Lightning stroke fault tracing method for wind farm power collection system based on comprehensive feature analysis

Abstract

The wind farm power collection system lightning stroke fault tracing method based on comprehensive feature analysis disclosed in the application comprises feature quantity calculation on voltage / current waveform data under lightning stroke of real-time monitoring of high and low voltage side sensors of a box-type transformer to obtain key feature parameters representing lightning stroke faults; transformation into dimensionless evaluation scores by interval mapping functions to obtain feature scores under three types of faults; weight allocation to feature scores under each type of fault and calculation of total scores of corresponding fault types, and judgment of fault types to which lightning stroke events belong according to the total scores. The application can directly distinguish three types of faults, i.e. side flash, direct lightning and ground net counterattack, by quantitative criteria, solve the misjudgment problem caused by fault mixing in traditional methods, help improve the accuracy of lightning protection and fault diagnosis, and then quickly locate fault points according to fault types, reduce downtime of wind farms, and provide strong guarantee for safe and stable operation of power systems.

Description

Technical Field

[0001] This invention belongs to the technical field of wind farm operation and maintenance methods in mountainous areas, specifically involving a method for tracing the source of lightning strike faults in wind farm power collection systems based on comprehensive feature analysis. Background Technology

[0002] With the rapid expansion of wind farms worldwide, the failure rate of box-type transformers in power collection systems due to lightning strikes has increased significantly. Accurate tracing of lightning-induced faults has become crucial for the intelligent operation and maintenance of wind farms. Traditional lightning strike location technologies (such as those based on traveling wave time difference) typically rely on single signal analysis for fault inference, failing to quantify the dynamic correlations of multi-dimensional features such as waveform steepness, frequency, and polarity. This results in insufficient tracing accuracy under complex operating conditions. Furthermore, traditional criteria have fixed weights, making it impossible to dynamically adjust classification thresholds based on environmental factors such as changes in wind farm grounding impedance and soil moisture fluctuations, thus hindering operational efficiency. Therefore, there is an urgent need for a lightning strike fault tracing method that can integrate multiple signal features and quantify the correlations between differences to improve the reliability of lightning protection systems in mountainous wind farms. Summary of the Invention

[0003] The purpose of this invention is to provide a method for tracing the source of lightning strike faults in wind farm collection systems based on comprehensive feature analysis, which solves the problem of insufficient tracing accuracy under complex operating conditions of existing lightning protection systems for mountainous wind farms.

[0004] The technical solution adopted in this invention is: a method for tracing the source of lightning strike faults in wind farm power collection systems based on comprehensive feature analysis, comprising the following steps:

[0005] Step 1: Based on the time-frequency characteristic differences, perform characteristic quantity calculation on the voltage / current waveform data under lightning strike monitored in real time by the sensors on the high and low voltage sides of the box-type transformer to obtain the key characteristic parameters characterizing the lightning strike fault;

[0006] Step 2: The key feature parameters obtained in Step 1 are converted into dimensionless evaluation scores through an interval mapping function to obtain the feature scores of the high and low voltage side voltage / current waveforms of the box transformer under three fault types (flashover fault, direct lightning fault and shared grounding grid backflash fault).

[0007] Step 3: Assign weights to the feature scores of each fault type obtained in Step 2 and calculate the total score of the corresponding fault type. Determine the fault type of the lightning strike event based on the total scores of the three fault types.

[0008] The invention is further characterized in that,

[0009] In step 1, the electrical characteristics of the high- and low-voltage sides of the box-type transformer under lightning strike are identified and calibrated by analyzing the voltage / current waveforms. Then, the key characteristic parameters characterizing the lightning strike fault are calculated using the following formulas. P1. P n , C , f m , I g , R :

[0010] 1) The ratio of the peak value of a certain phase voltage to the peak values ​​of the other two phases. P 1:

[0011]

[0012] In the formula, It is the maximum value of the three-phase voltage peak value; It is the peak value of a single-phase voltage. A, B, and C are used to distinguish the different three phases in the voltage waveform.

[0013] 2) Multiphase interference ratio P n :

[0014]

[0015] In the formula, σ The standard deviation of the three-phase voltage amplitude is calculated using the following formula:

[0016]

[0017] In the formula, n A value of 3 indicates three-phase voltage; U peak,i This represents the amplitude of a certain phase voltage.

