An overhead line tree contact fault detection method based on zero sequence current segmentation characteristics

Abstract

The application belongs to the technical field of power system distribution network relay protection, and provides an overhead line tree contact fault detection method based on zero sequence current segmentation characteristics, which comprises the following steps: firstly, the most similar modal component of the zero sequence current signal is extracted by using variational mode decomposition; secondly, based on extreme value distribution, the fault process is divided into four physical stages of tree line contact, water evaporation, carbonization formation and open fire discharge, and a dynamic characterization model of the fault development process is established; then, the standard deviation feature sequence and the slope feature sequence are respectively constructed for the first two key stages; the application accurately extracts the fault signal features by using variational mode decomposition, combines the phased feature extraction and the double-parameter fusion criterion, effectively solves the problems of poor anti-interference ability and low initial sensitivity of the traditional method, significantly improves the accuracy and robustness of the early detection of the overhead line tree contact fault, and reduces the false alarm rate, thereby providing a more reliable and timely fault early warning mechanism for the power system.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of power system distribution network relay protection, and particularly relates to an overhead line tree contact fault detection method based on zero sequence current segmentation characteristics. BACKGROUND

[0002] In a power system distribution network, an overhead line is an important part of power transmission, and its safe operation is directly related to the stability and reliability of the entire power grid. However, in mountainous areas, due to the natural growth of vegetation, the safety distance between the overhead line and the trees may be reduced due to the growth of trees, or even directly contacted, thereby causing single-phase tree contact fault. In the early stage of this fault, due to the high contact impedance between the tree and the line, the fault current is often weak, and the characteristic is short-lived, so that the traditional detection method based on the high-impedance grounding fault model cannot effectively identify it.

[0003] Tree contact fault is a dynamic and slowly changing process. With the continuation of the fault, the tree may be carbonized under the action of the arc, or the internal water may evaporate due to high temperature, resulting in a gradual decrease in contact impedance, and the fault current characteristics also evolve. In this process, if the fault cannot be detected and handled in time, the high temperature generated by the continuous leakage current or intermittent arc may ignite the surrounding vegetation, thereby causing forest fires, which poses a major threat to the safety of the power grid, the ecological environment, and people's life and property.

[0004] Existing fault detection methods mostly rely on a single fault characteristic quantity, such as transient characteristics or steady-state characteristics. These methods often have poor anti-interference ability, low initial sensitivity, and other problems when facing complex and variable fault environments, and are difficult to achieve early warning and accurate identification of tree contact faults. Therefore, it is of great significance to develop a method that can accurately and reliably detect overhead line tree contact faults by deeply studying the dynamic characteristics of tree contact faults in the long-term development process, which can improve the safety and stability of the power system. SUMMARY

[0005] The purpose of the present application is to provide an overhead line tree contact fault detection method based on zero sequence current segmentation characteristics, which solves the problem of weak fault characteristics in the prior art when the overhead line tree contact fault occurs, and the fault signal is easily submerged by noise such as load fluctuation and harmonics, resulting in detection difficulty.

[0006] The purpose of the present application is achieved by the following scheme:

[0007] In a first aspect, the present application provides an overhead line tree contact fault detection method based on zero sequence current segmentation characteristics, comprising the following steps:

[0008] Step 1, zero-sequence current effective value signal decomposition: the zero-sequence current effective value signal is decomposed by using variational mode decomposition (VMD), a plurality of intrinsic mode functions (IMF) are obtained, and then the cosine similarity of each IMF component is calculated, and the IMF with the maximum cosine similarity is selected as the most similar mode;

[0009] Step 2, fault process stage division: the local extreme point detection is performed on the most similar mode to obtain the extreme sequence {P1, P2, …, P k}; the most similar mode is divided into four fault development stages according to the extreme point distribution: interval represents the tree line contact stage, interval represents the water evaporation stage, interval represents the carbonization formation stage, interval represents the open fire discharge stage; the fault process stage is the initial stage of the fault, the tree just contacts the overhead line, and the fault current is weak and unstable due to the large contact impedance; the local extreme point detection shows that the current fluctuation in this stage is small, but there is an obvious starting contact point, which marks the beginning of the fault;

[0010] Step 3, feature extraction of the most similar mode signal: for the tree line contact stage , it is divided into continuous subintervals by using the equal time window, and the time window length is ; according to the standard deviation formula, the standard deviation of the most similar mode current waveform in each subinterval is calculated, and a standard deviation feature sequence is constructed, which is used to quantify the discrete feature of the signal amplitude of the tree line contact stage ; for the water evaporation stage , it is divided into b equal-length subintervals by using the equal time window, and the time window length is ; according to the slope formula, the time-domain slope of the most similar mode current waveform in each subinterval is calculated, and a slope feature sequence is generated to represent the dynamic trend of the signal change in the water evaporation stage ;

[0011] Step 4, overhead line tree fault detection: based on the standard deviation feature sequence and the slope feature sequence extracted in step 3, a double-parameter fusion criterion is constructed, and the criterion content is: when the standard deviation of all subintervals of the tree line contact stage satisfies and the water evaporation stage All sub-intervals satisfy , an overhead line tree line contact fault alarm is triggered; otherwise, it is determined that the line is in a normal operating state. and is a set threshold value.

