Power distribution network power transmission line fault positioning method for large-scale distributed power supply access

Abstract

The invention relates to the technical field of power transmission line fault positioning, in particular to a power distribution network power transmission line fault positioning method for large-scale distributed power supply access, and solves the technical problem of poor fault point positioning effect of a complex power distribution network in the prior art. The method comprises the following steps: acquiring an actually measured voltage sequence and an actually measured current sequence after a power distribution network power transmission line connected with a large-scale distributed power supply has a fault; determining a linear reference current sequence based on the actually measured voltage sequence and the actually measured current sequence; generating a reference energy density sequence based on the linear reference current sequence, and generating an actually measured energy density sequence based on the actually measured current sequence; extracting an instantaneous phase lag sequence from the reference energy density sequence and the actually measured energy density sequence according to a discrete optimal transmission model; determining a lag fluctuation index based on the instantaneous phase lag sequence; and determining a network fault type according to the lagging fluctuation index, and determining a fault position based on a corresponding fault positioning mode.

Description

Technical Field

[0001] This application relates to the field of transmission line fault location technology, specifically to a method for locating transmission line faults in distribution networks with large-scale distributed power source access. Background Technology

[0002] With the large-scale integration of inverter-type distributed power sources, such as photovoltaics and wind power, into distribution networks, the topology and fault characteristics of these networks are becoming increasingly complex, placing higher demands on the rapid and accurate location of transmission line faults. Traditional fault location methods are mainly based on single-ended electrical quantity measurement and impedance calculation principles, which are reliable in radial networks with a single power source. However, when faced with complex distribution networks containing a high proportion of distributed power sources, especially under the constraint of only being able to utilize single-ended waveform recording data from the substation side, existing location methods show poor overall fault location performance, making it difficult to meet the requirements for location accuracy and reliability in actual operation and maintenance. Summary of the Invention

[0003] To address the poor fault location accuracy in existing technologies for complex distribution networks, this application aims to provide a fault location method for transmission lines in distribution networks with large-scale distributed power generation. The specific technical solution adopted is as follows: Obtain the measured voltage and current sequences after a fault in the transmission line of a distribution network with large-scale distributed power generation access; Based on the measured voltage and current sequences, a linear reference current sequence characterizing the response of a purely passive linear network is determined. A reference energy density sequence is generated based on a linear reference current sequence, and a measured energy density sequence is generated based on a measured current sequence. The instantaneous phase lag sequence is extracted from the reference energy density sequence and the measured energy density sequence based on the discrete optimal transmission model. The instantaneous phase lag sequence is used to characterize the change of the local time lag of the measured current waveform relative to the linear reference current waveform over time. Based on the instantaneous phase lag sequence, the lag volatility index is determined, which is used to quantify the volatility of the instantaneous phase lag sequence. The network fault type is determined based on the hysteresis fluctuation index, and the fault location is determined based on the fault location method corresponding to the network fault type.

[0004] In one possible implementation, a linear reference current sequence characterizing the response of a purely passive linear network is determined based on the measured voltage sequence and the measured current sequence. This includes: extracting the fundamental voltage amplitude within the first power frequency cycle after the fault from the measured voltage sequence, and extracting the fundamental current amplitude within the first power frequency cycle after the fault from the measured current sequence; calculating the ratio of the fundamental current amplitude to the fundamental voltage amplitude to obtain the fundamental equivalent admittance modulus; and performing a linear transformation on the measured voltage sequence based on the fundamental equivalent admittance modulus to determine the linear reference current sequence.

[0005] In one possible implementation, generating a reference energy density sequence based on a linear reference current sequence and generating a measured energy density sequence based on a measured current sequence include: calculating the square of the instantaneous values ​​of the sampling points in the linear reference current sequence to obtain a reference instantaneous energy sequence; normalizing the reference instantaneous energy sequence to obtain a reference energy density sequence; calculating the square of the instantaneous values ​​of the sampling points in the measured current sequence to obtain a measured instantaneous energy sequence; and normalizing the measured instantaneous energy sequence to obtain a measured energy density sequence.

[0006] In one possible implementation, the instantaneous phase lag sequence is extracted from the reference energy density sequence and the measured energy density sequence according to the discrete optimal transmission model, including: constructing a time difference penalty matrix, where each element of the time difference penalty matrix is ​​determined based on the time difference between the sampling time of the reference energy density sequence and the sampling time of the measured energy density sequence; solving for an energy transfer mapping matrix that satisfies the mass conservation constraint and minimizes the weighted total transmission cost, using the reference energy density sequence as the source distribution and the measured energy density sequence as the target distribution; and calculating the instantaneous phase lag sequence based on the energy transfer mapping matrix.

[0007] In one possible implementation, the instantaneous phase lag sequence is calculated based on the energy transfer mapping matrix, including: determining an energy availability threshold; determining whether the energy value at each sampling moment in the reference energy density sequence is greater than or equal to the energy availability threshold; if so, calculating the weighted center of gravity of the energy value at each sampling moment on the target time axis based on the energy transfer mapping matrix, and determining the instantaneous phase lag value based on the time difference between the center of gravity and the original moment; if not, assigning the instantaneous phase lag value determined by the previous valid sampling moment to the current moment.

[0008] In one possible implementation, the lag fluctuation index is determined based on the instantaneous phase lag sequence, including: determining the lag fluctuation index as the arithmetic mean of the absolute values ​​of the first-order differences of the instantaneous phase lag sequence.

[0009] In one possible implementation, the fault types include: passive network faults and active network faults; the fault location method for passive network faults is impedance ranging, and the fault location method for active network faults is feature matching.

