Power distribution network dispatching behavior analysis method, system, equipment and medium

Abstract

The invention discloses a power distribution network scheduling behavior analysis method, system and device, and a medium. The method comprises the following steps: setting a scheduling rule corresponding to a scheduling behavior; analyzing the scheduling behavior through the scheduling operation ticket to obtain each operation instruction and target equipment, and judging whether each operation instruction violates a corresponding scheduling rule or not; recording the scheduling behaviors violating the scheduling rule, and for the scheduling behaviors not violating the scheduling rule, obtaining the target equipment and various state data of a feeder line where the target equipment is located; performing data preprocessing on the state data, and establishing an effective data set; and combining the effective data set with the recorded scheduling behaviors, and establishing a normal and abnormal behavior library of power distribution network scheduling. According to the method, a preliminary dispatching operation order dispatching instruction rule is formed according to a power distribution network dispatching procedure, primary preliminary checking is firstly carried out on each instruction by utilizing analysis of a dispatching operation order, and a basic error of a power distribution network dispatching behavior is identified.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of industrial power, and in particular to a power distribution network dispatching behavior analysis method, system, device and medium. BACKGROUND

[0002] The power distribution network is a key link in the power system for distributing power from a high-voltage transmission network to end users, mainly composed of lines, distribution transformers, switching devices and protection devices, etc. Its core function is to step down the high-voltage power output by the substation and realize flexible power supply relying on the mesh or radial structure, covering different load demands in cities, rural areas and industrial areas, etc. It is the last kilometer of power supply.

[0003] With the large-scale access of distributed energy on the user side to the power distribution network, the power distribution network gradually changes to the operation mode of "source-network-load-storage" coordination and flexible interaction, and its structure, operation and control are increasingly complex. At the same time, the power grid company has basically completed the intensive reform, and the dispatching business of the power distribution network has grown rapidly. There are problems such as non-standard management and insufficient technical support in the dispatching of the power distribution network, and the risk of misdispatching and misoperation is huge. At present, there is still a lack of means to analyze the dispatching behavior of the power distribution network, and it is difficult to effectively evaluate the work effectiveness and risk of the relevant staff. SUMMARY

[0004] In view of the above existing problems, the present application is proposed.

[0005] Therefore, the present application provides a power distribution network dispatching behavior analysis method, system, device and medium to solve the problem of lack of analysis and control means for dispatcher behavior in the existing power distribution network dispatching work.

[0006] To solve the above technical problems, the present application provides the following technical solutions:

[0007] In a first aspect, the present application provides a power distribution network dispatching behavior analysis method, comprising the following steps:

[0008] Setting the dispatching rules corresponding to the dispatching behavior;

[0009] Analyzing the dispatching behavior through the dispatching operation ticket to obtain the operation instructions and target devices, and judging whether the operation instructions violate the corresponding dispatching rules;

[0010] Recording the dispatching behavior that violates the dispatching rules, and obtaining the target devices and the state data of the target devices and the feeder where the target devices are located for the dispatching behavior that does not violate the dispatching rules;

[0011] Data preprocessing is performed on the state data to establish an effective data set;

[0012] By combining the effective dataset with the recorded scheduling behavior, a database of normal and abnormal behaviors for distribution network scheduling is established.

[0013] As a preferred embodiment of the power distribution network dispatching behavior analysis method of the present invention, the step of parsing dispatching behavior through dispatching operation tickets includes:

[0014] When a certain operation ticket is obtained, the operation instructions of each pair of operation tickets are parsed to determine the target device;

[0015] Analyze whether each operation instruction violates the scheduling rules, filter out scheduling behaviors that violate the scheduling rules, and then record the filtered scheduling behaviors.

[0016] The beneficial effects of this preferred technical solution are as follows: it provides a specific implementation method for scheduling behavior analysis, ensures the accuracy of target equipment identification through operation ticket analysis, realizes automatic screening and recording of violations, and improves the efficiency and accuracy of scheduling supervision.

[0017] As a preferred embodiment of the power distribution network dispatching behavior analysis method of the present invention, the step of preprocessing the state data includes:

[0018] The state data is re-transformed, and any missing state data is added.

