Intelligent power distribution equipment disconnection fault detection method, system and device

Abstract

The invention discloses an intelligent power distribution equipment disconnection fault detection method, system and device. The method comprises the steps of obtaining a negative sequence current value of intelligent power distribution equipment; when the negative-sequence current value is greater than a current threshold value, a negative-sequence current signal, a negative-sequence voltage signal and a zero-sequence current signal of each line in the intelligent power distribution equipment are acquired, and the current threshold value is dynamically adjusted based on historical operation data and a real-time load state; determining a feature vector set corresponding to each line; determining a negative sequence current weight, a negative sequence voltage weight and a zero sequence current weight of each line; calculating a weighted Hausdorff distance between every two lines in the intelligent power distribution equipment; constructing a weighted Hausdorff distance matrix; and on the basis of the weighted Hausdorff distance matrix, constructing a feature space topology consistency criterion, and judging whether a branch or a bus in the intelligent power distribution equipment has a broken line fault or not. The method has higher fault detection accuracy and stronger anti-interference and adaptability, and is suitable for a modern power distribution network with high-proportion new energy access and multiple operation modes.

Description

Technical Field

[0001] This invention relates to the field of power grid technology, and more specifically, to a method, system, and device for detecting open circuit faults in intelligent power distribution equipment. Background Technology

[0002] Power distribution is a link in the power system that directly faces electricity users during power generation, transmission, and distribution. The distribution network is close to users and has a complex surrounding environment, resulting in diverse distribution network structures (hand-in, open-loop), various system grounding methods (ungrounded, arc suppression coil grounding, low-resistance grounding), a year-on-year increase in the rate of mixed cabling and overhead cable connections, and complex fault types (lightning strikes, tree flashovers, line breaks, electric shocks, etc.). Therefore, it is impossible to accurately determine whether a line break fault has occurred.

[0003] Currently, common methods for detecting open circuit faults in distribution networks include threshold comparison-based methods and negative sequence current-based methods. Threshold comparison-based methods, with their preset fixed current thresholds, struggle to adapt to the frequently changing load conditions, diverse operating modes, and complex network topologies of distribution networks. Under light load conditions, the normal phase current may be inherently small, easily confused with open circuit faults, leading to misjudgments. Furthermore, for open circuit faults accompanied by high-resistance grounding, current changes may be subtle, easily resulting in missed detections because the threshold is not reached. Negative sequence current-based methods suffer from a significant problem of feature simplification. Relying solely on the negative sequence current component makes it difficult to comprehensively and accurately characterize the complex characteristics of open circuit faults, especially when the open circuit is not grounded or the grounding resistance is high, the characteristic quantity may be very weak. Secondly, they have poor anti-interference capabilities. Nonlinear load fluctuations, transient short-circuit faults, and system asymmetry in distribution networks all generate negative sequence components, creating strong background interference, making the single-characteristic method highly susceptible to misjudgments.

[0004] Existing technical document 1 (CN118604513A) discloses a method, device, equipment, medium, and program product for selecting single-phase open-circuit faults. Its shortcomings are: it uses only negative-sequence current characteristics to determine the faulty line, making it difficult to comprehensively reflect the fault characteristics; the preset fixed current threshold cannot adapt to changes in system operating modes, easily leading to misjudgment or missed fault detection; under noise interference or load fluctuations, the Hausdorff distance matrix based on a single feature is prone to mismatch; the fault criterion is too simplistic, failing to analyze the relative positional relationships of various line characteristics from a holistic system perspective, and lacks a fault judgment verification process. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method, system, and device for detecting line breakage faults in intelligent power distribution equipment, solving the problems of overly simplistic line characteristics and insufficient anti-interference capabilities in existing line breakage fault detection methods.

[0006] The present invention adopts the following technical solution.

[0007] The first aspect of the present invention provides a method for detecting open circuit faults in intelligent power distribution equipment, comprising the following steps: Obtain the negative sequence current value of intelligent power distribution equipment; When the negative sequence current value is greater than the current threshold, the negative sequence current signal, negative sequence voltage signal and zero sequence current signal of each line in the intelligent power distribution equipment are acquired, wherein the current threshold is dynamically adjusted based on historical operating data and real-time load status. The feature vector set corresponding to each line is determined based on the negative sequence current signal, negative sequence voltage signal and zero sequence current signal of each line, wherein the feature vector set includes: negative sequence current signal feature vector, negative sequence voltage signal feature vector and zero sequence current signal feature vector; The negative sequence current weight, negative sequence voltage weight, and zero sequence current weight of each line are determined based on the feature vector set corresponding to each line. The weighted Hausdorff distance between every two lines in the intelligent power distribution equipment is calculated based on the feature vector set corresponding to each line and the negative sequence current weight, negative sequence voltage weight, and zero sequence current weight of each line. Using the weighted Hausdorff distance between every two lines as matrix elements, the weighted Hausdorff distances between the same line and other lines are set in the same row and column to form a weighted Hausdorff distance matrix. Based on the weighted Hausdorff distance matrix, a feature space topology consistency criterion is constructed to determine whether a branch or busbar in an intelligent power distribution device has a broken line fault.