[0010] In one possible implementation, the network fault type is determined based on the hysteresis fluctuation index, including: comparing the hysteresis fluctuation index with a fluctuation judgment threshold; if the hysteresis fluctuation index is less than the fluctuation judgment threshold, the network fault type is determined to be a passive network fault; if the hysteresis fluctuation index is greater than or equal to the fluctuation judgment threshold, the network fault type is determined to be an active network fault.

[0011] In one possible implementation, when the network fault type is determined to be a passive network fault, the fault location is determined based on the fault location method corresponding to the network fault type, including: calculating the equivalent reactance of the fault circuit based on the fundamental amplitude and phase difference of the measured voltage sequence and the measured current sequence; calculating the fault distance based on the equivalent reactance and the known reactance parameters per unit length of the line; and determining the fault location based on the fault distance.

[0012] In one possible implementation, when the network fault type is determined to be an active network fault, the fault location is determined based on the fault location method corresponding to the network fault type, including: calculating the weighted total time-shift cost based on the energy transfer mapping matrix; determining the dynamic time-shift feature vector based on the hysteresis index and the weighted total time-shift cost; the dynamic time-shift feature vector is a mathematical vector used to characterize the time-varying distortion features of the fault waveform; calculating the Euclidean distance between the dynamic time-shift feature vector and the representative fingerprint vectors of each branch in the branch fault feature sample library; the branch fault feature sample library is a database storing feature vectors pre-generated for each line branch under typical fault conditions; and determining the line branch corresponding to the representative fingerprint vector with the smallest Euclidean distance as the fault location.

[0013] This application has the following beneficial effects: It constructs a theoretical benchmark (linear reference current sequence) to characterize the response of a purely passive linear network, and uses a discrete optimal transmission model to extract time-series characteristics (instantaneous phase lag sequence and lag fluctuation index) that can keenly reflect control dynamics from the difference between measured data and the theoretical benchmark. Finally, based on these time-series characteristics, it realizes the adaptive discrimination of fault types and the selection of location methods. Thus, under the strict condition of using only single-end waveform data from substations, it effectively overcomes the pseudo-distance interference caused by distributed power source access, realizes accurate and reliable location of faults in complex active distribution networks, and significantly reduces the system transformation cost. Attached Figure Description

[0014] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 A flowchart illustrating a method for locating faults in transmission lines of a distribution network with large-scale distributed power source access, provided as an embodiment of this application. Figure 1 ; Figure 2 A flowchart illustrating a method for locating faults in transmission lines of a distribution network with large-scale distributed power source access, provided as an embodiment of this application. Figure 2 ; Figure 3 A flowchart illustrating a method for locating faults in transmission lines of a distribution network with large-scale distributed power source access, provided as an embodiment of this application. Figure 3 ; Figure 4 A flowchart illustrating a method for locating faults in transmission lines of a distribution network with large-scale distributed power source access, provided as an embodiment of this application. Figure 4 ; Figure 5 A flowchart illustrating a method for locating faults in transmission lines of a distribution network with large-scale distributed power source access, provided as an embodiment of this application. Figure 5 ; Figure 6 A flowchart illustrating a method for locating faults in transmission lines of a distribution network with large-scale distributed power source access, provided as an embodiment of this application. Figure 6 ; Figure 7 A flowchart illustrating a method for locating faults in transmission lines of a distribution network with large-scale distributed power source access, provided as an embodiment of this application. Figure 7 . Detailed Implementation

[0016] To further illustrate the technical means and effects adopted by this application to achieve the intended purpose of the invention, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a method for locating faults in transmission lines of a distribution network with large-scale distributed power source access proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0017] 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 pertains.

[0018] Unless otherwise specified, the normalization function Norm() mentioned in this application uses maximum and minimum value normalization. The maximum and minimum values ​​are preset empirical extreme values ​​derived from a large amount of historical experimental data. If the calculation result exceeds the [0,1] interval, a truncation function is used to limit it to the [0,1] range (i.e., if the result is less than 0, it is taken as 0; if it is greater than 1, it is taken as 1) to eliminate the influence of outliers on the evaluation index.

[0019] The following description, in conjunction with the accompanying drawings, details the specific scheme of the fault location method for large-scale distributed power source access in the power distribution network provided in this application.

[0020] Please see Figure 1 This document illustrates a flowchart of a method for locating faults in transmission lines of a distribution network with large-scale distributed power source access, according to an embodiment of this application. Figure 1 As shown, the method includes the following steps: Step 101: Obtain the measured voltage and current sequences after a fault in the transmission line of the distribution network with large-scale distributed power source access.

[0021] Optionally, when a short-circuit fault occurs in a distribution network transmission line, the electrical quantity data related to the fault are collected by the protection device deployed at one end of the substation. Based on the fault trigger time, a continuous data segment containing specific time periods before and after the fault is extracted to construct a fault waveform analysis window. The fault recording analysis window It contains two sets of strictly synchronously sampled discrete-time series, namely the measured voltage series. and measured current sequence ,in A unified time sampling index for the entire period. , The total number of sampling points is and the sampling interval is . (For example ).at this time, and In the same index The two sets of sequences correspond to the same physical moment, ensuring time synchronization.

[0022] The above measured voltage sequence and measured current sequence To provide raw data support for the subsequent construction of linear reference current sequences, generation of energy density sequences, and extraction of fault features, the sampling frequency and data accuracy must meet the engineering requirements for subsequent fundamental parameter extraction and discrete optimal transmission model calculation, ensuring the effective capture of fault features and the accuracy of location results.

[0023] Step 102: Based on the measured voltage sequence and measured current sequence, determine the linear reference current sequence that characterizes the response of the pure passive linear network.