[0018] 3) Correlation coefficients between three-phase voltage waveforms C :

[0019]

[0020] In the formula, r The correlation coefficient between two phase voltages is calculated using the following formula:

[0021]

[0022] In the formula, N This represents the number of instantaneous voltage values ​​taken from the fault voltage waveform. X i and Y i This represents the instantaneous values ​​of the voltages taken in different two phases. , These represent the average values ​​of all instantaneous voltage values ​​taken for the corresponding phase.

[0023] 4) The dominant frequency of the low-voltage side voltage waveform f m :

[0024]

[0025] In the formula, It is the spectral amplitude of the low-voltage side voltage signal, calculated using Fourier transform:

[0026]

[0027] In the formula, u ( t ) indicates the low-voltage side voltage waveform.

[0028] 5) Peak grounding current I g The maximum value of the grounding grid current under lightning strike is directly taken.

[0029] 6) The attenuation consistency ratio between the voltage difference across the insulator and the waveform on the high-voltage side of the box-type transformer. R :

[0030]

[0031] In the formula, α The attenuation exponent is obtained by fitting the waveform attenuation process:

[0032]

[0033] In the formula, U 0 represents the initial value of the voltage difference across the insulator on the tower. U ( t () indicates the voltage difference across the insulator.

[0034] Step 2 specifically involves: calculating the key characteristic parameters representing lightning strike faults obtained in Step 1. P 1. P n , C , f m , I g , R After normalization using different thresholds and compression to the range of [0,1], feature scores of the high-voltage and low-voltage side voltage / current waveforms of the box-type transformer under three fault types are obtained, i.e., high-voltage side scores. S 1. Low-pressure side score S 2. Score at the grounding grid S 3. Score at the insulator S 4:

[0035] High-voltage side scores under three fault typesS The formulas for calculating 1 are as follows:

[0036]

[0037] Low-voltage side scores under three fault types S The formulas for calculating 2 are as follows:

[0038]

[0039] Scores at the grounding grid under three fault types S The formulas for calculating 3 are as follows:

[0040]

[0041] Scores at the insulator under three fault types S The formulas for calculating 4 are all:

[0042]

[0043] Step 3 yields the feature scores for each fault type. S 1. S 2. S 3. S 4. Feature scores for each fault type. S 1. S 2. S 3. S 4. Assign weights ω The total score for each fault type is calculated using the following formula:

[0044]

[0045] In the formula, ω Feature Score S 1. S 2. S 3. S A weight of 4.

[0046] The beneficial effects of this invention are as follows: The method for tracing the source of lightning strikes in wind farm collection systems based on comprehensive feature analysis can directly distinguish between three types of faults—flashover, direct lightning strike, and grounding grid backflash—through quantitative criteria. This solves the problem of misjudgment caused by mixed faults in traditional methods, helps improve the accuracy of lightning protection and fault diagnosis, and enables rapid location of fault points based on fault type, reducing wind farm downtime and providing strong protection for the safe and stable operation of the power system. Attached Figure Description

[0047] Figure 1This is a schematic diagram of the coupling path between the tower and the box-type transformer in the wind farm collector system and the conduction through the grounding grid during lightning current intrusion in an embodiment of the present invention;

[0048] Figure 2 This is a simulation waveform diagram of the bus voltage in an embodiment of the present invention;

[0049] Figure 3 This is a simulated waveform diagram of the fault current of wind turbine line No. 1 in an embodiment of the present invention;

[0050] Figure 4 This is a simulation waveform diagram of the fault current of the No. 2 wind turbine line in an embodiment of the present invention. Detailed Implementation

[0051] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0052] Example 1

[0053] This invention provides a method for tracing the source of lightning strike faults in wind farm power collection systems based on comprehensive feature analysis. It is specifically used to identify three types of lightning strike faults that frequently occur in the power collection systems of mountainous wind farms: lightning strike around the source (fault 1), lightning strike directly (fault 2), and backflashover fault of shared grounding grid (fault 3).

[0054] Example 2

[0055] This invention provides a method for tracing the source of lightning strike faults in wind farm power collection systems based on comprehensive feature analysis. By analyzing multiple features of the voltage / current waveforms on the high and low voltage sides of the box-type transformer during a lightning strike, and combining characteristic indicators such as single-phase peak ratio, multi-phase interference ratio, low-voltage side spectral characteristics, and peak grounding current, a weighted scoring method is used to comprehensively determine the type of lightning strike fault. Through quantitative criteria, it can effectively distinguish between fault types such as lightning strikes around phase conductors, direct strikes to towers / lightning conductors, and backflashover from the shared grounding grid, providing a scientific basis for fault determination and helping to improve the accuracy of lightning protection and fault diagnosis.