[0019] The supplemented status data is associated with the corresponding scheduling rules.

[0020] As a preferred embodiment of the power distribution network dispatching behavior analysis method described in this invention, the step of establishing an effective dataset includes:

[0021] The state data is classified using a clustering algorithm;

[0022] Several cluster centers are randomly selected, and data from the effective dataset is assigned to the nearest cluster center;

[0023] For each cluster, if the number of data in cluster k is less than the minimum number of data, then cluster k and its corresponding cluster center are cancelled, and the data in cluster k are redistributed to the nearest cluster center.

[0024] The formula is used to correct each cluster center.

[0025] The beneficial effects of this preferred technical solution are as follows: it realizes intelligent classification of state data by using a clustering algorithm, improves data processing efficiency, avoids the instability of small sample clustering by controlling the minimum number of data, and ensures the reliability of clustering results.

[0026] As a preferred embodiment of the power distribution network dispatching behavior analysis method of the present invention, the clustering algorithm further includes an iterative optimization step:

[0027] Based on the current clustering results, the position of each cluster center is recalculated;

[0028] Determine if the cluster center positions have changed; if they have changed, continue iterating; otherwise, stop iterating.

[0029] When the maximum number of iterations is reached, the iteration process stops and the final clustering result is output.

[0030] The beneficial effects of this preferred technical solution are: the introduction of an iterative optimization mechanism makes the clustering results more accurate and stable, and the avoidance of invalid iterations through convergence judgment improves the efficiency of the algorithm.

[0031] As a preferred embodiment of the power distribution network dispatching behavior analysis method of the present invention, the clustering algorithm further includes a splitting operation step:

[0032] when Then it is determined that a split should be performed, N C This represents the current number of clusters;

[0033] When N C If the number of iterations is ≥2K and the current iteration count is even, then a split is determined.

[0034] When σ max >θ S If so, then a split is determined, σ max This represents the maximum standard deviation of the sample distances from each cluster in the current clustering result;

[0035] When performing a split, the split operation is expressed mathematically as follows:

[0036] Z′1=Z i +f actor ×σ max ;

[0037] Z'2=Z i -f actor ×σ max ;

[0038] In the formula, Z1' is a cluster center after splitting, Z i f is the cluster center before splitting. actor Z'2 is the splitting parameter, and Z'2 is the other cluster center after the split.

[0039] The beneficial effects of this preferred technical solution are: the splitting operation can automatically identify the clusters that need to be subdivided, improving the precision of the clustering, and multiple judgment conditions ensure the rationality of the splitting operation.

[0040] As a preferred embodiment of the power distribution network dispatching behavior analysis method described in this invention, the clustering algorithm further includes a merging operation:

[0041] When D ij <θ C Then the i-th cluster and the j-th cluster are merged, where D ij Let be the distance between the i-th cluster and the j-th cluster;

[0042] If the number of data in cluster i is less than θ N If the cluster i is the closest to the adjacent cluster center in terms of radial distance, then the cluster i will be merged with the cluster j that is the closest to the adjacent cluster center.

[0043] The merge operation is represented as:

[0044]

[0045] Among them, Z' k N is the new cluster center after the merger. i N represents the number of samples in cluster i. j Z represents the number of samples in cluster j. i Z is the cluster center of cluster i. j Let be the cluster center of cluster j.

[0046] The beneficial effects of this preferred technical solution are: the merging operation can integrate similar clusters, avoid over-segmentation, and ensure the scientific nature of the merging operation through the dual constraints of distance threshold and minimum number of samples.

[0047] Secondly, the present invention provides a power distribution network dispatching behavior analysis system, comprising:

[0048] The semantic analysis module is used to parse scheduling procedures and scheduling operation tickets;

[0049] The rule generation and verification module is used to establish the scheduling rules that each scheduling instruction must follow and to verify whether each instruction on the operation ticket conforms to the rules.

[0050] The behavior analysis module is used to process and analyze the dispatching behavior of the power distribution network;

[0051] The status recording module is used to record information about dispatch operation tickets and various operational status data of the distribution network that are violated.

