A detection method based on the differentiated function adaptation of power distribution terminals

Abstract

This application discloses a detection method based on differentiated function adaptation of distribution terminals, relating to the field of power system automation; aiming to improve the detection efficiency and security of distribution networks. The method first constructs a differentiated detection scheme based on terminal feature weights, acquiring feature information and historical detection data of the distribution automation terminals to be detected, and filtering out mandatory detection items to form a basic detection item set. Subsequently, based on the constructed feature importance evaluation model, the influence weight of each terminal feature on the detection anomaly results is calculated, and the priority of candidate detection items is evaluated. Secondly, a terminal-related detection scheme based on state-function adaptation is generated. When an anomaly is detected in the target terminal, the comprehensive correlation degree with other terminals is calculated using a three-dimensional correlation feature library, dynamically generating a detection scheme. By executing the detection scheme and updating it in real time, comprehensive coverage of potential hazards is ensured, thereby improving the security and reliability of the distribution network.

Description

Technical Field

[0001] This application relates to the field of power system automation, and in particular to a detection method based on the differentiated functional adaptation of distribution terminals. Background Technology

[0002] With the comprehensive advancement of smart grid construction, distribution automation systems have become a core infrastructure for ensuring the reliability of power supply and improving power quality. In this system, distribution automation terminals (such as feeder terminals (FTUs), distribution transformer terminals (TTUs), and substation terminals (DTUs) are key devices for sensing the grid's operating status and executing control commands. Their installed capacity is continuously expanding with the modernization of the power grid, and their functions are becoming increasingly integrated and intelligent. To ensure the stability and safety of these terminals after they are put into operation, the power industry has established a complete testing system. Based on relevant national and industry standards, the system conducts rigorous grid entry testing and full lifecycle management of terminal equipment. Current testing work extensively covers basic performance indicators such as power adaptability and electromagnetic compatibility, as well as complex logic function verification and communication protocol testing. It is a crucial link in ensuring equipment quality and maintaining the safe and stable operation of the power grid during the digital transformation of the distribution network.

[0003] However, existing power distribution terminal testing technologies still suffer from significant drawbacks when faced with massive, heterogeneous terminal devices, including rigid testing models and incomplete hazard identification. Specifically, traditional testing methods often employ a static, "one-size-fits-all" approach, ignoring the individual differences in historical operating data and characteristic attributes among different terminals. Regardless of the terminal's criticality or historical failure rate, the same testing process is applied, resulting in insufficient resources allocated to high-risk projects while consuming significant time for low-risk projects, leading to low testing efficiency and a lack of targeted approach. More seriously, existing testing systems generally suffer from the limitation of "isolated handling of single-point problems." When a fault is detected in a terminal, repairs or retests are often only performed on that device itself, without extending the analysis to other terminals with adjacent topologies, similar operating conditions, or strong functional logic coupling, nor conducting collaborative analysis on these terminals. This isolated testing approach severs the systemic connections between power distribution network devices, easily overlooking cascading hazards caused by functional interactions or state transmission, making it difficult to achieve system-level hazard mitigation. Summary of the Invention

[0004] This invention addresses the technical problems existing in the background art by proposing a detection method based on differentiated functional adaptation of distribution terminals. First, a dynamic generation mechanism for differentiated detection schemes based on feature weights is constructed. Standard specifications are embedded in the rule engine layer to screen mandatory inspection items. Simultaneously, historical data is mined using the random forest algorithm at the machine learning layer to train the model and output the priority of inspection items. Core feature items are configured for repeated detection, while auxiliary feature items are sampled, thereby achieving precise resource allocation. Second, to address the limitations of single-point detection, a "three-dimensional associated feature library" is constructed, including topology, operating status, and functional adaptation dimensions. After a source terminal fault is detected, its comprehensive correlation with other terminals in the system is automatically calculated, and functional modules with strong dependencies (such as FTU fault detection and DTU data transmission) are located, generating associated detection schemes. Finally, during execution, online real-time iteration is performed based on newly discovered problems, dynamically updating the detection queue. This invention not only significantly improves the detection efficiency of distribution terminals but also effectively eliminates systemic hidden dangers through the "discover one point, investigate a whole area" correlation mechanism, significantly improving the overall operational safety of the distribution network.

[0005] To solve the technical problem, the technical solution of the present invention is as follows:

[0006] A detection method based on the differentiated function adaptation of power distribution terminals, the method comprising:

[0007] S1. The dynamic generation steps of the differential detection scheme based on terminal feature weights include:

[0008] Obtain the set of distribution automation terminals to be tested, and collect the terminal feature information and historical test data corresponding to each distribution automation terminal;

[0009] The set of power distribution automation terminals is filtered based on preset detection rules to determine the mandatory detection items to be performed by each power distribution automation terminal, thus forming a basic set of detection items.