[0012] Preferably, the step 1 includes the following specific steps:

[0013] Step 1.1, using variational mode decomposition (VMD) to perform m layer decomposition on the zero sequence current effective value signal to obtain m intrinsic mode functions (IMFs), denoted as IMF1, IMF2, …, IMF m ; wherein the variational mode decomposition is a new multi-resolution adaptive non-recursive signal decomposition technology, which can extract multiple modal components from the original signal at the same time, and match the center frequency for each component through iterative optimization;

[0014] Step 1.2, respectively calculating the cosine similarity of each IMF component and the original signal, selecting the IMF with the maximum cosine similarity as the most similar mode, and the cosine similarity is calculated as follows:

[0015] ;

[0016] In the formula, cos( θ ) is the cosine similarity, n is the number of sampling points, IMF p ( n ) is the p th IMF component, p =1,2,…, m , S ( n ) is the original signal, and IMF pq ( n ) is the current value of the p th sampling point in the q th IMF component, S q ( n ) is the current value of the q th sampling point in the original signal, q =1,2,…, n .

[0017] Preferably, the step 3 includes the following specific steps:

[0018] Step 3.1, for the tree line contact stage , is divided into continuous sub-intervals with equal time windows ; the standard deviation of the most similar modal current waveform in each sub-interval is calculated according to the standard deviation formula, and a standard deviation feature sequence is constructed to quantify the discrete feature of the signal amplitude of the tree line contact stage ; the length of the time window is:

[0019] ;

[0020] The standard deviation calculation formula is:

[0021] ;

[0022] In the formula, n is the number of sampling points, is the most similar modal signal current value corresponding to the q th sampling point, is the mean value of the most similar modal signal current, ;

[0023] Step 3.2, for the water evaporation stage , is divided into b equal-length sub-intervals by equal time windows ; the slope of the most similar modal current waveform in each sub-interval is calculated according to the slope formula, and a slope feature sequence is generated by calculating the time-domain slope of the waveform in each sub-interval to represent the dynamic trend of the signal change in the water evaporation stage ; the length of the time window is:

[0024] ;

[0025] The slope calculation formula is:

[0026] ;

[0027] In the formula, is the change amount of the most similar modal signal current value in the j th segment, is the time j change amount in the t th segment, j =1,2,…, b .

[0028] The application firstly adopts a variational mode decomposition (VMD) to extract a most similar modal component of a zero sequence current signal, and effectively removes non-fault interference; secondly, based on extreme value distribution, divides a fault process into four physical stages of tree line contact, water evaporation, carbonization formation and open fire discharge, and establishes a dynamic characterization model of the fault development process; then, constructs a standard deviation feature sequence and a slope feature sequence respectively for the first two key stages, and realizes double-dimensional feature extraction through quantifying signal discreteness and dynamic trend; finally, designs a double-parameter fusion criterion, and realizes fault recognition through a joint determination mechanism of a standard deviation threshold and a slope threshold.

[0029] In a second aspect, the application provides an overhead line tree contact fault detection device based on zero sequence current segmented features, which is suitable for the overhead line tree contact fault detection method based on zero sequence current segmented features.

[0030] A zero sequence current effective value signal decomposition module is configured to decompose the zero sequence current effective value signal by using the variational mode decomposition to obtain a plurality of intrinsic mode functions, and select an IMF with the maximum cosine similarity as the most similar mode by calculating the cosine similarity of each IMF component.

[0031] A fault process stage division module is configured to detect local extreme points of the most similar mode to obtain an extreme value sequence, and divide the fault development process into four characteristic stages based on the extreme value distribution: a tree line contact stage, a water evaporation stage, a carbonization formation stage, and an open fire discharge stage.

[0032] A most similar mode signal feature extraction module is configured to extract signal features for the tree line contact stage and the water evaporation stage, wherein the tree line contact stage adopts equal time window segmentation and calculates the standard deviation of the current waveform in each sub-interval to construct a standard deviation feature sequence, and the water evaporation stage adopts equal time window segmentation and calculates the time domain slope of the waveform in each sub-interval to generate a slope feature sequence.

[0033] An overhead line tree contact fault detection module is configured to construct a double-parameter fusion criterion based on the extracted standard deviation feature sequence and the slope feature sequence: when the standard deviations of all sub-intervals in the tree line contact stage satisfy , and the slopes of all sub-intervals in the water evaporation stage satisfy , an overhead line tree line contact fault alarm is triggered; otherwise, it is determined that the line is in a normal operating state; wherein and are set threshold values.

[0034] In a third aspect, the application provides a computer device including a memory and a processor, the memory stores a computer program, and the processor implements the above method when executing the computer program.

[0035] In a fourth aspect, the present application provides a computer readable storage medium, having stored thereon a computer program, wherein the computer program, when executed by a processor, implements the method described above.

