A method for identifying typical insulation defect discharge patterns under complex electrical stress on the valve side of converter transformers

Abstract

This invention discloses a method for identifying discharge patterns of typical insulation defects under complex electrical stress on the valve side of a converter transformer. The method includes the following steps: first, collecting partial discharge signals from the valve side of the converter transformer, recording the amplitude and phase of each partial discharge with a period of T. Acquisition is stopped when a discharge characteristic exceeds the normal value required by on-site operation and maintenance. Second, the discharge amplitude is divided into intervals and the number of discharges in each interval is calculated. Third, the number of discharges in each interval is normalized. Fourth, the data features of the normalized number of discharges are extracted. Fifth, the discharge pattern is identified. This method, while taking into account the complex electrical stresses of power-frequency AC, DC, and high-proportion harmonics on the valve side of the converter transformer, identifies discharge patterns of typical insulation defects by extracting PRPD spectrum features, facilitating targeted measures for different insulation defects during actual operation.

Description

Technical Field

[0001] The present invention belongs to the field of high voltage and insulation technology, and particularly relates to a method for identifying discharge patterns of typical insulation defects under complex electrical stress on the valve side of a converter transformer. Background Art

[0002] Converter transformers are core equipment in ultra-high voltage direct current (UHVDC) transmission projects and crucial components connecting the AC and DC systems. Their primary (grid-side) windings connect to the AC transmission system, while their secondary (valve-side) windings connect to the DC converter valves. They perform the crucial tasks of voltage conversion and isolation between the AC and DC systems. Therefore, the reliable operation of converter transformers is crucial for the safety and stability of the entire transmission system. State Grid Corporation of China conducted statistics on converter transformer failures at various voltage levels from 2006 to 2015 and found that valve-side winding failures accounted for 76% of all failures, with partial discharge (PD) caused by insulation defects being a significant contributing factor. Researchers at home and abroad have developed a series of reliable pattern recognition methods using machine learning methods such as artificial neural networks and support vector machines, as well as cluster analysis (fuzzy clustering, Markov clustering, and K-means clustering) and fuzzy recognition.

[0003] However, existing research faces the following major challenges: First, existing pattern recognition methods primarily target partial discharges under DC, power-frequency AC, and AC / DC combined voltages, failing to consider the complex electrical stresses associated with high harmonic overlays. Therefore, they are difficult to apply to discharge scenarios involving valve-side insulation defects in converter transformers. Second, in actual engineering sites, the longitudinal insulation voltage waveform on the valve side of converter transformers cannot be directly obtained, making it difficult to establish a phase reference for discharge data, which complicates pattern recognition. Typical valve-side insulation defects in oil-immersed converter transformers include oil-gap discharges without solid insulation, oil-paper insulation discharges with oil gaps, and oil-paper insulation discharges without oil gaps. Therefore, effectively identifying specific discharge patterns is crucial. Summary of the Invention

[0004] To solve the problems existing in the above-mentioned prior art, the purpose of the present invention is to provide a method for identifying discharge patterns of typical insulation defects under complex electrical stress on the valve side of a converter transformer. The method of the present invention realizes the identification of discharge patterns of typical insulation defects by extracting PRPD spectrum features under the complex electrical stress of power frequency AC, DC, and high-proportion harmonics on the valve side of the converter transformer, thereby facilitating the adoption of targeted measures for different insulation defects in actual operation.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A method for identifying typical insulation defect discharge patterns under complex electrical stress on the valve side of a converter transformer comprises the following steps:

[0007] Step 1: Collect partial discharge information

[0008] Collect partial discharge signals on the valve side of the converter transformer, record the amplitude and phase of each partial discharge with a period of T, and stop collecting when a discharge characteristic exceeds the normal value required by on-site operation and maintenance;

[0009] Step 2: Divide the discharge amplitude interval and calculate the number of discharges in each interval

[0010] Extract the maximum discharge amplitude U among all partial discharges m , with U m / N a Divide the discharge amplitude interval into spans and calculate the number of discharges N in each interval i ,i=1,2,....N a , N a Maximum interval number;

[0011] Step 3: Normalize the number of discharges in each interval;

[0012] Extract the maximum value N of the discharge times in each interval im , the normalized number of discharges n i for:

[0013]

[0014] Obtain the normalized data n of the number of discharges in each discharge amplitude interval i ,i=1,2,....N a ;

[0015] Step 4: Extract data features of normalized discharge times

[0016] With the kth group as the boundary, the discharge times of group 1 to group k are normalized data n i， The maximum value in is n z , where i=1,2,...k; group k to group N a Normalized data of discharge times of group n i The maximum value in is n y , where i = k,...N a ; k∈[2,N a -1], and k is the group number with the smallest normalized discharge number except the first and last groups;

[0017] Step 5: Identify the discharge pattern

[0018] When n z When ≠1, the discharge mode is pure oil gap discharge;

[0019] When n z =1, and F <n yWhen <1, the discharge mode is oil-paper insulation discharge with oil gap;

[0020] When n z =1, and 0 <n y When ≤F, the discharge mode is oil-paper insulation discharge without oil gap.