[0056] Example 3

[0057] This invention provides a method for tracing the source of lightning strike faults in wind farm power collection systems based on comprehensive feature analysis, comprising the following steps:

[0058] Step 1: When a lightning strike occurs, the voltage / current waveform data monitored in real time by the sensors on the high and low voltage sides of the box-type transformer are used to calculate the characteristic quantities based on the time-frequency characteristic differences, so as to obtain the key characteristic parameters characterizing the lightning strike fault.

[0059] Step 2: Construct a standardized scoring system. Use an interval mapping function to convert each feature parameter into a dimensionless evaluation score (normalized to the [0,1] interval) to eliminate the impact of dimensional differences on subsequent fault classification.

[0060] Step 3: Use a multi-dimensional evaluation model to identify faults, dynamically allocate feature weight coefficients, and combine them with standardized scoring formulas for weighted comprehensive evaluation, so as to achieve accurate identification of three types of lightning strike faults.

[0061] Example 4

[0062] This invention provides a method for tracing the source of lightning strike faults in wind farm power collection systems based on comprehensive feature analysis, comprising the following steps:

[0063] Step 1: By conducting a detailed analysis of the high and low voltage / current waveforms of the box-type transformer under lightning strike, identify and calibrate the electrical characteristics under lightning strike, and calculate the corresponding values ​​(characteristic parameters) of the corresponding characteristics.

[0064] Step 2: After obtaining the electrical characteristic parameters related to lightning strike faults, these parameters are converted into standardized numerical scores. To avoid unfair impacts on the final judgment results due to differences in the dimensions of the features, each characteristic parameter is normalized so that they are compared under the same standard. The normalized characteristic parameters are compressed to the range of [0,1] to ensure they have the same quantization scale. The calculation of feature scores aims to quantify the contribution of each feature to the lightning strike fault determination, thereby providing support for subsequent fault classification.

[0065] Step 3: In the previous steps, all characteristics of the voltage / current waveforms on the high and low voltage sides of the box-type transformer were converted into standardized scores. These scores reflect the electrical characteristics under different lightning fault types on a uniform quantitative scale. However, different characteristics have varying degrees of influence on lightning fault determination; some characteristics may be more discriminative under certain fault types, while others may be less important. Therefore, it is necessary to assign different weights to each characteristic to reflect its contribution to the determination result.

[0066] Example 5

[0067] This invention provides a method for tracing the source of lightning strike faults in wind farm power collection systems based on comprehensive feature analysis, which is implemented according to the following steps:

[0068] Step 1: Analyze the voltage / current transient waveforms captured by the high and low voltage side sensors of the transformer substation through the feature calculation module, and calculate the fault waveforms according to the feature formula to obtain the core parameters.

[0069] Step 1.1: Calculate the single-phase peak ratio P 1: The ratio of the peak value of a certain phase voltage on the high-voltage side to the peak values ​​of the other two phases reflects whether the single-phase characteristics are significant.

[0070] (1)

[0071] in: It is the maximum value of the three-phase voltage peak value; It is the peak value of a single-phase voltage. A, B, and C are used to distinguish the different three phases in the voltage waveform.

[0072] Step 1.2: Calculate the multiphase interference ratio P n The amplitude difference between the three-phase voltages on the high-voltage side reflects the degree of multiphase interference.

[0073] (2)

[0074] in σ The standard deviation of the three-phase voltage amplitude is calculated using the following formula:

[0075] (3)

[0076] in, n A value of 3 indicates three-phase voltage; U peak,i This represents the amplitude of a certain phase voltage.

[0077] Step 1.3: Calculate the three-phase voltage synchronization. C The correlation coefficient between the three-phase voltage waveforms on the high-voltage side is used to assess whether the three phases are synchronously offset.

[0078] (4)

[0079] In the formula r The correlation coefficient between the two-phase voltages is calculated using the following formula:

[0080] (5)

[0081] in, N This represents the number of instantaneous voltage values ​​taken from the fault voltage waveform. X i and Y i Indicates the instantaneous values ​​of the voltages taken in two different phases (e.g.) U A and U B ), , These represent the average values ​​of all instantaneous voltage values ​​taken for the corresponding phase.

