A method and device for identifying faults in high-voltage transmission lines

Abstract

This invention relates to the field of electrical signal processing technology and discloses a method and device for fault identification in high-voltage transmission lines through forests, aiming to solve the problem of rapid and accurate identification of tree-line faults and wildfire faults in 110kV / 220kV high-voltage transmission lines. The method includes: acquiring the voltage waveform of the faulty line relative to ground, extracting the time window before breakdown and normalizing it; determining the first breakdown time through abrupt change detection; extracting the fundamental phase of the normalized waveform and constructing a phase offset feature; calculating the fundamental residual before breakdown, statistically analyzing the residual energy according to the power frequency half-cycle and constructing a half-cycle distortion asymmetry feature; and combining the criteria of the two features to output a fault type judgment. This invention weakens the influence of voltage level differences, has strong feature complementarity, good anti-interference ability, can be implemented using only existing voltage acquisition links, is computationally simple, suitable for real-time engineering applications, and significantly improves the accuracy and efficiency of identifying tree-line faults and wildfire faults.

Description

Technical Field

[0001] This application relates to the field of electrical signal processing technology, specifically to a method and device for fault identification of high-voltage transmission lines in forest areas. Background Technology

[0002] 110kV and 220kV transmission lines, as the main arteries of the power grid, bear the critical task of transmitting large amounts of electricity over long distances. Their safe and stable operation is directly related to the reliability of the power grid and the quality of power supply. However, transmission corridors often traverse complex geographical environments, including mountains, farmland, and areas with diverse vegetation, posing severe external environmental challenges to line operation. In recent years, with the improvement of the ecological environment and vegetation growth, tree obstructions (trees growing too tall and touching conductors) and wildfires (fires occurring beneath the lines) have become the two main causes of high-voltage transmission line tripping. Statistics show that these two types of faults account for a continuously increasing proportion of transmission line tripping accidents, seriously threatening the safe operation of the power grid.

[0003] Accurately identifying the cause of a power transmission line tripping due to a wildfire is of significant legal and economic importance. In cases where a tree-line fault causes a wildfire, the close proximity of trees to the conductor leads to a discharge, and the high-temperature arc ignites the vegetation below, creating a wildfire. In this situation, the transmission line maintenance unit may bear corresponding responsibility for failing to promptly trim obstructing trees. In cases where a wildfire is caused by a non-line-related issue, an external fire source (such as human activity, lightning strikes, or spontaneous combustion) ignites the fire, and the high temperature of the flames reduces the air insulation strength, leading to a line discharge and tripping. In this case, the transmission line is the victim, and the party responsible for the fire is the source of the fire. The responsible parties in these two scenarios are drastically different, but under current technological conditions, on-site investigation and handling of the cause of the accident are difficult and time-consuming. Because wildfire sites are often severely damaged, traditional investigation methods based on on-site traces are insufficient to quickly and accurately reconstruct the truth of the accident, often leading to disputes over liability, affecting the efficiency of accident handling, and protecting the legitimate rights and interests of power grid companies.

[0004] In existing technologies, some transmission line fault identification methods rely on multi-dimensional data collection, such as current signals and leakage currents, requiring additional data acquisition equipment and increasing engineering application costs. Other methods rely on only a single feature for fault identification, exhibiting poor anti-interference capabilities and prone to misjudgment under noise disturbances and fluctuating operating conditions. Furthermore, they fail to consider the differences in measurement scales between lines of different voltage levels, making them difficult to apply universally between 110kV and 220kV lines. Therefore, there is an urgent need for a method and device capable of rapidly, accurately, and objectively identifying faults in forest-related high-voltage transmission lines. Summary of the Invention

[0005] To address the above problems, this invention provides a method and apparatus for fault identification in high-voltage transmission lines in forest areas. The method reduces the impact of voltage level differences by normalizing a preset time window before the first breakdown occurs. Abrupt change detection is used in the normalized waveform to determine the first breakdown moment as a time reference. Furthermore, the fundamental voltage phase is extracted at a preset time before the breakdown moment, and a phase offset feature is constructed. Simultaneously, within the preset time window before breakdown, a half-cycle distortion asymmetry feature is constructed based on the positive and negative half-cycle statistics of the fundamental residual. Finally, a tree-line fault or wildfire fault judgment is output based on a combination criterion of the above two types of features, thereby improving the accuracy and engineering applicability of fault type identification.