[0008] Preferably, the current threshold is dynamically adjusted based on historical operating data and real-time load status, including the following steps: Continuously record and learn the historical negative sequence current data and corresponding positive sequence current data of intelligent power distribution equipment at different time periods under normal operating conditions; By using moving averages, corresponding negative sequence current benchmark thresholds are established for different typical load ranges and updated as the intelligent power distribution equipment operates. Real-time monitoring of the current positive sequence current; and based on the load range into which the current positive sequence current falls, calling the corresponding negative sequence current reference threshold. The final dynamic current threshold is calculated using the following formula:

[0009] In the formula, I 2_base It is based on the current positive sequence current I 1_real The determined negative sequence current reference threshold, σ I2 It is the standard deviation of historical negative sequence current data within the corresponding load range. K 1 and K2 represents the reliability coefficient.

[0010] Preferably, the step of determining the feature vector set corresponding to each line based on the negative-sequence current signal, negative-sequence voltage signal, and zero-sequence current signal of each line includes: By taking the negative sequence current signal, negative sequence voltage signal and zero sequence current signal as time series, negative sequence current data sequence, negative sequence voltage data sequence and zero sequence current data sequence are obtained; The time-shifting scale analysis method is used to process the negative sequence current data sequence, negative sequence voltage data sequence, and zero sequence current data sequence to obtain the negative sequence current time-shifted subsequence set, the negative sequence voltage time-shifted subsequence set, and the zero sequence current time-shifted subsequence set, respectively. The first entropy value corresponding to the negative sequence current signal, the first entropy value corresponding to the negative sequence voltage signal, and the first entropy value corresponding to the zero sequence current signal are determined respectively based on the negative sequence current time shift subsequence set, the negative sequence voltage time shift subsequence set, and the zero sequence current time shift subsequence set; The corresponding multi-scale standard deviations of negative sequence current, negative sequence voltage, and zero sequence current are determined based on the sets of negative sequence current time-shifted subsequences, negative sequence voltage time-shifted subsequences, and zero sequence current time-shifted subsequences, respectively. The second entropy value corresponding to the negative-sequence current signal is determined based on the first entropy value and the multi-scale standard deviation of the negative-sequence current signal, and is used as the feature vector of the negative-sequence current signal; the second entropy value corresponding to the negative-sequence voltage signal is determined based on the first entropy value and the multi-scale standard deviation of the negative-sequence voltage signal, and is used as the feature vector of the negative-sequence voltage signal; the second entropy value corresponding to the zero-sequence current signal is determined based on the first entropy value and the multi-scale standard deviation of the zero-sequence current signal, and is used as the feature vector of the zero-sequence current signal. The feature vector set is obtained by concatenating the feature vectors of the negative sequence current signal, the negative sequence voltage signal, and the zero sequence current signal.

[0011] Preferably, the step of determining the negative-sequence current signal feature weight, negative-sequence voltage signal feature weight, and zero-sequence current signal feature weight for each line based on the feature vector set corresponding to each line includes: The set of feature vectors corresponding to each line is input into a preset data model to obtain the negative sequence current weight, negative sequence voltage weight, and zero sequence current weight of each line.

[0012] Preferably, the weighted Hausdorff distance between line i and line j Expressed as follows:

[0013] In the formula: H ij I The first Hausdorff distance component is calculated based on the eigenvector of the negative sequence current signal; H ij U The second Hausdorff distance component is calculated based on the eigenvector of the negative-sequence voltage signal. H ij I0 The third Hausdorff distance component is calculated based on the eigenvector of the zero-sequence current signal. ω i I ω i U ω i I0 These are the negative sequence current weight, negative sequence voltage weight, and zero sequence current weight for line i, respectively.

[0014] Preferably, the step of constructing the feature space topology consistency criterion includes: Based on the weighted Hausdorff distance matrix, the relative density factor of each line is calculated to form a relative density vector. The calculation of the relative density factor includes: calculating the average Hausdorff distance of a certain line to all other lines, taking the reciprocal of the average Hausdorff distance as the density factor of the line, and normalizing the density factor to a probability distribution to obtain the relative density factor. By comparing the information entropy of the relative density vector with a preset entropy threshold, it can be determined whether the line breakage fault occurred on a branch or a bus.

[0015] Preferably, the relative density factor is expressed by the following formula:

[0016] In the formula: n is the total number of lines, ρ i Density factor.

[0017] Preferably, the information entropy of the relative density vector is expressed by the following formula: .

[0018] In the formula: n is the total number of lines, D i is the relative density factor.

[0019] Preferably, the entropy threshold is set to a certain percentile of the entropy value under normal operating conditions.

[0020] Preferably, when the information entropy of the relative density vector is less than a preset entropy threshold, it is determined to be a branch line breakage fault, and the faulty line is the line corresponding to the largest relative density factor. When the information entropy of the relative density vector is greater than or equal to the preset entropy threshold, it is determined to be a bus line breakage fault.

[0021] A second aspect of the present invention provides an intelligent power distribution equipment open circuit fault detection system, wherein the method for detecting the open circuit fault of the intelligent power distribution equipment includes: The current acquisition module is used to acquire the negative sequence current value of the intelligent power distribution equipment; The signal acquisition module is used to acquire the negative sequence current signal, negative sequence voltage signal and zero sequence current signal of each line in the intelligent power distribution equipment when the negative sequence current value is greater than the current threshold. The feature extraction module is used to extract the feature vector set corresponding to each line; The weight determination module is used to determine the negative sequence current weight, negative sequence voltage weight, and zero sequence current weight for each line. The distance calculation module is used to calculate the weighted Hausdorff distance between every two lines in the intelligent power distribution equipment; The matrix building module is used to construct the weighted Hausdorff distance matrix; The fault diagnosis module is used to construct a feature space topology consistency criterion and determine whether a branch or busbar in the intelligent power distribution equipment has a broken line fault.