[0036] Compared with the prior art, the present application has the following beneficial effects:

[0037] The present application accurately extracts fault signal features through variational mode decomposition, effectively solves the problems of poor anti-interference ability and low initial sensitivity of traditional methods by combining phased feature extraction (standard deviation quantization of tree contact phase discreteness, slope description of dynamic trend in water evaporation phase) and double parameter fusion criterion, significantly improves the accuracy and robustness of early detection of overhead line tree contact fault, reduces the false alarm rate, and provides a more reliable and timely fault warning mechanism for the power system. BRIEF DESCRIPTION OF DRAWINGS

[0038] Figure 1 The flow chart of the overhead line tree contact fault detection method based on zero sequence current segmentation features of the present application;

[0039] Figure 2 The simulation model schematic diagram of the single-phase tree contact fault of the overhead line of the embodiment of the present application without branch 10kV system;

[0040] Figure 3 The simulation result schematic diagram of the effective value of zero sequence current in the normal state of the embodiment of the present application;

[0041] Figure 4 The simulation result schematic diagram of the effective value of zero sequence current when the tree contact fault occurs in phase B at a distance of 10km from the bus of the embodiment of the present application;

[0042] Figure 5 The mode schematic diagram of the effective value of zero sequence current after VMD decomposition when the tree contact fault occurs of the embodiment of the present application;

[0043] Figure 6 The cosine similarity comparison schematic diagram of each mode after VMD decomposition of the effective value of zero sequence current when the tree contact fault occurs of the embodiment of the present application;

[0044] Figure 7 The extreme point distribution of the most similar mode and the waveforms of the tree contact phase segmentation and the water evaporation phase segmentation when the tree contact fault occurs of the embodiment of the present application. DETAILED DESCRIPTION

[0045] In order to facilitate the understanding of the present application, the present application will be described more fully below with reference to the related drawings. The preferred embodiments of the present application are shown in the drawings. However, the present application can be realized in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the application more thorough and comprehensive.

[0046] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in the description herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.

[0047] Embodiment: The present application provides an overhead line tree contact fault detection method based on zero sequence current segmentation characteristics, as shown in the method applied to the terminal for example, it can be understood that the method can also be applied to the server, but also can be applied to the system including the terminal and the server, and is realized through the interaction of the terminal and the server, including the following steps: Figure 1

[0048] Step 1, zero sequence current effective value signal decomposition: the zero sequence current effective value signal is decomposed by using variational mode decomposition (VMD), and a plurality of intrinsic mode functions (IMF) are obtained; the cosine similarity of each IMF component is calculated, and the IMF with the maximum cosine similarity is selected as the most similar mode.

[0049] The zero sequence current effective value signal is analyzed in depth by using variational mode decomposition (VMD) technology, and m intrinsic mode functions (IMF) are obtained by m-layer decomposition, which are respectively denoted as IMF1, IMF2, …, IMF m .

[0050] The above steps are aimed at decomposing the complex zero sequence current signal into a plurality of modal components with different frequency characteristics, so as to more accurately capture the fault characteristics in the signal.

[0051] Specifically, the step 1 includes the following specific steps:

[0052] Step 1.1, the zero sequence current effective value signal is decomposed by using variational mode decomposition (VMD) m Layer decomposition, obtaining m Intrinsic mode functions (IMF), denoted as IMF1, IMF2, …, IMF m ; the layered analysis of the signal is realized; ​

[0053] Step 1.2, calculate the cosine similarity of each IMF component and the original signal respectively, select the IMF with the maximum cosine similarity as the most similar mode, and the cosine similarity is calculated as follows:

[0054] ;

[0055] In the formula, cos( θ ) is the cosine similarity, n is the number of sampling points, IMF p ( n ) is the p th IMF component, p =1,2,…, m , S ( n ) is the original signal, IMF pq ( n ) is the current value of the p th sampling point in the q th IMF component, S q ( n ) is the current value of the q th sampling point in the original signal, q =1,2,…, n ;

[0056] By calculating the cosine similarity of each IMF component and the original signal, the IMF component that best matches the characteristics of the original signal is further selected, that is, the IMF with the maximum cosine similarity is selected as the most similar mode.

[0057] The calculation of the cosine similarity takes into account the current value of the signal at each sampling point, and by quantifying the similarity between the IMF component and the original signal, it ensures that the selected mode can accurately reflect the key characteristics of the original signal, providing a reliable foundation for subsequent fault stage division and feature extraction.

[0058] Step 2, fault process stage division: local extreme point detection is performed on the most similar mode to obtain the extreme value sequence {P1, P2, …, P k}; according to the distribution of extreme points, the most similar mode is divided into four fault development stages: interval represents the tree contact stage, interval represents the water evaporation stage, interval represents the carbonization formation stage, interval represents the open fire discharge stage.

[0059] Specifically, in the fault process stage division, first, local extreme point detection is performed on the most similar mode to generate the extreme value sequence {P1, P2, …, P k} and then according to the distribution law of extreme points, the fault development process is divided into four continuous physical stages: tree line contact stage (characterizing weak current fluctuations of initial contact), water evaporation stage (reflecting water vaporization caused by temperature rise of contact point), carbonization formation stage (reflecting resistance change of high-temperature carbonization of tree tissue), and open fire discharge stage (representing severe current release caused by continuous arc ignition).