[0010] Based on the historical detection data, a feature importance evaluation model is constructed to calculate the influence weight of each terminal feature on the detection anomaly results;

[0011] Based on the terminal feature information and the influence weight, priority evaluation is performed on the candidate detection items other than the basic detection item set, and a differentiated detection scheme for each power distribution automation terminal is generated based on the priority evaluation results.

[0012] S2. Steps for generating a terminal-related detection scheme based on state-function adaptation, including:

[0013] During the execution of the differentiated detection scheme, when an anomaly is detected in the target distribution automation terminal, the correlation feature information between the target distribution automation terminal and other distribution automation terminals is obtained.

[0014] Based on the associated feature information, the comprehensive correlation degree between the target power distribution automation terminal and other power distribution automation terminals is calculated, and a set of target terminals associated with the target power distribution automation terminal is selected according to the comprehensive correlation degree.

[0015] Based on the abnormal function modules of the target power distribution automation terminal and the function modules of each terminal in the target terminal set, a list of associated function modules that need to be linked for detection is determined, and a terminal linked detection scheme is generated.

[0016] The terminal-related detection scheme is executed, and the detection scheme to be executed is dynamically updated based on the newly detected abnormal results during the related detection process, so as to realize the online iteration of the detection scheme.

[0017] The associated feature information includes physical topology associated features, operating status associated features, and functional logic associated features.

[0018] The power distribution terminal mentioned in this application refers to power distribution automation terminal equipment used in power distribution automation systems, including feeder terminal unit (FTU), power distribution transformer terminal unit (TTU), station terminal unit (DTU), etc.

[0019] This application also proposes a detection system based on the differentiated function adaptation of power distribution terminals, wherein the system is applied to any of the methods described above, and the system includes:

[0020] A dynamic generation module for differential detection schemes based on terminal feature weights, the module comprising:

[0021] Used to acquire a set of distribution automation terminals to be detected, and to collect terminal feature information and historical detection data corresponding to each distribution automation terminal;

[0022] This is used to filter the set of power distribution automation terminals based on preset detection rules, determine the mandatory detection items to be performed by each power distribution automation terminal, and form a basic detection item set.

[0023] This is used to construct a feature importance evaluation model based on the historical detection data, and to calculate the influence weight of each terminal feature on the detection anomaly results;

[0024] It is used to prioritize candidate detection items other than the basic detection item set based on the terminal feature information and the influence weight, and generate differentiated detection schemes for each power distribution automation terminal based on the priority evaluation results.

[0025] A terminal-linked detection scheme generation module based on state-function adaptation is provided, the module comprising:

[0026] During the execution of the differentiated detection scheme, when an anomaly is detected in the target distribution automation terminal, the correlation feature information between the target distribution automation terminal and other distribution automation terminals is obtained.

[0027] Based on the associated feature information, the system calculates the comprehensive correlation degree between the target power distribution automation terminal and other power distribution automation terminals, and filters out the set of target terminals associated with the target power distribution automation terminal according to the comprehensive correlation degree.

[0028] Based on the abnormal function modules of the target power distribution automation terminal and the function modules of each terminal in the target terminal set, determine the list of associated function modules that need to be linked for detection, and generate a terminal linked detection scheme.

[0029] This is used to execute the terminal-related detection scheme and dynamically update the detection scheme to be executed based on newly detected abnormal results during the related detection process, thereby realizing online iteration of the detection scheme.

[0030] This application has the following advantages:

[0031] To address the shortcomings of existing scheduling models that rely on expert experience, leading to insufficient scenario coverage and limited optimization for unknown operating conditions, this application proposes an exploration model training method based on situation confidence. First, a scheduling mechanism model that distinguishes between long-term safety constraints and short-term hard constraints is constructed based on expert experience. A deep reinforcement learning model with exploration neurons in the input layer is then established, ensuring output safety through mechanism verification during the early shaping phase. Second, training is conducted in a simulation environment, using clustering algorithms to classify historical operating conditions. The exploration coefficients are adaptively adjusted based on the similarity between the current condition and typical scenarios, as well as historical performance. Finally, during exploration strategy verification, only short-term hard constraints are locked, ignoring long-term constraints to expand the search space, and model parameters are updated through simulation feedback. This method effectively overcomes the shortcomings of traditional expert experience models, such as excessively large safety domains and insufficient scenario coverage. By using situation confidence to guide the model to conduct targeted and efficient exploration within the safety boundary, it significantly improves the proactive adaptability and optimization efficiency of distribution network scheduling models in the face of unknown and complex operating conditions.