[0024] Optionally, the linear reference current sequence is the ideal current response that a purely passive linear network should produce when the fault circuit completely follows Ohm's law and the initial impedance angle deviation is ignored. It is the core theoretical benchmark for measuring the nonlinear distortion of the measured current waveform. An ideal response scale for a purely passive network is established based on the linear reference current sequence, which enables the accurate identification of dynamic distortion characteristics in the fault waveform and lays a reliable foundation for subsequent extraction of instantaneous phase lag sequences and fault type determination.

[0025] As one possible implementation, this application extracts the fundamental voltage amplitude and fundamental current amplitude within the first power frequency cycle after the fault from the measured voltage sequence and measured current sequence, respectively. The ratio of the two is used to obtain the fundamental equivalent admittance modulus, which can reflect the impedance characteristics of the passive network. Then, the measured voltage sequence is linearly transformed using the fundamental equivalent admittance modulus as a link to finally generate a linear reference current sequence.

[0026] Step 103: Generate a reference energy density sequence based on a linear reference current sequence, and generate a measured energy density sequence based on a measured current sequence.

[0027] Optionally, the reference energy density sequence is a non-negative energy distribution statistic obtained by point-by-point squaring and full-time normalization based on a linear reference current sequence. It is the core source distribution data adapted to the discrete optimal transmission model. An ideal energy distribution scale for pure passive networks is constructed based on the reference energy density sequence, providing a unified benchmark for comparative analysis with measured energy distribution and helping to accurately capture the energy temporal distortion characteristics caused by faults.

[0028] The measured energy density sequence is a non-negative energy distribution statistic obtained from the measured current sequence through point-by-point squaring and full-time normalization. It is the target distribution data reflecting the actual fault state. Based on the measured energy density sequence, the true energy time-series distribution under fault conditions is objectively presented. By comparing it with the reference energy density sequence, intuitive data support is provided for analyzing the nonlinear distortion of the fault waveform.

[0029] Step 104: Extract the instantaneous phase lag sequence from the reference energy density sequence and the measured energy density sequence according to the discrete optimal transmission model.

[0030] The instantaneous phase lag sequence is used to characterize the change of the local time lag of the measured current waveform relative to the linear reference current waveform over time. Optionally, the instantaneous phase lag sequence is calculated based on the energy transfer mapping matrix of the discrete optimal transmission model, characterizing the time-series characteristics of the change of the local time lag of the measured current waveform relative to the linear reference current waveform over time. This sequence can effectively separate the constant lag of the line impedance from the dynamic lag of the inverter control, providing a highly discriminative core feature for calculating the lag fluctuation index and determining the fault type.

[0031] The Discrete Optimal Transmission Model (DITM) is a mathematical model constrained by time difference penalties and based on energy conservation. It is used to analyze the microscopic temporal transfer relationship between a reference and measured energy density sequences. By solving the energy transfer mapping matrix with minimum cost, this model achieves a refined deconstruction of waveform distortion, providing core algorithmic support for the extraction of instantaneous phase lag sequences.

[0032] As one possible implementation, this step uses the reference energy density sequence as the source distribution and the measured energy density sequence as the target distribution to construct a time-difference-based constraint mechanism, solving for the energy transfer mapping matrix that satisfies energy conservation and minimizes cost. Following this, combined with energy effectiveness filtering logic, the corresponding lag value is calculated for effective energy moments, and a reasonable compensation strategy is adopted for ineffective energy moments, ultimately obtaining the instantaneous phase lag sequence characterizing the local time lag change of the measured current waveform relative to the linear reference current waveform.

[0033] Step 105: Determine the hysteresis index based on the instantaneous phase lag sequence.

[0034] The lag volatility index is used to quantify the volatility of instantaneous phase lag sequences.

[0035] Optionally, this application determines the hysteresis fluctuation index as the arithmetic mean of the absolute values ​​of the first-order differences of the instantaneous phase lag sequence. This index is low in sensitivity to the constant phase lag of passive networks and highly discriminative of the dynamic phase fluctuations of active networks, providing a reliable and intuitive quantitative basis for the binary determination of fault types.

[0036] Step 106: Determine the network fault type based on the hysteresis fluctuation index, and determine the fault location based on the fault location method corresponding to the network fault type.

[0037] As one possible implementation, the fault types include passive network faults and active network faults. The fault location method for passive network faults is impedance ranging, while the fault location method for active network faults is feature matching.

[0038] In other words, when the hysteresis fluctuation index determines that the network fault type is a passive network fault, fault location is based on impedance ranging; when the hysteresis fluctuation index determines that the network fault type is an active network fault, fault location is based on feature matching.

[0039] Based on the above technical solution, this application constructs a theoretical benchmark (linear reference current sequence) to characterize the response of a purely passive linear network, and uses a discrete optimal transmission model to extract time-series characteristics (instantaneous phase lag sequence and lag fluctuation index) that can keenly reflect control dynamics from the difference between measured data and the theoretical benchmark. Finally, based on these time-series characteristics, adaptive discrimination of fault types and selection of location methods are realized. Thus, under the strict condition of using only single-end waveform data from substations, the pseudo-distance interference caused by distributed power source access is effectively overcome, and accurate and reliable location of faults in complex active distribution networks is achieved, significantly reducing system transformation costs.

[0040] like Figure 2 As shown, in one possible implementation, the process of determining the linear reference current sequence characterizing the response of a purely passive linear network based on the measured voltage sequence and the measured current sequence in step 102 above can be specifically implemented through the following steps: Step 201: Extract the fundamental voltage amplitude within the first power frequency cycle after the fault from the measured voltage sequence, and extract the fundamental current amplitude within the first power frequency cycle after the fault from the measured current sequence.