[0052] Thirdly, the present invention provides an electronic device, comprising:

[0053] Memory and processor;

[0054] The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the power distribution network dispatch behavior analysis method are implemented.

[0055] Fourthly, the present invention provides a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the power distribution network dispatch behavior analysis method.

[0056] Compared with existing technologies, the beneficial effects of this invention are as follows: Based on the distribution network dispatching regulations, preliminary dispatching operation tickets and dispatching instruction rules are formed. By parsing the dispatching operation tickets, a preliminary verification (pre-verification) is first performed on each instruction to identify basic errors in distribution network dispatching behavior. For dispatching behaviors that pass the verification, this invention records and analyzes the distribution network operation status after executing the relevant instructions, thereby identifying abnormal dispatching behaviors and achieving a secondary verification (post-verification) of distribution network dispatching behavior. This enables full-process analysis of distribution network dispatching behavior, facilitating the dispatching agency's assessment, analysis, and training of dispatching personnel. Attached Figure Description

[0057] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0058] Figure 1 This is a schematic diagram of the overall process of the power distribution network dispatching behavior analysis method according to an embodiment of the present invention.

[0059] Figure 2 This is a structural diagram of a power distribution network dispatching behavior analysis method according to an embodiment of the present invention. Detailed Implementation

[0060] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0061] Example 1, referring to Figure 1 As an embodiment of the present invention, a method for analyzing the dispatching behavior of a power distribution network is provided, comprising the following steps S1 to S5:

[0062] S1. Set the scheduling rules corresponding to the scheduling behavior;

[0063] S2. Parse the scheduling behavior through the scheduling operation ticket to obtain various operation instructions and target devices, and determine whether each operation instruction violates the corresponding scheduling rules;

[0064] S3. Record scheduling behaviors that violate scheduling rules, and for scheduling behaviors that do not violate scheduling rules, obtain the status data of the target device and the feeder where the target device is located;

[0065] S4. Perform data preprocessing on the state data to establish an effective dataset;

[0066] S5. Combine the effective dataset with the recorded scheduling behavior to establish a database of normal and abnormal behaviors for power distribution network scheduling.

[0067] It should be noted that distribution network dispatching is a crucial aspect of power system operation, requiring dispatchers to execute various operational instructions based on the real-time status of the power grid and dispatching rules. In actual dispatching processes, due to the complexity of the power grid structure and the variability of operating conditions, dispatchers may face issues such as information asymmetry and tight decision-making timelines, making them prone to violations or inappropriate dispatching behaviors. Traditional dispatching supervision relies primarily on manual review and post-event analysis, which suffers from poor real-time performance and low efficiency. Furthermore, the lack of effective behavioral analysis mechanisms makes it difficult to quantitatively assess and provide early warnings of dispatcher actions, hindering the timely detection and correction of potential risks. Moreover, with the continuous expansion of the distribution network and its increasing intelligence, the complexity and frequency of dispatching operations are increasing, placing higher demands on the accuracy and real-time performance of dispatching behavior analysis.

[0068] Therefore, to address the aforementioned issues in dispatch supervision and behavior analysis, a comprehensive dispatch rule system and behavior analysis mechanism are established through steps S1-S5 to achieve intelligent analysis of dispatch operation tickets and automatic identification of violations; through the preprocessing of status data and the establishment of effective datasets, a reliable data foundation is provided for dispatch behavior analysis; based on the construction of a database of normal and abnormal behaviors, accurate classification and intelligent early warning of dispatch behaviors are achieved, thereby improving the safety and reliability of distribution network dispatch.

[0069] Example 2, refer to Figures 1-2 As an embodiment of the present invention, a method for analyzing the dispatching behavior of a power distribution network is provided based on the above embodiment.

[0070] In this embodiment of the application, step S2, the step of parsing the scheduling behavior through the scheduling operation ticket, includes A1 to A2:

[0071] A1. When an operation ticket is obtained, the operation instructions of each operation ticket are parsed to determine the target device.