[0032] To address the issues of poor economic efficiency and lack of evolutionary capability caused by the conservative nature of traditional mechanistic models in their safety domains, this application proposes a distribution network scheduling optimization method based on exploratory decision-making and cross-questioning of the mechanistic model. First, the same grid condition is simultaneously input into both the basic model and the exploratory model, and the strategy deviation is calculated. If the deviation exceeds a threshold, a challenge score is calculated based on condition similarity, historical challenge success rate, and hard constraint safety distance to determine whether to initiate a challenge. Second, after triggering the challenge, both strategies are substituted into a high-precision simulation environment for full-cycle verification. If the exploratory strategy outperforms the basic strategy in terms of economic efficiency while satisfying all long-term and short-term constraints, the challenge is considered successful. Finally, aggregate weights are calculated based on the performance improvement and hard constraint safety margin, and the basic model is updated with weighted averages. This application, through a closed-loop iterative mechanism, can break through the conservatism of traditional mechanistic models while strictly ensuring system operational safety, achieving continuous evolution and self-improvement of the scheduling strategy, and effectively balancing the safety and economy of distribution network scheduling. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0034] Figure 1 This is a main technical roadmap for a detection method based on the differentiated function adaptation of power distribution terminals, provided in the embodiments of this application. Detailed Implementation

[0035] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0036] Example 1:

[0037] This application proposes a detection method based on differentiated function adaptation of distribution terminals. This method solves the problems of inefficiency caused by the "one-size-fits-all" approach and missed potential hazards caused by "single-point detection" in traditional distribution terminal detection by constructing a dynamic generation mechanism for differentiated detection schemes based on feature weights and a terminal-related detection scheme generation mechanism based on state-function adaptation. The specific process is as follows: Figure 1 As shown.

[0038] S1: Dynamic generation of differential detection schemes based on feature weights;

[0039] To address the issues of neglecting individual differences in terminals and uneven allocation of testing resources in existing testing processes, this application combines a rule engine with machine learning to achieve dynamic priority ranking and scheme generation for testing projects.

[0040] S1.1: Rule engine layer embedding and basic rule base construction;

[0041] Establish a basic testing rule base for distribution terminals, transforming the "Distribution Automation Terminal Testing Procedure" and related enterprise standards into computer-executable logical rules. Define the set of distribution terminals to be tested as follows: , Represents the set of power distribution terminals to be tested The first in One power distribution terminal.

[0042] The rules engine filters the endpoints and outputs a set of mandatory checks that must be performed. This set consists of all mandatory test items (such as the power adaptability test that all terminals must perform), denoted as... , Representative of the set of mandatory testing items The first in There are [number] testing items. Among them... The final detection plan will be generated later. It is called as the basic set.

[0043] S1.2: Calculation of feature weights for the machine learning layer;

[0044] A feature importance evaluation model is constructed using the random forest algorithm. Based on historical detection data, it mines the implicit correlation between features and detection results, thereby outputting feature weights used to determine the priority of detection items. First, a historical detection dataset is constructed. As a training set, this dataset consists of a large number of "terminal features-detection items-detection results" triplet data.

[0045] Random forest model is The system consists of several decision trees, trained using the Bootstrap resampling technique. During the node splitting process of each tree, the contribution of features to reducing impurity is calculated, thereby quantifying the influence of features on the detection results.

[0046] First, the impurity of a node is calculated using the Gini index. Assume that in a node of the decision tree... In the middle, the sample belongs to the first The probability of class (pass / fail) is Then the node Gini index The calculation formula is as follows:

[0047] (1)

[0048] In the formula, This represents the total number of categories.

[0049] During the splitting process of a decision tree, features At the node The decrease in the Gini index brought about by the location The (Variable Importance Measure) is calculated as follows:

[0050] (2)

[0051] In the formula, Representation of features At the node Information gain after splitting; Pre-split node The total number of samples; and These represent the number of samples for the left and right child nodes after the split, respectively. and These are the Gini indices for the left and right child nodes, respectively.

[0052] The output features are derived by combining the computational results of all decision trees in the random forest. Global importance weights This weight will serve as the core basis for prioritizing subsequent testing items, and the calculation formula is as follows:

[0053] (3)

[0054] In the formula, Features Normalized global weights; This represents the total number of decision trees in the random forest. Indicates the first The set of all nodes in a decision tree; Indicates the first Nodes in a tree Features Importance rating; In the random forest, the first... In a decision tree, nodes Utilizing features The decrease in the Gini index resulting from the splitting process; This represents the total dimension of the input features.