[0021] Where F is the pattern discrimination threshold, F∈(0,1).

[0022] Preferably, in the first step, a certain discharge characteristic quantity exceeds the normal value required by on-site operation and maintenance, which means that the cumulative number of discharges exceeds the cumulative limit value N of the discharge number, the cumulative discharge time exceeds the cumulative limit value t of the discharge time, or the acetylene content exceeds the acetylene content limit value P.

[0023] Preferably, in the first step, the method for collecting the partial discharge signal on the valve side of the converter transformer adopts the pulse current method, the ultra-high frequency method or the ultrasonic method. The pulse current method has the advantages of high sensitivity and strong anti-interference ability; the ultra-high frequency method has good anti-corona performance and is suitable for online monitoring of partial discharge; the ultrasonic method is simple to operate on site and convenient to use; the partial discharge signal collection form adopts continuous collection or triggered collection. Continuous collection can obtain all subtle discharge signals and improve the sensitivity of pattern recognition; triggered collection can filter out part of the signal interference caused by environmental reasons and improve the reliability of pattern recognition.

[0024] Preferably, in the first step, T is an integer multiple of 20ms; N ≥ 100. T being an integer multiple of 20ms, i.e., an integer multiple of the power frequency period, facilitates dividing the amplitude range of the discharge signal within one or more cycles; N ≥ 100, i.e., only performing judgment and identification when there are more than 100 discharge points, is intended to reduce the impact of randomness of partial discharges and improve the reliability of the identification results.

[0025] Preferably, the maximum interval number N a ∈[5,20], when the number of amplitude divisions is greater than or equal to 5, the characteristics of the overall discharge amplitude distribution can be better reflected, and N a The larger the amplitude distribution feature, the more obvious it is, and the higher the recognition accuracy is; i is the interval number, and the discharge amplitude interval of the i-th group is [U m / N a *(i-1),U m / N a *i].

[0026] Compared with the prior art, the present invention has the following advantages:

[0027] By extracting PRPD spectrum features, this method identifies typical insulation defect discharge patterns, taking into account the complex electrical stresses of power-frequency AC, DC, and high-proportion harmonics on the valve side of converter transformers. This facilitates targeted measures for different insulation defects during actual operation. Compared with existing technologies, this method does not require a large number of training samples, requiring only a small amount of data to determine the field pattern discrimination threshold F. Furthermore, this method takes into account the complex electrical stresses of high-proportion harmonics on the valve side of converter transformers, making it more suitable for converter transformer valve-side insulation defect discharge scenarios than existing technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a flow chart of the method of the present invention.

[0029] Figure 2 It is the normalized data change curve of the number of discharges, where (a) is oil insulation, (b) is oil-paper insulation without oil gap, and (c) is oil-paper insulation with oil gap. DETAILED DESCRIPTION

[0030] The present invention will be further described in detail below with reference to the accompanying drawings and examples. The following examples are only used to more clearly illustrate the technical solutions of the present invention and are not intended to limit the scope of protection of the present invention.

[0031] The following introduces an implementation case using the longitudinal insulation discharge of the high-end YY connection on the valve side of the converter transformer as an example.

[0032] like Figure 1 As shown, the present invention provides a method for identifying typical insulation defect discharge patterns under complex electrical stress on the valve side of a converter transformer, comprising the following steps:

[0033] Step 1: Collect partial discharge information

[0034] Collect partial discharge signals from the valve side of the converter transformer, recording the amplitude and phase of each partial discharge with a period of T. When a discharge characteristic exceeds the normal value required by on-site operation and maintenance (including but not limited to the cumulative number of discharges exceeding the cumulative discharge number limit N, the cumulative discharge time exceeding the cumulative discharge time limit t, and the acetylene content exceeding the acetylene content limit P), stop collecting and proceed to the second step. Partial discharge signal collection methods include but are not limited to pulse current, ultra-high frequency, and ultrasonic methods. Partial discharge signal collection methods include but are not limited to continuous and triggered collection. T is an integer multiple of 20ms, and N ≥ 100.

[0035] In this example, T=20ms, and the discharge characteristic value limit is selected as N=1000.

[0036] Step 2: Divide the discharge amplitude interval and calculate the number of discharges in each interval

[0037] Extract the maximum discharge amplitude U among all partial discharges m , with U m / N a Divide the discharge amplitude interval into spans and calculate the number of discharges N in each interval i ,(i=1,2,....N a ). Among them, N a ∈[5,20], i is the interval number, the discharge amplitude interval of the i-th group is [U m / N a *(i-1),U m / N a *i].

[0038] In this example: N a =10.

