Power distribution network operation mode optimization decision-making method, system, equipment and medium

Abstract

The invention discloses a power distribution network operation mode optimization decision-making method, system and device and a medium, and belongs to the technical field of power systems, and the method comprises the steps: obtaining power distribution network operation safety monitoring data, and carrying out the independent component analysis processing, and obtaining a power distribution network operation safety monitoring data matrix; mapping the power distribution network operation safety monitoring data matrix to a high-dimensional feature space, and extracting feature representation of the power distribution network operation safety monitoring data matrix in the high-dimensional feature space; calculating a kernel divergence matrix based on the feature representation, wherein the kernel divergence matrix comprises a kernel intra-class divergence matrix and a kernel inter-class divergence matrix; and determining an optimal discrimination vector according to the intra-kernel-class divergence matrix and the inter-kernel-class divergence matrix, carrying out similarity matching on the optimal discrimination vector and an optimal kernel discrimination vector library of a pre-established power grid operation fault data set, and identifying a power distribution network operation fault type. According to the invention, accurate identification of the operation fault of the power distribution network is realized, and the accuracy and real-time performance of monitoring the operation state of the power distribution network are obviously improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and in particular to a distribution network operation mode optimization decision-making method, system, equipment and medium. Background Art

[0002] In an increasingly complex and volatile electricity market, the lack of an effective operational optimization decision-making system for distribution networks leads to frequent equipment failures, which in turn impacts the stable operation of the entire power system. To address this challenge, researchers have begun developing a distribution network operation optimization decision-making system. By applying modern information technology and optimization algorithms, this system enables real-time monitoring, analysis, and optimization of distribution network operation status.

[0003] While existing technologies for optimizing distribution network operation have made some progress, they still face the following shortcomings: First, real-time monitoring capabilities are insufficient. Traditional monitoring systems often fail to monitor the operating status of distribution networks in real time, resulting in delayed fault detection and inefficient handling. Second, analytical methods are limited to simple data analysis and statistics, lacking in-depth analysis and optimization capabilities, and unable to fully tap into and utilize the potential value of distribution network operation data. Finally, decision support is insufficient. Existing decision support systems often lack intelligence and automation capabilities, failing to provide timely and accurate decision support to operations and maintenance personnel. This leads to inefficient decision-making and may even result in misjudgments. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by the present invention is: how to realize real-time monitoring of the operating status of the distribution network and accurate fault identification through high-dimensional feature space mapping and kernel discriminant analysis technology, while improving the accuracy and real-time performance of fault diagnosis, reducing the misjudgment rate and optimizing the fault handling process, thereby effectively improving the stability and reliability of the distribution network operation.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for optimizing the decision-making of the distribution network operation mode, which comprises the following steps:

[0007] Obtain distribution network operation safety monitoring data and perform independent component analysis to obtain the distribution network operation safety monitoring data matrix;

[0008] Mapping the distribution network operation safety monitoring data matrix to a high-dimensional feature space, and extracting a feature representation of the distribution network operation safety monitoring data matrix in the high-dimensional feature space;

[0009] Calculating a kernel divergence matrix based on the feature representation, wherein the kernel divergence matrix includes a kernel intra-class divergence matrix and a kernel inter-class divergence matrix;

[0010] An optimal discriminant vector is determined according to the intra-core scatter matrix and the inter-core scatter matrix, and similarity matching is performed between the optimal discriminant vector and the optimal core discriminant vector library of the pre-established power grid operation fault data set to identify the distribution network operation fault type.

[0011] As a preferred solution of the distribution network operation mode optimization decision method described in the present invention, obtaining the distribution network operation safety monitoring data and performing independent component analysis processing includes:

[0012] Preprocessing the distribution network operation safety monitoring data to form preprocessed data;

[0013] Performing transformation processing on the preprocessed data to obtain a weight matrix;

[0014] The pre-processed data is decoupled using the weight matrix to obtain the distribution network operation safety monitoring data matrix. The beneficial effect of this preferred technical solution is that it uses a combination of data pre-processing and transformation processing to achieve effective decoupling of distribution network operation safety monitoring data, eliminate correlation interference between data, and improve the accuracy and efficiency of subsequent analysis.

