Operation and maintenance data uploading method and system of power distribution terminal

Abstract

The invention discloses an operation and maintenance data uploading method and system for a power distribution terminal, and the method comprises the steps: obtaining a real-time operation state and an operation and maintenance task instruction of the power distribution terminal, constructing a dynamic association network based on a historical fault mode database and real-time data, and recognizing an association monitoring parameter which is strongly associated with a target monitoring parameter; adopting a multi-dimensional matching score and a dynamic threshold value of resource awareness to screen out candidate operation and maintenance strategies covering the target and the associated parameters at the same time from a strategy library; dividing the candidate strategies into strategy clusters with different association strengths by using a spectral clustering algorithm based on the condition anomaly probability between the strategy coverage parameters, and storing the strategy clusters into a queue to be executed; performing priority ranking on the strategies in the queue; and finally, generating an uploading execution sequence by adopting a dynamic priority scheduling mechanism based on a sliding time window. According to the method, associated perception, dynamic optimization and intelligent scheduling of operation and maintenance data uploading are realized, and the data value density and the system response efficiency are remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of power distribution network operation and maintenance management technology, and particularly relates to a method and system for uploading operation and maintenance data of power distribution terminals. Background Technology

[0002] With the continuous improvement of smart grid and distribution network automation, the number of distribution terminals connected to the network has surged, generating massive amounts of operational data, including status updates, fault recordings, and power quality data. Efficiently and reliably uploading this data to the master station system is crucial for accurate equipment status assessment, rapid fault location, and optimized network operation.

[0003] Currently, the uploading of operation and maintenance data from distribution terminals mostly adopts fixed strategies such as timed reporting, change reporting, or simple priority reporting. These methods have significant shortcomings: First, they do not adequately utilize data correlation: the uploading strategies often treat individual parameters in isolation, ignoring the inherent correlations between parameters in fault modes or operating states. This can lead to uploaded datasets that may omit key contextual information or contain a large amount of redundant data, consuming valuable communication bandwidth. Second, the strategies lack dynamic adaptability: traditional methods struggle to adjust the content, order, and frequency of data uploads in real time based on dynamic changes in real-time grid conditions (such as load fluctuations and channel quality), operation and maintenance task objectives (such as fault diagnosis and health assessment), and the terminal's own status. Finally, the level of scheduling intelligence is low: facing numerous terminals and a complex strategy library, how to intelligently filter and prioritize the most effective and urgent execution sequence from a massive amount of potential upload actions is an optimization problem that urgently needs to be solved.

[0004] Therefore, there is an urgent need for a method that can understand the inherent relationships between data, dynamically adapt to environmental changes, and intelligently schedule upload strategies to improve the accuracy, efficiency, and practicality of power distribution operation and maintenance data uploads. Summary of the Invention

[0005] To address the problems existing in the background technology, the present invention provides a method and system for uploading operation and maintenance data of power distribution terminals, aiming to realize the associated perception, dynamic weighting and intelligent scheduling of operation and maintenance data upload.

[0006] In a first aspect, the present invention provides a method for uploading operation and maintenance data of a power distribution terminal, comprising:

[0007] Acquire a set of real-time operating status data of the power distribution terminal and preset operation and maintenance task instructions, wherein the operation and maintenance task instructions contain at least one target monitoring parameter;

[0008] Based on the real-time operating status data set and the preset historical fault mode database, at least one associated monitoring parameter that has a potential correlation with the at least one target monitoring parameter is identified.

[0009] In the preset operation and maintenance strategy library, select candidate operation and maintenance strategies that simultaneously cover the at least one target monitoring parameter and the associated monitoring parameters corresponding to the at least one target monitoring parameter, and use them as a candidate operation and maintenance strategy set.

[0010] Based on the correlation strength between all monitoring parameters covered by each candidate operation and maintenance strategy in the candidate operation and maintenance strategy set, the candidate strategy set is divided into multiple strategy clusters, each strategy cluster corresponds to a correlation strength interval, and each strategy cluster is stored in a different queue to be executed.