[0072] Specifically, the parsing process of the operation ticket includes: parsing the text content of the operation ticket using natural language processing technology to extract key information such as operation type, equipment name, and operation time; establishing the correspondence between operation instructions and target equipment using an equipment coding mapping table to ensure the accuracy of equipment identification; and standardizing the ambiguous expressions in the operation ticket to unify the format and expression of operation instructions.

[0073] A2. Analyze whether each operation instruction violates the scheduling rules, filter out scheduling behaviors that violate the scheduling rules, and then record the filtered scheduling behaviors.

[0074] Specifically, violation judgment of scheduling rules can be achieved in the following ways: establish a scheduling rule base, which includes multi-dimensional rules such as safe operating procedures, equipment constraints, and timing requirements; match the parsed operation instructions with the rule base one by one, and make automatic judgments through the rule engine; classify and process violations, and mark and record serious violations, general violations, and potential risk operations respectively.

[0075] It should be noted that a multi-verification mechanism is used in the operation ticket parsing process. First, precise matching is performed through equipment codes. Second, correlation verification is performed by combining equipment location and electrical connection relationship to ensure the accuracy of target equipment identification. In the rule violation judgment stage, a hierarchical rule system is established, including mandatory safety rules, operation sequence rules and equipment status constraint rules. Rule conflicts are avoided by prioritizing rules, thereby improving the reliability of violation judgment.

[0076] In an optional implementation, the operation ticket parsing in step S2 can also be combined with historical operation data for intelligent analysis, and operation patterns and abnormal behaviors can be identified through machine learning algorithms to improve the accuracy and intelligence of the parsing.

[0077] Specifically, intelligent analysis of historical operation data includes: establishing a historical operation database to collect and store past operation ticket information, operation results, equipment responses, and other data to form a complete operation record chain; using time series analysis methods to mine historical operation data and identify operation patterns and frequency characteristics for different time periods and equipment types; using clustering algorithms to classify historical operation behaviors and establish a typical operation mode library, including routine operation modes, emergency operation modes, and maintenance operation modes; and using anomaly detection algorithms (such as isolated forests and local anomaly factors) to identify abnormal behaviors that deviate from historical operation modes and set corresponding anomaly thresholds.

[0078] In another optional implementation, the scheduling rule violation judgment in step S2 can also adopt fuzzy logic reasoning method to intelligently evaluate boundary conditions and complex operation scenarios, thereby reducing the occurrence of misjudgment and omission.

[0079] In this embodiment of the application, step S3, the step of preprocessing the state data, includes B1 to B2:

[0080] B1. Transform the state data and fill in any missing state data.

[0081] Specifically, the process of re-transforming state data and supplementing missing data includes: using standardization transformation to convert different types of state data into a unified numerical format to eliminate the influence of differences in dimensions and inconsistent numerical ranges; supplementing short-term missing data through time series interpolation algorithms, using linear interpolation or spline interpolation based on state values ​​at previous and subsequent times; and estimating long-term missing key state data using a state mapping method based on similar equipment, and reconstructing the data by combining equipment operating patterns and historical statistical characteristics.

[0082] B2. Associate the supplemented status data with the corresponding scheduling rules.

[0083] Specifically, the association between status data and scheduling rules can be established in the following ways: construct a status-rule mapping matrix to establish a correspondence between each status parameter and the relevant scheduling rule; classify and label status data according to device type and operation category to ensure the accuracy of rule association; and establish a dynamic weight allocation mechanism to adjust the weight coefficient of status data in rule judgment according to its importance and reliability.

[0084] For example, state data normalization transformation can be performed using the following formula:

[0085]

[0086] In the formula, X norm X represents the standardized state value; X represents the original state value; X min X max These are the minimum and maximum values ​​of the original data; U min U max These represent the minimum and maximum values ​​within the target numerical range.

[0087] It should be noted that a multi-layered data quality inspection mechanism was adopted in the data preprocessing process. First, outlier data was identified and processed through an outlier detection algorithm. Second, data consistency verification was used to ensure the compatibility of data from different sources. Finally, integrity assessment was used to ensure the availability of key status parameters. In the process of associating status data with scheduling rules, a rule applicability assessment system was established to dynamically adjust the scope of application of the rules according to the current status and operating conditions of the equipment, so as to avoid judgment errors caused by misuse of rules.