[0022] A third aspect of the present invention provides an intelligent power distribution device equipped with the aforementioned intelligent power distribution equipment disconnection fault detection system.

[0023] A fourth aspect of the present invention provides an electronic device comprising: A memory that stores computer instructions; The processor executes the steps of the intelligent power distribution equipment disconnection fault detection method when executing the computer instructions.

[0024] A fifth aspect of the present invention provides a storage medium having computer instructions stored thereon, which, when executed by a processor, perform the steps of the intelligent power distribution equipment disconnection fault detection method.

[0025] Compared with the prior art, the beneficial effects of the present invention include at least the following: 1. This invention introduces negative sequence voltage and zero sequence current signals, and fully considers multi-source information such as voltage characteristics and transient characteristics, to accurately identify open circuit faults in complex power distribution networks.

[0026] 2. This invention uses an adaptive current threshold instead of a fixed setting value for fault identification, automatically adapting to various power distribution network operation scenarios and reducing the workload of manually adjusting the setting value.

[0027] 3. This invention uses a weighted Hausdorff distance matrix, which assigns different weights based on feature importance, to enhance the sensitivity to key features. It actively strengthens key features when there is noise interference or load fluctuations, further reducing the risk of misjudgment.

[0028] 4. This invention improves the fault criteria by quantifying the degree of isolation of each line in the feature space through the relative density factor, which is applicable to fault judgment with multiple features. This invention adopts the feature space topology consistency criterion, which not only considers the Hausdorff distance of a single feature between lines, but also further analyzes the distribution density and relative distribution relationship of multiple line features in multidimensional space, and performs topology consistency verification, thereby increasing the reliability of fault identification.

[0029] 5. In summary, this invention has higher fault detection accuracy, stronger anti-interference and adaptability, and is suitable for modern power distribution networks with a high proportion of new energy access and multiple operating modes. Attached Figure Description

[0030] Figure 1 This is a flowchart of a method for detecting open circuit faults in intelligent power distribution equipment according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the determination of the feature vector set corresponding to each line according to an embodiment of the present invention; Figure 3 This is a flowchart for determining the first entropy value according to an embodiment of the present invention; Figure 4 This is a flowchart of determining disconnection faults by constructing a feature space topology consistency criterion according to an embodiment of the present invention; Figure 5 This is a schematic diagram illustrating the principle of the Hausdorff distance matrix provided in accordance with an embodiment of the present invention; Figure 6 This is a schematic block diagram of an intelligent power distribution equipment open circuit fault detection system provided in accordance with an embodiment of the present invention; Figure 7 This is a schematic block diagram of an electronic device provided according to an embodiment of the present invention. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The embodiments described in this application are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this invention.

[0032] This invention acquires the negative-sequence current signal, negative-sequence voltage signal, and zero-sequence current signal of the line, then extracts the feature vectors of each signal, constructs a weighted Hausdorff distance matrix based on multiple feature vectors, and finally analyzes the Hausdorff distance matrix. When the feature space topology consistency criterion is met, it is determined that a line fault has occurred in the intelligent power distribution equipment, thus realizing the detection of single-phase open-circuit faults in intelligent power distribution equipment.

[0033] like Figure 1 As shown, Embodiment 1 of the present invention provides a method for detecting open circuit faults in intelligent power distribution equipment, comprising the following steps: Step 1: Obtain the negative sequence current value of the intelligent power distribution equipment in real time when it is working.

[0034] Step 2: When the negative sequence current value is greater than the current threshold, collect the negative sequence current signal, negative sequence voltage signal and zero sequence current signal of each line in the intelligent power distribution equipment.

[0035] In a preferred but non-limiting embodiment of the present invention, the negative sequence current value is compared with the current threshold. When the negative sequence current value is greater than the current threshold, it is determined that the intelligent power distribution equipment has malfunctioned, and it is necessary to determine which line in the intelligent power distribution equipment the malfunction occurred in. When the negative sequence current value is less than or equal to the current threshold, it is determined that the intelligent power distribution equipment is in normal working condition.

[0036] It is understandable that the negative sequence current signal refers to the current component in a three-phase symmetrical voltage system that has a phase angle difference of 120 degrees from the positive sequence current when the three-phase currents are unbalanced. In other words, when a fault occurs in intelligent power distribution equipment, a large number of negative sequence current signals will appear. The methods for obtaining negative sequence current signals, negative sequence voltage signals, and zero sequence current signals can be the symmetrical component method and the instantaneous power method.

[0037] In a preferred but non-limiting embodiment of the present invention, the current threshold corresponds to the load status of the intelligent power distribution equipment and is dynamically adjusted based on historical operating data and real-time load status.

[0038] More preferably, the step of dynamically adjusting the current threshold based on historical operating data and real-time load status specifically includes: Step A.1: Establish and dynamically update the negative sequence current reference threshold. More preferably, step A.1 includes: Step A.1.1: Continuously record and learn the historical negative sequence current data I of the intelligent power distribution equipment at different time periods under normal operating conditions. 2_history and the corresponding positive sequence current data I 1_history .