[0060] The above division accurately locates the dynamic evolution path of the fault from germination to deterioration through the time distribution of extreme points and signal characteristics.

[0061] The stage division breaks through the dependence of traditional methods on single fault characteristics through characteristic analysis driven by physical processes, and realizes dynamic tracking of the whole life cycle of tree contact fault.

[0062] Specifically, the extreme point of the tree line contact stage marks the starting time of the fault, the extreme density change of the water evaporation stage reveals the decreasing trend of contact impedance, the extreme fluctuation of the carbonization formation stage reflects the mutation of resistance, and the extreme cluster of the open fire discharge stage corresponds to the outbreak of high-temperature arc.

[0063] This multi-stage analysis not only improves the sensitivity of fault detection (can identify weak contact signals), but also enhances the accuracy of fault location (through stage characteristic matching), providing a phased and differentiated basis for the active defense of power systems.

[0064] Step 3, most similar modal signal feature extraction: for the tree line contact stage , it is divided into continuous subintervals by equal time window ; according to the standard deviation formula, the standard deviation of the most similar modal current waveform in each subinterval is calculated respectively, and the standard deviation feature sequence is constructed to quantify the discrete feature of the signal amplitude of the tree line contact stage ; for the water evaporation stage , it is divided into b equal length subintervals by equal time window ; according to the slope formula, the slope of the most similar modal current waveform in each subinterval is calculated respectively, and the slope feature sequence is generated to represent the dynamic trend of signal change in the water evaporation stage .

[0065] Specifically, the step 3 includes the following specific steps:

[0066] Step 3.1, for the tree line contact stage , it is divided into continuous sub-intervals , the time window length is ; the standard deviation of the current waveform in each sub-interval is calculated respectively to construct a standard deviation feature sequence for quantifying the tree line contact stage the discreteness feature of signal amplitude; the time window length is:

[0067] ;

[0068] The standard deviation calculation formula is:

[0069] ;

[0070] In the formula, n is the number of sampling points, is the current value of the most similar modal signal corresponding to the q th sampling point, is the mean value of the most similar modal signal current, ;

[0071] Step 3.1 focuses on the tree line contact stage, which is divided into continuous sub-intervals by equal time window segmentation, and the standard deviation of the current waveform in each sub-interval is calculated to construct a standard deviation feature sequence to quantify the discreteness of signal amplitude; because the initial fault current of the tree line contact is weak and the fluctuation is random, the standard deviation can effectively capture the discreteness of the current amplitude, reflecting the dynamic characteristics of the unstable contact;

[0072] Step 3.2, for the water evaporation stage , it is divided into b equal-length sub-intervals by equal time window, the time window length is , by calculating the time domain slope of the waveform in each sub-interval, a slope feature sequence is generated to represent the dynamic trend of signal change in the water evaporation stage ; the time window length is:

[0073] ;

[0074] The slope calculation formula is:

[0075] ;

[0076] In the formula, is the change amount of the most similar modal signal current value in the j th segment, is the time j of the tChange j =1,2,…, b .

[0077] Step 3.2 focuses on the moisture evaporation stage. It also uses an equal time window to divide the signal into b sub-intervals, calculates the time-domain slope of the waveform in each sub-interval, and generates a slope feature sequence to characterize the dynamic trend of signal changes. As the fault develops, moisture evaporation leads to a decrease in contact impedance, and the change in current slope can reflect the aggravation of the fault. The slope feature sequence can accurately characterize this process.

[0078] This step, through phased feature extraction, achieves a refined characterization of the dynamic evolution of tree-contact faults; the standard deviation feature sequence of the tree-line contact stage can sensitively identify weak initial fault signals, avoiding interference from load fluctuations or noise; the slope feature sequence of the moisture evaporation stage accurately reflects the deepening process of the fault through dynamic trend analysis; the combination of the two forms a multi-dimensional, hierarchical fault feature system, which not only improves the sensitivity of fault detection (capturing weak contact signals) but also enhances the reliability of fault identification (verified through dynamic trends).

[0079] The aforementioned feature extraction method provides a quantitative basis for subsequent two-parameter fusion criteria, significantly improving the accuracy and robustness of early warning of tree-touching faults on overhead lines.

[0080] Step 4, Overhead line tree contact fault detection: Based on the standard deviation feature sequence extracted in Step 3. With slope feature sequence A two-parameter fusion criterion is constructed; the criterion content is: when the tree line touches the stage All sub-intervals The standard deviation satisfies And the water evaporation stage All sub-intervals The slope satisfies If the above-ground line tree-line contact fault alarm is triggered, the line is considered to be in normal operating condition. and The threshold value is set.

[0081] Specifically, in step 4, the standard deviation feature sequence of the tree line contact stage extracted in step 3 is used... Slope characteristic sequence of water evaporation stage A two-parameter fusion criterion is constructed.