[0055] S1.3: Dynamic detection scheme generation and configuration;

[0056] Input the feature vector of the terminal to be detected, and first use the rule engine to filter out the mandatory detection items (such as the power adaptability detection described in S1.1) to form the basic detection set. Subsequently, based on the feature weights calculated in Section S1.2, the remaining non-mandatory inspection items are prioritized. The set of strongly relevant feature indices for a given inspection item is defined as follows: The priority score for each candidate detection item is calculated by combining feature weights. The calculation formula is as follows:

[0057] (4)

[0058] In the formula, To determine the priority score for each item in the detection project; These are the model weight coefficients; Features The global importance weight; For a feature to exist, there is an indicator function. This indicates that the terminal to be tested possesses certain characteristics. If the terminal has characteristics , The value is 1 if it is 1, otherwise it is 0. This represents the number of anomaly detections for this project in historical data. This represents the total number of historical tests conducted for the project.

[0059] Score based on priority Based on the ranking of features, non-mandatory inspection items are dynamically divided into core feature items and auxiliary feature items, and different detection strategies are configured for each: core feature items are prioritized and subjected to 3 repeated tests, while auxiliary feature items are prioritized and subjected to 10% random sampling tests. The final generated set of differentiated detection schemes... The definition is as follows:

[0060] (5)

[0061] In the formula, The final list of tests to be performed; This is the complete set of candidate test items, excluding mandatory test items; and Each project and Priority score; To differentiate between core and auxiliary projects, a score threshold is set. A random function between 0 and 1, used to implement 10% sampling of auxiliary items.

[0062] This formula takes the union of three different detection items to form the final execution list. The first category of testing items is the set of mandatory testing items. These are mandatory detection items directly defined by the rule engine, such as power adaptability. Regardless of the score calculated by machine learning, this category must be detected; the second category of detection items... This refers to the candidate pool Priority scores selected from the middle Greater than or equal to the set threshold These are the projects that machine learning considers to be the riskiest and most important, and they must be fully inspected; the third category of inspection projects. This refers to the remaining items with lower scores. This represents the logical "AND", which is the score. Less than the threshold and random function Values ​​less than 0.1 (meaning only a 10% probability of being selected) are considered unimportant by the model, but they cannot be completely ignored. Instead, they are covered by low-cost sampling, which saves resources while retaining the possibility of discovering unexpected problems.

[0063] Configure different numbers of duplicate checks for different project types in the solution. The configuration rules are as follows:

[0064] (6)

[0065] In the formula, For the project The number of repeated tests; 3 represents the number of times the mandatory items are tested as specified in the standard; 3 represents three repeated tests on the core feature items to ensure accuracy; 1 represents a single test on the selected auxiliary feature items.

[0066] S2: Generation of terminal-related detection scheme based on state-function adaptation;

[0067] To address the systemic hidden dangers and defects caused by the "isolated treatment of single-point problems" in traditional detection methods, this step constructs a terminal correlation analysis model to automatically generate related detection solutions.

[0068] S2.1: Construction of Terminal Association Feature System;

[0069] A three-dimensional correlation feature library is constructed to quantify the correlation between terminals from three dimensions: physical topology, operating status, and functional logic.

[0070] First, topology correlation is calculated. Based on the distribution network topology map, the physical connection relationships between terminals are extracted and quantified using connection distance weights. The source terminal is defined as... The target terminal is The topological correlation between the two The calculation is as follows:

[0071] (7)

[0072] In the formula, Indicates source terminal With the target terminal The degree of topological association; This is the distance attenuation coefficient; Weighted by the number of electrical connection hops or physical distance between two terminals.

[0073] Secondly, calculate the correlation of operational status. The data acquisition terminal over a period of time... The internal operating parameters (including voltage, current, and load rate) are used to construct an operating parameter vector. Terminals are defined. The runtime parameter vector is The source-terminal correlation coefficient was calculated using the Pearson correlation coefficient. With the target terminal operational status correlation The formula is as follows:

[0074] (8)

[0075] In the formula, The correlation coefficient is the operating status coefficient. and Terminals and exist Normalized runtime parameter values ​​at any given time; , They are vectors and The time series mean. If the calculated correlation coefficient... If so, it is determined that the two terminals have a high correlation in their operating states, where This is the threshold for determining high correlation.

[0076] Finally, calculate the functional adaptation relevance. Based on the terminal's functional interaction logic, construct a functional dependency matrix and label the dependency levels. Define the terminal's functional module set as follows: , for The first in Functional modules, functional dependency matrix elements in It is expressed as follows:

[0077] (9)

[0078] In the formula, For functional modules right The dependence strength value; For strong dependency weights (such as FTU fault detection depending on DTU data transmission); It is a weakly dependent weight (e.g., TTU power statistics are not directly related to FTU).