[0088] In an optional implementation, the state data preprocessing in step S3 can also employ deep learning methods to improve data quality. By constructing a data completion model based on a Long Short-Term Memory (LSTM) network, intelligent prediction is performed using the temporal characteristics of historical state data and equipment operating modes, enabling high-precision reconstruction of missing data. This method can capture the nonlinear changes and complex time dependencies of equipment states, offering better accuracy and adaptability compared to traditional interpolation methods, and is particularly suitable for the dynamic characteristic analysis of power equipment state data.

[0089] In another optional implementation, the data preprocessing in step S3 can also be combined with multi-source data fusion technology to collaboratively process state data from different monitoring systems. By establishing a multi-dimensional descriptive model of equipment state, the integrity and reliability of the data can be improved. Specifically, this includes: using a Kalman filter algorithm to fuse multi-sensor data to eliminate measurement noise and system errors; using a Bayesian network to establish probabilistic dependencies between equipment state parameters and supplementing missing correlation data through inference mechanisms; and introducing an expert knowledge base to verify and correct the fusion results, ensuring that the data preprocessing results conform to the physical constraints and engineering realities of power system operation.

[0090] In a third optional implementation, the association between state data and scheduling rules in step S3 can also employ semantic mapping technology. This involves constructing an ontology knowledge graph of distribution network equipment and operations to establish semantic associations between state data and scheduling rules. This method includes: defining a semantic concept system for equipment states, operational behaviors, and safety rules; establishing a standardized knowledge representation framework; extracting key semantic information from scheduling rules using natural language processing technology and automatically matching it with the semantic tags of state data; and using an inference engine to determine the applicability of rules in complex scenarios, supporting rule association analysis with multiple constraints and logical combinations, thereby improving the intelligence level and processing complexity of scheduling behavior analysis.

[0091] In the embodiments of this application, step S4, the step of establishing a valid dataset, includes C1 to C4:

[0092] C1. Classify the state data using a clustering algorithm;

[0093] Specifically, the clustering algorithm uses an improved K-means algorithm to classify state data, and determines the affiliation relationship by calculating the Euclidean distance from the data point to the cluster center; it introduces a density weight factor to give higher weight to the cluster centers of dense data areas, thereby improving the stability of clustering; and it dynamically determines the optimal number of clusters by combining the elbow rule and the silhouette coefficient to avoid classification bias caused by manually setting the number of clusters.

[0094] The clustering algorithm also includes iterative optimization steps C1.1 to C1.4:

[0095] C1.1 Based on the current clustering results, recalculate the position of each cluster center;

[0096] C1.2 Determine if the cluster center position has changed. If it has changed, continue iterating; if it has not changed, stop iterating.

[0097] C1.3 When the maximum number of iterations is reached, stop the iteration process and output the final clustering result.

[0098] C2. Randomly select several cluster centers and assign the data in the effective dataset to the nearest cluster center;

[0099] C3. Determine the clusters. If the number of data in cluster k is less than the minimum number of data, cancel cluster k and its corresponding cluster center, and redistribute the data in cluster k to the nearest cluster center.

[0100] C4. Use formulas to correct each cluster center.

[0101] Non-violation operations are classified using clustering algorithms. Taking the ISODATA algorithm as an example, the implementation steps are as follows:

[0102] Randomly select several cluster centers z i Let K(x,z) be the center of the i-th cluster. Assign data from the valid dataset to the nearest cluster center. The distance is calculated using the formula K(x,z). i )=exp(-γ||xz i || 2 ), K(x,z) i Let x be the data from the i-th cluster center z. i Radial basis kernel function distance; ||xz i || 2 For data X, the cluster center z is the cluster center. i The Euclidean distance; γ is the width control parameter of the kernel function;

[0103] For each cluster obtained, determine: if the number of data in cluster k is less than the minimum number of data θN If the cluster k and its corresponding cluster center are not selected, the data in cluster k will be redistributed to the nearest cluster center.