[0041] Specifically, the measured voltage sequence within the fault recording analysis window and measured current sequence Full-wave Fourier Transform (FFT) algorithms are applied to extract the fundamental components of each signal within the first complete power frequency cycle after the fault. Here, n is the sampling point index, n=1,2,...,N, and N is the total number of sampling points within one power frequency cycle. Using this algorithm, the fundamental voltage phasor and fundamental current phasor can be calculated, and their fundamental voltage amplitudes can then be obtained. and fundamental current amplitude Based on this, this step uses Fourier transform to accurately separate the amplitude of the core electrical quantity representing power frequency energy transmission from transient fault waveforms that may contain harmonics and noise, providing key input parameters for constructing a linear reference.

[0042] Step 202: Calculate the ratio of the fundamental current amplitude to the fundamental voltage amplitude to obtain the fundamental equivalent admittance mode.

[0043] Specifically, the fundamental current amplitude extracted in step S201 With fundamental voltage amplitude By dividing, the equivalent admittance mode of the fundamental wave is calculated. As an example, the fundamental equivalent admittance modulus Satisfy the following formula:

[0044] In this formula, This represents the fundamental equivalent admittance mode, and its unit is Siemens (S). Indicates the amplitude of the fundamental current; This represents the amplitude of the fundamental voltage. It should be noted that the fundamental voltage amplitude... The voltage validity threshold must be greater than a preset threshold value. This threshold value can be determined based on a preset percentage (e.g., 1%) of the system's rated voltage to adapt to distribution networks with different voltage levels. If it is lower than this threshold value, it is judged as an extreme short-circuit fault, and the fault is directly applied. Set to the preset fundamental equivalent admittance mode value. As an example, the preset fundamental equivalent admittance mode value... .

[0045] Step 203: Perform a linear transformation on the measured voltage sequence based on the fundamental equivalent admittance modulus to determine the linear reference current sequence.

[0046] Specifically, the fundamental equivalent admittance mode value calculated in step S202 is... As a scaling factor that maps voltage to a linear reference current, and compared with the measured voltage sequence A linear transformation involving point-by-point multiplication is performed to generate a linear reference current sequence. As an example, a linear reference current sequence It satisfies the following formula:

[0047] In this formula, This represents the value of the linear reference current sequence at the nth sampling point; This represents the value of the measured voltage sequence at the nth sampling point; This represents the equivalent admittance mode of the fundamental wave.

[0048] Based on the above technical solution, this embodiment of the invention successfully establishes a stable and reliable comparison benchmark in the time domain by accurately extracting the fundamental electrical quantities in the initial stage of a fault and constructing an ideal linear reference current sequence accordingly. This benchmark effectively separates the inherent linear impedance component in the fault response, laying a solid foundation for the subsequent accurate quantification of the nonlinearity and time-varying components introduced by distributed power source control using the discrete optimal transmission model, thereby solving the fundamental problem of unclear benchmark in single-ended electrical quantity analysis.

[0049] like Figure 3As shown, in one possible implementation, the process of generating the reference energy density sequence based on the linear reference current sequence and the process of generating the measured energy density sequence based on the measured current sequence in step 103 above can be specifically implemented through the following steps: Step 301: Calculate the square of the instantaneous value of the sampling point in the linear reference current sequence to obtain the reference instantaneous energy sequence.

[0050] Specifically, for the linear reference current sequence Squaring each sample value generates a reference instantaneous energy sequence. As an example, refer to the instantaneous energy sequence Satisfy the following formula:

[0051] In this formula, This represents the value of the reference instantaneous energy sequence at the nth sampling point; This represents the value of the linear reference current sequence at the nth sampling point.

[0052] Step 302: Normalize the reference instantaneous energy sequence to obtain the reference energy density sequence.

[0053] Specifically, the reference instantaneous energy sequence obtained in step S301 The algorithm employs a summation-normalization method, which first calculates the sum of energy values ​​at all sampling points within the entire analysis window, and then divides the energy value of each point in the sequence by this sum to obtain the reference energy density sequence. As an example, a reference energy density sequence Satisfy the following formula:

[0054] In this formula, This represents the value of the reference energy density sequence at the nth sampling point; This represents the value of the reference instantaneous energy sequence at the nth sampling point; N is the total number of sampling points; the denominator is... This represents the sum of all sampled values ​​of the reference instantaneous energy sequence. In this formula... These are parameter tuning coefficients, and their values ​​should be very small integers (e.g., 0.01) to avoid denominators of 0.

[0055] Step 303: Calculate the square of the instantaneous value of the sampling point in the measured current sequence to obtain the measured instantaneous energy sequence.

[0056] Specifically, for the measured current sequence The measured instantaneous energy sequence is generated by squaring each sample value. It satisfies the following formula:

[0057] In this formula, This represents the value of the measured instantaneous energy sequence at the nth sampling point; This represents the value of the measured current sequence at the nth sampling point.

[0058] Step 304: Normalize the measured instantaneous energy sequence to obtain the measured energy density sequence.

[0059] Specifically, the measured instantaneous energy sequence obtained in step S303 Similarly, the summation and normalization algorithm is applied to obtain the measured energy density sequence. It satisfies the following formula:

[0060] In this formula, This represents the value of the measured energy density sequence at the nth sampling point; The denominator represents the value of the measured instantaneous energy sequence at the nth sampling point; This represents the sum of all sampled values ​​in the measured instantaneous energy sequence. In this formula... These are parameter tuning coefficients, and their values ​​should be very small integers (e.g., 0.01) to avoid denominators of 0.