[0082] Step 1.4, Main Frequency f m The dominant frequency component of the low-voltage side voltage waveform reflects the spectral characteristics of lightning strike faults.

[0083] (6)

[0084] It is the spectral amplitude of the low-voltage side voltage signal, which can be calculated using Fourier Transform (FFT):

[0085] (7)

[0086] in, u ( t ) indicates the low-voltage side voltage waveform.

[0087] Step 1.5, Peak grounding current I g The maximum value of the grounding grid current reflects the amount of energy of the lightning current passing through the grounding grid.

[0088] Step 1.6, Attenuation Consistency Ratio R The degree of matching between the voltage difference across the insulator and the waveform attenuation characteristics of the high-voltage side of the transformer is mainly used as an auxiliary indicator for fault identification.

[0089] The attenuation index is obtained by fitting the waveform attenuation process. α :

[0090] (8)

[0091] (9)

[0092] in, U 0 represents the initial value of the voltage difference across the insulator on the tower. U ( t () indicates the voltage difference across the insulator.

[0093] Step 2: Convert each extracted electrical feature parameter into a quantifiable score, normalize each feature parameter to compress its range to [0,1], for subsequent scientific evaluation and fault classification; the specific process is as follows:

[0094] Step 2.1: Calculate the high-voltage side score. S 1:

[0095] (10)

[0096] Step 2.2: Calculate the low-pressure side score S2:

[0097] (11)

[0098] Step 2.3: Calculate the score at the grounding grid. S 3:

[0099] (12)

[0100] Step 2.4: Calculate the score at the insulator. S 4:

[0101] (13)

[0102] Step 3: Conduct a comprehensive assessment of lightning strike faults using a weighted scoring method. This step involves assigning weights to each feature score and calculating the total score for each fault type to accurately distinguish between different lightning strike fault types. The specific process is as follows:

[0103] (14)

[0104] ω The weights of each characteristic can be dynamically adjusted according to environmental factors such as changes in grounding impedance of the corresponding wind farm and fluctuations in soil moisture, thereby improving operation and maintenance efficiency.

[0105] After the total score is calculated, the system can determine the fault type of the lightning strike event based on the total score under different faults.

[0106] Example 6

[0107] This invention provides a method for tracing the source of lightning strike faults in wind farm power collection systems based on comprehensive feature analysis. A simulation model of the wind farm power collection system is built on the ATP-EMTP platform. A schematic diagram of lightning current intrusion is shown below. Figure 1 From simulation Figure 2 It can be observed that when the lightning strike occurred, the bus voltage waveform exhibited a significant abrupt change with a large amplitude, and gradually returned to normal power frequency after a period of time. The waveform shows that before the fault occurred (40ms), the three-phase voltage waveform of the bus was a typical sinusoidal waveform, with stable fluctuations and no abnormalities.

[0108] Figure 3 and Figure 4 The data shows that, simultaneously, in both lines 1 and 2, there was almost no current in the turbine lines before the fault occurred, and the current loop had not yet been formed. At the moment of the fault (40ms), two-phase short circuits occurred in both circuits. Due to the busbar connection method and the fault current flow path, this two-phase short circuit caused the system to produce an effect similar to a three-phase short circuit on the busbar. From Figure 2 As can be seen, the bus voltage drops to almost zero at the fault point, with phase B voltage being the most affected.

[0109] From the simulation results, we can conclude that:

[0110]

[0111] First, the corresponding values ​​for different electrical characteristics are calculated using the following formula.

[0112]

[0113]

[0114]

[0115]

[0116] for f m Take 450Hz. C Take 0.1; for unknown quantities ( I g , R The input is set to -1, and the corresponding weights are ( ω 3. ω 4) Set to 0.

[0117] Secondly, each extracted feature parameter is converted into a quantifiable score and weighted to obtain the total score.

[0118] Fault 1:

[0119]

[0120] Fault 2:

[0121]

[0122] Fault 3:

[0123]

[0124] In summary, using the comprehensive score calculation of different faults using this invention: Fault 1 scores 0.25926, Fault 2 scores 0.55952, Fault 3 scores 0.2. Fault 2 has the highest score, so the source of the fault is most likely Fault 2, i.e., a direct lightning strike.