[0082] The specific criterion is: during the tree line contact stage. All sub-intervals Standard deviation All are greater than the set threshold (Reflecting a significant increase in signal amplitude dispersion), and the water evaporation stage All sub-intervals Slope k j are less than a set threshold (indicating that the signal change trend tends to be flat or declining), it is determined that the overhead line tree line contact fault occurs, and an alarm is triggered; if the above conditions are not met, it is determined that the line is in a normal operating state.

[0083] wherein, and determined by fault simulation and historical data statistics optimization, to balance the detection sensitivity and false alarm rate.

[0084] The dual-parameter fusion criterion realizes multi-dimensional verification of the tree contact fault by combining the static discrete feature (standard deviation) and the dynamic change trend (slope); the high standard deviation in the tree line contact stage can capture the random fluctuations of the initial weak fault and avoid load noise interference; the low slope in the water evaporation stage can verify the stability of the fault development and exclude transient interference; the synergistic effect of the two can significantly improve the accuracy and robustness of fault detection, reduce the false alarm risk of a single parameter criterion, and ensure the sensitive identification of early faults through threshold optimization, thereby providing a reliable and timely fault warning mechanism for the power system.

[0085] As can be seen from the above, the method realizes accurate layering of the fault signal through variational mode decomposition, combines phased feature extraction and dual-parameter fusion criterion, and constructs a complete closed loop from signal deconstruction to fault determination: the most relevant mode is selected by using VMD and cosine similarity to ensure the relevance of feature extraction; the signal discreteness is quantified by the standard deviation in the tree line contact stage, and the dynamic trend is described by the slope in the water evaporation stage, forming a multi-dimensional fault representation; finally, based on the dual-parameter collaborative threshold determination, the sensitivity and anti-interference ability of fault detection are effectively balanced; the scheme significantly improves the accuracy of early identification of overhead line tree contact faults, reduces the false alarm rate, and provides efficient and reliable fault warning and disposal basis for the power system.

[0086] The working principle of the overhead line tree contact fault detection method based on zero sequence current segmented features is as follows:

[0087] 1. Variational mode decomposition (VMD)

[0088] Variational mode decomposition (VMD) is a new multi-resolution adaptive non-recursive signal decomposition technology; multiple modal components can be extracted from the original signal at the same time, and the center frequency of each component is matched through iterative optimization; the main steps are as follows:

[0089] 1) Let the original signal decomposition bem a modal component, p =1,2,…, m , the first p modal component IMF p ( t ) is:

[0090] ;

[0091] wherein: A p ( t ) is an amplitude function; Φ p ( t ) is a phase function; IMF p is a decomposed modal component.

[0092] 2) Perform Hilbert transform, spectral modulation and gradient two-norm operation on each modal component in turn to estimate the bandwidth of each modal component, and construct a constraint variational model as:

[0093] ;

[0094] wherein: m is the number of decomposed modes (positive integer), IMF p , ω p correspond to the first p modal component and the center frequency after decomposition, is the Dirac function, is the convolution operator.

[0095] 3) Make full use of the advantages of quadratic penalty factor and Lagrange operator , and convert the above constrained variational problem into an unconstrained variational problem, that is:

[0096] ;

[0097] 4) Use the alternating direction multiplier method to iteratively solve the unconstrained variational problem, u i , ω i and The iteration process is:

[0098] ;

[0099] wherein: is the residual obtained by Wiener filtering; This represents the number of iterations. The centroid of the power spectrum of the current mode function; Update the step size for the Lagrange operator; Performing an inverse Fourier transform, its real part is... .

[0100] 2. Cosine similarity

[0101] Cosine similarity is a metric used to measure the similarity between two vectors, specifically the angle between them. It measures the similarity between two vectors and is defined as follows:

[0102] ;

[0103] In the formula, cos( θ () represents the cosine similarity. n The number of sampling points, IMF p ( n ) is the first p One IMF component, p =1,2,…, m , S ( n () represents the original signal, IMF pq ( n ) is the first p The first IMF component q The current value at each sampling point, S q ( n ) is the first in the original signal q Current values ​​at each sampling point q =1,2,…, n ;

[0104] The values ​​of cosine similarity have the following meanings: 1 indicates that the two vectors are exactly the same; 0 indicates that the two vectors are orthogonal, that is, they have no similarity; -1 indicates that the two vectors are completely opposite.

[0105] Application example: Using PSCAD to build a simulation model of a single-phase tree contact fault in a 10kV overhead line system without branches, such as... Figure 2 As shown; the system consists of four lines: It is a 25km pure overhead line. It is a 5km pure cable line. It is a hybrid line consisting of a 6km cable and a 10km overhead line. It is a hybrid line consisting of a 5km overhead line and a 4km cable. The parameters of the overhead line and the cable are shown in Table 1. The simulation system was used to simulate the tree contact fault of the overhead line. The total simulation time was 55s, and the early fault was set to occur at 0s, with the fault duration being 55s.

