Method for extracting early fault features of elbow type head of ring main unit

Abstract

The invention discloses a ring main unit elbow type head early fault feature extraction method, and relates to the elbow type head fault detection technology field, and the method comprises the following steps: decomposing a fault current waveform through wavelet transformation, and determining whether an electric arc is generated; and multiple characteristics of the zero break duration in the fault current waveform are taken as comprehensive criteria, and the early fault of the elbow-shaped head is identified in a combined manner. According to the method, the occurrence frequency, the polarity effect and the time length change trend characteristics of the zero-break duration are used as comprehensive criteria to identify the early fault of the elbow-shaped head, and other types of early fault waveform data measured in a real-model test site and elbow-shaped head early fault real-model test data are used for identification, so that the effectiveness of the method is verified; and a foundation is laid for online monitoring of early faults of the elbow-shaped head.

Description

Technical Field

[0001] The invention relates to the technical field of elbow head fault detection, and more particularly to a method for extracting early fault features of an elbow head of a ring network cabinet. Background Art

[0002] In cable power supply systems, ring network power supply systems with high reliability, high power supply quality and low line loss are often used. Ring network cabinets are widely used due to their small size, compact structure, simple installation and operation, stable operation, etc. The elbow head of the ring network cabinet is made of silicone rubber material. As a matching accessory of the cable line, it is installed at the connection between the cable and the casing to play an insulating protection role and prevent the cable from discharging to the cabinet wall. Its failure has an important impact on the reliable operation of the ring network cabinet and the cable system.

[0003] The early failure of the elbow head of the ring main unit occurs in the fault transition period. During this stage, the equipment has problems such as aging of insulation materials and degradation of insulation performance due to long-term operation, resulting in some defects that are not enough to cause permanent failures. Due to the environment in which the equipment is located and the fault development mechanism, the arcs caused by these defects can often extinguish themselves. At the same time, the instantaneous arc is not enough to trigger the protection action. The equipment can still operate normally after the arc is extinguished, but the damage caused by the arc to the insulation material cannot be restored by itself. With the increase in the number of arcs in the early faults, the performance of the insulation material continues to decline, and eventually the equipment enters the permanent failure period.

[0004] At present, there are two main types of arc detection methods. One is to detect the sound, light, heat, electricity and other signals generated synchronously when the arc is burning, and the other is to detect the arc by judging and identifying the waveform. From the practical experience on the spot, the detection methods for the sound, light, heat, electricity and other signals of arc burning have a limited measurement range and are only effective for specific faults of specific equipment. However, for low-cost, large-scale and widely distributed power distribution equipment, the waveform judgment and identification method can be better applied in practice.

[0005] The development process of the early fault of the elbow head is divided into three stages: "water hose approaching-water hose boiling-water hose drying up", and the arc occurrence location and intensity in different stages are different. Through the mechanism analysis, simulation calculation and artificial experiment of the early fault of the elbow head of the ring network cabinet, it can be found that the early fault of the elbow head has the following characteristics: the zero-break duration of the fault current waveform in the water hose approaching stage gradually decreases until it disappears and turns into a sine waveform; in the water hose boiling stage, the fault current waveform has multiple "no zero break-zero break-no zero break" transitions, and the zero break duration change trend is first increased and then decreased; in the water hose drying stage, the fault current waveform changes from "no zero break" to "zero break", and the zero break duration gradually increases.

[0006] Therefore, how to extract the early fault characteristics of the fault current based on the above characteristics and design an identification method for the early fault of the elbow head is a technical problem that technical personnel in this field urgently need to solve. Summary of the invention

[0007] In view of this, the present invention provides a method for extracting early fault features of an elbow head of a ring main unit, which solves the problems existing in the background technology.

[0008] In order to achieve the above object, the present invention provides the following technical solutions: A method for extracting early fault features of a ring main unit elbow head comprises the following steps: Decompose the fault current waveform through wavelet transform to determine whether an arc has occurred; The multiple characteristics of zero-break duration in fault current waveform are used as comprehensive criteria to jointly identify early-stage elbow faults.