[0039] Step A.1.2: Establish corresponding negative sequence current reference thresholds I for different typical load ranges using moving averages.2_base For example, during peak load periods, due to increased three-phase imbalance, the negative sequence current reference threshold I... 2_base The correspondingly higher value is also observed during periods of low load; the negative sequence current reference threshold I is also observed during off-peak periods. 2_base The answer is lower.

[0040] Step A.1.3: Continuously update the negative sequence current reference threshold I as the intelligent power distribution equipment operates. 2_base This is to track the impact of equipment aging and slow changes in network architecture.

[0041] Step A.2: Calculate the final dynamic current threshold based on the negative sequence current reference threshold. More preferably, step A.2 includes: Step A.2.1, monitor the current positive sequence current I. 1_real This is to reflect the real-time load.

[0042] Step A.2.2, based on the current positive sequence current I 1_real The load range falls within, and the corresponding negative sequence current reference threshold I is invoked. 2_base .

[0043] Step A.2.3: Calculate the final dynamic current threshold I. set It can be expressed by the following formula:

[0044] In the formula: I 2_base It is based on the current positive sequence current I 1_real A defined negative sequence current reference threshold; σ I2 It is the standard deviation of historical negative sequence current data within the corresponding load range, representing the normal fluctuation range of negative sequence current under that load; K 1 and K 2 represents the reliability factor, typically... K The value of 1 is between 1.5 and 2.5. K The value of 2 is between 2 and 3. These two coefficients can be adjusted according to the sensitivity and reliability requirements of different power distribution areas.

[0045] Step 3: Determine the feature vector set corresponding to each line based on the negative sequence current signal, negative sequence voltage signal, and zero sequence current signal of each line, such as... Figure 2 As shown.

[0046] In a preferred but non-limiting embodiment of the present invention, step 3 specifically includes: Step 3.1: Due to the time-domain continuity of the negative-sequence current signal, negative-sequence voltage signal, and zero-sequence current signal, they can be considered as a time series, i.e., the negative-sequence current data sequence x. iI (i=1, 2, 3, ..., N), negative sequence voltage data sequence x iU and zero-sequence current data sequence x iI0 (i=1, 2, 3, ..., N).

[0047] Step 3.2: The time-shifting scale analysis method is used to process the negative sequence current data sequence, negative sequence voltage data sequence and zero sequence current data sequence to obtain the negative sequence current time-shifted subsequence set, the negative sequence voltage time-shifted subsequence set and the zero sequence current time-shifted subsequence set respectively.

[0048] In a further preferred embodiment of the present invention, step 3.2 specifically includes: Step 3.2.1, set the time shift scaling factor sequence τ (τ=1,2,..,τ) a ,..,τ max ), where τ max The preset maximum time shift scale is determined based on the signal sampling frequency and the characteristic period to be analyzed.

[0049] Step 3.2.2, based on the set time shift scaling factor τ a Perform intervals of τ on a given data sequence. a The sampling generates a set of time-shifted subsequences. Specifically, the first... j A time-shifted subsequence y j (τ) It can be expressed by the following formula:

[0050] In the formula: m The length of the subsequence (embedding dimension) must satisfy j+(m-1)τ≤N.

[0051] j The index of the starting point of the subsequence is 1≤j≤N-(m-1)τ.

[0052] Step 3.2.3, for each data sequence x in step 3.1 iI x iU and x iI0 For each time shift scale factor in the time shift scale factor sequence τ, repeat step 3.2.2 to generate all time shift subsequences at that scale, and obtain the negative sequence current time shift subsequence set, the negative sequence voltage time shift subsequence set, and the zero sequence current time shift subsequence set, respectively.

[0053] Step 3.3: Determine the first entropy value corresponding to the negative sequence current signal, the first entropy value corresponding to the negative sequence voltage signal, and the first entropy value corresponding to the zero sequence current signal based on the negative sequence current time shift subsequence set, the negative sequence voltage time shift subsequence set, and the zero sequence current time shift subsequence set, respectively.

[0054] In a further preferred embodiment of the present invention, such as Figure 3 As shown, the first entropy value is the fine time-shifted multi-scale diverse entropy (RTSMDE), and step 3.3 specifically includes: Step 3.3.1: Calculate the state probability of each negative sequence current time shift subsequence, negative sequence voltage time shift subsequence, and zero sequence current time shift subsequence.

[0055] More preferably, the state probability is obtained by calculating the proportion of similar vectors in the sequence.

[0056] Step 3.3.2: Determine the first entropy value corresponding to the negative sequence current signal based on the state probabilities of multiple negative sequence current time-shifted subsequences; determine the first entropy value corresponding to the negative sequence voltage signal based on the state probabilities of multiple negative sequence voltage time-shifted subsequences; determine the first entropy value corresponding to the zero sequence current signal based on the state probabilities of multiple zero sequence current time-shifted subsequences.

[0057] Step 3.4: Determine the corresponding multi-scale standard deviations of negative-sequence current, negative-sequence voltage, and zero-sequence current based on the negative-sequence current time-shifted subsequence set, negative-sequence voltage time-shifted subsequence set, and zero-sequence current time-shifted subsequence set from Step 3.2, that is, calculate the multi-scale standard deviation (MSD) of each time-shifted subsequence set.