[0061] Based on the above technical solution, this embodiment of the invention successfully transforms the original voltage and current signal analysis problem into a comparison problem of two energy density sequences with probability distributions by converting the current sequence into an instantaneous energy sequence and normalizing it. This key transformation eliminates polarity interference of AC signals and ensures that the data format strictly conforms to the input requirements of discrete optimal transmission theory, thus providing the possibility for subsequent application of this mathematical model to accurately quantify the time-varying distortion characteristics between waveforms.

[0062] like Figure 4 As shown, in one possible implementation, the process of extracting the instantaneous phase lag sequence from the reference energy density sequence and the measured energy density sequence according to the discrete optimal transmission model in step 104 above can be specifically implemented through the following steps: Step 401: Construct the time difference penalty matrix.

[0063] In this context, each element of the time difference penalty matrix is ​​determined based on the time difference between the sampling time of the reference energy density sequence and the sampling time of the measured energy density sequence.

[0064] Specifically, the time index of the reference energy density sequence is defined as k (source index), and the time index of the measured energy density sequence is defined as m (sink index), where... , Where N is the sequence length. Construct Time difference penalty matrix The first of the matrix Line number Column elements Used for quantization, transferring unit energy from the reference time. Transfer to the measured time The physical cost required. This embodiment uses the square of the time difference as the cost function to penalize drastic time jumps and ensure the convexity of the optimization problem. As an example, the first... Line number Column elements Satisfy the following formula:

[0065] In this formula, This represents the cost required to transfer a unit of energy from time k of the reference sequence to time m of the measured sequence; k and m are the sampling point indices of the reference sequence and the measured sequence, respectively. The sampling interval is set to 100 μs (the specific value can be determined based on the actual scenario requirements). It should be noted that the formula uses the square of the time difference as the cost function, which aims to more severely punish energy matching of non-adjacent times, thereby strengthening the model's constraint on local time alignment of the waveform and ensuring that the optimization problem is a strictly convex programming problem, which is easy to solve.

[0066] Step 402: Using the reference energy density sequence as the source distribution and the measured energy density sequence as the target distribution, solve for the energy transfer mapping matrix that satisfies the mass conservation constraint and minimizes the weighted total transmission cost.

[0067] Specifically, this step involves establishing a discrete-optimal transmission linear programming model. This model aims to find a... dimensional energy transfer mapping matrix The objective is to minimize the total weighted time-shift cost. As an example, the transport linear programming model is expressed as follows:

[0068] The optimization problem must satisfy the following constraints: Source energy conservation constraint: the reference sequence at any time The energy must be fully allocated, that is... ; The sink energy satisfies the constraint: the measured sequence at any time... The energy must be fully replenished, that is... ; Non-negativity constraint: The transmission ratio cannot be negative, i.e. .

[0069] The above model is solved using standard linear programming algorithms such as the simplex method or interior point method (or approximate solution methods such as the Sinkhorn iterative algorithm) to obtain the optimal energy transfer mapping matrix. .

[0070] Energy transfer mapping matrix elements in Used to characterize the reference waveform What percentage of the energy at time step 1 was transferred to the measured waveform? The time frame is thus used to fully describe the microscopic temporal structure of waveform distortion.

[0071] Step 403: Calculate the instantaneous phase lag sequence based on the energy transfer mapping matrix.

[0072] As one possible implementation, this step can be specifically implemented as follows: determine the energy availability threshold; determine whether the energy value at each sampling moment in the reference energy density sequence is greater than or equal to the energy availability threshold; if so, calculate the weighted center of gravity of the energy value at each sampling moment on the target time axis according to the energy transfer mapping matrix, and determine the instantaneous phase lag value according to the time difference between the center of gravity and the original moment; if not, assign the instantaneous phase lag value determined by the previous valid sampling moment to the current moment.

[0073] Optionally, after obtaining the energy transfer mapping matrix, it is necessary to convert the probability distribution information it contains into an intuitive time series to analyze the dynamic evolution of the phase lag. Since AC signals have zero-crossing points, calculating the delay at moments with extremely low energy density can lead to numerical instability or even division by zero errors. Therefore, this embodiment employs a protection strategy with energy threshold judgment to extract local delays.

[0074] Specifically, a threshold for energy availability is set. (For example, taking the maximum value of the reference energy density sequence) This percentage (e.g., 1%) aims to effectively filter out energy points with significantly lower energy than the valid signal due to measurement noise and quantization errors near zero crossings, while ensuring that the energy values ​​of the vast majority of valid signal points are included. The initial length is... Instantaneous phase lag sequence Time index of the reference sequence from to Perform point-by-point traversal calculations: Step 1: Energy Efficiency Assessment: Check the energy density at the current reference time. Is it greater than the threshold? .

[0075] Step 2: Centroid Calculation (Effective Area): like This indicates that the signal energy at that moment is significant and possesses physical significance for calculating statistical characteristics. At this point, the reference time is calculated. The weighting center of energy on the measured time axis As an example, weighting the center of gravity moment Satisfy the following formula:

[0076] Subsequently, the physical time of the centroid and the original reference time are calculated. The difference in physical time is used to obtain the instantaneous phase lag value at that moment, which serves as an example instantaneous phase lag value. Satisfy the following formula:

[0077] Step 3: Zero-order hold (invalid region): like This indicates that the signal is near its zero-crossing point at this moment, and direct calculation would introduce a large error. Therefore, a zero-order hold strategy is adopted, setting the lag value at the current moment equal to the calculated value at the previous valid moment, i.e., setting... It should be noted that invalid points at the beginning of the sequence can be set to... Alternatively, search backwards for the first valid value.