[0009] Optionally, determining whether an arc is generated comprises the following steps: Perform wavelet transform on the fault current waveform to obtain basic signals and high-frequency detail signals; The high-order harmonic information in the high-frequency detail signal is analyzed, and by comparing the energy intensity of the normal sinusoidal current and the fault current in different high-frequency bands, it is determined whether an arc has occurred.

[0010] Optionally, the specific process of performing wavelet transform on the fault current waveform is: The original waveform Discretize into ,in n is a positive integer, a discrete signal The Mallat algorithm is used to solve the problem; The discrete signal After passing through low-pass filters and high-pass filters, the signal is decomposed into basic signals and detail signals. The basic signals are further refined and decomposed into higher-level basic signals and detail signals according to research needs. The final decomposition results are shown in equations (1)-(2): (1); (2); Where: S represents the original signal, Decomposition to the n The basic signal after the layer, Decomposition to the n The detail signal after the layer.

[0011] Optionally, a db2 wavelet basis is used for wavelet transformation.

[0012] Optionally, the basis for judging whether an arc is generated is The specific judgment method for high-order harmonic signals within the frequency band is as follows: If equation (3) is satisfied, that is, the normalized current peak value of the high-frequency detail signal within a half-cycle is higher than the corresponding value of the sine wave, and the current zero-break phenomenon caused by the arc causes a sudden change in the high-frequency detail signal, then it is considered that an arc has occurred; (3); Where: It represents the peak value of each half cycle in the fifth layer of high-frequency detail signal of fault current wavelet decomposition. is the corresponding half-cycle peak value of the fault current; It represents the peak value of each half cycle in the fifth layer of high-frequency detail signal of the sine wave wavelet decomposition under normal operation. is the corresponding peak value of the sine wave half cycle.

[0013] Optionally, the early fault identification method of the elbow head is: After judging the occurrence of arc, take the fault mutation point as the reference point, detect the current zero break shoulder in the fault current in the direction of time earlier than the reference point and time later than the reference point, until the current zero break phenomenon disappears, and record the zero break duration and the number of zero break occurrences; The number of zero-rest flat shoulders that appear during the occurrence and extinction of an arc is greater than 4 times, which is used as one of the criteria for the early stage fault of an elbow head. At the same time, the quadratic function fitting of the zero-rest duration during the primary arc process is used to determine whether the arc fault is an early stage fault of an elbow head.

[0014] Optionally, whether the arc fault is an early-stage elbow fault is determined by fitting a quadratic function of the zero-break duration in the primary arc process, specifically: In the stage of water belt approaching, the duration of current zero rest is continuously reduced, the main trend of the curve is a straight line with a negative slope, and the first-order coefficient term is a negative value; In the drying stage of the water belt, the duration of the current zero rest continues to increase, the main trend of the curve is a straight line with a positive slope, and the first-order coefficient term is a positive value; In the boiling stage of water belt, the overall trend of the current zero rest duration is a parabola opening downward, and the quadratic coefficient term is a negative value; The zero-rest duration at the end of the water belt boiling stage is shorter than that at the middle stage, and the zero-rest duration at the end of the water belt drying stage is longer than that at the middle stage; Therefore, the corresponding fault stage is determined by judging the fitting result parameters of a current zero-break duration and the relationship between the zero-break durations in the final and middle stages; among them, formula (4) is used to identify the water belt approaching stage, formula (5) is used to identify the water belt boiling stage, and formula (6) is used to identify the water belt drying stage: (4); (5); (6); Where: A , B are the parameters obtained after fitting the current zero-off duration, corresponding to the quadratic coefficient term and the linear coefficient term respectively; Indicates the duration of the mid-term zero break. Indicates the duration of the final zero rest period.