[0006] On the one hand, the present invention provides a method for fault identification of high-voltage transmission lines in forest areas, characterized by comprising the following steps: Step 1: Collect the phase-to-ground voltage waveforms before and after the power grid fault, recorded by the transmission line fault recorder. Capture the preset time window before the first breakdown occurs. For the time window Calculate the effective value and with right Normalization is performed to obtain the per-unit waveform. ; Step 2: Standardize the waveform In the middle, mutation detection is used to determine the first breakdown time. and the As a time reference for subsequent feature extraction; Step 3: Standardize the waveform Extract the fundamental component and obtain the fundamental phase. ;exist Previous preset time point Read the fundamental phase The value of the phase offset feature is determined by the phase distance relationship between the phase and the positive peak phase and the negative peak phase; Step 4: Set a time window before breakdown. Internal computation The residual of its fundamental component The time window is divided into a positive half-cycle set and a negative half-cycle set according to the power frequency half-cycle; the residual energy statistics are calculated for the positive half-cycle set and the negative half-cycle set respectively, and the value of the half-cycle distortion asymmetry feature is determined by the ratio of the two. Step 5: Make a category decision based on the phase offset feature and the half-cycle distortion asymmetry feature: when the phase offset feature corresponds to the negative peak side breakdown tendency and the half-cycle distortion asymmetry feature meets the preset threshold relationship, output the tree line fault decision; otherwise, output the wildfire fault decision.

[0007] Preferably, the formula for per-unit processing in step 1 is:

[0008] Preferably, in step 2, mutation detection is used to determine the first breakdown time. The specific process is as follows: Construction of mutation detection indicators:

[0009] in, The sampling period of the fault recorder; These are the weighting coefficients; when First time exceeding the threshold And during the duration When the persistence condition is met, the earliest moment when the condition is met is determined as the first breakdown moment. ;where the threshold Within the preset time window before breakdown The statistic is adaptively set to , Within the time window The mean and standard deviation, These are preset coefficients.

[0010] Preferably, the specific steps for determining the phase offset feature value in step 3 are as follows: Step 31: Standardize the waveform Extract the fundamental component and obtain the fundamental phase. The fundamental component is obtained by bandpass filtering near the power frequency. And perform a Hilbert transform on it to obtain the analytic signal. Define the fundamental phase

[0011] Step 32: Obtain the phase before breakdown ,in The duration is 0.2ms to 2ms; Step 33, Calculation The wraparound distance to the positive and negative peak phases:

[0012] in Used to normalize phase difference ;when When it is determined to be a negative peak side breakdown tendency, when The tendency is determined to be a positive peak side breakdown tendency, and this tendency is used as a phase offset feature.

[0013] Preferably, in step 4, the specific steps for determining the asymmetric feature value of the half-cycle distortion are as follows: Step 41, after breaking through the front window Internal calculation of fundamental residual in For the fundamental component, Duration; Step 42: According to power frequency zero crossover or according to phase The breakdown window is divided into several sets of positive and negative half-cycles, and the mean residual energy of the positive and negative half-cycles is calculated respectively. And define the half-cycle distortion asymmetry ratio

[0014] And As a characteristic of asymmetry in half-cycle distortion.

[0015] Preferably, the method for classifying based on phase offset features and half-cycle distortion asymmetry features in step 5 is as follows: When the phase bias characteristic is a negative peak side breakdown tendency and If the fault is detected in time, it is determined to be a treeline fault; otherwise, it is determined to be a wildfire fault; where the threshold is... Based on the per-unit statistical settings, it is applicable to lines with different rated voltage levels.

[0016] Preferably, when a tree-line fault is determined, the following conditions must be met for M consecutive power frequency half-cycles. ≥ , where M is a preset integer, M=2-6.