[0011] Obtain the dynamic weight coefficients of the target monitoring parameters and each of the associated monitoring parameters, and sort the candidate operation and maintenance strategies in the same queue for execution within a cluster according to the dynamic weight coefficients. The intra-cluster sorting rules include:

[0012] Calculate the weight evaluation value of each candidate operation and maintenance strategy, wherein the weight evaluation value is the sum of the dynamic weight coefficients corresponding to all monitoring parameters covered by the candidate operation and maintenance strategy;

[0013] Candidate operation and maintenance strategies whose weight evaluation values ​​exceed the preset threshold are marked as high-weight candidate operation and maintenance strategies and stored in the first priority sub-queue; the remaining candidate operation and maintenance strategies are marked as regular candidate operation and maintenance strategies and stored in the second priority sub-queue.

[0014] Within the first priority sub-queue and the second priority sub-queue, a secondary sorting is performed according to the weight evaluation value from high to low;

[0015] Based on the correlation strength intervals corresponding to each queue to be executed, the execution order of the queues to be executed is determined, and a predetermined number of operation and maintenance strategies are selected from the queues to be executed in sequence to generate the final upload execution sequence, and the power distribution terminal is controlled to upload data in sequence.

[0016] Secondly, the present invention provides a power distribution terminal operation and maintenance data uploading system, comprising:

[0017] The acquisition module is configured to acquire a set of real-time operating status data of the power distribution terminal and a preset operation and maintenance task instruction, wherein the operation and maintenance task instruction contains at least one target monitoring parameter.

[0018] The identification module is configured to identify at least one associated monitoring parameter that has a potential correlation with the at least one target monitoring parameter based on the real-time operating status data set and a preset historical fault mode database.

[0019] The selection module is configured to select candidate operation and maintenance strategies from a preset operation and maintenance strategy library that simultaneously cover the at least one target monitoring parameter and the associated monitoring parameters corresponding to the at least one target monitoring parameter, as a candidate operation and maintenance strategy set.

[0020] The partitioning module is configured to divide the candidate operation and maintenance strategy set into multiple strategy clusters based on the correlation strength between all monitoring parameters covered by each candidate operation and maintenance strategy in the candidate operation and maintenance strategy set. Each strategy cluster corresponds to a correlation strength interval, and each strategy cluster is stored in a different queue to be executed.

[0021] The sorting module is configured to obtain the dynamic weight coefficients of the target monitoring parameters and each of the associated monitoring parameters, and to perform intra-cluster sorting of each candidate operation and maintenance strategy in the same queue to be executed according to the dynamic weight coefficients. The intra-cluster sorting rules include:

[0022] Calculate the weight evaluation value of each candidate operation and maintenance strategy, wherein the weight evaluation value is the sum of the dynamic weight coefficients corresponding to all monitoring parameters covered by the candidate operation and maintenance strategy;

[0023] Candidate operation and maintenance strategies whose weight evaluation values ​​exceed the preset threshold are marked as high-weight candidate operation and maintenance strategies and stored in the first priority sub-queue; the remaining candidate operation and maintenance strategies are marked as regular candidate operation and maintenance strategies and stored in the second priority sub-queue.

[0024] Within the first priority sub-queue and the second priority sub-queue, a secondary sorting is performed according to the weight evaluation value from high to low;

[0025] The generation module is configured to determine the execution order of the queues to be executed based on the relative strength of the association intervals corresponding to each queue, and to select a predetermined number of operation and maintenance strategies from the queues in order to generate the final upload execution sequence, thereby controlling the power distribution terminal to upload data in sequence.

[0026] Thirdly, an electronic device is provided, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of the operation and maintenance data uploading method for a power distribution terminal according to any embodiment of the present invention.

[0027] Fourthly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the steps of the power distribution terminal operation and maintenance data uploading method according to any embodiment of the present invention.