[0125] Based on the analysis of the external propagation characteristics of lightning overvoltage and the actual situation on site, this accident was caused by lightning directly striking the double-circuit tower, resulting in two different two-phase short-circuit faults on the two lines. The lightning strike caused a phase-to-phase short circuit between phases A and B on line 1, forming a new current loop, which caused the current in phase A to increase rapidly. Due to the phase-to-phase short circuit in line 1, the current in phase B also increased significantly, which is consistent with the characteristic of the current rapidly establishing up and reaching a stable value after a two-phase short circuit. Line 2 was similar to line 1, with a phase-to-phase short circuit between phases B and C caused by a lightning strike.

Claims

1. A method for tracing the source of lightning strike faults in wind farm power collection systems based on comprehensive feature analysis, characterized in that, Includes the following steps: Step 1: Based on the time-frequency characteristic differences, perform characteristic quantity calculation on the voltage / current waveform data under lightning strike monitored in real time by the sensors on the high and low voltage sides of the box-type transformer to obtain the key characteristic parameters characterizing the lightning strike fault; Specifically, by analyzing the voltage / current waveforms of the high and low voltage sides of a box-type transformer under lightning strikes, the electrical characteristics under lightning strikes are identified and calibrated. Then, the key characteristic parameters characterizing lightning strike faults are calculated using the following formulas. P 1. P n , C , f m , I g , R : 1) The ratio of the peak value of a certain phase voltage to the peak values ​​of the other two phases. P 1: In the formula, It is the maximum value of the three-phase voltage peak value; It represents the peak value of a single-phase voltage. A, B, and C are used to distinguish the different three phases in the voltage waveform. 2) Multiphase interference ratio P n : In the formula, σ The standard deviation of the three-phase voltage amplitude is calculated using the following formula: In the formula, n A value of 3 indicates three-phase voltage; U peak,i The amplitude of a certain phase voltage; 3) Correlation coefficients between three-phase voltage waveforms C : In the formula, r The correlation coefficient between two phase voltages is calculated using the following formula: In the formula, N This represents the number of instantaneous voltage values ​​taken from the fault voltage waveform. X i and Y i This represents the instantaneous values ​​of the voltages taken in different two phases. , These represent the average values ​​of all instantaneous voltage values ​​taken for the corresponding phase; 4) The dominant frequency of the low-voltage side voltage waveform f m : In the formula, It is the spectral amplitude of the low-voltage side voltage signal, calculated using Fourier transform: In the formula, u ( t () indicates the low-voltage side voltage waveform; 5) Peak grounding current I g The maximum value of the grounding grid current under lightning strike is directly taken; 6) The attenuation consistency ratio between the voltage difference across the insulator and the waveform on the high-voltage side of the box-type transformer. R : In the formula, α The attenuation exponent is obtained by fitting the waveform attenuation process: In the formula, U 0 represents the initial value of the voltage difference across the insulator on the tower. U ( t () indicates the voltage difference across the insulator; Step 2: The key feature parameters obtained in Step 1 are converted into dimensionless evaluation scores through an interval mapping function to obtain the feature scores of the high and low voltage / current waveforms of the box-type transformer under three fault types. Specifically, this involves calculating the key characteristic parameters representing lightning strike faults obtained in step 1. P 1. P n , C , f m , I g , R After normalization using different thresholds and compression to the range of [0,1], feature scores of the high-voltage and low-voltage side voltage / current waveforms of the box-type transformer under three fault types are obtained, i.e., high-voltage side scores. S 1. Low-pressure side score S 2. Score at the grounding grid S 3. Score at the insulator S 4; Among them, the high-voltage side scores under the three fault types S The formulas for calculating 1 are as follows: Low-voltage side scores under three fault types S The formulas for calculating 2 are as follows: Scores at the grounding grid under three fault types S The formulas for calculating 3 are as follows: Scores at the insulator under three fault types S The formulas for calculating 4 are all: Step 3: Assign weights to the feature scores of each fault type obtained in Step 2 and calculate the total score of the corresponding fault type. Determine the fault type of the lightning strike event based on the total scores of the three fault types.

2. The method for tracing the source of lightning strike faults in wind farm power collection systems based on comprehensive feature analysis as described in claim 1, characterized in that, In step 3, feature scores are obtained for each fault type. S 1. S 2. S 3. S 4. Feature scores for each fault type. S 1. S 2. S 3. S 4. Assign weights ω The total score for each fault type is calculated using the following formula: In the formula, ω Feature Score S 1. S 2. S 3. S A weight of 4.

Images