[0017] On the other hand, the present invention also provides a fault identification device for high-voltage transmission lines in forest areas, characterized in that the device comprises: At least one processor; and a memory communicatively connected to said at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform any of the method steps described above.

[0018] The beneficial effects of this invention are as follows: By standardizing the voltage waveform within a preset time window before breakdown, the influence of different rated voltage levels and measurement scale differences on feature extraction can be weakened, improving the applicability of the method between 110kV / 220kV lines; by using abrupt change detection to determine the first breakdown moment as a time reference, feature extraction is performed around the critical moment of fault triggering, enhancing the specificity and consistency of the features; the phase offset feature constructed based on the fundamental voltage phase and the half-cycle distortion asymmetric feature constructed based on the fundamental residual half-cycle statistics complement each other, improving the distinguishability between tree-line faults and wildfire faults, and enhancing the anti-interference capability under noise disturbances and operating condition fluctuations; this invention only uses the voltage waveform relative to ground recorded by the power grid fault recorder as input, without the need for additional deployment of leakage current, current signal, and other acquisition equipment, directly utilizing existing power grid measurement conditions, reducing engineering application costs, and facilitating online deployment and promotion. Furthermore, this invention only relies on the voltage waveform to complete the identification, facilitating online deployment using existing voltage acquisition links, and the calculation process is simple, making it suitable for real-time engineering applications. Attached Figure Description

[0019] Figure 1 This is a step diagram illustrating an embodiment of the present invention. Detailed Implementation

[0020] The present invention will now be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Per-unit value is a common network equation used to establish systems with multiple voltage levels. It is widely used in circuit breaking calculations and relay protection settings. Phase offset is used to characterize the correspondence between the occurrence of breakdown and the peak position of the power frequency waveform, thereby reflecting the bias of the breakdown triggering process on the power frequency phase. This feature is based on the phase relationship and is relatively independent of the absolute amplitude. It can maintain good stability under different voltage levels and certain noise disturbance conditions. Half-cycle distortion asymmetry is used to quantify the degree of difference in distortion between the positive and negative half-cycles before breakdown, improving the separability of waveform distortion mode differences under different fault types.

[0022] In the process of identifying wildfire flame discharge and treeline discharge, the breakdown voltage signal often has a significant polarity effect. Flame breakdown often occurs in the positive half-wave, while treeline breakdown often occurs in the negative half-wave. By using two characteristics, namely phase offset characteristics and half-cycle distortion asymmetry characteristics, the polarity effect of the breakdown voltage waveform signal can be more clearly displayed, making it easier to judge wildfire faults and treeline faults, thus helping investigators to determine the cause of wildfire accidents.

[0023] like Figure 1 As shown in the figure, an embodiment of the present invention provides a method for fault identification of high-voltage transmission lines in forest areas, which includes the following steps: Step 1: Acquire the voltage waveform of the measured phase to ground, and select the preset time window before the first breakdown as the analysis interval before breakdown, and perform per-unit processing on the voltage waveform.

[0024] In this step, the voltage waveforms relative to ground before and after the fault are first collected by a fault recorder on the power grid transmission line. Since fault recorders typically trigger recording after a fault is detected, the acquired waveforms contain data from several cycles before the fault occurred to several cycles after. To eliminate the influence of different rated voltage levels (110kV and 220kV) and different measurement scales on feature extraction, This invention captures a preset time window before the first breakdown occurs. As the pre-breakdown analysis interval, this time window typically covers 1–3 power frequency cycles before the fault occurs.

[0025] Within the time window Calculate the effective value and with right Normalization is performed to obtain the per-unit waveform. ;

[0026] By standardizing the voltages of different voltage levels, the line voltages are normalized to a similar range, thereby eliminating the impact of voltage level differences on subsequent feature extraction and enhancing the versatility of the method.

[0027] Step 2: Determine the first breakdown moment in the voltage waveform.

[0028] When a high-voltage transmission line experiences tree obstruction or wildfire faults, insulation breakdown typically manifests as a rapid voltage drop. To accurately pinpoint the onset of the breakdown, this invention uses a per-unit waveform... The algorithm used is a mutation detection algorithm based on first-order and second-order differences.