[0015] As a preferred solution of the distribution network operation mode optimization decision method described in the present invention, mapping the distribution network operation safety monitoring data matrix into a high-dimensional feature space includes:

[0016] Selecting a mapping function to map the distribution network operation safety monitoring data matrix to the high-dimensional feature space through the mapping function;

[0017] Features are extracted from the high-dimensional feature space and a feature representation matrix is ​​constructed to obtain the feature representation. The beneficial effect of this preferred technical solution is that the data is mapped to the high-dimensional feature space through a mapping function, which fully utilizes the advantages of the kernel method in nonlinear data processing, so that the complex distribution network operating status can be better expressed and distinguished in the high-dimensional space.

[0018] As a preferred solution of the distribution network operation mode optimization decision method described in the present invention, wherein: calculating the kernel divergence matrix based on the feature representation includes:

[0019] determining a kernel sampling vector based on the feature representation;

[0020] Calculating a mean parameter based on the kernel sampling vector;

[0021] The intra-kernel divergence matrix and the inter-kernel divergence matrix are calculated based on the mean parameter. The beneficial effect of this preferred technical solution is that the method of calculating the kernel divergence matrix based on feature representation can effectively capture the differences and similarities between different categories of data, provide a reliable mathematical basis for fault identification, and improve the accuracy of fault classification.

[0022] As a preferred solution of the distribution network operation mode optimization decision method described in the present invention, wherein: determining the optimal discriminant vector according to the intra-kernel scatter matrix and the inter-kernel scatter matrix includes:

[0023] Establish characteristic equation;

[0024] The characteristic equation is solved to obtain the optimal discriminant vector, which is used for similarity matching with the optimal kernel discriminant vector library of the power grid operation fault data set.

[0025] As a preferred solution of the distribution network operation mode optimization decision method described in the present invention, wherein: performing similarity matching between the optimal discriminant vector and the optimal kernel discriminant vector library of the pre-established power grid operation fault data set includes:

[0026] Calculating the cosine of the vector intersection angle between the optimal discriminant vector and each vector in the optimal kernel discriminant vector library of the power grid operation fault data set;

[0027] evaluating the magnitude of the vector intersection angle cosine according to a preset threshold value, and when the vector intersection angle cosine is greater than the preset threshold value, determining that the distribution network operation fault type corresponding to the current distribution network operation state is the same as the fault type associated with the corresponding vector in the optimal kernel discriminant vector library of the power grid operation fault data set;

[0028] The optimal kernel discriminant vector library for the power grid operation fault dataset stores various types of distribution network operation fault modes, including equipment faults, line faults, abnormal voltage faults, and abnormal load faults. The beneficial effects of this preferred technical solution include: using the cosine of the vector intersection angle as a similarity metric, combined with a preset threshold judgment mechanism, achieves precise matching and identification of fault types, effectively reducing the misjudgment rate and improving the reliability of fault diagnosis.

[0029] As a preferred solution of the method for optimizing the decision-making of the distribution network operation mode described in the present invention, it further includes the step of optimizing the distribution network operation fault type, and the optimization process includes:

[0030] Formulate a fault handling strategy based on the distribution network operation fault type, isolate the faulty equipment, and prevent the fault from expanding;

[0031] Carry out emergency repairs or replacements for equipment failures in the distribution network operation failure types, including repairs of transformer failures, switchgear failures, and protection device failures;

[0032] Carry out inspection and maintenance on line faults in the distribution network operation fault type, formulate a power supply restoration plan, restore power supply according to the user importance level, and prioritize the normal power supply of important users. The beneficial effect of this preferred technical solution is: formulate differentiated processing strategies for different types of distribution network faults, and combine power supply restoration plans based on user importance level, which not only ensures the priority power supply of important users, but also minimizes power outage losses and improves the stability and reliability of distribution network operation.

[0033] Another object of the present invention is to provide a distribution network operation mode optimization decision system.

[0034] To solve the above technical problems, the present invention provides the following technical solutions: a distribution network operation mode optimization decision system, comprising: a data processing module for acquiring distribution network operation safety monitoring data and performing independent component analysis processing to obtain a distribution network operation safety monitoring data matrix;

[0035] A feature mapping module is used to map the distribution network operation safety monitoring data matrix to a high-dimensional feature space and extract the feature representation of the distribution network operation safety monitoring data matrix in the high-dimensional feature space;

[0036] A divergence calculation module is used to calculate the kernel divergence matrix based on the feature representation. The kernel divergence matrix includes the intra-kernel divergence matrix and the inter-kernel divergence matrix.