[0079] Based on the functional dependency matrix, the source terminal With the target terminal Functional compatibility The calculation is as follows:

[0080] (10)

[0081] In the formula, To adapt the relevance of functions; and Terminals and A subset of functional modules; , These are specific interactive function modules.

[0082] S2.2: Problem Terminal Correlation Calculation:

[0083] When a problem is detected by a terminal (denoted as the "source terminal"), the three-dimensional correlation feature library is invoked, and a weighted summation method is used to calculate the comprehensive correlation degree between the source terminal and other terminals (denoted as the "target terminal"). The calculation of the source terminal... With any other terminal in the system Comprehensive correlation The formula is as follows:

[0084] (11)

[0085] In the formula, The overall relevance score; Indicates the fault source terminal; Indicates the candidate target terminal; The weight coefficients are respectively for the topology, state, and function dimensions, and satisfy the following conditions: ; The topological correlation degree calculated in step S2.1; The absolute value of the operating status correlation coefficient calculated in step S2.1; The functional adaptation correlation degree calculated in step S2.1.

[0086] S2.3: Location of related functions and generation of list;

[0087] The target terminal set is selected based on comprehensive correlation, and the specific related functional modules that need to be detected are further identified. Based on the fault modules of the source terminal (such as the fault detection module of the FTU), the functional dependency matrix is ​​queried to locate the directly related functional modules (such as the data transmission module of the DTU and the instruction receiving module of the superior master station); at the same time, combined with the operating status data, the target terminal functional modules that are highly correlated with the fault parameters of the source terminal (such as the sudden change in current during a fault) are selected.

[0088] The generated list of associated functions The definition is as follows:

[0089] (12)

[0090] In the formula, This represents the final list of associated functions that require joint detection. Represents a set of highly correlated terminals For each target terminal in the process, the union operation is performed, where The condition satisfied in step S2.2 The terminal configuration of the conditions, This represents the threshold for determining the overall correlation. This indicates that the terminal is closely related to the source of the fault, and it is identified as a highly correlated terminal, thus being included in [the list of terminals]. If it does not, proceed to the next step of detailed function investigation; otherwise, it means that the relevance is not enough and it is determined to be irrelevant, so ignore it directly. Indicates the target terminal The middle belongs to the functional set The specific functional modules to be determined; As a preset functional logic dependency indicator, determined based on the functional dependency matrix, if the faulty module of the source terminal affects the functional module of the target terminal... If a clear strong or weak dependency exists, then Otherwise, it is 0; Indicates the logical "OR"; The sequence of characteristic parameters representing the moment of source terminal failure With the target terminal functional module Synchronous operating parameter sequence The covariance between them is used to quantify the dynamic correlation between the two in time series data; This is a preset threshold for determining parameter correlation.

[0091] The physical meaning and function of this formula lies in locking down the scope of associated detection through a combination of logical rules and data-driven approaches. The system iterates through each highly correlated terminal and selects functional modules that meet any of the following conditions to include in the list: ① Logical correlation: Based on a preset knowledge base, it determines that there is a direct functional dependency relationship (i.e., ); ② Correlation in data dimensions: Although there is no explicit logical dependency, the operating parameters of the two show a statistical correlation above the threshold during the fault occurrence (i.e. By using the union operation, this method can effectively avoid missing implicit association faults caused by the lag in updating the expert rule base, and achieve comprehensive coverage of potentially risky functions.

[0092] S2.4: Online iteration of the terminal-related detection scheme;

[0093] When implementing the terminal-related detection scheme, online real-time iteration is performed based on newly detected problems to achieve the effect of "discovering one point and investigating a whole area" in hazard management. Let the... The queue of detection schemes in round detection is The updated solution queue Represented as:

[0094] (13)

[0095] In the formula, This is a collection of completed detection tasks; This is a set of incremental solutions generated based on newly discovered problems. This is set subtraction; This represents taking the union of sets. This formula defines the online rolling update and closed-loop iteration mechanism for the terminal-related detection scheme queue, its function being: using the current detection scheme queue... Based on this, the set of tasks that have completed detection is eliminated through set difference operations. It also dynamically incorporates incremental detection task sets generated based on newly discovered fault features using set union operations. This allows for the construction of the next round of optimized execution queues. This ensures that the detection system can adaptively adjust the detection range based on real-time investigation results.