[0015] Optionally, the method further includes: comparing the zero-rest duration of the positive and negative half cycles, and using the lack of polarity effect as a basis for determining an early-stage failure of the elbow head.

[0016] It can be seen from the above technical solution that compared with the prior art, the present invention discloses a method for extracting the early fault characteristics of the elbow head of a ring network cabinet, which decomposes the fault current waveform by wavelet and calculates The normalized current peak of the half-cycle high-frequency detail signal within the frequency band is compared with the sinusoidal waveform to determine the occurrence of arc; then the continuous fault current zero-rest shoulder times, the difference between positive and negative half-cycles, and the zero-rest duration fitting parameters in different stages are used as fault identification features to accurately identify the early faults of the elbow head, laying the foundation for further realizing the online monitoring of the early faults of the elbow head, timely discovery and replacement of defective elbow heads, which is of great significance to improving the operational reliability of the distribution network. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0018] Figure 1 A flow chart of a method for extracting early fault features of a ring main unit elbow head provided by the present invention; Figure 2 A schematic diagram of the wavelet decomposition waveform process provided by the present invention; Figure 3 This is the wavelet decomposition result of the current signal provided by the present invention. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] The embodiment of the present invention discloses a method for extracting early fault features of a ring main unit elbow head, comprising the following steps: Decompose the fault current waveform through wavelet transform to determine whether an arc has occurred; The multiple characteristics of zero-break duration in fault current waveform are used as comprehensive criteria to jointly identify early-stage elbow faults.

[0021] Next, through Figure 1 The specific process shown is used to explain the technical solution of the present invention in detail.

[0022] 1. Fault feature identification basis When an AC arc is struck, the voltage will produce an arcing spike, and when the arc is extinguished, the voltage will produce an arcing spike. At the same time, the arc current will have a short zero shoulder when it passes through zero. Compared with a sine wave, the arc current will produce a large number of high-order harmonics during the zero shoulder period and the arcing spike and arcing spike. Therefore, the specific method of judging whether an arc is generated given in this embodiment is: Perform wavelet transform on the fault current waveform to obtain basic signal and high-frequency detail signal; analyze the high-order harmonic information in the high-frequency detail signal, and judge whether an arc is generated by comparing the energy intensity of normal sinusoidal current and fault current in different high-frequency bands.

[0023] Based on the mechanism analysis of the early-stage fault of the elbow head and the results of equivalent tests and true-type tests, the arc in the early-stage fault of the elbow head occurs in the air gap between the elbow head and the bushing. The arc burning environment is relatively stable and is less affected by external factors. Therefore, in the artificial test results, the arc will continue to burn for more than two power frequency cycles after it appears. The number of consecutive zero-rest shoulders in the fault current waveform can be used as one of the bases for identifying the early-stage fault of the elbow head.

[0024] Through the mechanism analysis, simulation calculation and artificial experiment of the early fault of elbow head, it is found that the fault current waveform can be roughly divided into three stages. At the same time, the zero-break duration of the fault current waveform in each stage has certain characteristics: the zero-break duration gradually decreases until it disappears in the water belt approaching stage, and there are multiple transitions of "no zero break-zero break-no zero break" in the water belt boiling stage, and the zero-break duration change trend is first increased and then decreased; in the water belt drying stage, the fault current waveform changes from "no zero break" to "zero break", and the zero-break duration gradually increases. Therefore, by fitting the current zero-break flat shoulder time, the development trend of the zero-break flat shoulder time can be displayed, and this feature can be used as one of the bases for identifying the early fault of elbow head.

[0025] In addition, the materials on both sides of the arc in the early fault of the elbow head are conductive rubber and water hose or water hose and water hose. The material properties of the anode and cathode on both sides of the arc are not much different. Therefore, in the fault current waveform, the zero-rest shoulder will appear in both the positive and negative half-cycles, and the duration will not show obvious difference. The characteristics of the early faults of tree branches colliding with overhead bare wires and the pin insulators of distribution lines are significantly different from them. Therefore, the number of occurrences of the zero-rest shoulder and the duration characteristics in the positive and negative half-cycles can be used as one of the bases for identifying the early faults of the elbow head.