[0037] The fault identification module is used to determine the optimal discriminant vector based on the intra-core scatter matrix and the inter-core scatter matrix, perform similarity matching between the optimal discriminant vector and the optimal kernel discriminant vector library of the pre-established power grid operation fault data set, and identify the distribution network operation fault type.

[0038] The present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of the distribution network operation mode optimization decision method are implemented.

[0039] The present invention provides a computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of the distribution network operation mode optimization decision-making method are implemented.

[0040] The beneficial effects of the present invention are as follows: the present invention can accurately identify potential safety hazards in the operation of the distribution network, can improve the real-time monitoring capability of the distribution network operation, and improve the stability of the distribution network operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0042] Figure 1 The present invention provides an overall flow chart of a method for optimizing decision-making in a distribution network operation mode according to an embodiment of the present invention. DETAILED DESCRIPTION

[0043] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0044] Example 1, with reference to Figure 1 , which is the first embodiment of the present invention, provides a method for optimizing decision-making of distribution network operation mode, comprising:

[0045] S100: Acquire distribution network operation safety monitoring data and perform independent component analysis to obtain a distribution network operation safety monitoring data matrix;

[0046] S200: Mapping the distribution network operation safety monitoring data matrix to a high-dimensional feature space, and extracting feature representations of the distribution network operation safety monitoring data matrix in the high-dimensional feature space;

[0047] S300: Calculating a kernel divergence matrix based on the feature representation, where the kernel divergence matrix includes a kernel intra-class divergence matrix and a kernel inter-class divergence matrix;

[0048] S400: Determine an optimal discriminant vector based on the intra-core scatter matrix and the inter-core scatter matrix, perform similarity matching between the optimal discriminant vector and the optimal kernel discriminant vector library of the pre-established power grid operation fault data set, and identify the distribution network operation fault type.

[0049] It should be noted that in the increasingly complex and volatile power market environment, distribution network equipment failures are frequent, impacting the stable operation of the entire power system. Existing distribution network operation optimization technologies suffer from insufficient real-time monitoring capabilities, limited analytical methods, and insufficient decision support. Traditional monitoring systems often fail to monitor the operating status of distribution networks in real time, resulting in delayed fault detection and inefficient handling. Furthermore, existing analytical methods are often limited to simple data analysis and statistics, lacking in-depth analysis and optimization capabilities, and are unable to fully tap and utilize the potential value of distribution network operating data. Furthermore, existing decision support systems often lack intelligence and automation capabilities, failing to provide timely and accurate decision support to operations and maintenance personnel, leading to inefficient decision-making and even the possibility of misjudgment.

[0050] Therefore, to address the above-mentioned problems, this embodiment, through steps S100-S400, constructs a distribution network operation mode optimization decision-making method, which achieves efficient processing and analysis of distribution network operation safety monitoring data, can accurately identify safety hazards in distribution network operation, and significantly improves the real-time monitoring capability of distribution network operation and the accuracy of fault identification. At the same time, due to the use of advanced technologies such as high-dimensional feature space mapping and kernel discriminant analysis, the present invention has strong adaptability to complex and nonlinear distribution network operation states, can effectively respond to fault identification needs under various complex working conditions, and improve the stability and reliability of distribution network operation.

[0051] Example 2, reference Figure 1 , which is the second embodiment of the present invention, is an embodiment of the present invention. Based on the above embodiments, a distribution network operation mode optimization decision method is provided.

[0052] In the embodiment of the present invention, obtaining the distribution network operation safety monitoring data and performing independent component analysis in step S100 includes the following steps A1-A3:

[0053] A1: Preprocess the distribution network operation safety monitoring data to generate preprocessed data;

[0054] A2: Transform the preprocessed data to obtain the weight matrix;

[0055] A3: Decouple the preprocessed data through the weight matrix to obtain the distribution network operation safety monitoring data matrix.

[0056] Specifically, in A1, the preprocessing of distribution network operation safety monitoring data includes data centering and whitening. Centering involves subtracting the mean of each column of the observed data to achieve a mean of zero. Whitening ensures that the dimensions of the new data matrix are independent of each other and have unit variance. This preprocessing method effectively eliminates noise interference in the data and provides a more stable data foundation for subsequent independent component analysis.