[0078] As one possible implementation, after calculating the instantaneous phase lag sequence, To further eliminate the impact of high-frequency quantization noise on subsequent difference calculations and enhance engineering robustness, after obtaining the original... After processing the sequence, a moving average filter is applied. Specifically, this includes setting the filter window length to... (For example, take) (Point), calculate the hysteresis value after filtering. It satisfies the following formula:

[0079] For boundary points, the window size is adjusted accordingly or edge filling is used.

[0080] After the above processing, the final sequence is obtained. This describes the trajectory of the phase lag of the measured current relative to the linear reference current over time within the fault analysis window. For passive linear faults, the sequence approximates a horizontal straight line; for active nonlinear faults, the sequence exhibits significant fluctuations or trend terms.

[0081] Based on the above technical solution, this embodiment of the invention successfully deconstructs the overall morphological difference between the reference waveform and the measured waveform into a series of time-varying micro-time offsets by constructing a time difference penalty matrix, solving the optimal energy transfer mapping, and calculating the instantaneous phase lag sequence based on an energy threshold protection strategy. This method can accurately capture and quantify the non-uniform time distortion effect introduced by the dynamic control of the inverter's phase-locked loop, providing high-resolution timing characteristic data for subsequently distinguishing between constant impedance delay and dynamic control fluctuations.

[0082] In one possible implementation, the process of determining the lag fluctuation index based on the instantaneous phase lag sequence in step 105 above can be specifically implemented as follows: Calculate the first-order difference statistic of an instantaneous phase-lag sequence. Define the lag fluctuation index. Instantaneous phase lag sequence (or after smoothing) The arithmetic mean of the absolute values ​​of the first differences of the latitudinal volatility index. As an example, the latitudinal volatility index... Satisfy the following formula:

[0083] In this formula, for passive linear faults, although the waveform may be distorted, its phase lag is mainly determined by the constant line impedance. It is approximately constant on the time axis, and its difference value is close to zero. The value is extremely small.

[0084] For active nonlinear faults, the inverter's phase-locked loop control continuously adjusts the phase of the output current according to real-time changes in the grid voltage, leading to... It exhibits continuous fluctuations on the timeline. The value has increased significantly.

[0085] therefore, It can effectively remove the constant line impedance delay component, retaining only the dynamic change component caused by control behavior.

[0086] like Figure 5 As shown, in one possible implementation, the process of determining the network fault type based on the hysteresis index in step 106 above can be specifically implemented through the following steps: Step 501: Compare the lagged volatility index with the volatility judgment threshold.

[0087] As one possible approach, the process of determining the fluctuation threshold includes: setting up a mainline fault condition containing only passive loads in the simulation model and performing Monte Carlo random simulation (including random noise superposition).

[0088] The maximum value of the hysteresis fluctuation index calculated under all passive fault conditions is denoted as . .

[0089] Set fluctuation judgment threshold The safety factor for this maximum value is calculated using the following formula: .in, For the reliability coefficient, a recommended value range is [value range missing]. to This ensures that measurement noise and slight fluctuations caused by line distributed capacitance can be reliably filtered out, preventing misjudgments.

[0090] Step 502: If the hysteresis fluctuation index is less than the fluctuation judgment threshold, the network fault type is determined to be a passive network fault.

[0091] Specifically, if This indicates that the instantaneous phase lag sequence is approximately constant on the time axis, and the fluctuation amplitude is within the noise margin. This means that the impedance parameters of the fault circuit satisfy the linear time-invariant characteristic, and the current lag is determined only by the fixed line inductance, without being significantly affected by the dynamic adjustment of the inverter's phase-locked loop. Therefore, the fault is determined to occur in the passive main line or a pure load branch, and the fault circuit satisfies Ohm's law.

[0092] Step 503: If the hysteresis fluctuation index is greater than or equal to the fluctuation judgment threshold, the network fault type is determined to be an active network fault.

[0093] like This indicates that the instantaneous phase lag sequence exhibits significant time-varying fluctuations, exceeding the natural response range of the passive network. This implies that the faulty loop contains a controlled voltage source, and the current waveform is modulated by the transient control behavior of power electronic devices, leading to nonlinear distortion of the apparent impedance. Therefore, the fault is determined to occur in an active branch containing distributed power sources, at which point the traditional impedance ranging method fails.

[0094] It should be noted that, in this application embodiment, the network failure scenarios include: Scenario 1, where the network failure type is a passive network failure; and Scenario 2, where the network failure type is an active network failure. The fault location methods used for Scenario 1 and Scenario 2 are different, and will be explained separately below: Scenario 1: The network fault type is a passive network fault. like Figure 6 As shown, in this scenario, the process of determining the fault location based on the fault location method corresponding to the network fault type can be implemented as follows: Step 601: Calculate the equivalent reactance of the fault circuit based on the fundamental amplitude and phase difference of the measured voltage and current sequences.

[0095] Specifically, the extracted fundamental voltage amplitude was retrieved by backtracking. and fundamental current amplitude and the phase difference between the two (That is, voltage phase minus current phase). The equivalent reactance of the fault circuit is calculated using the fundamental impedance formula. As an example, the equivalent reactance of a faulty circuit Satisfy the following formula:

[0096] Step 602: Calculate the fault distance based on the equivalent reactance and the known reactance parameters per unit length of the line.

[0097] Specifically, based on the known reactance parameters per unit length of the line (This parameter is determined by the line model and is a known constant.) Calculate the fault distance. As an example, fault distance Satisfy the following formula:

[0098] Step 603: Determine the fault location based on the fault distance.