[0058] Step 3.5: Determine the second entropy value of the negative-sequence current signal based on the first entropy value and the multi-scale standard deviation of the negative-sequence current signal, and use the second entropy value of the negative-sequence current signal as the feature vector of the negative-sequence current signal. The second entropy value of the negative-sequence voltage signal is determined based on the first entropy value and the multi-scale standard deviation of the negative-sequence voltage signal. This second entropy value is then used as the feature vector of the negative-sequence voltage signal. The second entropy value of the zero-sequence current signal is determined based on the first entropy value and the multi-scale standard deviation of the zero-sequence current signal. This second entropy value is then used as the feature vector of the zero-sequence current signal. .

[0059] More preferably, the second entropy value is the fine time-shifted multi-scale standard deviation entropy (RTSMSDDE), which is obtained by multiplying the first entropy value and the multi-scale standard deviation, and is expressed by the following formula:

[0060] Step 3.6: Concatenate the feature vectors of the negative-sequence current signal, negative-sequence voltage signal, and zero-sequence current signal to obtain a feature vector set, i.e. The set of eigenvectors can be viewed as a high-dimensional eigenvector.

[0061] Step 4: Determine the negative sequence current weight, negative sequence voltage weight, and zero sequence current weight for each line based on the feature vector set corresponding to each line.

[0062] The negative sequence current weight, negative sequence voltage weight, and zero sequence current weight change with the negative sequence current signal, negative sequence voltage signal, and zero sequence current signal, and the negative sequence current weight, negative sequence voltage weight, and zero sequence current weight are different for different lines.

[0063] In a preferred but non-limiting embodiment of the present invention, the feature vector set corresponding to each line is input into a preset data model to obtain the negative sequence current weight, negative sequence voltage weight, and zero sequence current weight of each line. The preset data model is a fully connected neural network based on an attention mechanism. Step 4 specifically includes: Step 4.1: Construct a pre-defined data model by inputting the feature vector set from Step 3.6 into a fully connected layer to learn the non-linear interactions between different features and generate higher-level feature representations, as shown in the following formula:

[0064] In the formula, W1 and b1 are the weight matrix and bias vector of the fully connected layer, respectively, ReLU is the activation function, and H... i This represents the features of the generated hidden layer.

[0065] Step 4.2: Through an attention layer, calculate an attention score for each of the three feature vectors in the feature vector set to evaluate their importance in the current fault scenario.

[0066] More preferably, the attention score is generated through another fully connected layer, expressed as follows:

[0067] In the formula, W2 and b2 are the parameters of the attention layer, and the output is... It is a three-dimensional vector, where each element corresponds to the initial score of a feature vector.

[0068] Step 4.3: Normalize the attention score using the Softmax function to ensure that the sum of the three weights is 1, thus obtaining the final negative-sequence current weight ω. i I Negative sequence voltage weight ω i U and zero-sequence current weight ωi I0 It can be expressed by the following formula:

[0069] Step 4.4: Supervised training of the preset data model using historical line breakage fault data. During training, samples that accurately identify faulty lines are taken as positive examples. The optimization objective is whether the weighted Hausdorff distance matrix, calculated using the weights of the model output, can correctly identify faulty lines. Gradient descent is used to minimize the line selection error rate, thereby determining all parameters W1, b1, W2, b2 in the model.

[0070] Understandably, this invention determines weights by using a pre-set data model instead of manually setting them, reducing subjective bias and enabling dynamic adjustment of the weights. Through the aforementioned construction and training process, the pre-set data model can dynamically allocate appropriate weights based on the real-time, multi-source feature vectors of each line, thereby highlighting key features and suppressing secondary features in the subsequent weighted Hausdorff distance calculation, thus improving the accuracy of fault line selection.

[0071] Step 5: Calculate the weighted Hausdorff distance between every two lines in the intelligent power distribution equipment based on the feature vector set corresponding to each line, the negative sequence current weight, the negative sequence voltage weight, and the zero sequence current weight of each line.

[0072] Hausdorff distance is a metric that measures the distance between two subsets in a space. It can be used to compare differences in shape, structure, or features between different sets. Its principle is as follows: Figure 5 As shown, this represents two finite point sets A = (a1, ..., a2) n B = (b1, ..., b) n The maximum difference between ).

[0073] In a preferred but non-limiting embodiment of the present invention, a first Hausdorff distance component is calculated based on the negative-sequence current signal feature vectors of the two lines, a second Hausdorff distance component is calculated based on the negative-sequence voltage signal feature vectors of the two lines, and a third Hausdorff distance component is calculated based on the zero-sequence current signal feature vectors of the two lines. Then, the three components are weighted and summed according to the negative-sequence current weight, the negative-sequence voltage weight, and the zero-sequence current weight to obtain the final weighted Hausdorff distance between each pair of lines. The weighted Hausdorff distance between line i and line j is shown below. Expressed as follows:

[0074] In the formula: Hij I The first Hausdorff distance component is calculated based on the eigenvector of the negative sequence current signal; H ij U The second Hausdorff distance component is calculated based on the eigenvector of the negative-sequence voltage signal. H ij I0 The third Hausdorff distance component is calculated based on the eigenvector of the zero-sequence current signal. ω i I ω i U ω i I0 These are the negative sequence current weight, negative sequence voltage weight, and zero sequence current weight for line i, respectively.

[0075] More preferably, the negative sequence current weight can be 0.5, the negative sequence voltage weight can be 0.3, and the zero sequence current weight can be 0.2.

[0076] Step 6: Use the weighted Hausdorf distance between each pair of lines. As matrix elements, construct a weighted Hausdorff distance matrix.