[0057] In A2, the preprocessed data is transformed, including two steps: linear transformation and nonlinear transformation. Linear transformation uses the current weight matrix to linearly transform the observed data to obtain the transformed data; nonlinear transformation applies a nonlinear function to the transformed data to make it more similar to the independent components. The weight matrix is ​​then updated based on the results of the nonlinear transformation and an optimization algorithm, such as the fixed-point iteration method and the gradient ascent method. Finally, the updated weight vector is normalized to ensure its length is 1. At the same time, the weight vector is decorrelated using the Gram-Schmidt orthogonalization method or other orthogonalization techniques to ensure that the newly obtained independent components are orthogonal to the previously obtained independent components.

[0058] In A3, the weight vector is iteratively updated to check whether the change in the weight vector before and after the update is less than a preset threshold. If the convergence condition is met, the algorithm is considered converged and iterations are stopped; otherwise, iterations continue. Once all independent components have converged, the final weight matrix is ​​used to decouple the preprocessed data, resulting in a distribution network operation safety monitoring data matrix. The components in this matrix are statistically independent, which can more accurately reflect the different aspects of distribution network operation and provide a good data foundation for subsequent high-dimensional feature space mapping.

[0059] In an optional embodiment, the independent component analysis process performed in step S100 can also be implemented using the information maximization principle. This method maximizes the mutual information of the output signals and uses the maximum entropy principle to find independent components that best preserve the original data information. This method is suitable for processing non-stationary signals and can better cope with sudden changes in the operating state of the distribution network.

[0060] In another optional embodiment, the independent component analysis process performed in step S100 can also be performed based on time-frequency analysis. This method first performs a time-frequency transform (such as a wavelet transform or a Hilbert-Huang transform) on the distribution network operation safety monitoring data, and then applies independent component analysis in the time-frequency domain. This method can simultaneously capture the time and frequency characteristics of the data and is particularly effective for identifying transient faults and harmonic interference in the power grid.

[0061] In an embodiment of the present invention, mapping the distribution network operation safety monitoring data matrix to a high-dimensional feature space in step S200 includes the following steps B1-B2:

[0062] B1: Select a mapping function to map the distribution network operation safety monitoring data matrix into a high-dimensional feature space through the mapping function;

[0063] B2: Extract features in the high-dimensional feature space and construct a feature representation matrix to obtain feature representation.

[0064] Specifically, in B1, a nonlinear mapping function is used The distribution network operation safety monitoring data matrix X is mapped to the high-dimensional feature space H, that is, This mapping can transform the nonlinear relationship in the original space into a linear relationship in the high-dimensional feature space, which is convenient for subsequent analysis and processing.

[0065] In B2, the pattern vector of the distribution network operation safety monitoring data matrix X in the high-dimensional feature space is extracted Since the dimension of high-dimensional feature space may be very high or even infinite, directly calculating It is usually not feasible. Therefore, in actual operation, the kernel technique is used to directly calculate the inner product through the kernel function to avoid explicit mapping to high-dimensional space. Based on the pattern vector of matrix X in high-dimensional feature space, Calculate the covariance matrix of the sample matrix X in the high-dimensional linear space

[0066]

[0067] in, represents the eigenvector matrix after centering, It is expressed as the mean of the sample after being mapped to the dimensional linear space, express The transpose of , n represents the number of samples.

[0068] Then, the kernel matrix K∈R is defined according to the kernel function n×n , where each element K of the kernel matrix K ij for:

[0069]

[0070] In this way, the present invention obtains the feature representation of the distribution network operation safety monitoring data in a high-dimensional feature space, laying the foundation for the subsequent kernel divergence matrix calculation.

[0071] In an optional implementation, the mapping function selected in step S200 may be a Gaussian kernel function, which is in the form of:

[0072]

[0073] Where σ is the kernel width parameter, ‖x i -x j ‖ is the Euclidean distance between two sample vectors. The Gaussian kernel function can map samples into an infinite-dimensional feature space and is suitable for processing complex nonlinear relationships. It is particularly effective in distinguishing multiple fault modes in distribution network operation.

[0074] In another optional implementation, the mapping function selected in step S200 may be a polynomial kernel function, which is in the form of:

[0075] K(x i ,x j )=(x i ·x j +c) d ;

[0076] Where d is the order of the polynomial, and c is a constant. Polynomial kernel functions are suitable for processing data with polynomial relationships between features and perform well in scenarios such as distribution network load forecasting and line loss analysis.