[0099] The aforementioned fault distance refers to the physical path length from the substation to the fault point, which maintenance personnel can use to troubleshoot faults.

[0100] Based on the above technical solutions, this application adopts impedance ranging for passive network faults. It makes full use of the fundamental parameters of the measured voltage and current sequences, and realizes the location by calculating the equivalent reactance and fault distance. This method is consistent with the characteristic that the fault current of passive network follows the linear time-invariant law. The calculation process is simple and efficient, without relying on additional sample libraries or complex algorithms. It can directly output the physical distance from the substation to the fault point. The location results are intuitive and easy to understand, which makes it easy for operation and maintenance personnel to quickly locate the fault range and carry out troubleshooting and repair work.

[0101] Scenario 2: The network fault type is an active network fault. like Figure 7 As shown, in this scenario, the process of determining the fault location based on the fault location method corresponding to the network fault type can be implemented as follows: Step 701: Calculate the weighted total time shift cost based on the energy transfer mapping matrix.

[0102] Specifically, to help determine the overall distortion level of the waveform, the weighted total time shift cost is calculated based on the solved optimal transmission objective function value. This metric reflects the total physical work required to transform a linear reference waveform into a measured waveform; as an example, the weighted total time-shift cost Satisfy the following formula:

[0103] in, For elements of the time difference penalty matrix, These are the elements of the optimal energy transfer mapping matrix.

[0104] Step 702: Determine the dynamic time-shift feature vector based on the hysteresis fluctuation index and the weighted total time-shift cost.

[0105] Among them, the dynamic time-shift feature vector is a mathematical vector used to characterize the time-varying distortion features of the fault waveform.

[0106] As an example, dynamic time-shifted feature vectors Satisfy the following formula:

[0107] Based on this, the dynamic time-shifted feature vector It incorporates the dynamic control characteristics of fault waveforms, which can be directly used for fault nature determination and branch matching. This is the transpose symbol.

[0108] Step 703: Calculate the Euclidean distance between the dynamic time-shifted feature vector and the fingerprint vectors representing each branch in the branch fault feature sample library.

[0109] Among them, the branch fault feature sample library is a database that stores feature vectors pre-generated for each line branch under typical fault conditions.

[0110] Optionally, iterate through the representative fingerprint vectors of all branches in the sample database. Calculate the dynamic time-shifted feature vector Euclidean distance between each fingerprint vector As an example, dynamic time-shifted feature vectors Euclidean distance between the fingerprint vector of any ID Satisfy the following formula:

[0111] in, and The corresponding samples in the sample library The fluctuation exponential component and total cost component of the fingerprint vector of the branch.

[0112] As one possible implementation, the construction process of the branch fault feature sample library includes: traversing every branch node containing distributed generation in the digital simulation model of the distribution network. For each branch node, three typical short-circuit fault conditions are set, along with their corresponding fault transition resistances. Set to: low resistance (e.g.) ), medium resistance (e.g.) ) and high resistance (e.g. These three operating conditions are designed to cover the main resistance variation range when a fault occurs.

[0113] For each simulated fault waveform under different operating conditions, the following online processing procedure is followed to calculate the corresponding dynamic time-shift feature vector. .

[0114] For three feature vectors at the same branch node (corresponding to low, medium, and high resistance respectively), calculate their geometric center (i.e., the arithmetic mean of each component), and define this center vector as the representative fingerprint vector of that branch. Since the hysteresis fluctuation index extracted in this invention mainly reflects the dynamic texture structure of the waveform over time, rather than the absolute amplitude, this feature has strong robustness to changes in the magnitude of the fault transition resistance. Using the geometric center as a representative fingerprint can effectively characterize the fault attributes of this branch.

[0115] Number all branches Its corresponding representative fingerprint vector Establish index relationships and store them in the sample database. This yields a sample library of branch fault features.

[0116] Step 704: Determine the line branch corresponding to the fingerprint vector with the smallest Euclidean distance as the fault location.

[0117] Optionally, the fault location determined in this step is the branch number where the fault occurs. And the corresponding geographical location description.

[0118] It should be noted that, due to the lagged volatility index used in this application This feature primarily reflects the relative rate of change of the waveform over time (i.e., the dynamic texture structure of the waveform), rather than the absolute amplitude. Therefore, it is highly robust to changes in the magnitude of the fault transition resistance. That is, although the amplitudes of the fault waveforms differ under different resistances, the dynamic adjustment pattern (ripple characteristics) caused by the inverter's phase-locked loop remains similar. Therefore, even if the measured fault transition resistance is not entirely consistent with the preset typical resistances in the sample library, the measured feature vector will still fall within the neighborhood of the corresponding branch fingerprint vector, thus ensuring the effectiveness of matching and localization based on the nearest neighbor principle.

[0119] Based on the above technical solutions, this application adopts an impedance ranging method for passive network faults. It makes full use of the fundamental parameters of the measured voltage and current sequences and realizes the location by calculating the equivalent reactance and fault distance. This method is consistent with the characteristic that the fault current of passive networks follows the linear time-invariant law. The calculation process is simple and efficient and does not rely on additional sample libraries or complex algorithms.

