Fault section location method for power distribution network based on dynamic time warping and structural similarity

Abstract

The application discloses a power distribution network fault section positioning method based on dynamic time warping and structural similarity. First, the transient zero sequence current sequence of each monitoring point is collected after the fault occurs; then, the dynamic time warping similarity and the time-frequency structural similarity between the transient zero sequence current sequences of adjacent monitoring points are calculated; if the section corresponding to the minimum value of the dynamic time warping similarity between the transient zero sequence current sequences of adjacent monitoring points is the same section as the section corresponding to the minimum value of the time-frequency structural similarity between the transient zero sequence current sequences of adjacent monitoring points, the section is the fault section; otherwise, the comprehensive similarity between the transient zero sequence current sequences of adjacent monitoring points is further calculated, and the section corresponding to the minimum value of the comprehensive similarity is taken as the fault section. The method can effectively solve the problems of weak high-resistance grounding fault characteristics, signal susceptible to noise interference and the like, and improves the reliability, robustness and engineering applicability of the power distribution network fault section positioning.

Description

Technical Field

[0001] This invention belongs to the field of distribution network fault location technology, and specifically relates to a method for locating fault sections in distribution networks based on dynamic time warping and structural similarity. Background Technology

[0002] The distribution network is a crucial component of the power transmission system. Its complex network structure results in a high probability of feeder faults, with single-phase grounding faults accounting for a significant proportion. To ensure the safe and stable operation of the distribution network, it is essential to quickly and accurately locate faulty sections, providing a reliable basis for subsequent fault isolation, maintenance, and power restoration. For low-resistance grounding faults, existing detection and location technologies offer high reliability due to their significant fault characteristics. However, when high-resistance grounding faults occur in the distribution network, the subtle fault characteristics present a significant challenge to achieving accurate fault location.

[0003] Currently, existing methods for fault location in distribution networks mainly include the traveling wave method, the matrix method, and the artificial intelligence method. The traveling wave-based fault location method boasts high accuracy and fast response speed. However, in medium-voltage distribution networks with multiple branches and complex topologies, traveling waves are easily affected by the superposition of reflected waves and wavefront identification errors. In engineering applications, extensive deployment of fault points is often required to improve ranging accuracy, leading to increased construction and maintenance costs and hindering widespread adoption in medium-voltage distribution networks. With the continuous development of distribution networks, their complex network structure makes the construction of the matrix method even more difficult, making it difficult to adapt to the flexible and changing operating conditions of distribution networks. Artificial intelligence-based fault location methods offer good accuracy and fault tolerance, but their parameter settings are sensitive, computational complexity is high, and their fault tolerance for information distortion is limited. Therefore, existing fault location methods are significantly affected by complex topologies, changes in operating conditions, and information distortion, making it difficult to guarantee location stability and accuracy during actual distribution network operation. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the technical problem this invention aims to solve is to provide a method for locating fault sections in distribution networks based on dynamic time warping and structural similarity.

[0005] The present invention solves the aforementioned technical problem by adopting the following technical solution: A method for locating fault sections in a distribution network based on dynamic time warping and structural similarity, characterized by the following steps: Step 1: After a high-resistance grounding fault occurs in the distribution network, collect the transient zero-sequence current sequence at each monitoring point on the line; Step 2: Perform dynamic time warping on the transient zero-sequence current sequences of adjacent monitoring points, and calculate the dynamic time warping similarity between the transient zero-sequence current sequences of adjacent monitoring points; perform wavelet transform on the transient zero-sequence current sequences of each monitoring point to obtain the time-frequency feature matrix; calculate the structural similarity between the time-frequency feature matrices of adjacent monitoring points to obtain the time-frequency structural similarity between the transient zero-sequence current sequences of adjacent monitoring points; Step 3: If the segment corresponding to the minimum dynamic time warping similarity between the transient zero-sequence current sequences of adjacent monitoring points is the same segment as the segment corresponding to the minimum time-frequency structure similarity between the transient zero-sequence current sequences of adjacent monitoring points, then the segment is the fault segment; otherwise, calculate the comprehensive similarity between the transient zero-sequence current sequences of adjacent monitoring points according to formula (13), and take the segment corresponding to the minimum comprehensive similarity as the fault segment; (13) In the formula, Adjacent monitoring points and Comprehensive similarity between transient zero-sequence current sequences and These are the weighting coefficients, Adjacent monitoring points and Dynamic time-warped similarity between transient zero-sequence current sequences Adjacent monitoring points and Time-frequency structural similarity between transient zero-sequence current sequences.