[0104] Using formulas The cluster centers are modified, where z′ j N represents the corrected j-th cluster center; j S represents the total number of data points in the j-th cluster center; j Let be the data set in the j-th cluster center.

[0105] In an alternative implementation, the recalculation of cluster center locations in step C1.1 can also employ a weighted average method, including the following steps:

[0106] The weight coefficient is calculated based on the distance of each data point to the current cluster center, with closer points receiving larger weights. A Gaussian kernel function or exponential decay function is used to adjust the weights to ensure reasonable weight allocation. The cluster center coordinates are recalculated using a weighted average formula: C new =∑(w i ×x i ) / ∑(w i ), where w i Let x be the weight of the i-th data point. i Let be the coordinates of the i-th data point; for outliers or noise points, filter them by setting a weight threshold to avoid their adverse effects on the calculation of cluster centers; introduce a time decay factor in the weight calculation process so that newer scheduling data has higher weights, thereby improving the adaptability of the clustering results to the current scheduling mode.

[0107] In another optional implementation, the determination of changes in cluster center location in step C1.2 can also employ a multi-indicator comprehensive judgment mechanism, including the following steps:

[0108] A threshold ε1 is set for the Euclidean distance of cluster center position changes. When the displacement distance of all cluster centers is less than ε1, the position is considered not to have changed significantly. The cluster internal density index is calculated. When the improvement of the internal density of clusters is less than the set threshold ε2 in several consecutive iterations, convergence is considered. A cluster stability assessment is introduced. By calculating the change ratio of the cluster affiliation of data points in two adjacent iterations, the clustering result is considered stable when the change ratio is less than the threshold ε3. A comprehensive judgment strategy is adopted. Iteration stops only when the three indicators of position change, density improvement and stability assessment simultaneously meet the convergence conditions. A dynamic threshold adjustment mechanism is established to adaptively adjust the various judgment thresholds according to the size and complexity of the dataset, thereby improving the accuracy and adaptability of convergence judgment.

[0109] Based on the current clustering results, a splitting operation is performed, with the following splitting rules:

[0110] like Then it is determined that a split will occur; N C This represents the current total number of clusters;

[0111] If N C If the number of iterations is ≥2K and the current iteration count is even, then a split is determined.

[0112] If σ max >θ S If σ max This represents the maximum distance from the standard deviation of samples in each cluster in the current clustering results;

[0113] The splitting operation is performed using the following formula:

[0114] Z′1=Z i +f actor ×σ max ;

[0115] Z'2=Z i -f actor ×σ max ;

[0116] In the formula, Z1' is a cluster center after splitting; Z i The cluster centers before splitting; f actor Z' is the splitting parameter; Z'2 is the other cluster center after the split.

[0117] Based on the current clustering results, a merging operation is performed, and the merging rules are as follows:

[0118] If D ij <θ C Then, the i-th cluster and the j-th cluster are merged; where D ij Let be the distance between the i-th cluster and the j-th cluster;

[0119] If the number of data in cluster i is less than θ N If the cluster i is the closest to the adjacent cluster center radial basis distance, then the cluster i will be merged with the cluster j.

[0120] The following formula is used for combining operations:

[0121]

[0122] In the formula Z' k The new cluster centers after the merger; N i N is the number of samples in cluster i; j Z represents the number of samples in cluster j; i Z is the cluster center of cluster i; jLet be the cluster center of cluster j.

[0123] Clustering stops when the set total number of iterations I is reached; otherwise, the next round of iterations begins.

[0124] Clustering algorithms can be used to classify scheduling behavior. For each category, methods such as isolated forests can be used for anomaly detection, thereby enabling the analysis and summarization of power distribution network scheduling behavior.