[0120] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0121] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for fault location in transmission lines of a distribution network with large-scale distributed power source integration, characterized in that, The method includes: Obtain the measured voltage and current sequences after a fault in the transmission line of a distribution network with large-scale distributed power generation access; Based on the measured voltage sequence and the measured current sequence, a linear reference current sequence characterizing the response of a purely passive linear network is determined. A reference energy density sequence is generated based on the linear reference current sequence, and a measured energy density sequence is generated based on the measured current sequence. The instantaneous phase lag sequence is extracted from the reference energy density sequence and the measured energy density sequence according to the discrete optimal transmission model; the instantaneous phase lag sequence is used to characterize the change of the local time lag of the measured current waveform relative to the linear reference current waveform over time. Based on the instantaneous phase lag sequence, a lag fluctuation index is determined, which is used to quantify the degree of fluctuation of the instantaneous phase lag sequence; The network fault type is determined based on the hysteresis fluctuation index, and the fault location is determined based on the fault location method corresponding to the network fault type.

2. The method for fault location of power transmission lines in a distribution network with large-scale distributed power source access according to claim 1, characterized in that, Based on the measured voltage sequence and the measured current sequence, a linear reference current sequence characterizing the response of a purely passive linear network is determined, including: Extract the fundamental voltage amplitude within the first power frequency cycle after the fault from the measured voltage sequence, and extract the fundamental current amplitude within the first power frequency cycle after the fault from the measured current sequence; Calculate the ratio of the fundamental current amplitude to the fundamental voltage amplitude to obtain the fundamental equivalent admittance mode. The measured voltage sequence is linearly transformed based on the fundamental equivalent admittance modulus to determine the linear reference current sequence.

3. The method for fault location of power transmission lines in a distribution network with large-scale distributed power source access according to claim 1, characterized in that, Generating a reference energy density sequence based on the linear reference current sequence, and generating a measured energy density sequence based on the measured current sequence, include: The reference instantaneous energy sequence is obtained by calculating the square of the instantaneous values ​​at the sampling points in the linear reference current sequence; The reference instantaneous energy sequence is normalized to obtain the reference energy density sequence; The square of the instantaneous value at each sampling point in the measured current sequence is calculated to obtain the measured instantaneous energy sequence. The measured instantaneous energy sequence is normalized to obtain the measured energy density sequence.

4. The method for fault location of power transmission lines in a distribution network with large-scale distributed power source access according to claim 1, characterized in that, Based on the discrete optimal transmission model, instantaneous phase lag sequences are extracted from the reference energy density sequence and the measured energy density sequence, including: A time difference penalty matrix is ​​constructed, wherein each element of the time difference penalty matrix is ​​determined based on the time difference between the sampling time of the reference energy density sequence and the sampling time of the measured energy density sequence; Using the reference energy density sequence as the source distribution and the measured energy density sequence as the target distribution, solve for the energy transfer mapping matrix that satisfies the mass conservation constraint and minimizes the weighted total transmission cost. The instantaneous phase lag sequence is calculated based on the energy transfer mapping matrix.

5. The method for fault location of power transmission lines in a distribution network with large-scale distributed power source access according to claim 4, characterized in that, The instantaneous phase lag sequence is calculated based on the energy transfer mapping matrix, including: Determine the energy availability threshold; Determine whether the energy value at each sampling time in the reference energy density sequence is greater than or equal to the energy validity threshold; If so, the weighted center of the energy value at each sampling moment on the target time axis is calculated according to the energy transfer mapping matrix, and the instantaneous phase lag value is determined according to the time difference between the center of gravity and the original moment. If not, the instantaneous phase lag value determined at the previous valid sampling time is assigned to the current time.

6. The method for fault location of distribution network transmission lines with large-scale distributed power source access according to claim 1, characterized in that, Based on the instantaneous phase lag sequence, the lag fluctuation index is determined, including: The arithmetic mean of the absolute values ​​of the first-order differences of the instantaneous phase lag sequence is determined as the lag fluctuation index.

7. The method for fault location of distribution network transmission lines with large-scale distributed power source access according to claim 4, characterized in that, The fault types include: passive network faults and active network faults; The fault location method for the passive network fault is impedance ranging, and the fault location method for the active network fault is feature matching.

8. The method for fault location of power transmission lines in a distribution network with large-scale distributed power source access according to claim 7, characterized in that, The network fault type is determined based on the hysteresis fluctuation index, including: Compare the hysteresis volatility index with the volatility judgment threshold; If the hysteresis fluctuation index is less than the fluctuation determination threshold, then the network fault type is determined to be a passive network fault. If the hysteresis fluctuation index is greater than or equal to the fluctuation determination threshold, then the network fault type is determined to be an active network fault.

9. The method for fault location of power transmission lines in a distribution network with large-scale distributed power source access according to claim 8, characterized in that, If the network fault type is determined to be a passive network fault, the fault location is determined based on the fault location method corresponding to the network fault type, including: The equivalent reactance of the fault circuit is calculated based on the fundamental amplitude and phase difference of the measured voltage sequence and the measured current sequence. Calculate the fault distance based on the equivalent reactance and the known reactance parameters per unit length of the line; The fault location is determined based on the fault distance.

10. The method for fault location of distribution network transmission lines with large-scale distributed power source access according to claim 8, characterized in that, If the network fault type is determined to be an active network fault, the fault location is determined based on the fault location method corresponding to the network fault type, including: The weighted total time-shift cost calculated based on the energy transfer mapping matrix; Based on the hysteresis fluctuation index and the weighted total time shift cost, a dynamic time shift feature vector is determined; the dynamic time shift feature vector is a mathematical vector used to characterize the time-varying distortion features of the fault waveform. Calculate the Euclidean distance between the dynamic time-shifted feature vector and the fingerprint vectors representing each branch in the branch fault feature sample library; the branch fault feature sample library is a database that stores feature vectors pre-generated for each line branch under typical fault conditions; The line branch corresponding to the fingerprint vector with the smallest Euclidean distance is determined as the fault location.

Images