[0106] Table 1

[0107]

[0108] Simulation analysis

[0109] The simulation setting of the overhead line tree contact fault is specifically shown in Figure 2 If the overhead line tree contact fault is not disposed in time, it may cause the high-resistance short-circuit current to continue to flow, and in severe cases, it can cause the line protection to act, leading to outage, even causing equipment damage, system collapse, and even major safety accidents such as fire. Therefore, accurate modeling and real-time monitoring of TSF are crucial for improving the fault diagnosis capability of the power system and ensuring the safe and stable operation of power facilities; the overhead line tree contact fault is simulated and analyzed, the effective value of the zero-sequence current is extracted, the VMD is used to decompose the effective value of the zero-sequence current into two layers, and the decomposition result is shown in Figure 5 The cosine similarity of each IMF component and the original signal is calculated, the local features of the effective value of the zero-sequence current are removed, and the global feature change trend of the fault evolution process is accurately captured, and the cosine similarity results corresponding to the two modes are shown in Figure 6 Secondly, the most similar mode is subjected to local extreme point detection, and the fault development process is divided into four characteristic stages, namely, the tree line contact stage (0, 14.77s), the water evaporation stage (14.77s, 26.73s), the carbonization formation stage (26.73s, 29.61s), and the open fire discharge stage (29.61s, 55s); thirdly, the most similar mode signals of the tree line contact stage and the water evaporation stage are respectively divided into 3 small segments according to the time axis (the time window length of the tree line contact stage sub-interval is 4.92s, and the time window length of the water evaporation stage sub-interval is 3.98s), the standard deviation of each small segment in the tree line contact stage is calculated, and the slope of each small segment in the water evaporation stage is calculated, and the extreme point distribution of the most similar mode during the tree contact fault and the waveforms of the tree line contact stage segmentation and the water evaporation stage segmentation are shown in Figure 7 Finally, the standard deviation feature and the slope feature are used to construct a double-parameter fusion criterion: when the standard deviations of all sub-intervals in the tree line contact stage satisfy , and the slopes of all sub-intervals in the water evaporation stage satisfy , the overhead line tree line contact fault alarm is triggered; otherwise, it is determined that the line is in a normal operating state; 0 is a set threshold; taking the C-phase fault as an example, the analysis result is as follows:

[0110] The simulation result of the zero-sequence current effective value when the line is in a normal operating state is shown in Figure 3The simulation result of the zero sequence current effective value when a single-phase tree contact fault occurs at a distance of 10 km from the bus on the pure overhead line l1 is shown in Table 2. Figure 4 The detection result of the zero sequence current effective value is shown in Table 2.

[0111] Table 2

[0112]

[0113] According to the result analysis of Table 2, the detection method of the application successfully realizes the accurate detection of the overhead line tree contact fault by analyzing the standard deviation of the most similar modal component signal of the zero sequence current effective value in the tree line contact stage and the slope characteristic in the water evaporation stage; according to Table 2, the standard deviation and the slope k j <0> in the fault state form a significant difference from the normal state, verifying the sensitivity and reliability of the criterion for the early tree line contact fault, and providing effective technical support for the active defense of the power system.

[0114] The application first decomposes the zero sequence current effective value signal by using VMD, screens out the most relevant modal function to the fault characteristics by using the cosine similarity, and accurately extracts the fault characteristic component; secondly, based on the extreme value distribution law of the most similar modal, the fault process is innovatively divided into four physical stages of tree line contact, water evaporation, carbonization formation and open fire discharge, and the fault evolution mechanism is revealed; the standard deviation characteristic sequence is quantized according to the standard deviation characteristic sequence of the tree line contact stage, and the time domain slope characteristic sequence is constructed according to the water evaporation stage to depict the dynamic trend, realizing the multi-dimensional feature collaborative representation; finally, the fault alarm is triggered through the double-parameter fusion criterion of the standard deviation and the slope , and the detection reliability is improved in combination with the logic and mechanism; the method provides a solution with theoretical explanatory and engineering practicality for tree line fault detection by driving feature engineering through physical process, and provides an extensible technical framework for complex fault diagnosis of distribution networks.

[0115] In summary: the beneficial effects of the application are:

[0116] (1) Line feature analysis: based on the extreme value distribution, the fault process is divided into four physical stages of tree line contact, water evaporation, carbonization formation and open fire discharge, and differential indexes are designed according to the characteristics of each stage: the standard deviation sequence is used to capture the current fluctuation dispersion in the early contact stage, and the time domain slope sequence is used to represent the resistance gradient trend in the water evaporation stage, realizing the deep association of the feature parameters and the physical mechanism; the method breaks through the limitation of traditional single feature analysis, and provides accurate quantitative basis for early warning and evolution tracking of faults.

[0117] (2) Signal processing: VMD decomposition combined with an adaptive mode selection strategy is adopted to effectively solve the mode aliasing problem of traditional EMD methods; the cosine similarity of IMF components is optimized to accurately capture the global feature change trend of the fault evolution process; the introduction of the extreme value sequence realizes the dynamic division of the fault stage; compared with the fixed threshold segmentation method, this method can adapt to the change of line parameters and still maintain high robustness under the scene of load fluctuation or intermittent arc.