[0077] In a preferred but non-limiting embodiment of the present invention, after obtaining multiple weighted Hausdorff distances, the line numbers i and j in the corresponding intelligent power distribution equipment are placed in their respective positions, and the multiple weighted Hausdorff distances are combined to obtain a Hausdorff distance matrix. Specifically, the weighted Hausdorff distances between the same line and other lines are set in the same row and column, thereby forming the weighted Hausdorff distance matrix.

[0078] More preferably, the weighted Hausdorff distance matrix Expressed as follows: ; In the formula: i and j are the line numbers in the intelligent power distribution equipment; n is the total number of lines; This represents the weighted Hausdorff distance between line i and line j.

[0079] It is understandable that the main diagonal elements of the Hausdorff distance matrix are all 0, and it is a real symmetric matrix. Furthermore, the route corresponding to the Hausdorff distance can be easily determined from the Hausdorff distance matrix, thus facilitating the subsequent identification of faulty routes.

[0080] Step 7: Based on the weighted Hausdorff distance matrix, construct the feature space topology consistency criterion to determine whether a branch or busbar in the intelligent power distribution equipment has a broken line fault.

[0081] In a preferred but non-limiting embodiment of the invention, such as Figure 4 As shown, step 7 specifically includes: Step 7.1: Based on the weighted Hausdorff distance matrix from Step 6, calculate the relative density factor for each line to form a relative density vector. In a further preferred embodiment, Step 7.1 specifically includes: Step 7.1.1, for line i, calculate its average Hausdorff distance to all other lines, expressed by the following formula:

[0082] Step 7.1.2: The larger the average Hausdorff distance, the more "isolated" the line is in the feature space, and the lower its density. The reciprocal of the average Hausdorff distance is used as the density factor, expressed by the following formula:

[0083] Step 7.1.3: To eliminate the influence of absolute values, the density factor ρ calculated in step 7.1.2 is... i Normalized to relative density factor D i Specifically, the Softmax function is used to transform it into a probability distribution, which directly reflects the "relative probability" that line i is a faulty line, expressed by the following formula:

[0084] Step 7.1.4: Calculate the relative density factor of all lines in the system. D i Combined into relative density vector D It can be expressed by the following formula:

[0085] Step 7.2, compare the relative density vector from Step 7.1. D The information entropy is compared with the preset entropy threshold to determine whether the line breakage fault occurred on a branch or the bus.

[0086] More preferably, the relative density vector is calculated. D The information entropy is expressed by the following formula:

[0087] More preferably, the entropy threshold E th Set to a certain percentile of the entropy value under normal operating conditions.

[0088] More preferably, the branch fault criterion is as follows: if E D < E th If so, it is determined to be a branch line open circuit fault. A low entropy value implies a relative density vector... D One of them D i Significantly greater than other values, meaning there is one and only one "isolated" line in the system, which meets the characteristics of a branch fault. In this case, the faulty line is... D The largest D i The corresponding route.

[0089] More preferably, the bus fault criterion is as follows: if E D ≥ E th If this occurs, it is determined to be a busbar open circuit fault. A high entropy value means that all lines... D i The values ​​are not significantly different, no line is significantly isolated, and all lines exhibit similar characteristics, which is a typical feature of bus faults.

[0090] It is understandable that the feature space topology consistency criterion not only considers the Hausdorff distance of a single feature between lines, but also further analyzes the distribution density and relative distribution relationship of multiple line features in multidimensional space. This invention takes into account that the features of faulty lines form a relatively isolated cluster in the feature space, while the features of non-faulty lines cluster in another region. It quantifies the degree of isolation of each line feature in the feature space through a relative density factor, making it suitable for fault judgment based on multiple features. Furthermore, based on the improved criterion, topology consistency verification can be performed, leading to more reliable fault identification.

[0091] like Figure 6 As shown, Embodiment 2 of the present invention provides an intelligent power distribution equipment open circuit fault detection system, which operates the intelligent power distribution equipment open circuit fault detection method as described in Embodiment 1. The intelligent power distribution equipment open circuit fault detection system 70 includes: The current acquisition module 702 is used to acquire the negative sequence current value of the intelligent power distribution equipment. The signal acquisition module 704 is used to acquire the negative sequence current signal, negative sequence voltage signal and zero sequence current signal of each line in the intelligent power distribution equipment when the negative sequence current value is greater than the current threshold. The feature extraction module 706 is used to extract the feature vector set corresponding to each line; The weight determination module 708 is used to determine the negative sequence current weight, negative sequence voltage weight and zero sequence current weight for each line. Distance calculation module 710 is used to calculate the weighted Hausdorff distance between every two lines in intelligent power distribution equipment; Matrix construction module 712 is used to construct the weighted Hausdorff distance matrix; The fault diagnosis module 714 is used to construct a feature space topology consistency criterion and determine whether a branch or busbar in the intelligent power distribution equipment has a broken line fault.

[0092] Embodiment 3 of the present invention provides an intelligent power distribution device equipped with an intelligent power distribution equipment disconnection fault detection system as described in Embodiment 2.

[0093] The intelligent power distribution device provided by the present invention can achieve the technical effects of any of the above embodiments, which will not be elaborated further here.

[0094] like Figure 7 As shown, Embodiment 4 of the present invention provides an electronic device. The electronic device 80 includes a memory 802, a processor 804, and a computer program stored in the memory 802 and executable on the processor 804. When the processor 804 executes the computer program, it implements the steps of the intelligent power distribution equipment disconnection fault detection method as described in Embodiment 1.