[0029] First, construct mutation detection indicators.

[0030]

[0031] The sampling interval Δt is the sampling period of the fault recorder. Considering the need to capture the power frequency voltage waveform (50Hz) and the transient characteristics of breakdown, its value range is Δt=20μs-200μs, preferably Δt=50μs, corresponding to a sampling frequency of 20kHz, which can ensure the effective capture of the power frequency voltage waveform and the transient characteristics of breakdown.

[0032] The first item The second term reflects the first-order difference (rate of change) of the signal. The weighting coefficients reflect the second-order difference (curvature change) of the signal. Used to balance the contributions of the two.

[0033] Because the initial conditions and environmental noise vary for different faults, using a fixed threshold may lead to missed detections or false detections. Therefore, this invention employs an adaptive threshold method. The threshold is calculated within a preset time window before breakdown. Statistics, including the mean and standard deviation Set the adaptive threshold to ; These are preset coefficients, determined based on the required detection sensitivity and false alarm rate. The value range is 3-4.

[0034] To avoid misjudgment caused by instantaneous jumps due to noise, this invention introduces a continuous condition.

[0035] when First time exceeding the threshold Afterwards, it is necessary to check the duration. Does the internal condition continue to be satisfied? Duration Take a range of 0.05ms to 2ms.

[0036] The earliest moment when the persistence condition is met is defined as the first breakdown moment. and the This serves as a time benchmark for subsequent feature extraction, ensuring temporal consistency in feature extraction.

[0037] Step 3: At a preset time point before the first breakdown moment, obtain the voltage fundamental phase and extract the phase bias characteristics accordingly.

[0038] Phase offset characteristics are used to characterize the correspondence between the breakdown occurrence time and the peak position of the power frequency waveform, reflecting the bias of the breakdown triggering process towards the power frequency phase. Since tree line faults typically exhibit negative peak-side breakdown (negative half-cycle), while wildfire faults typically exhibit positive peak-side breakdown (positive half-cycle), this characteristic can effectively distinguish between the two types of faults.

[0039] The specific extraction steps are as follows: Step 31, fundamental component extraction and phase calculation; Standardized perturbation waveform Perform bandpass filtering to extract the fundamental frequency component near the power frequency. The passband was set to 45Hz-55Hz to filter out high-frequency noise and low-frequency drift. Subsequently, a Hilbert transform was performed on the fundamental component to construct the analytic signal. ,in for Hilbert transform; Fundamental phase Defined as the phase angle of an analytic signal: The positive half-cycle is defined as when 0 ≤ ϕ < π, and the negative half-cycle is defined as when −π ≤ ϕ < 0.

[0040] Step 32, Phase acquisition before breakdown; To avoid the impact of voltage waveform distortion on phase calculation after breakdown, this invention calculates the phase at the breakdown time. Previous preset time point Read the fundamental phase ,in The interval is 0.2ms to 2ms to ensure that the stable phase information before breakdown is obtained.

[0041] Step 33, Phase offset determination; calculate Circumference to the positive and negative peak phases

[0042] in The formula for the function is:

[0043] Used to normalize phase difference ; By comparison and The magnitude relationship determines the phase offset characteristic F. φ ; when When it is determined to be a negative peak side breakdown tendency, when The time was determined to be a positive peak side breakdown tendency, and the tendency was used as the phase bias feature F. φ This feature is based on phase relationship and is relatively independent of absolute amplitude, maintaining good stability under different voltage levels and certain noise disturbances.

[0044] Step 4: Within a preset time window before the first breakdown moment, the voltage distortion degree is statistically analyzed for positive and negative half cycles according to the power frequency half cycle, and the half-cycle distortion asymmetry feature is extracted accordingly.

[0045] The half-cycle distortion asymmetry feature is used to quantify the degree of difference in distortion between the positive and negative half-cycles before breakdown. Tree-line faults and wildfire faults exhibit different nonlinear characteristics in their pre-discharge processes (such as corona and leader development) before breakdown, leading to differences in the degree of voltage waveform distortion between the positive and negative half-cycles. This feature complements the phase bias feature in step S3, improving the separability of waveform distortion mode differences under different fault types.

