitatively judge whether there is a change in the equivalent electrical parameters of the insulation of the A phase system to the ground. The normalized U N The movement law of the parallel resonance point of the amplitude-frequency characteristic is used to quantitatively evaluate the defect location and aging degree. The frequency and amplitude of the resonance point are denoted as f′ respectively. z and M′ z .

[0054] Experimental Results

[0055] First, the effectiveness of phase-by-phase monitoring using the neutral point response voltage in the DM-CM switching sequence is verified. In the (011→111) switching sequence, the A / B / C phase windings of the test transformer are respectively subjected to Tap3-ΔC. g (2nF)-Gnd aging simulation test. Figure 7 As shown, under healthy conditions, U N The time domain waveform is black, and there is a damped oscillation with a small amplitude. In the 111 switch combination state, there is an aging defect in the insulation of the A phase to the ground, which leads to the U N The damped oscillation of the phase B or phase C has obvious observable changes; in the case of aging defects, U N The damped oscillation is not significantly different from the healthy condition. Therefore, by specifying the switching timing within U N The time domain characteristics can directly determine the defect phase, and no frequency domain transformation is required. After the defect phase is clearly identified, the switch timing is controlled according to the CM-DM of the corresponding phase. NThe amplitude-frequency characteristics are used to determine the defect distribution location along the winding and the degree of aging.

[0056] Secondly, the location and aging degree of local defects are quantitatively evaluated based on the amplitude-frequency characteristics of the neutral point response voltage in the CM-DM switching sequence. In the (000→100) switching sequence, the local aging defects of the A-phase winding to the ground main insulation are evaluated at different locations along the winding and at the same aging degree (ΔC g =2nF) aging simulation test. Figure 8 Six groups of tests were given. N Time domain waveform of Tap1-ΔC g (2nF)-Gnd and Tap2-ΔC g u corresponding to (2nF)-Gnd N Waveform and U in health status N The waveforms are basically the same. Under the same fault severity, the closer the defect location is to the neutral point, the greater the U N The longer the transient oscillation time is. N Time domain features cannot distinguish the influence of local defect location and aging degree, which means that time domain features cannot be used for positioning and quantitative evaluation.

[0057] Next, the CM-DM switch timing is N The time domain waveform is denoised and then analyzed in the frequency domain. Fig. 9 The normalized U values ​​of Tap3, Tap4 and Tap5 at different aging degrees are given respectively. N Spectrum comparison results. Considering the dispersion of the switching transient voltage analysis results, Fig. 9 The spectrum presented is in average mode, i.e., the voltage transients under the same switching state are measured 10 times, and the 10 times are normalized to U N The spectrum is averaged and the obviously abnormal data is removed. When the position of the local defect remains unchanged, as the aging degree deepens, f′ z Gradually decreases, M′ z ′ gradually increases. After the local insulation defect of the converter transformer winding to the ground occurs, the distribution position along the winding is fixed, and the aging degree of the defect gradually deepens during operation. Therefore, according to U N Parallel resonance information (f′ z , M′ z ) change rules, it can distinguish the respective influences of defect location and defect development degree, and can further realize positioning and quantitative evaluation.

[0058] like Fig.10 As shown in the figure, after the occurrence of three groups of local defects, as the aging degree deepens, (f′ z ,M′ z )The moving curve is linearly distributed. According to the real-time measurement (f′ z ,M′z ), use a linear function for linear fitting, written as f′ z =-aM′ z +b form. The coefficient a is related to the defect location information. The larger a is, the closer it is to the neutral point. The coefficient b is related to the aging degree information. The smaller b is, the deeper the aging degree is. During the operation of the converter transformer, by analyzing the U N Amplitude-frequency characteristics, according to (f′ z ,M′ z )The continuous change trajectory can realize the quantitative evaluation of the location of local defects and the degree of aging development of the winding-to-ground main insulation.

[0059] This embodiment is only a preferred specific implementation of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for online phase-by-phase monitoring and evaluation of local insulation aging defects of a converter transformer, comprising the following steps: Step 1: monitor the voltage between the start end of the three-phase winding and the voltage between the neutral point and the ground after the differential mode-common mode DM-CM switch sequence switching and the common mode-differential mode CM-DM switch sequence switching of the converter transformer; Step 2: After the differential mode-common mode DM-CM switch timing of the converter transformer is switched, the transient time domain waveform of the neutral point voltage to the ground is analyzed to determine the phase where the local insulation aging defect of the converter transformer occurs; under a healthy state, the transient time domain waveform of the neutral point voltage to the ground has a damped oscillation with a small amplitude. When the damped oscillation of the transient time domain waveform of the neutral point voltage to the ground after the differential mode-common mode DM-CM switch timing is switched has a significant observable change compared to the healthy state, it is determined that the phase has a main insulation defect to the ground; when the damped oscillation of the transient time domain waveform of the neutral point voltage to the ground after the differential mode-common mode DM-CM switch timing is switched does not change significantly compared to the healthy state, then the phase does not have a main insulation defect to the ground; Step 3: Based on the determination of the phase of the local insulation aging defect in step 2, a fast Fourier analysis is performed on the neutral point-to-ground voltage after the common mode-differential mode CM-DM switch sequence is switched, and the parallel resonance point information (f z ′,M z ′) A linear function is used for linear fitting to obtain its amplitude-frequency characteristics in a wide frequency range, based on which the location and degree of aging of local insulation aging defects of the converter transformer are evaluated; the fitting curve is f z ′=-aM z ′+b,f z ′ and M z ′ is the frequency and amplitude of the resonance point; the coefficient a is related to the defect location information, the larger a is, the closer to the neutral point; the coefficient b is related to the aging degree information, the smaller b is, the deeper the aging degree is.

2. The method for online phase-by-phase monitoring and evaluation of local insulation aging defects of converter transformers according to claim 1 is characterized in that: In step 1, phase-by-phase monitoring of the three-phase winding is implemented within one monitoring cycle, and the voltage between the starting end of the three-phase winding and the ground and the voltage between the neutral point and the ground after the switch state is switched are obtained by executing interruption and insertion control switch timing.

3. The method for online phase-by-phase monitoring and evaluation of local insulation aging defects of converter transformers according to claim 2 is characterized in that: Each monitoring cycle includes six control cycles, each control cycle includes several modulation cycles, and the switch timing control is performed by a counter.

4. The method for online phase-by-phase monitoring and evaluation of local insulation aging defects of converter transformers according to claim 3 is characterized in that: In each control cycle, when the counter steps to a specified number, an interrupt is executed after detecting that the upper bridge arm of one of the first three phases is simultaneously turned off and the lower bridge arm is simultaneously turned on, and a DM-CM switch timing sequence of the specified phase is inserted; in the next control cycle, an interrupt is executed again after detecting that the upper bridge arm of the phase is simultaneously turned off and the lower bridge arm is simultaneously turned on, and a CM-DM switch timing sequence of the specified phase is inserted; Similarly, another four interruptions are executed in sequence to insert the specified switching timings for monitoring the other two phases.

5. The method for online phase-by-phase monitoring and evaluation of local insulation aging defects of converter transformers according to claim 3 is characterized in that: The switch combination states inserted into the control sequence are switched in sequence. Each time the switch state of only one phase bridge arm changes, the normal conduction of each bridge arm of the converter is not affected.

Images