[0095] The electronic device 80 provided by the present invention, when the processor 804 executes the computer program, implements the steps of the above-mentioned intelligent power distribution equipment disconnection fault detection method, and can achieve the technical effects of any of the above embodiments, which will not be repeated here.

[0096] Embodiment 5 of the present invention provides a storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the intelligent power distribution equipment disconnection fault detection method as described in Embodiment 1.

[0097] The storage medium provided by this invention enables the computer program to implement the steps of the above-described intelligent power distribution equipment disconnection fault detection method when executed by a processor, and can achieve the technical effects of any of the above embodiments, which will not be elaborated further.

[0098] Compared with the prior art, the beneficial effects of the present invention include at least the following: 1. This invention introduces negative sequence voltage and zero sequence current signals, and fully considers multi-source information such as voltage characteristics and transient characteristics, to accurately identify open circuit faults in complex power distribution networks.

[0099] 2. This invention uses an adaptive current threshold instead of a fixed setting value for fault identification, automatically adapting to various power distribution network operation scenarios and reducing the workload of manually adjusting the setting value.

[0100] 3. This invention uses a weighted Hausdorff distance matrix, which assigns different weights based on feature importance, to enhance the sensitivity to key features. It actively strengthens key features when there is noise interference or load fluctuations, further reducing the risk of misjudgment.

[0101] 4. This invention improves the fault criteria by quantifying the degree of isolation of each line in the feature space through the relative density factor, which is applicable to fault judgment with multiple features. This invention adopts the feature space topology consistency criterion, which not only considers the Hausdorff distance of a single feature between lines, but also further analyzes the distribution density and relative distribution relationship of multiple line features in multidimensional space, and performs topology consistency verification, thereby increasing the reliability of fault identification.

[0102] 5. In summary, this invention has higher fault detection accuracy, stronger anti-interference and adaptability, and is suitable for modern power distribution networks with a high proportion of new energy access and multiple operating modes.

[0103] In the description of this specification, the terms "one embodiment," "some embodiments," "specific embodiment," etc., refer to a specific feature, structure, material, or characteristic described in connection with that embodiment or example, which is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0104] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for detecting open circuit faults in intelligent power distribution equipment, characterized in that, Includes the following steps: Obtain the negative sequence current value of intelligent power distribution equipment; When the negative sequence current value is greater than the current threshold, the negative sequence current signal, negative sequence voltage signal and zero sequence current signal of each line in the intelligent power distribution equipment are acquired, wherein the current threshold is dynamically adjusted based on historical operating data and real-time load status. The feature vector set corresponding to each line is determined based on the negative sequence current signal, negative sequence voltage signal and zero sequence current signal of each line, wherein the feature vector set includes: negative sequence current signal feature vector, negative sequence voltage signal feature vector and zero sequence current signal feature vector; The negative sequence current weight, negative sequence voltage weight, and zero sequence current weight of each line are determined based on the feature vector set corresponding to each line. The weighted Hausdorff distance between every two lines in the intelligent power distribution equipment is calculated based on the feature vector set corresponding to each line and the negative sequence current weight, negative sequence voltage weight, and zero sequence current weight of each line. Using the weighted Hausdorff distance between every two lines as matrix elements, the weighted Hausdorff distances between the same line and other lines are set in the same row and column to form a weighted Hausdorff distance matrix. Based on the weighted Hausdorff distance matrix, a feature space topology consistency criterion is constructed to determine whether a branch or busbar in an intelligent power distribution device has a broken line fault.

2. The method for detecting open circuit faults in intelligent power distribution equipment according to claim 1, characterized in that: The current threshold is dynamically adjusted based on historical operating data and real-time load status, including the following steps: Continuously record and learn the historical negative sequence current data and corresponding positive sequence current data of intelligent power distribution equipment at different time periods under normal operating conditions; By using moving averages, corresponding negative sequence current benchmark thresholds are established for different typical load ranges and updated as the intelligent power distribution equipment operates. Real-time monitoring of the current positive sequence current; and based on the load range into which the current positive sequence current falls, calling the corresponding negative sequence current reference threshold. The final dynamic current threshold is calculated using the following formula: In the formula, I 2_base It is based on the current positive sequence current I 1_real The determined negative sequence current reference threshold, σ I2 It is the standard deviation of historical negative sequence current data within the corresponding load range. K 1 and K 2 represents the reliability coefficient.