[0046] The specific extraction steps are as follows: Step 41: Residual calculation; After shattering the front window Internally, calculate the per-unit waveform. Its fundamental component The residual, The residual It reflects the harmonics, noise, and distortion components in the voltage waveform other than the fundamental frequency. The duration T1 is the continuity condition for sudden change detection, and its value ranges from T1=0.05ms to 2ms, usually selected as 1-2 power frequency cycles.

[0047] Step 42: Semi-circle division and energy statistics; Based on power frequency zero crossover or based on phase The breakdown window is divided into several sets of positive and negative half-cycles, and the mean residual energy of the positive and negative half-cycles is calculated respectively. And define the half-cycle distortion asymmetry ratio

[0048] when A larger value indicates a relatively stronger negative half-cycle distortion energy; when The smaller the value, the stronger the distortion energy in the positive half-cycle, and the more likely it is to cause distortion. F, as a characteristic of half-cycle distortion asymmetry h Based on extensive statistical data, treeline faults typically manifest as follows: Wildfire malfunctions typically manifest as or .

[0049] Step 5: Make a decision based on the phase offset feature and the half-cycle distortion asymmetry feature.

[0050] The classification is determined based on the phase offset feature and the half-cycle distortion asymmetry feature: when the phase offset feature corresponds to the negative peak side breakdown tendency and the half-cycle distortion asymmetry feature meets the preset threshold relationship, the tree line fault judgment is output; otherwise, the wildfire fault judgment is output.

[0051] Specifically, based on the phase offset characteristic F φ Asymmetric features of half-cycle distortion F h Perform a category decision. Preferably, set the half-cycle distortion asymmetry threshold as θ. A A treeline fault is identified when the following conditions are met: (F φ (for negative peak side breakdown tendency)∧( ≥θ A Otherwise, it is determined to be a flame malfunction. The threshold θ A The preset threshold can preferably be adaptively set based on the standardized statistics, for example, calculating the residual energy ratio within the statistical time window before breakdown. mean μ A With standard deviation σ A And set: θ A =μ A +γσ A γ is a preset coefficient, preferably γ is 1-4, so as to be applicable to lines with different rated voltage levels and different measurement scales.

[0052] To improve robustness, it can be required that the condition is met for M consecutive half-cycles. ≥θ A Only then is the tree fault output, where M is a preset integer, preferably M=2-6.

[0053] In this embodiment, the voltage waveforms relative to ground before and after the fault in the power grid transmission line are first acquired using a fault recorder, and the waveforms are normalized within a preset time window before the first breakdown. The first breakdown time is determined from the normalized waveform. The fundamental voltage phase is extracted at a preset time point before the breakdown time, and a phase offset feature is constructed. Within the preset time window before breakdown, a half-cycle distortion asymmetry feature is constructed based on the positive and negative half-cycle statistics of the fundamental residual. A decision is made based on the phase offset feature and the half-cycle distortion asymmetry feature, and the identification result of tree-line fault or wildfire fault is output. This improves the accuracy and engineering deployability of fault type identification for 110kV and 220kV transmission lines.

[0054] Using the voltage waveforms relative to ground recorded by the fault recorder of the power grid transmission line before and after the accident as input, there is no need to introduce additional measurement links such as leakage current, which facilitates the direct use of the voltage measurement conditions of the existing power grid fault recorder; by standardizing and aligning with the breakdown time, the influence of voltage level differences and changes in operating conditions on the identification results can be reduced; the phase offset feature and the half-cycle distortion asymmetry feature describe the waveform characteristics before breakdown from the two perspectives of "phase relationship" and "half-cycle distortion energy difference", respectively, with strong complementarity, and can maintain good discrimination stability under certain noise interference conditions; the overall calculation amount is small, which is suitable for online real-time implementation.

[0055] In another embodiment, the present invention also provides a fault identification device for high-voltage transmission lines in forest areas, characterized in that the device comprises: At least one processor; and a memory communicatively connected to said at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform any of the method steps described above.