[0077] In the embodiment of the present invention, calculating the kernel divergence matrix based on the feature representation in step S300 includes the following steps C1-C3:

[0078] C1: Determine the kernel sampling vector based on feature representation;

[0079] C2: Calculate the mean parameter based on the kernel sampling vector;

[0080] C3: Calculates the intra-class scatter matrix and inter-class scatter matrix based on the mean parameter.

[0081] Specifically, in C1, for the observation vector x in the original space, calculate its kernel decoupling sampling vector σ in the high-dimensional feature space x :

[0082] σ x =(K(x1, x), ..., K(x M , x)) T ;

[0083] Among them, K(x1, x) represents the kernel function value of vector x1 and vector x, M represents the number of reference vectors, σ x It is an M-dimensional vector that captures the distribution characteristics of the original observation vector x in the high-dimensional feature space.

[0084] In C2, based on the kernel decoupling sampling vector σ x Calculate the kernel sampling mean μ within the class i:

[0085]

[0086] in, Represents the category x corresponding to the matrix X in the high-dimensional feature space M The jth sampling vector of the i-th subcategory and the reference vector x M The kernel function value of .

[0087] At the same time, the overall kernel sampling mean μ0 needs to be calculated:

[0088]

[0089] Among them, K(x M , x i ) represents the category x corresponding to the matrix X in the high-dimensional feature space M With category x i The kernel function value of the sampling vector .

[0090] In C3, based on the kernel sampling mean μ within the class i The kernel intra-class scatter matrix K is calculated by the overall kernel sampling mean μ0 W :

[0091]

[0092] At the same time, based on the kernel sampling mean μ within the class i The kernel inter-class scatter matrix K is calculated by the overall kernel sampling mean μ0 b :

[0093]

[0094] Among them, (μ i -μ0) T Indicates (μ i -μ0). These two scatter matrices jointly describe the class distinction characteristics of distribution network operation status data in the high-dimensional feature space. The kernel intra-class scatter matrix K W Measures the degree of dispersion of data of the same category in high-dimensional feature space, the kernel inter-class scatter matrix K b Measures the distance between data centers of different categories.

[0095] In an optional embodiment, a nonparametric kernel density estimation method can also be used in step S300 to enhance the calculation accuracy of the kernel divergence matrix. This method does not assume that the data follows a specific distribution, but instead estimates the probability density based on the density of local data points. This method can more accurately characterize the complex distribution characteristics of the distribution network operating status data.

[0096] In another optional embodiment, step S300 may also utilize an adaptive kernel bandwidth selection technique to optimize kernel divergence matrix calculation. This technique dynamically adjusts kernel function parameters based on local data density, using a larger bandwidth in data-sparse areas and a smaller bandwidth in data-dense areas, thereby improving the ability to identify fault modes of varying sizes.

[0097] In the embodiment of the present invention, determining the optimal discriminant vector according to the intra-core scatter matrix and the inter-core scatter matrix in step S400 includes the following steps D1-D2:

[0098] D1: Establish characteristic equation;

[0099] D2: Solve the characteristic equation to obtain the optimal discriminant vector, which is used for similarity matching with the optimal kernel discriminant vector library of the power grid operation fault dataset.

[0100] Specifically, in D1, based on the kernel intra-class divergence matrix K W and the kernel inter-class divergence matrix K b Establish the characteristic equation:

[0101] K b β=λK W β;

[0102] Where λ is the eigenvalue matrix of the kernel matrix K, and β is the optimal discriminant vector to be solved in the present invention. This characteristic equation aims to find the projection direction that maximizes the ratio of the between-class divergence to the within-class divergence, i.e., the optimal discriminant vector.

[0103] In D2, the characteristic equation is solved using numerical methods (such as power iteration or QR decomposition) to obtain the eigenvalue λ and the corresponding eigenvector β. The eigenvector corresponding to the largest eigenvalue is selected as the optimal discriminant vector. This vector maximizes the separability between classes in the high-dimensional feature space, providing the best discriminant basis for subsequent fault type identification.

[0104] The optimal discriminant vector β is then similarly matched against a pre-established library of optimal kernel discriminant vectors from a grid operation fault dataset. The cosine of the vector intersection angle between β and each vector in the library is calculated. If the cosine of the vector intersection angle is greater than a preset threshold, the distribution network operation fault is determined to be of the same type as the fault corresponding to the optimal kernel discriminant vector in the library.