3. The method for detecting open circuit faults in intelligent power distribution equipment according to claim 1, characterized in that: The step of determining the feature vector set corresponding to each line based on the negative sequence current signal, negative sequence voltage signal, and zero sequence current signal of each line includes: By taking the negative sequence current signal, negative sequence voltage signal and zero sequence current signal as time series, negative sequence current data sequence, negative sequence voltage data sequence and zero sequence current data sequence are obtained; The time-shifting scale analysis method is used to process the negative sequence current data sequence, negative sequence voltage data sequence, and zero sequence current data sequence to obtain the negative sequence current time-shifted subsequence set, the negative sequence voltage time-shifted subsequence set, and the zero sequence current time-shifted subsequence set, respectively. The first entropy value corresponding to the negative sequence current signal, the first entropy value corresponding to the negative sequence voltage signal, and the first entropy value corresponding to the zero sequence current signal are determined respectively based on the negative sequence current time shift subsequence set, the negative sequence voltage time shift subsequence set, and the zero sequence current time shift subsequence set; The corresponding multi-scale standard deviations of negative sequence current, negative sequence voltage, and zero sequence current are determined based on the sets of negative sequence current time-shifted subsequences, negative sequence voltage time-shifted subsequences, and zero sequence current time-shifted subsequences, respectively. The second entropy value corresponding to the negative-sequence current signal is determined based on the first entropy value and the multi-scale standard deviation of the negative-sequence current signal, and is used as the feature vector of the negative-sequence current signal; the second entropy value corresponding to the negative-sequence voltage signal is determined based on the first entropy value and the multi-scale standard deviation of the negative-sequence voltage signal, and is used as the feature vector of the negative-sequence voltage signal; the second entropy value corresponding to the zero-sequence current signal is determined based on the first entropy value and the multi-scale standard deviation of the zero-sequence current signal, and is used as the feature vector of the zero-sequence current signal. The feature vector set is obtained by concatenating the feature vectors of the negative sequence current signal, the negative sequence voltage signal, and the zero sequence current signal.

4. The method for detecting open circuit faults in intelligent power distribution equipment according to claim 1, characterized in that: The step of determining the negative-sequence current signal feature weight, negative-sequence voltage signal feature weight, and zero-sequence current signal feature weight for each line based on the feature vector set corresponding to each line includes: The set of feature vectors corresponding to each line is input into a preset data model to obtain the negative sequence current weight, negative sequence voltage weight, and zero sequence current weight of each line.

5. The method for detecting open circuit faults in intelligent power distribution equipment according to claim 1, characterized in that: Weighted Hausdorff distance between line i and line j Expressed as follows: In the formula: H ij I The first Hausdorff distance component is calculated based on the eigenvector of the negative sequence current signal. H ij U The second Hausdorff distance component is calculated based on the eigenvector of the negative-sequence voltage signal. H ij I0 The third Hausdorff distance component is calculated based on the eigenvector of the zero-sequence current signal. ω i I ω i U ω i I0 These are the negative sequence current weight, negative sequence voltage weight, and zero sequence current weight for line i, respectively.

6. The method for detecting open circuit faults in intelligent power distribution equipment according to claim 1, characterized in that: The steps for constructing the feature space topology consistency criterion include: Based on the weighted Hausdorff distance matrix, the relative density factor of each line is calculated to form a relative density vector. The calculation of the relative density factor includes: calculating the average Hausdorff distance of a certain line to all other lines, taking the reciprocal of the average Hausdorff distance as the density factor of the line, and normalizing the density factor to a probability distribution to obtain the relative density factor. By comparing the information entropy of the relative density vector with a preset entropy threshold, it can be determined whether the line breakage fault occurred on a branch or a bus.

7. The method for detecting open circuit faults in intelligent power distribution equipment according to claim 6, characterized in that: The relative density factor is expressed by the following formula: In the formula: n is the total number of lines, ρ i This is the density factor.

8. The method for detecting open circuit faults in intelligent power distribution equipment according to claim 6, characterized in that: The information entropy of the relative density vector is expressed by the following formula: In the formula: n is the total number of lines, D i is the relative density factor.

9. The method for detecting open circuit faults in intelligent power distribution equipment according to claim 6, characterized in that: The entropy threshold is set to a certain percentile of the entropy value under normal operating conditions.

10. The method for detecting open circuit faults in intelligent power distribution equipment according to claim 6, characterized in that: When the information entropy of the relative density vector is less than the preset entropy threshold, it is determined to be a branch line breakage fault, and the faulty line is the line corresponding to the largest relative density factor. When the information entropy of the relative density vector is greater than or equal to the preset entropy threshold, it is determined to be a bus line breakage fault.

11. A system for detecting open circuit faults in intelligent power distribution equipment, comprising the method for detecting open circuit faults in intelligent power distribution equipment as described in any one of claims 1 to 10, characterized in that, include: The current acquisition module is used to acquire the negative sequence current value of the intelligent power distribution equipment; The signal acquisition module is used to acquire the negative sequence current signal, negative sequence voltage signal and zero sequence current signal of each line in the intelligent power distribution equipment when the negative sequence current value is greater than the current threshold. The feature extraction module is used to extract the feature vector set corresponding to each line; The weight determination module is used to determine the negative sequence current weight, negative sequence voltage weight, and zero sequence current weight for each line. The distance calculation module is used to calculate the weighted Hausdorff distance between every two lines in the intelligent power distribution equipment; The matrix building module is used to construct the weighted Hausdorff distance matrix; The fault diagnosis module is used to construct a feature space topology consistency criterion and determine whether a branch or busbar in the intelligent power distribution equipment has a broken line fault.

12. An intelligent power distribution device, characterized in that, Also includes: The intelligent power distribution device is equipped with the intelligent power distribution equipment disconnection fault detection system as described in claim 11.

13. An electronic device, characterized in that, include: A memory that stores computer instructions; A processor, which executes the steps of the intelligent power distribution equipment disconnection fault detection method as described in any one of claims 1 to 10 when executing the computer instructions.

14. A storage medium storing computer instructions thereon, characterized in that: When the computer instructions are executed by the processor, they perform the steps of the intelligent power distribution equipment disconnection fault detection method as described in any one of claims 1 to 10.

Images