[0096] Among them, the newly discovered faulty terminals Taking into account the strength of its association with potential terminals and its historical risk level, an incremental solution set is constructed. The specific screening formula is as follows:

[0097] (14)

[0098] In the formula, This represents the set of incremental detection schemes determined after screening. This indicates potential associated distribution terminals to be evaluated; Indicates a newly discovered faulty terminal With potential terminals The overall correlation between them is calculated using the same logic as in step S2.2; Indicates terminal The historical failure rate risk value is based on the historical operating data of the terminal and is used to quantify the health status or failure tendency of the terminal. This represents the risk gain coefficient, used to adjust the weight of historical risk factors on the final association determination result; This indicates a dynamic threshold, which is adjusted in real time based on the remaining computing power or remaining detection time of the current detection system.

[0099] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0100] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A detection method based on the differentiated functional adaptation of power distribution terminals, characterized in that, The method includes: S1. The dynamic generation steps of the differential detection scheme based on terminal feature weights include: Obtain the set of distribution automation terminals to be tested, and collect the terminal feature information and historical test data corresponding to each distribution automation terminal; The set of power distribution automation terminals is filtered based on preset detection rules to determine the mandatory detection items to be performed by each power distribution automation terminal, thus forming a basic set of detection items. Based on the historical detection data, a feature importance evaluation model is constructed to calculate the influence weight of each terminal feature on the detection anomaly results; Based on the terminal feature information and the influence weight, priority evaluation is performed on the candidate detection items other than the basic detection item set, and a differentiated detection scheme for each power distribution automation terminal is generated based on the priority evaluation results. S2. Steps for generating a terminal-related detection scheme based on state-function adaptation, including: During the execution of the differentiated detection scheme, when an anomaly is detected in the target distribution automation terminal, the correlation feature information between the target distribution automation terminal and other distribution automation terminals is obtained; Based on the associated feature information, the comprehensive correlation degree between the target power distribution automation terminal and other power distribution automation terminals is calculated, and a set of target terminals associated with the target power distribution automation terminal is selected according to the comprehensive correlation degree. Based on the abnormal function modules of the target power distribution automation terminal and the function modules of each terminal in the target terminal set, a list of associated function modules that need to be linked for detection is determined, and a terminal linked detection scheme is generated. The terminal-related detection scheme is executed, and the detection scheme to be executed is dynamically updated based on the newly detected abnormal results during the related detection process, so as to realize the online iteration of the detection scheme.

2. The detection method based on the differentiated function adaptation of power distribution terminals according to claim 1, characterized in that, The associated feature information includes physical topology associated features, operating status associated features, and functional logic associated features.

3. The detection method based on the differentiated function adaptation of power distribution terminals according to claim 1, characterized in that, The dynamic generation step of the differential detection scheme based on terminal feature weights includes: Establish a basic testing rule base for distribution terminals, transforming distribution automation terminal testing procedures and related enterprise standards into computer-executable logical rules; define the set of distribution terminals to be tested as follows: , Represents the set of power distribution terminals to be tested The first in One power distribution terminal; Based on the aforementioned logical rules, the rule engine filters the terminals and outputs a set of mandatory detection items that must be executed. The set consists of all mandatory inspection items, denoted as , Representative of the set of mandatory testing items The first in One testing item; among which The final detection plan will be generated later. It is called as the basic set.

4. The detection method based on the differentiated function adaptation of power distribution terminals according to claim 3, characterized in that, In the dynamic generation step of the differentiated detection scheme based on terminal feature weights, the calculation of the influence weight of each terminal feature on the detection anomaly result includes: A feature importance evaluation model is constructed using the random forest algorithm. Based on historical detection data, it mines the implicit correlation between features and detection results, and outputs feature weights to determine the priority of detection items. First, a historical detection dataset is constructed. As a training set; Random forest model is The system consists of several decision trees, trained using Bootstrap resampling technology. During the node splitting process of each tree, the contribution of features to reducing impurity is calculated, thereby quantifying the influence of features on the detection results. First, the Gini index is used to calculate the impurity of a node, assuming that in the node of the decision tree... In the middle, the sample belongs to the first The probability of class is Then the node Gini index The calculation formula is as follows: （1） In the formula, The total number of categories; During the splitting process of a decision tree, features At the node The decrease in the Gini index brought about by the location The calculation is as follows: （2） In the formula, Representation of features At the node Information gain after splitting; Pre-split node The total number of samples; and These represent the number of samples for the left and right child nodes after the split, respectively. and These are the Gini indices for the left and right child nodes, respectively. The output features are derived by combining the computational results of all decision trees in the random forest. Global importance weights The weight This will serve as the core basis for prioritizing subsequent testing items, and the calculation formula is as follows: （3） In the formula, Features Normalized global weights; This represents the total number of decision trees in the random forest. Indicates the first The set of all nodes in a decision tree; Indicates the first Nodes in a tree Features Importance rating; In the random forest, the first... In a decision tree, nodes Utilizing features The decrease in the Gini index resulting from the splitting process; This represents the total dimension of the input features.