[0026] In summary, the characteristic information that can be used to identify the early fault of the elbow head includes: the mutation intensity of the decomposed high-order harmonic signal, the difference in zero-rest time between the positive and negative half-cycles, the number of consecutive zero-rest flat shoulders in the current waveform, and the fitting results of the zero-rest flat shoulder duration.

[0027] 2. Wavelet Decomposition In the field of arc waveform analysis, the general traditional method is Fourier transform, but this method can only reflect the characteristics of the waveform signal from the global characteristics, and it is difficult to observe the local characteristics of the waveform. Wavelet transform takes into account the advantages of the improved Fourier algorithm in extracting the typical characteristics of the waveform, and can also refine the advantages of the multi-scale detail characteristics of the waveform in different frequency bands. Therefore, this embodiment uses wavelet transform to decompose the fault current waveform, analyzes the high-order harmonics caused by the arc ignition, extinguishing, and zero rest, and uses the high-order harmonic current intensity displayed in the decomposition result as the feature for identifying the arc.

[0028] The schematic diagram of wavelet transform decomposition waveform is as follows Figure 2 As shown, the specific process is: The original waveform Discretize into ,in n is a positive integer, a discrete signal The Mallat algorithm is used to solve the problem; The discrete signal After passing through low-pass filters and high-pass filters, the signal is decomposed into basic signals and detail signals. If the research needs are not met, the basic signal is further refined and decomposed into higher-level basic signals and detail signals. The final decomposition results are shown in equations (1)-(2): (1); (2); Where: S represents the original signal, Decomposition to the n The basic signal after the layer, Decomposition to the n The detail signal after the layer.

[0029] 3. Feature Information Discrimination When performing wavelet decomposition, the wavelet basis used for decomposition will have an impact on the decomposition result. After comprehensively considering the support length, vanishing moment order, regularity and symmetry of each wavelet, this embodiment uses the db2 wavelet basis for wavelet decomposition. At the same time, if the wavelet decomposition order is too small, it is impossible to obtain sufficiently fine high-frequency detail signals; if the wavelet decomposition order is too large, the high-frequency detail signals will be significantly affected by noise. Therefore, when selecting the wavelet decomposition order, this embodiment performs 4, 5, and 6 wavelet decompositions on the normal sinusoidal current waveform obtained in the true test, and the fault current waveform fragments with current zero shoulder phenomenon in the water hose approaching stage, water hose boiling stage, and water hose drying stage. It is found that the high-frequency detail signal of the 5th layer obtained by using 5-layer decomposition is significantly different from that of other layers. Therefore, this embodiment uses the 5-layer wavelet decomposition results as the identification criterion. The decomposition results of the current waveform with zero shoulder are similar. Therefore, this embodiment takes the decomposition results obtained in the water hose approaching stage as an example. The results obtained after decomposition of the sinusoidal current waveform and the zero shoulder current waveform are as follows. Figure 3 As shown, (a) is the wavelet decomposition result of the sinusoidal current, and (b) is the wavelet decomposition result of the current in the water belt approaching stage.

[0030] from Figure 3 The results show that due to the zero-shoulder phenomenon of the arc current, a large amount of high-order harmonic information is generated in different frequency bands after wavelet decomposition. Figure 3 The decomposition results shown are due to The current signal amplitude in the frequency band is large and the clutter is small. The high-order harmonic signal corresponding to the arc is obviously prominent. Therefore, this embodiment comprehensively considers the amplitude height and vanishing order of each decomposition result and uses The high-order harmonic signals within the frequency band are used as the basis for judging the arcing of the fault current. The specific judgment method is: If equation (3) is satisfied, that is, the normalized current peak value of the high-frequency detail signal within a half-cycle is higher than the corresponding value of the sine wave, and the current zero-break phenomenon caused by the arc causes a sudden change in the high-frequency detail signal, then it is considered that an arc has occurred; (3); Where: It represents the peak value of each half cycle in the fifth layer of high-frequency detail signal of fault current wavelet decomposition. is the corresponding half-cycle peak value of the fault current; It represents the peak value of each half cycle in the fifth layer of high-frequency detail signal of the sine wave wavelet decomposition under normal operation. is the corresponding peak value of the sine wave half cycle.