[0039] Step 3: Normalize the number of discharges in each interval

[0040] Extract the maximum value N of the discharge times in each interval im , the normalized number of discharges n i for:

[0041]

[0042] Obtain the normalized data n of the number of discharges in each discharge amplitude interval i ,(i=1,2,....N a ).

[0043] In this example: the normalized data of discharge times for each insulation defect n i (i=1,2,...,10)such as Figure 2 As shown in (a), (b) and (c).

[0044] Step 4: Extract data features of normalized discharge times

[0045] With the kth group as the boundary, the discharge times of group 1 to group k are normalized data n i The maximum value among (i=1,2,...k) is n z , group k to group N a Normalized data of discharge times of group n i (i=k,...N a ) is the maximum value n y . Where: k∈[2,N a -1], and k is the group number with the smallest normalized discharge number except the first and last groups.

[0046] In this example: For oil insulation, k = 2, and we get n z =0.02308, n y=1;

[0047] For oil-paper insulation without oil gap, k = 9, and n z =1,n y =0.00357;

[0048] For oil-paper insulation with oil gap, k = 7, and n is obtained z =1,n y =0.12367.

[0049] Step 5: Identify the discharge pattern

[0050] When n z When ≠1, the discharge mode is pure oil gap discharge;

[0051] When n z =1, and F <n y When <1, the discharge mode is oil-paper insulation discharge with oil gap;

[0052] When n z =1, and 0 <n y When ≤F, the discharge mode is oil-paper insulation discharge without oil gap.

[0053] Where F is the mode discrimination threshold, F∈(0,1). In this example: F=0.12, for oil insulation, n z =0.02308≠1, so the identification result is pure oil gap discharge;

[0054] For oil-paper insulation without oil gap, n z =1, and 0 <n y =0.00357≤0.12, so the identification result is oil-paper insulation discharge without oil gap;

[0055] For oil-paper insulation with oil gap, n z =1, and 0.12 <n y =0.12367<1, so the identification result is oil-paper insulation discharge with oil gap.

Claims

1. A method for identifying typical insulation defect discharge patterns under complex electrical stress on the valve side of a converter transformer, characterized in that: The steps include: Step 1: Collect partial discharge information Collect partial discharge signals on the valve side of the converter transformer, record the amplitude and phase of each partial discharge with a period of T, and stop collecting when a discharge characteristic exceeds the normal value required by on-site operation and maintenance; Step 2: Divide the discharge amplitude interval and calculate the number of discharges in each interval Extract the maximum discharge amplitude U among all partial discharges m , with U m / N a Divide the discharge amplitude interval into spans and calculate the number of discharges N in each interval i ,i=1,2,....N a , N a Maximum interval number; Step 3: Normalize the number of discharges in each interval; Extract the maximum value N of the discharge times in each interval im , the normalized number of discharges n i for: Obtain the normalized data n of the number of discharges in each discharge amplitude interval i ,i=1,2,....N a ; Step 4: Extract the data features of normalized discharge times; With the kth group as the boundary, the discharge times of group 1 to group k are normalized data n i， The maximum value in is n z , where i=1,2,...k; group k to group N a Normalized data of discharge times of group n i The maximum value in is n y , where i = k,...N a ; k∈[2,N a -1], and k is the group number with the smallest normalized discharge number except the first and last groups; Step 5: Identify the discharge pattern; When n z When ≠1, the discharge mode is pure oil gap discharge; When n z =1, and F <n y When <1, the discharge mode is oil-paper insulation discharge with oil gap; When n z =1, and 0 <n y When ≤F, the discharge mode is oil-paper insulation discharge without oil gap; Where F is the pattern discrimination threshold, F∈(0,1).

2. The method for identifying typical insulation defect discharge patterns under complex electrical stress on the valve side of a converter transformer according to claim 1 is characterized in that: In the first step, a discharge characteristic quantity exceeds the normal value required by on-site operation and maintenance, which means that the cumulative number of discharges exceeds the cumulative limit value N of the discharge number, the cumulative discharge time exceeds the cumulative limit value t of the discharge time, or the acetylene content exceeds the acetylene content limit value P.

3. The method for identifying typical insulation defect discharge patterns under complex electrical stress on the valve side of a converter transformer according to claim 1 is characterized in that: In the first step, the method for collecting the partial discharge signal on the valve side of the converter transformer adopts the pulse current method, the ultra-high frequency method or the ultrasonic method; the partial discharge signal collection form adopts continuous collection or triggered collection.

4. The method for identifying typical insulation defect discharge patterns under complex electrical stress on the valve side of a converter transformer according to claim 1 is characterized in that: In the first step, T is an integer multiple of 20ms; N ≥ 100.

5. The method for identifying typical insulation defect discharge patterns under complex electrical stress on the valve side of a converter transformer according to claim 1 is characterized in that: Maximum interval number N a ∈[5,20], i is the interval number, the discharge amplitude interval of the i-th group is [U m / N a *(i-1),U m / N a *i].

Images