5. The detection method based on the differentiated function adaptation of power distribution terminals according to claim 4, characterized in that, The dynamic generation step of the differential detection scheme based on terminal feature weights also includes: The system first uses a rule engine to filter out mandatory detection items from the input feature vector of the terminal to be detected, forming a basic detection set. Subsequently, based on the obtained feature weights, the remaining non-mandatory inspection items are prioritized, and the set of strongly correlated feature indices for a given inspection item is defined as follows: The priority score for each candidate detection item is calculated by combining feature weights. The calculation formula is as follows: （4） In the formula, To determine the priority score for each item in the detection project; These are the model weight coefficients; Features The global importance weight; For a feature to exist, there is an indicator function. This indicates that the terminal to be tested possesses certain characteristics. If the terminal has characteristics , The value is 1 if it is 1, otherwise it is 0. This represents the number of anomaly detections for this project in historical data. This refers to the total number of historical tests conducted for this project. Score based on priority Based on the ranking of non-mandatory inspection items, non-mandatory inspection items are dynamically divided into core feature items and auxiliary feature items, and different detection strategies are configured for each: core feature items are ranked first and undergo three repeated inspections, while auxiliary feature items are ranked last and undergo 10% random sampling inspections, ultimately generating a set of differentiated detection schemes. The definition is as follows: （5） In the formula, The final list of tests to be performed; This is the complete set of candidate test items, excluding mandatory test items; and Each project and Priority score; To differentiate between core and auxiliary projects, a score threshold is set. A random function between 0 and 1; Formula (5) takes the union of the three different detection items to form the final execution list. The first category of testing items is a set of mandatory testing items. Category II testing items This refers to the candidate pool Priority scores selected from the middle Greater than or equal to the set threshold Projects, Category III testing projects This refers to the remaining items with lower scores. Logical expression: AND, i.e., score Less than the threshold and random function The value is less than 0.1; Configure different number of duplicate checks for different project types. The configuration rules are as follows: （6） In the formula, For the project The number of repeated tests; 3 represents the number of times the mandatory items are tested as specified in the standard; 3 represents three repeated tests on the core feature items to ensure accuracy; 1 represents a single test on the selected auxiliary feature items.

6. The detection method based on the differentiated function adaptation of power distribution terminals according to claim 5, characterized in that, The terminal-linked detection scheme generation step includes: A three-dimensional correlation feature library is constructed to quantify the correlation between terminals from three dimensions: physical topology, operating status, and functional logic. First, topological correlation is calculated; based on the distribution network topology map, the physical connection relationships between terminals are extracted and quantified using connection distance weights; the source terminal is defined as... The target terminal is The topological correlation between the two The calculation is as follows: （7） In the formula, Indicates source terminal With the target terminal The degree of topological association; This is the distance attenuation coefficient; Weighted by the number of electrical connection hops or physical distance between two terminals; Secondly, calculate the correlation of operating status; the data acquisition terminal over a period of time The runtime parameters within are used to construct a runtime parameter vector; the terminal is defined. The runtime parameter vector is ; Calculate the source and terminal using the Pearson correlation coefficient With the target terminal operational status correlation The formula is as follows: （8） In the formula, The correlation coefficient is the operating status coefficient. and Terminals and exist Normalized runtime parameter values ​​at any given time; , They are vectors and The time series mean; if the calculated correlation coefficient If so, it is determined that the two terminals have a high correlation in their operating states, where The threshold for determining high correlation; Finally, calculate the functional adaptation relevance; construct a functional dependency matrix based on the terminal's functional interaction logic and label the dependency levels; define the terminal's functional module set as follows. , for The first in Functional modules, functional dependency matrix elements in It is expressed as follows: （9） In the formula, For functional modules right The dependence strength value; Strongly dependent weights; Weakly dependent weights; Based on the functional dependency matrix, the source terminal With the target terminal Functional compatibility The calculation is as follows: （10） In the formula, To adapt the relevance of functions; and Terminals and A subset of functional modules; , These are specific interactive function modules.

7. The detection method based on the differentiated function adaptation of power distribution terminals according to claim 6, characterized in that, When a problem is detected in a certain terminal, i.e., the source terminal, the three-dimensional correlation feature library is invoked, and a weighted summation method is used to calculate the comprehensive correlation degree between other terminals, i.e., the target terminal, and the source terminal; the source terminal is then calculated. With any other terminal in the system Comprehensive correlation The formula is as follows: （11） In the formula, The overall relevance score; Indicates the fault source terminal; Indicates the candidate target terminal; The weight coefficients are respectively for the topology, state, and function dimensions, and satisfy the following conditions: ; The calculated topological affinity; This represents the absolute value of the calculated operational state correlation coefficient. To adapt the correlation degree to the calculation function.