[0031] Table 1 lists the fault current waveforms in the three stages of the elbow head early fault calculated according to formula (3) after wavelet decomposition. The maximum normalized current peak value of the half-cycle high-frequency detail signal in the frequency band and the corresponding zero-off time. As can be seen from Table 1, the longer the zero-off time, the more serious the current waveform distortion, the greater the mutation of the decomposed high-frequency signal, and the greater the normalized current peak value of the half-cycle high-frequency detail signal, which can be used to detect the occurrence of arcs.

[0032] Table 1 Three stages of early fault and corresponding values ​​of sinusoidal current Next, after determining that an arc has occurred, take the fault mutation point as the reference point, and detect the current zero-break shoulder in the fault current in the direction of time earlier than the reference point and time later than the reference point until the current zero-break phenomenon disappears, and record the zero-break duration and the number of zero-break occurrences.

[0033] Generally speaking, the number of zero-rest flat shoulders in the process of arc appearance and extinction in an elbow-type early-stage fault is greater than 4 times, so this rule can be used as one of the criteria for judging the elbow-type early-stage fault; at the same time, by fitting the quadratic function of the zero-rest duration in the primary arc process, it is judged whether the arc fault is an elbow-type early-stage fault, and the zero-rest duration fitting results shown in Table 2 are obtained.

[0034] Table 2 Zero rest duration fitting results Furthermore, whether the arc fault is an early-stage elbow fault is determined by fitting the quadratic function of the zero-break duration in the primary arc process, specifically: In the stage of water belt approaching, the duration of current zero rest is continuously reduced, the main trend of the curve is a straight line with a negative slope, and the first-order coefficient term is a negative value; In the drying stage of the water belt, the duration of the current zero rest continues to increase, the main trend of the curve is a straight line with a positive slope, and the first-order coefficient term is a positive value; In the boiling stage of water belt, the overall trend of the current zero rest duration is a parabola opening downward, and the quadratic coefficient term is a negative value; The zero-rest duration at the end of the water belt boiling stage is shorter than that at the middle stage, and the zero-rest duration at the end of the water belt drying stage is longer than that at the middle stage; Therefore, the corresponding fault stage is determined by judging the fitting result parameters of a current zero-break duration and the relationship between the zero-break durations in the final and middle stages; among them, formula (4) is used to identify the water belt approaching stage, formula (5) is used to identify the water belt boiling stage, and formula (6) is used to identify the water belt drying stage: (4); (5); (6); Where: A , B are the parameters obtained after fitting the current zero-off duration, corresponding to the quadratic coefficient term and the linear coefficient term respectively; Indicates the duration of the mid-term zero break. Indicates the duration of the final zero rest period.

[0035] In addition, according to the changing trend of the zero-break duration, it can be seen that in the early fault of the elbow head, the arc occurs between materials with similar properties, without the involvement of metal, and there is no obvious difference in the zero-break duration in the positive and negative half-cycles. Therefore, by comparing the zero-break duration in the positive and negative half-cycles, the lack of polarity effect can be used as a basis for judging the early fault of the elbow head.

[0036] 4. Test Data Verification In order to verify the accuracy of the method of the present invention, this embodiment conducts tree branch touching overhead bare conductor, early fault of pin insulator of distribution line, and cable creepage test at the same real test site, and the measured fault waveform data and the fault waveform data obtained by the elbow head real test conducted in this embodiment are used together to verify the practicality of the proposed identification method. The fault current waveform acquisition method of various tests is the same as the elbow head early fault equivalent test and real test conducted in this embodiment.