[0118] (3) Fault detection criterion: the dual-parameter fusion criterion significantly improves the accuracy and reliability of detection, and the standard deviation and slope features are used to quantify the dispersion of signal amplitude in the tree line contact stage and the dynamic trend of signal change in the water evaporation stage, respectively, to fully capture the characteristics of the entire fault development process; at the same time, the multi-stage feature extraction method enhances the sensitivity to the early stage of the fault, and combined with the flexible threshold setting, it reduces the false positive rate and the false negative rate, and adapts to the needs of different operating environments; in addition, the criterion can realize early warning, which saves valuable time for operation and maintenance personnel and avoids fault deterioration, thereby improving the safety and stability of line operation, and its theory is clear and logical, easy to be applied in engineering, which provides an important guarantee for the safe and stable operation of the power system.

[0119] The application provides an overhead line tree contact fault detection device based on segmented features of zero sequence current, which is suitable for the overhead line tree contact fault detection method based on segmented features of zero sequence current.

[0120] A zero sequence current effective value signal decomposition module is used to decompose the zero sequence current effective value signal by using variational mode decomposition to obtain a plurality of intrinsic mode functions, and select the IMF with the maximum cosine similarity as the most similar mode by calculating the cosine similarity of each IMF component.

[0121] A fault process stage division module is used to detect local extreme points of the most similar mode to obtain an extreme value sequence, and divide the fault development process into four feature stages: tree line contact stage, water evaporation stage, carbonization formation stage and open fire discharge stage based on the extreme value distribution.

[0122] A most similar mode signal feature extraction module is used to extract signal features for the tree line contact stage and the water evaporation stage, wherein the tree line contact stage adopts equal time window segmentation and calculates the standard deviation of the current waveform in each sub-interval to construct a standard deviation feature sequence, and the water evaporation stage divides the time domain slope of the waveform in each sub-interval by equal time window segmentation to generate a slope feature sequence.

[0123] ​The overhead line tree contact fault detection module is configured to construct a double-parameter fusion criterion based on the extracted standard deviation feature sequence and the slope feature sequence: when the standard deviations of all subintervals in the tree line contact stage satisfy and the slopes of all subintervals in the water evaporation stage satisfy , the overhead line tree line contact fault alarm is triggered; otherwise, it is determined that the line is in a normal operating state; wherein and are threshold values.

[0124] As can be seen from the above, the zero-sequence current effective value signal decomposition module accurately extracts the most relevant modal function to the fault characteristics, the fault process stage division module innovatively divides the fault development into four physical stages, the most similar modal signal feature extraction module extracts the standard deviation and slope feature sequence for different stages, and finally the overhead line tree contact fault detection module constructs a double-parameter fusion criterion to realize early warning and accurate identification of the overhead line tree contact fault, thereby significantly improving the safety and stability of the power system.

[0125] Embodiments of the present application provide an electronic device suitable for the overhead line tree contact fault detection method based on zero-sequence current segmented features, comprising:

[0126] a memory configured to protect computer programs and data;

[0127] a processor configured to run system programs.

[0128] Embodiments of the present application provide a computer storage medium suitable for the overhead line tree contact fault detection method based on zero-sequence current segmented features, and the system and data are subjected to hierarchical security management according to security management requirements.

[0129] Those skilled in the art will appreciate that embodiments of the present application can be provided as a system or a computer program product. Therefore, the present application can be in the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can be in the form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program code.

[0130] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure One one or more flow or blocks in the flowchart block or blocks. Figure One one or more flow or blocks in the flowchart block or blocks.

[0131] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart block or blocks. Figure One one or more flow or blocks in the flowchart block or blocks. Figure One one or more flow or blocks in the flowchart block or blocks.

[0132] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure One one or more flow or blocks in the flowchart block or blocks. Figure One Figure One one or more flow or blocks in the flowchart block or blocks.

[0133] In one typical configuration, the computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0134] The memory can include non-persistent memory and / or volatile memory, such as a random access memory (RAM) including a cache area for the temporary storage of data. The memory can also include non-volatile memory, such as read only memory (ROM), electrically programmable read only memory (EPROM), or electrically erasable programmable read only memory (EEPROM), for the storage of software that is read during runtime. The memory is an example of computer readable media.

[0135] Computer-readable media includes permanent and non-permanent, movable and non-movable media, which can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette, magnetic tape disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information accessible by a computing device. According to the definition herein, computer-readable media does not include transitory media such as modulated data signals and carriers.

[0136] It should also be noted that the terms "comprising", "including", or any other variant thereof are intended to cover non-exclusive inclusions, so that a process, product or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or inherent to such a process, product or device. Without more limitations, the element defined by the statement "including a" does not exclude the presence of other identical elements in the process, product or device including the element.