[0105] In an optional embodiment, regularization technology can also be used in step S400 to improve the stability of solving the characteristic equation. W Add a regularization term on the diagonal, that is, K W Replace with K W+αI (where α is the regularization parameter and I is the identity matrix) to reduce the impact of the singularity problem on the solution process and improve the generalization ability of the optimal discriminant vector.

[0106] In another optional embodiment, an incremental eigenvalue solution method can also be used in step S400. This method does not need to process all historical data at one time, but can gradually update the optimal discriminant vector as new data arrives. It is suitable for online learning scenarios in real-time monitoring of distribution networks.

[0107] In the embodiment of the present invention, the similarity matching of the optimal discriminant vector with the pre-established optimal core discriminant vector library of the power grid operation fault data set in step S400 includes the following steps E1-E3:

[0108] E1: Calculate the cosine of the vector intersection angle between the optimal discriminant vector and each vector in the optimal kernel discriminant vector library of the power grid operation fault data set;

[0109] E2: Evaluate the magnitude of the vector intersection angle cosine according to a preset threshold. When the vector intersection angle cosine is greater than the preset threshold, determine that the distribution network operation fault type corresponding to the current distribution network operation state is the same as the fault type associated with the corresponding vector in the optimal kernel discriminant vector library of the power grid operation fault dataset;

[0110] E3: Among them, the optimal kernel discriminant vector library of the power grid operation fault data set stores various types of distribution network operation fault modes, including equipment failure, line failure, voltage abnormality failure and load abnormality failure.

[0111] Specifically, in E1, the formula for calculating the cosine of the vector intersection angle is:

[0112]

[0113] Among them, β is the current optimal discriminant vector, β i is the i-th reference vector in the vector library, ‖β‖ represents the Euclidean norm of vector β, ‖β i ‖ represents the vector β i The Euclidean norm of β·β i Represents the dot product of two vectors. The closer the cosine of the vector intersection angle is to 1, the more similar the directions of the two vectors are, and the closer the corresponding fault types are.

[0114] In E2, the system evaluates the cosine of the vector intersection angle based on a preset threshold. When the cosine value exceeds the threshold, the system determines that the current distribution network operating state is the same fault type as the corresponding vector in the optimal kernel discriminant vector library. This similarity threshold-based judgment method ensures both pattern matching accuracy and a certain degree of fault tolerance.

[0115] In E3, the optimal kernel discriminant vector library of the power grid operation fault dataset stores feature vectors of various typical fault modes, including:

[0116] 1. Equipment failure: such as aging of transformer insulation, poor contact of switchgear, malfunction of protection device, etc.

[0117] 2. Line fault: such as line short circuit, disconnection, grounding, etc.;

[0118] 3. Abnormal voltage failure: such as overvoltage, undervoltage, voltage fluctuation, etc.;

[0119] 4. Abnormal load failure: such as sudden load increase, low power factor, three-phase imbalance, etc.

[0120] By matching these pre-calibrated fault modes, the system can quickly identify the fault type in the current distribution network operation.

[0121] In an optional embodiment, a weighted cosine similarity calculation method can also be used in step S400 to assign different weights to different feature dimensions, highlight the role of key features in similarity calculation, and improve the recognition accuracy of specific types of faults. The weighted cosine similarity can be expressed as:

[0122]

[0123] Among them, w j is the weight of the j-th dimension feature, β j and β i,j are vectors β and β respectively. i The jth component of , where d is the dimension of the vector.

[0124] In another optional embodiment, step S400 can also combine multiple similarity indicators for comprehensive judgment. In addition to the cosine of the vector intersection angle, multiple similarity metrics such as Euclidean distance and Mahalanobis distance can also be calculated simultaneously. The final similarity score is obtained through weighted fusion to improve the robustness of recognition.

[0125] In addition, the embodiment of the present invention further includes the step of optimizing the distribution network operation fault type, and the optimization process includes the following steps F1-F3:

[0126] F1: Develop a fault handling strategy based on the type of distribution network operation fault and isolate the faulty equipment;

[0127] F2: Emergency repair or replacement of equipment failures in distribution network operation, including transformer failures, switchgear failures, and protection device failures;

[0128] F3: Carry out inspection and maintenance on line faults in distribution network operation fault types, formulate power supply restoration plans, and restore power supply according to user importance levels.