[0056] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

Claims

1. A method for fault identification of high-voltage transmission lines in forest areas, characterized in that, Includes the following steps: Step 1: Collect the phase-to-ground voltage waveforms before and after the power grid fault, recorded by the transmission line fault recorder. Capture the preset time window before the first breakdown occurs. For the time window Calculate the effective value and with right Normalization is performed to obtain the per-unit waveform. ; Step 2: Standardize the waveform In the middle, mutation detection is used to determine the first breakdown time. and the As a time reference for subsequent feature extraction; Step 3: Standardize the waveform Extract the fundamental component and obtain the fundamental phase. exist Previous preset time point Read the fundamental phase The value of the phase offset feature is determined by the phase distance relationship between the phase and the positive peak phase and the negative peak phase; Step 4: Set a time window before breakdown. Internal computation The residual of its fundamental component The time window is divided into a positive half-cycle set and a negative half-cycle set according to the power frequency half-cycle; the residual energy statistics are calculated for the positive half-cycle set and the negative half-cycle set respectively, and the value of the half-cycle distortion asymmetry feature is determined by the ratio of the two. Step 5: Make a category decision based on the phase offset feature and the half-cycle distortion asymmetry feature: when the phase offset feature corresponds to the negative peak side breakdown tendency and the half-cycle distortion asymmetry feature meets the preset threshold relationship, output the tree line fault decision; otherwise, output the wildfire fault decision.

2. The method according to claim 1, characterized in that, The formula for per-unit processing in step 1 is:

3. The method according to claim 1, characterized in that, Step 2 uses mutation detection to determine the first breakdown time. The specific process is as follows: Construction of mutation detection indicators: in, The sampling period of the fault recorder; For weighting coefficients; when First time exceeding the threshold And during the duration When the persistence condition is met, the earliest moment when the condition is met is determined as the first breakdown moment. ;where the threshold Within the preset time window before breakdown The statistic is adaptively set to ， Within the time window The mean and standard deviation, These are preset coefficients.

4. The method according to claim 1, characterized in that, The specific steps for determining the phase offset feature value in step 3 are as follows: Step 31: Standardize the waveform Extract the fundamental component and obtain the fundamental phase. The fundamental component is obtained by bandpass filtering near the power frequency. And perform a Hilbert transform on it to obtain the analytic signal. Define the fundamental phase ; Step 32: Obtain the phase before breakdown ,in The duration is 0.2ms to 2ms; Step 33, Calculation The wraparound distance to the positive and negative peak phases: ,in Used to normalize phase difference ;when When it is determined to be a negative peak side breakdown tendency, when The tendency is determined to be a positive peak side breakdown tendency, and this tendency is used as a phase offset feature.

5. The method according to claim 1, characterized in that, The specific steps for determining the asymmetric feature value of half-cycle distortion in step 4 are as follows: Step 41, after breaking through the front window Internal calculation of fundamental residual ,in For the fundamental component, Duration; Step 42: According to power frequency zero crossover or according to phase The breakdown window is divided into several sets of positive and negative half-cycles, and the mean residual energy of the positive and negative half-cycles is calculated respectively. And define the half-cycle distortion asymmetry ratio And As a characteristic of asymmetry in half-cycle distortion.

6. The method according to claim 1, characterized in that, The method for class determination based on phase offset characteristics and half-cycle distortion asymmetry characteristics in step 5 is as follows: When the phase bias characteristic is a negative peak side breakdown tendency and If the fault is detected in time, it is determined to be a treeline fault; otherwise, it is determined to be a wildfire fault; where the threshold is... Based on the per-unit statistical settings, it is applicable to lines with different rated voltage levels.

7. The method according to claim 6, characterized in that, When a tree-line fault is identified, the following conditions must be met for M consecutive power frequency half-cycles. ≥ , where M is a preset integer, M=2-6.

8. A fault identification device for high-voltage transmission lines in forest areas, characterized in that, The device includes: At least one processor; and a memory communicatively connected to said at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to cause the at least one processor to... The processor is capable of performing the method as described in any one of claims 1-7.

Images