[0006] Compared with the prior art, the present invention has the following beneficial effects: To address the issues of time misalignment, frequency component dispersion, and weak features easily submerged by noise in transient signals during field measurements, this method extracts the transient zero-sequence current signal after a fault occurs and constructs a fault location criterion combining dynamic time warping similarity and time-frequency structure similarity. This enables accurate identification of high-resistance grounding fault sections. This method effectively solves the problems of weak high-resistance grounding fault features, susceptibility to noise interference, and insufficient accuracy of traditional location methods. It improves the reliability, robustness, and engineering applicability of fault section location in distribution networks, providing effective technical support for rapid fault isolation and safe and stable operation of distribution networks. Attached Figure Description

[0007] Figure 1 This is the overall flowchart; Figure 2 The diagram shows the amplitude of transient zero-sequence current upstream and downstream of the fault point; Figure 3 The time-frequency diagram shows the transient zero-sequence current upstream of the fault point. Figure 4The time-frequency diagram shows the transient zero-sequence current downstream of the fault point. Figure 5 This is a schematic diagram of the dynamic time warping principle; Figure 6 The faulty circuit topology diagram is shown in the example. Detailed Implementation

[0008] Specific embodiments are given below with reference to the accompanying drawings. These specific embodiments are only used to describe the technical solution of the present invention in detail and are not intended to limit the scope of protection of this application.

[0009] like Figure 1 As shown, this invention provides a method for locating fault sections in a distribution network based on dynamic time warping and structural similarity, comprising the following steps: Step 1: After a high-resistance grounding fault occurs in the distribution network, collect the transient zero-sequence current sequence at each monitoring point on the line; The transient fault process is mainly concentrated within half to one cycle after the fault occurs and has a short duration. At the moment the fault occurs, the characteristics of the transient zero-sequence current are very obvious. Therefore, it is more effective to analyze the fault process using the transient zero-sequence current signal. Based on the zero-sequence equivalent network, the time-frequency domain difference analysis of the transient zero-sequence current is performed. Analyzing the fault characteristics is actually equivalent to analyzing the zero-state response generated by the fault point under the excitation of the virtual power source.

[0010] Let the equivalent voltage at the fault point be... for: (1) In the formula, This represents the phase voltage amplitude at the fault location during normal operation. It is the power frequency angular frequency. The initial phase angle at the fault point; For a neutral-point ungrounded system, Kirchhoff's laws give the equivalent voltage at the fault point as: (2) In the formula, This refers to the zero-sequence bus voltage of a neutral-point ungrounded system. For transition resistance, This is the sum of the capacitances of all lines to ground.

[0011] Substituting equation (1) into equation (2), we get: (3) Solving the differential equation (3), we can obtain the zero-sequence voltage of the busbar in the neutral-point ungrounded system as follows: (4) In the formula, The phase lag angle caused by the zero-sequence equivalent network to the power frequency sinusoidal quantity; Based on the bus zero-sequence voltage of a neutral-point ungrounded system, the transient zero-sequence current upstream of the fault point can be obtained. for: (5) Downstream transient zero-sequence current at the fault point for: (6) The ratio of the upstream and downstream transient zero-sequence currents at the fault point is: (7) In the formula, This is the capacitance to ground of the faulty line. To take the absolute value; From equation (7) and Figure 2 It can be seen that the zero-sequence current amplitude is higher upstream of the fault point than downstream of the fault point. Therefore, there are obvious differences in the time domain characteristics of the transient zero-sequence current on both sides of the fault point, which have strong distinguishability.

[0012] At the same time, such as Figure 3 , 4 As shown, frequency domain characteristic analysis of the transient zero-sequence current sequence at the monitoring point reveals significant differences in the frequency component distribution and amplitude of the transient zero-sequence current upstream and downstream of the fault point, indicating that the frequency domain characteristics of the transient zero-sequence current on both sides of the fault point also have strong distinguishability.

[0013] The time-frequency domain analysis results of the transient zero-sequence currents on both sides of the above-mentioned fault point provide a theoretical basis for using dynamic time regularity and time-frequency structure similarity as criteria to locate the fault section.

[0014] Step 2: Perform Dynamic Time Warping (DTW) on the transient zero-sequence current sequences of adjacent monitoring points, and calculate the DTW similarity between the transient zero-sequence current sequences of adjacent monitoring points; simultaneously, perform wavelet transform on the transient zero-sequence current sequences of each monitoring point to obtain the time-frequency feature matrix; calculate the structural similarity between the time-frequency feature matrices of adjacent monitoring points to obtain the time-frequency structural similarity between the transient zero-sequence current sequences of adjacent monitoring points; Set up two adjacent monitoring points , transient zero-sequence current sequence , They are respectively: (8) (9) In the formula, For the first Monitoring points at each sampling time The transient zero-sequence current, ; For the first Monitoring points at each sampling time The transient zero-sequence current, ; Construct the distance matrix between the transient zero-sequence current sequences of adjacent monitoring points according to equation (10): (10) In the formula, The distance matrix is ​​the first... Line 1 Column elements; like Figure 5 As shown, the principle of dynamic time warping is to find the path with the minimum cumulative distance in the distance matrix to achieve adaptive alignment of two transient zero-sequence current sequences; the dynamic time warping distance is calculated based on the distance matrix and normalized to obtain the dynamic time warping similarity of transient zero-sequence current sequences of adjacent monitoring points. (11) (12) In the formula, Adjacent monitoring points and The dynamic time warp distance between transient zero-sequence current sequences Adjacent monitoring points and Dynamic time-warped similarity between transient zero-sequence current sequences The path with the minimum cumulative distance; Wavelet transform is performed on the transient zero-sequence current sequences of each monitoring point to obtain the time-frequency feature matrix of each monitoring point; the structural similarity index (SSIM) algorithm is used to calculate the structural similarity between the time-frequency feature matrices of adjacent monitoring points to obtain the time-frequency structural similarity between the transient zero-sequence current sequences of adjacent monitoring points.