[0125] In an optional implementation, the clustering algorithm in step S4 can also employ a hybrid strategy combining hierarchical clustering and the K-means algorithm. First, agglomerated hierarchical clustering is used to determine the approximate location of the initial cluster centers, avoiding local optima caused by random initialization. Then, the K-means algorithm is used for fine-tuning. This method includes: calculating the distance matrix between all data points; performing hierarchical clustering using the Ward connectivity criterion; determining a reasonable number of clusters through dendrogram analysis; using the hierarchical clustering results as the initial centers for the K-means algorithm; further optimizing the starting positions using an improved K-means++ initialization strategy; and introducing clustering quality evaluation metrics, such as the Davies-Bouldin index and the Calinski-Harabasz index, to quantitatively compare different clustering schemes and select the optimal clustering configuration.

[0126] In another optional implementation, the effective dataset establishment in step S4 can also employ the density-based clustering algorithm DBSCAN, which is particularly suitable for processing outliers and noise data in distribution network status data. It adaptively adjusts the clustering based on the distribution characteristics of distribution network equipment status data; identifies core points, boundary points, and noise points through density reachability analysis, automatically determining the shape and number of clusters; establishes an outlier filtering mechanism to process identified noise points separately, avoiding their interference with normal scheduling behavior analysis; and semantically annotates the clustering results using domain knowledge, mapping different clusters to specific equipment status types and operating scenarios, improving the practicality and interpretability of the dataset.

[0127] In a third optional implementation, the clustering algorithm in step S4 can also employ a graph-based spectral clustering method, which discovers complex nonlinear clustering structures by constructing a similarity graph of the state data. This method includes: calculating the similarity matrix between state data points, using a Gaussian kernel function or cosine similarity to measure the correlation between data; constructing a Laplace matrix and performing eigenvalue decomposition to map high-dimensional state data to a low-dimensional feature space; applying a traditional clustering algorithm in the feature space to complete the final classification, reconstructing the clustering results of the original data space through a linear combination of feature vectors; and introducing multi-scale analysis techniques to perform clustering analysis at different resolutions, capturing the pattern characteristics of distribution network state data at different time scales, and providing data support for multi-level analysis of dispatching behavior.

[0128] In summary, based on the distribution network dispatching regulations, preliminary dispatching operation tickets and dispatching instruction rules were formed. By parsing the dispatching operation tickets, a preliminary verification (pre-verification) of each instruction was first performed to identify basic errors in distribution network dispatching behavior. For dispatching behaviors that passed the verification, this invention recorded and analyzed the distribution network operation status after executing the relevant instructions, thereby identifying abnormal dispatching behaviors and realizing a secondary verification (post-verification) of distribution network dispatching behavior. This enabled full-process analysis of distribution network dispatching behavior, and facilitated the dispatching agency's assessment, analysis, and training of dispatching personnel.

[0129] Example 3 illustrates a schematic scheme for a distribution network dispatching behavior analysis method. It should be noted that the technical solution of this distribution network dispatching behavior analysis system belongs to the same concept as the technical solution of the aforementioned distribution network dispatching behavior analysis method. Details not described in detail in this embodiment can be found in the description of the technical solution of the aforementioned distribution network dispatching behavior analysis method.

[0130] This embodiment also provides a power distribution network dispatching behavior analysis system, including a semantic analysis module for parsing dispatching procedures and dispatching operation tickets;

[0131] The rule generation and verification module is used to establish the scheduling rules that each scheduling instruction must follow and to verify whether each instruction on the operation ticket conforms to the rules.

[0132] The behavior analysis module is used to process and analyze the dispatching behavior of the power distribution network;

[0133] The status recording module is used to record information about dispatch operation tickets and various operational status data of the distribution network that are violated.

[0134] This embodiment also provides an electronic device suitable for distribution network dispatch behavior analysis, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the distribution network dispatch behavior analysis method proposed in the above embodiment.

[0135] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, it implements the distribution network dispatch behavior analysis method proposed in the above embodiments.