[0129] Specifically, in F1, the system automatically generates a corresponding fault handling strategy based on the identified distribution network fault type. Based on the nature and location of the fault, the faulty equipment is isolated to prevent further expansion, and the faulty section is located through segmented isolation and step-by-step power testing.

[0130] In F2, for equipment failures, the system provides detailed repair instructions and replacement suggestions:

[0131] 1. Transformer failure: The system will recommend corresponding emergency repair plans based on the fault type (such as insulation breakdown, winding short circuit, oil overtemperature, etc.), including emergency measures and backup equipment deployment plans;

[0132] 2. Switchgear failure: Provide specialized maintenance solutions for different types of switches (such as circuit breakers, disconnectors, load switches, etc.) and different fault causes (such as mechanical failure, arc extinguishing failure, control circuit failure, etc.);

[0133] 3. Protection device failure: Provide solutions for resetting, adjusting parameters, or replacing protection devices to ensure the continuity of grid protection functions.

[0134] In F3, the system conducts comprehensive inspections and maintenance for line faults, including inspections, switching operations, and tripping operations. Simultaneously, it adjusts operating modes based on dispatch commands, restoring power to healthy users and prioritizing critical users to minimize power outage losses.

[0135] In an optional implementation, fault handling can also incorporate the distribution network's topology reconfiguration capabilities. Based on the current network state, the system calculates the optimal reconfiguration plan. By adjusting switch states, the network topology is altered to achieve load transfer and fault isolation, minimizing the impacted area. This approach is particularly suitable for distribution networks with redundant power supply paths, enabling fault isolation and repair without disrupting user power supply.

[0136] Another optional implementation method can also integrate predictive maintenance strategies. Based on historical fault data and equipment operating parameters, potential failure risks can be predicted and maintenance can be scheduled in advance, transforming reactive repairs into proactive prevention, significantly improving the reliability and operational efficiency of the distribution network. Predictive maintenance strategies can be implemented by establishing an equipment health index model. This model comprehensively considers multiple factors such as equipment operating time, load factor, environmental factors, and historical failure rates to calculate the health status and remaining life of the equipment, providing a scientific basis for maintenance decisions.

[0137] In summary, the distribution network operation mode optimization decision-making method provided by the present invention processes the distribution network operation safety monitoring data through independent component analysis, maps the data to a high-dimensional feature space, and realizes the accurate identification of distribution network operation faults through kernel divergence matrix analysis and optimal discriminant vector matching. This method can cope with the high-dimensional, nonlinear and time-varying characteristics of distribution network operation data, effectively reduce the misjudgment rate, and improve the reliability of fault diagnosis. At the same time, the system also provides optimization processing solutions for different fault types, including fault isolation, equipment repair and power supply restoration strategies, forming a complete set of distribution network operation mode optimization decision-making technology system. Compared with traditional methods, the present invention has significant advantages in improving the stability and reliability of distribution network operation, and provides a strong guarantee for the safe, economical and efficient operation of the power system.

[0138] Example 3 is the third embodiment of the present invention, which differs from the first two embodiments in that:

[0139] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0140] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a computer-readable medium can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0141] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0142] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0143] Embodiment 4 is the fourth embodiment of the present invention, which provides a distribution network operation mode optimization decision system, including:

[0144] The data processing module is used to obtain the distribution network operation safety monitoring data and perform independent component analysis to obtain the distribution network operation safety monitoring data matrix;

[0145] A feature mapping module is used to map the distribution network operation safety monitoring data matrix to a high-dimensional feature space and extract the feature representation of the distribution network operation safety monitoring data matrix in the high-dimensional feature space;

[0146] A divergence calculation module is used to calculate the kernel divergence matrix based on the feature representation. The kernel divergence matrix includes the intra-kernel divergence matrix and the inter-kernel divergence matrix.

[0147] The fault identification module is used to determine the optimal discriminant vector based on the intra-core scatter matrix and the inter-core scatter matrix, perform similarity matching between the optimal discriminant vector and the optimal kernel discriminant vector library of the pre-established power grid operation fault data set, and identify the distribution network operation fault type.