[0015] Step 3: Locate the fault segment based on dynamic time warping similarity and time-frequency structure similarity; If the segment corresponding to the minimum dynamic time warping similarity between the transient zero-sequence current sequences of adjacent monitoring points is the same segment as the segment corresponding to the minimum time-frequency structure similarity between the transient zero-sequence current sequences of adjacent monitoring points, then the segment is the fault segment; otherwise, the comprehensive similarity between the transient zero-sequence current sequences of adjacent monitoring points is further calculated according to formula (13), and the segment corresponding to the minimum comprehensive similarity is taken as the fault segment. (13) In the formula, Adjacent monitoring points and Comprehensive similarity between transient zero-sequence current sequences and These are the weighting coefficients, satisfying... , Adjacent monitoring points and Time-frequency structural similarity between transient zero-sequence current sequences.

[0016] Example This embodiment takes a phase A ground fault caused by foreign object contact downstream of tower No. 60 on a power distribution network in North China as an example. The topology is as follows: Figure 6 As shown, a waveform measuring device was installed at towers 1, 31, 45, 60, 61, 104, and 134 on the faulty line, and these were recorded as monitoring points in sequence. ~ The sampling frequency is 4096Hz, and the transient zero-sequence current sequence of each waveform measuring device at the corresponding monitoring point is used.

[0017] Dynamic time warping is performed on the transient zero-sequence current sequences of adjacent monitoring points, and the dynamic time warping similarity of adjacent sequences is calculated as follows:

[0018] Wavelet transform is performed on the transient zero-sequence current at each monitoring point to extract the time-frequency feature matrix. The time-frequency structural similarity of the fingertips of the transient zero-sequence current sequences at adjacent monitoring points is calculated as follows:

[0019] Since both the low dynamic time warp similarity segment and the low time-frequency structural similarity segment are segment 4 (i.e., between tower 60 and tower 61), segment 4 is identified as the fault segment, which is consistent with the location of the foreign object contact fault confirmed by the on-site inspection. This verifies the effectiveness of the method of the present invention in locating high-resistance grounding fault segments in actual power distribution networks.

[0020] In summary, the method of the present invention can achieve rapid and accurate location of high-resistance grounding fault sections in distribution networks, and to a certain extent overcomes the limitations of complex topology, changes in operating conditions and information distortion, thereby providing effective guidance information for subsequent rapid fault handling.

[0021] Any aspects not covered in this invention are applicable to existing technologies.

Claims

1. A method for locating fault sections in a distribution network based on dynamic time warping and structural similarity, characterized in that, Includes the following steps: Step 1: After a high-resistance grounding fault occurs in the distribution network, collect the transient zero-sequence current sequence at each monitoring point on the line; Step 2: Perform dynamic time warping on the transient zero-sequence current sequences of adjacent monitoring points, and calculate the dynamic time warping similarity between the transient zero-sequence current sequences of adjacent monitoring points; Wavelet transform is performed on the transient zero-sequence current sequence at each monitoring point to obtain the time-frequency feature matrix; Calculate the structural similarity between the time-frequency feature matrices of adjacent monitoring points to obtain the time-frequency structural similarity between the transient zero-sequence current sequences of adjacent monitoring points; Step 3: If the segment corresponding to the minimum dynamic time warping similarity between the transient zero-sequence current sequences of adjacent monitoring points is the same segment as the segment corresponding to the minimum time-frequency structure similarity between the transient zero-sequence current sequences of adjacent monitoring points, then the segment is the fault segment; otherwise, calculate the comprehensive similarity between the transient zero-sequence current sequences of adjacent monitoring points according to formula (13), and take the segment corresponding to the minimum comprehensive similarity as the fault segment; （13） In the formula, Adjacent monitoring points and Comprehensive similarity between transient zero-sequence current sequences and These are the weighting coefficients, Adjacent monitoring points and Dynamic time-warped similarity between transient zero-sequence current sequences Adjacent monitoring points and Time-frequency structural similarity between transient zero-sequence current sequences.

Images