[0037] Waveform recording devices have now developed to a relatively mature stage and can perform high-precision online monitoring of power supply lines in distribution networks. In practical applications, current information at the fault site can be obtained through waveform recording devices on both sides of the fault site. The power supply distance of the cable power supply system in a 10kV distribution network is generally within 3km~5km, so there is no need to consider the attenuation of traveling waves during transmission.

[0038] The tree itself has a large resistance, and the arc caused by the branch touching the overhead bare wire is a typical high-resistance grounding fault, which is similar to the characteristics of the high-resistance grounding fault due to the large resistance of the water hose in the early stage of the elbow-type fault. However, the main characteristics of the early stage of the tree-to-wire fault are that the arc gradually lengthens during the separation process of the branch and the wire, and the zero-break shoulder time of the fault current gradually increases. However, the zero-break duration of the fault current that occurs during the boiling stage of the water hose in the early stage of the elbow-type fault does not exist in the characteristics of the early stage of the tree-to-wire fault, and the trend of increasing first and then decreasing does not exist.

[0039] The early faults of pin insulators in distribution lines occur in rainy and foggy weather, when moisture seeps into the cracks of insulators and causes self-extinguishing arcs. The electrodes on both sides of the arc are aluminum conductors and electrolytes, so the fault waveform has a more obvious polarity effect, and the probability of arcing in the positive half cycle is higher than that in the negative half cycle. However, the electrodes on both sides of the arc in the early faults of elbow-type heads are conductive rubber and water hose or water hose and water hose, and the material properties are similar, so the fault current waveform does not have a polarity effect, and there is no obvious difference in the zero-break duration of the positive and negative half cycles.

[0040] The cable runs in the cable well. Due to the high humidity, the cable may be damp, which may cause water to enter the cable body and cable accessories such as the intermediate joints, resulting in creepage. This type of fault is affected by the generation of organic matter at high temperature of cross-linked polyethylene insulation and the narrow space. The fault current waveform has a large number of unstable transient characteristics, which is significantly different from the early fault current waveform of the elbow head.

[0041] By inputting 168 fault current waveform data obtained from tree branches touching overhead bare wires, early faults of pin insulators on distribution lines, cable creepage, and elbow head real-type tests, the fault current waveform data was obtained by using the fault current waveform data based on Figure 1 The process shown in the figure shows that 32 real test data of elbow head early faults are successfully identified, which verifies the effectiveness of the elbow head early fault identification method proposed in this paper.

[0042] V. Conclusion This embodiment combines mechanism analysis, simulation calculation, and artificial test results to design a feature extraction method for early-stage elbow head faults. First, the fault current waveform is decomposed by wavelet to calculate The normalized current peak of the half-cycle high-frequency detail signal within the frequency band is compared with the sinusoidal waveform to determine the occurrence of arc; then the number of consecutive fault current zero-rest shoulders, the difference between positive and negative half-cycles, and the zero-rest duration fitting parameters in different stages are taken as fault identification features, and the overall identification method process is designed; finally, combined with other types of early fault current waveforms in the experiment, the effectiveness of the designed method is verified.

[0043] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0044] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for extracting early fault features of elbow head of ring main unit, characterized in that: The following steps are involved: Decompose the fault current waveform through wavelet transform to determine whether an arc has occurred; The multiple characteristics of zero-break duration in fault current waveform are used as comprehensive criteria to jointly identify early-stage elbow faults.

2. The method for extracting early fault features of elbow head of ring main unit according to claim 1 is characterized in that: Determining whether an arc is generated includes the following steps: Perform wavelet transform on the fault current waveform to obtain basic signals and high-frequency detail signals; The high-order harmonic information in the high-frequency detail signal is analyzed, and by comparing the energy intensity of the normal sinusoidal current and the fault current in different high-frequency bands, it is determined whether an arc has occurred.