[0137] The embodiments of the present application are given for example and description, although the embodiments of the present application have been shown and described above, it is understood that the above embodiments are exemplary and cannot be understood as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A method for detecting tree fault on overhead line based on zero sequence current segment features, characterized in that, The method comprises the following steps: Step 1, decomposing the zero-sequence current effective value signal by using the variational mode decomposition to obtain a plurality of intrinsic mode functions (IMFs), and then selecting the IMF with the maximum cosine similarity as the most similar mode by calculating the cosine similarity of each IMF component; Step 2, local extremum point detection is performed on the most similar modal to obtain an extremum sequence {P1, P2, …, P k}; and the most similar modal is divided into four fault development stages according to the extremum point distribution: an interval representing a tree contact stage, an interval representing a water evaporation stage, an interval representing a carbonization formation stage, an interval representing a fire discharge stage; Step 3, most similar modality signal feature extraction: for the tree line contact stage , it is segmented into a continuous subintervals with equal time windows, and the length of the time window is ; According to the standard deviation formula, the standard deviation of the most similar modal current waveform in each sub-interval is calculated respectively, and the standard deviation feature sequence is constructed for quantifying the tree line contact stage the discrete feature of signal amplitude; For the water evaporation stage , it is divided into b equal-length subintervals by equal time windows ; according to the slope formula, the slope of the most similar modal current waveform of each subinterval is calculated respectively to generate a slope feature sequence to represent the dynamic trend of signal change in the water evaporation stage . The step 3 comprises the following specific steps: Step 3.1, for the tree-line contact phase , is divided into a consecutive sub-intervals of equal time windows ; The standard deviation of the current waveform in each sub-interval is calculated respectively to construct a standard deviation feature sequence for quantifying the tree line contact stage Discrete feature of signal amplitude; time window length is: ； The standard deviation calculation formula is: ； wherein n is the number of sampling points, is the number of the q is the most similar modality signal current value corresponding to the is the most similar modality signal current mean value, i = 1, 2, …, a ; Step 3.2, for the water evaporation stage , divide it into b equal-length sub-intervals by equal-time windows, the length of the time window is , generate a slope feature sequence by calculating the time-domain slope of each sub-interval waveform to represent the dynamic trend of signal change in the water evaporation stage , the length of the time window is : ； The slope calculation formula is: ； wherein is the most similar modal signal current value change of the first j segment, is the time j change of the first t segment, j = 1, 2, …, b； Step 4, overhead line tree contact fault detection: based on the standard deviation feature sequence extracted in step 3 and the slope feature sequence , a two-parameter fusion criterion is constructed; the criterion content is: when the standard deviation of all sub-intervals of the tree line contact stage satisfies , and the slope of all sub-intervals of the moisture evaporation stage satisfies , the overhead line tree line contact fault alarm is triggered; otherwise, it is determined that the line is in normal operation state; and are set threshold values.

2. The method for detection of treeing faults in overhead lines based on zero sequence current segmentation characteristics as claimed in claim 1 wherein: The step 1 comprises the following specific steps: Step 1.1, using the variational mode decomposition to perform layer decomposition on the zero sequence current effective value signal, obtaining m m m ;​​ Step 1.2, calculating the cosine similarity of each IMF component and the original signal respectively, selecting the IMF with the maximum cosine similarity as the most similar mode, and the cosine similarity is calculated as follows: ； where cos( ) is the cosine similarity, θ n is the number of sampling points, IMF p n is the i-th IMF component, p p = 1, 2, …, m S n is the original signal, IMF pq n is the i-th IMF component, p q is the current value of the i-th sampling point in the i-th IMF component, S q n is the current value of the i-th sampling point in the original signal, q q = 1, 2, …, n .​​​​​​​​​ 3. An overhead line tree contact fault detection device based on zero sequence current segment features, characterized by: The device is suitable for the overhead line tree fault detection method based on the segmented characteristics of the zero-sequence current as claimed in any one of claims 1-2, and the device comprises: A zero-sequence current effective value signal decomposition module is configured to decompose the zero-sequence current effective value signal by using the variational mode decomposition to obtain a plurality of intrinsic mode functions (IMFs), and then select the IMF with the maximum cosine similarity as the most similar mode by calculating the cosine similarity of each IMF component; A fault process stage division module is configured to perform local extreme point detection on the most similar mode to obtain an extreme sequence, and divide the fault development process into four characteristic stages based on the extreme distribution: tree line contact stage, water evaporation stage, carbonization formation stage, and open fire discharge stage. A most similar mode signal feature extraction module is configured to extract signal features for the tree line contact stage and the water evaporation stage, wherein the tree line contact stage adopts equal time window segmentation and calculates the standard deviation of the current waveform in each sub-interval to construct a standard deviation feature sequence, and the water evaporation stage adopts equal time window segmentation and calculates the time domain slope of the waveform in each sub-interval to generate a slope feature sequence. The overhead line tree contact fault detection module is used for constructing a double-parameter fusion criterion based on the extracted standard deviation feature sequence and the slope feature sequence: when the standard deviations of all subintervals in the tree line contact stage satisfy , and the slopes of all subintervals in the water evaporation stage satisfy , the overhead line tree line contact fault alarm is triggered; otherwise, it is determined that the line is in a normal operating state; wherein and are set threshold values.

4. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that, The computer program is executed by the processor to realize the method of any one of claims 1-2.

5. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to realize the method of any one of claims 1-2.

Images