[0136] The storage medium proposed in this embodiment and the method for analyzing distribution network dispatching behavior proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0137] Based on the above description of the implementation methods, those skilled in the art will clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0138] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for analyzing the dispatching behavior of a power distribution network, characterized in that, Includes the following steps: Define the scheduling rules corresponding to the scheduling behavior; By parsing the scheduling behavior through the scheduling operation ticket, we can obtain various operation instructions and target devices, and determine whether each operation instruction violates the corresponding scheduling rules. Record scheduling behaviors that violate scheduling rules, and for scheduling behaviors that do not violate scheduling rules, obtain the status data of the target device and the feeder where the target device is located; The state data is preprocessed to establish a valid dataset; By combining the effective dataset with the recorded scheduling behavior, a database of normal and abnormal behaviors for distribution network scheduling is established.

2. The distribution network dispatching behavior analysis method as described in claim 1, characterized in that, The steps for parsing scheduling behavior using scheduling operation tickets include: When a certain operation ticket is obtained, the operation instructions of each pair of operation tickets are parsed to determine the target device; Analyze whether each operation instruction violates the scheduling rules, filter out scheduling behaviors that violate the scheduling rules, and then record the filtered scheduling behaviors.

3. The distribution network dispatching behavior analysis method as described in claim 2, characterized in that, The steps for preprocessing the state data include: The state data is re-transformed, and any missing state data is added. The supplemented status data is associated with the corresponding scheduling rules.

4. The distribution network dispatching behavior analysis method as described in claim 3, characterized in that, The steps to build a valid dataset include: The state data is classified using a clustering algorithm; Several cluster centers are randomly selected, and data from the effective dataset is assigned to the nearest cluster center; For each cluster, if the number of data in cluster k is less than the minimum number of data, then cluster k and its corresponding cluster center are cancelled, and the data in cluster k are redistributed to the nearest cluster center. The formula is used to correct each cluster center.

5. The distribution network dispatching behavior analysis method as described in claim 4, characterized in that, The clustering algorithm also includes an iterative optimization step: Based on the current clustering results, the position of each cluster center is recalculated; Determine if the cluster center positions have changed; if they have changed, continue iterating; otherwise, stop iterating. When the maximum number of iterations is reached, the iteration process stops and the final clustering result is output.

6. The distribution network dispatching behavior analysis method as described in claim 5, characterized in that, The clustering algorithm also includes a splitting operation step: when Then it is determined that a split should be performed, N C This represents the current number of clusters; When N C If the number of iterations is ≥2K and the current iteration count is even, then a split is determined. When σ max >θ S If so, then a split is determined, σ max This represents the maximum standard deviation of the sample distances from each cluster in the current clustering result; When performing a split, the split operation is expressed mathematically as follows: Z'1=Z i +f actor ×σ max ; Z'2=Z i -f actor ×σ max ; In the formula, Z'1 is a cluster center after splitting, Z i f is the cluster center before splitting. actor Z'2 is the splitting parameter, and Z'2 is the other cluster center after the split.

7. The distribution network dispatching behavior analysis method as described in claim 6, characterized in that, The clustering algorithm also includes a merge operation: When D ij <θ C Then the i-th cluster and the j-th cluster are merged, where D ij Let be the distance between the i-th cluster and the j-th cluster; If the number of data in cluster i is less than θ N If the cluster i is the closest to the adjacent cluster center in terms of radial distance, then the cluster i will be merged with the cluster j that is the closest to the adjacent cluster center. The merge operation is represented as: Among them, Z' k N is the new cluster center after the merger. i N represents the number of samples in cluster i. j Z represents the number of samples in cluster j. i Z is the cluster center of cluster i. j Let be the cluster center of cluster j.

8. A power distribution network dispatching behavior analysis system, using the method described in any one of claims 1-7, characterized in that, include: The semantic analysis module is used to parse scheduling procedures and scheduling operation tickets; The rule generation and verification module is used to establish the scheduling rules that each scheduling instruction must follow and to verify whether each instruction on the operation ticket conforms to the rules. The behavior analysis module is used to process and analyze the dispatching behavior of the power distribution network; The status recording module is used to record information about dispatch operation tickets and various operational status data of the distribution network that are violated.

9. An electronic device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, they implement the steps of the power distribution network dispatch behavior analysis method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the power distribution network dispatch behavior analysis method according to any one of claims 1 to 7.

Images