3. A method for extracting early fault features of elbow head of ring main unit according to claim 2, characterized in that: The specific process of wavelet transform of fault current waveform is as follows: The original waveform Discretize into ,in n is a positive integer, a discrete signal The Mallat algorithm is used to solve the problem; The discrete signal After passing through low-pass filters and high-pass filters, the signal is decomposed into basic signals and detail signals. The basic signals are further refined and decomposed into higher-level basic signals and detail signals according to research needs. The final decomposition results are shown in equations (1)-(2): （1）； （2）； Where: S represents the original signal, Decomposition to the n The basic signal after the layer, Decomposition to the n The detail signal after the layer.

4. The method for extracting early fault features of elbow head of ring main unit according to claim 1 is characterized in that: The db2 wavelet basis is used for wavelet transformation.

5. The method for extracting early fault features of elbow head of ring main unit according to claim 2 is characterized in that: The basis for judging whether an arc is generated is The specific judgment method for high-order harmonic signals within the frequency band is as follows: If equation (3) is satisfied, that is, the normalized current peak value of the high-frequency detail signal within a half-cycle is higher than the corresponding value of the sine wave, and the current zero-break phenomenon caused by the arc causes a sudden change in the high-frequency detail signal, then it is considered that an arc has occurred; （3）； Where: It represents the peak value of each half cycle in the fifth layer of high-frequency detail signal of fault current wavelet decomposition. is the corresponding half-cycle peak value of the fault current; It represents the peak value of each half cycle in the fifth layer of high-frequency detail signal of the sine wave wavelet decomposition under normal operation. is the corresponding peak value of the sine wave half cycle.

6. The method for extracting early fault features of elbow head of ring main unit according to claim 1 is characterized in that: The identification method of early fault of elbow head is: After judging the occurrence of arc, take the fault mutation point as the reference point, detect the current zero break shoulder in the fault current in the direction of time earlier than the reference point and time later than the reference point, until the current zero break phenomenon disappears, and record the zero break duration and the number of zero break occurrences; The number of zero-rest flat shoulders that appear during the occurrence and extinction of an arc is greater than 4 times, which is used as one of the criteria for the early stage fault of an elbow head. At the same time, the quadratic function fitting of the zero-rest duration during the primary arc process is used to determine whether the arc fault is an early stage fault of an elbow head.

7. A method for extracting early fault features of elbow head of ring main unit according to claim 6, characterized in that: By fitting the quadratic function of the zero-break duration in the primary arc process, it is determined whether the arc fault is an early elbow fault, specifically: In the stage of water belt approaching, the duration of current zero rest is continuously reduced, the main trend of the curve is a straight line with a negative slope, and the first-order coefficient term is a negative value; In the drying stage of the water belt, the duration of the current zero rest continues to increase, the main trend of the curve is a straight line with a positive slope, and the first-order coefficient term is a positive value; In the boiling stage of water belt, the overall trend of the current zero rest duration is a parabola opening downward, and the quadratic coefficient term is a negative value; The zero-rest duration at the end of the water belt boiling stage is shorter than that at the middle stage, and the zero-rest duration at the end of the water belt drying stage is longer than that at the middle stage; Therefore, the corresponding fault stage is determined by judging the fitting result parameters of a current zero-break duration and the relationship between the zero-break durations in the final and middle stages; among them, formula (4) is used to identify the water belt approaching stage, formula (5) is used to identify the water belt boiling stage, and formula (6) is used to identify the water belt drying stage: （4）； （5）； （6）； Where: A , B are the parameters obtained after fitting the current zero-off duration, corresponding to the quadratic coefficient term and the linear coefficient term respectively; Indicates the duration of the mid-term zero break. Indicates the duration of the final zero rest period.

8. The method for extracting early fault features of elbow head of ring main unit according to claim 1 is characterized in that: Also includes: The zero-rest duration of the positive and negative half cycles is compared, and the lack of polarity effect is used as the basis for judging the early failure of the elbow head.

Images