8. The detection method based on the differentiated function adaptation of power distribution terminals according to claim 7, characterized in that, Based on the target terminal set filtered by the comprehensive correlation degree, the specific related functional modules that need to be detected are located; based on the fault modules of the source terminal, the functional dependency matrix is ​​queried to locate the directly related functional modules; at the same time, combined with the running status data, target terminal functional modules that are highly correlated with the fault parameters of the source terminal are filtered. The generated list of associated functions The definition is as follows: （12） In the formula, This represents the final list of associated functions that require joint detection. Represents a set of highly correlated terminals For each target terminal in the process, the union operation is performed, where By satisfying The terminal configuration of the conditions, This represents the threshold for determining the overall correlation. This indicates that the terminal is closely related to the source of the fault, and it is identified as a highly correlated terminal, thus being included in [the list of terminals]. If it does not, proceed to the next step of detailed function investigation; otherwise, it means that the relevance is not enough and it is determined to be irrelevant, so ignore it directly. Indicates the target terminal The middle belongs to the functional set The specific functional modules to be determined; As a preset functional logic dependency indicator, determined based on the functional dependency matrix, if the faulty module of the source terminal affects the functional module of the target terminal... If a clear strong or weak dependency exists, then Otherwise, it is 0; Logical expression: OR; The sequence of characteristic parameters representing the moment of source terminal failure With the target terminal functional module Synchronous operating parameter sequence The covariance between them is used to quantify the dynamic correlation between the two in time series data; This is a preset threshold for determining parameter correlation.

9. The detection method based on the differentiated function adaptation of power distribution terminals according to claim 8, characterized in that, The implementation of the terminal-related detection scheme specifically includes: When implementing the terminal-related detection scheme, online real-time iteration is performed based on newly detected issues; let the first... The queue of detection schemes in round detection is The updated solution queue Represented as: （13） In the formula, This is a collection of completed detection tasks; A set of incremental solutions generated based on newly discovered problems; This is set subtraction; This represents taking the union of sets; the formula defines the online rolling update and closed-loop iteration mechanism for the terminal-related detection scheme queue; Among them, the newly discovered faulty terminals Taking all factors into consideration Based on the strength of association with potential terminals and historical risk levels, construct an incremental solution set. The specific screening formula is as follows: （14） In the formula, This represents the set of incremental detection schemes determined after screening. This indicates potential associated distribution terminals to be evaluated; Indicates a newly discovered faulty terminal With potential terminals The overall correlation between them; Indicates terminal The historical failure rate risk value is obtained based on the historical operating data of the terminal and is used to quantify the health status or failure tendency of the terminal. This represents the risk gain coefficient, used to adjust the weight of historical risk factors on the final association determination result; This indicates a dynamic threshold, the... Adjustments are made in real time based on the remaining computing power or remaining detection time of the current detection system.

10. A detection system based on the differentiated function adaptation of power distribution terminals, characterized in that, The system is applied to the method according to any one of claims 1-9, and the system comprises: A dynamic generation module for differential detection schemes based on terminal feature weights, the module comprising: Used to acquire a set of distribution automation terminals to be detected, and to collect terminal feature information and historical detection data corresponding to each distribution automation terminal; This is used to filter the set of power distribution automation terminals based on preset detection rules, determine the mandatory detection items to be performed by each power distribution automation terminal, and form a basic detection item set. This is used to construct a feature importance evaluation model based on the historical detection data, and to calculate the influence weight of each terminal feature on the detection anomaly results; It is used to prioritize candidate detection items other than the basic detection item set based on the terminal feature information and the influence weight, and generate differentiated detection schemes for each power distribution automation terminal based on the priority evaluation results. A terminal-linked detection scheme generation module based on state-function adaptation is provided, the module comprising: During the execution of the differentiated detection scheme, when an anomaly is detected in the target distribution automation terminal, the correlation feature information between the target distribution automation terminal and other distribution automation terminals is obtained; Based on the associated feature information, the system calculates the comprehensive correlation degree between the target power distribution automation terminal and other power distribution automation terminals, and filters out the set of target terminals associated with the target power distribution automation terminal according to the comprehensive correlation degree. Based on the abnormal function modules of the target power distribution automation terminal and the function modules of each terminal in the target terminal set, determine the list of associated function modules that need to be linked for detection, and generate a terminal linked detection scheme. This is used to execute the terminal-related detection scheme and dynamically update the detection scheme to be executed based on newly detected abnormal results during the related detection process, thereby realizing online iteration of the detection scheme.

Images