[0148] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A distribution network operation mode optimization decision method, characterized by: include, Obtain distribution network operation safety monitoring data and perform independent component analysis to obtain the distribution network operation safety monitoring data matrix; Mapping the distribution network operation safety monitoring data matrix to a high-dimensional feature space, and extracting a feature representation of the distribution network operation safety monitoring data matrix in the high-dimensional feature space; Calculating a kernel divergence matrix based on the feature representation, wherein the kernel divergence matrix includes a kernel intra-class divergence matrix and a kernel inter-class divergence matrix; An optimal discriminant vector is determined according to the intra-core scatter matrix and the inter-core scatter matrix, and similarity matching is performed between the optimal discriminant vector and the optimal core discriminant vector library of the pre-established power grid operation fault data set to identify the distribution network operation fault type.

2. A distribution network operation mode optimization decision method according to claim 1, characterized in that: Obtaining the distribution network operation safety monitoring data and performing independent component analysis processing includes: Preprocessing the distribution network operation safety monitoring data to form preprocessed data; Performing transformation processing on the preprocessed data to obtain a weight matrix; The pre-processed data is decoupled through the weight matrix to obtain the distribution network operation safety monitoring data matrix.

3. A distribution network operation mode optimization decision method according to claim 2, characterized in that: Mapping the distribution network operation safety monitoring data matrix to a high-dimensional feature space includes: Selecting a mapping function to map the distribution network operation safety monitoring data matrix to the high-dimensional feature space through the mapping function; Features are extracted in the high-dimensional feature space and a feature representation matrix is ​​constructed to obtain the feature representation.

4. A distribution network operation mode optimization decision method according to claim 3, characterized in that: Calculating the kernel divergence matrix based on the feature representation includes: determining a kernel sampling vector based on the feature representation; Calculating a mean parameter based on the kernel sampling vector; The intra-kernel scatter matrix and the inter-kernel scatter matrix are calculated based on the mean parameter.

5. A distribution network operation mode optimization decision method according to claim 4, characterized in that: Determine the optimal discriminant vector according to the intra-kernel scatter matrix and the inter-kernel scatter matrix include: Establish characteristic equation; The characteristic equation is solved to obtain the optimal discriminant vector, which is used for similarity matching with the optimal kernel discriminant vector library of the power grid operation fault data set.

6. A distribution network operation mode optimization decision method according to claim 4, characterized in that: Performing similarity matching between the optimal discriminant vector and a pre-established optimal kernel discriminant vector library of a power grid operation fault data set includes: Calculating the cosine of the vector intersection angle between the optimal discriminant vector and each vector in the optimal kernel discriminant vector library of the power grid operation fault data set; evaluating the magnitude of the vector intersection angle cosine according to a preset threshold value, and when the vector intersection angle cosine is greater than the preset threshold value, determining that the distribution network operation fault type corresponding to the current distribution network operation state is the same as the fault type associated with the corresponding vector in the optimal kernel discriminant vector library of the power grid operation fault data set; The optimal kernel discriminant vector library of the power grid operation fault data set stores various types of distribution network operation fault modes, including equipment faults, line faults, abnormal voltage faults and abnormal load faults.

7. A distribution network operation mode optimization decision method according to claim 4, characterized in that: The method further includes the step of optimizing the distribution network operation fault type, wherein the optimization process includes: Formulate a fault handling strategy based on the distribution network operation fault type and isolate the faulty equipment; Carry out emergency repairs or replacements for equipment failures in the distribution network operation failure types, including repairs of transformer failures, switchgear failures, and protection device failures; Carry out inspection and maintenance on line faults in the distribution network operation fault types, formulate power supply restoration plans, and restore power supply according to user importance levels.

8. A distribution network operation mode optimization decision system, applying a distribution network operation mode optimization decision method according to any one of claims 1 to 7, characterized in that: include: The data processing module is used to obtain the distribution network operation safety monitoring data and perform independent component analysis to obtain the distribution network operation safety monitoring data matrix; A feature mapping module is used to map the distribution network operation safety monitoring data matrix to a high-dimensional feature space and extract feature representations of the distribution network operation safety monitoring data matrix in the high-dimensional feature space; A divergence calculation module, configured to calculate a kernel divergence matrix based on the feature representation, wherein the kernel divergence matrix includes a kernel intra-class divergence matrix and a kernel inter-class divergence matrix; A fault identification module is used to determine an optimal discriminant vector based on the intra-core scatter matrix and the inter-core scatter matrix, perform similarity matching between the optimal discriminant vector and the optimal core discriminant vector library of the pre-established power grid operation fault data set, and identify the type of distribution network operation fault.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the distribution network operation mode optimization decision method described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a distribution network operation mode optimization decision method according to any one of claims 1 to 7 are implemented.

Images