[0028] The distribution terminal operation and maintenance data upload method and system of this application, by constructing a dynamic correlation network and performing multi-step propagation search and real-time change filtering, intelligently identifies the monitoring parameter set strongly correlated with the target parameters, significantly improving the correlation and contextual integrity of data acquisition; furthermore, it adopts a multi-dimensional scoring mechanism that integrates target coverage, correlation coverage, and correlation pair bridging degree, and introduces a resource-aware dynamic matching threshold to accurately select a set of candidate strategies with high fit from the strategy library; through a spectral clustering algorithm based on the probability of historical condition anomalies, the strategies are divided into strategy clusters with different intrinsic correlation strengths and queued, laying the foundation for differentiated scheduling; by integrating inherent baseline weights, real-time operation layer weight adjustment factors, and task target preference coefficients, dynamic weights are calculated to drive the generation of strategy priorities that can sensitively reflect the real-time status of the system and operation and maintenance intentions; finally, a polling scheduling mechanism based on a sliding time window and incorporating multi-factor dynamic priority adjustment is adopted to generate an upload execution sequence that tightly couples business logic and communication resource constraints. This method achieves full-link intelligence from data association perception and intelligent policy selection to resource optimization and scheduling, effectively improving the value density of operation and maintenance data uploads, resource utilization efficiency, and system response agility to abnormal events. Attached Figure Description

[0029] 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 some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 A flowchart illustrating a method for uploading maintenance data from a power distribution terminal, as provided in an embodiment of the present invention;

[0031] Figure 2 This is a structural block diagram of a power distribution terminal operation and maintenance data upload system provided in an embodiment of the present invention;

[0032] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0033] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] Please see Figure 1 The diagram illustrates a flowchart of a method for uploading operation and maintenance data of a power distribution terminal according to this application.

[0035] like Figure 1 As shown, the method for uploading operation and maintenance data from the power distribution terminal specifically includes the following steps:

[0036] Step S101: Obtain the real-time operating status data set of the power distribution terminal and the preset operation and maintenance task instructions, wherein the operation and maintenance task instructions contain at least one target monitoring parameter.

[0037] In this step,

[0038] Step S102: Based on the real-time operating status data set and the preset historical fault mode database, identify at least one associated monitoring parameter that has a potential correlation with the at least one target monitoring parameter.

[0039] In this step, a dynamic correlation network is constructed, where the nodes of the dynamic correlation network represent different types of monitoring parameters, and the weights of the edges of the dynamic correlation network are determined based on the frequency and temporal relationship of simultaneous abnormal states in the historical fault mode database.

[0040] The real-time operating status data set is mapped to the dynamic association network, and parameter nodes that are currently in an abnormal state or close to the threshold state are activated.

[0041] Starting from the node corresponding to the at least one target monitoring parameter, perform a multi-step propagation search in the dynamic association network to identify other parameter nodes that are within a specified number of hops and are connected to the target node via a path whose edge weight exceeds a dynamic threshold.

[0042] Based on the real-time change rate of parameters in the real-time operating status data set, the searched related nodes are filtered: only those parameter nodes whose change rate is significantly positively or negatively correlated with the change rate of the target monitoring parameter within the current time window are retained.

[0043] The monitoring parameters represented by the remaining parameter nodes after filtering are identified as associated monitoring parameters that have a potential correlation with the target monitoring parameter.

[0044] In one specific embodiment, an intelligent correlation parameter discovery process is implemented. Its core lies in constructing and utilizing a dynamic correlation network to combine static historical experience with dynamic real-time status, thereby accurately locating the group of monitoring parameters most relevant to the current operational task objectives. The specific implementation method is as follows:

[0045] First, an initial correlation network skeleton is constructed offline based on a pre-defined historical fault mode database. This database stores records of abnormal states of various monitored parameters (such as voltage, current, temperature, switching quantities, harmonic content, etc.) and their occurrence time series during each fault event. Each node in the network uniquely corresponds to one monitored parameter. The weight of the edge between two parameter nodes is calculated by analyzing their co-occurrence relationship in all historical fault records. The calculation method not only considers the frequency of both being marked as abnormal (e.g., parameter A and parameter B are abnormal simultaneously in 80% of faults), but also introduces the temporal relationship as a weighting factor (e.g., if parameter A always occurs before parameter B, then the edge from A to B will receive a higher weight). This network construction based on historical data provides an experience-driven prior knowledge foundation for correlation discovery.

[0046] Upon receiving a set of real-time operational status data, the system maps it to the established interconnected network. This mapping process involves "activating" network nodes: it iterates through all monitored parameters, and if a node's current value exceeds a preset normal threshold or falls within a preset warning range, its corresponding network node is marked as "activated." This step transforms the abstract real-time data into "hotspots" within the network that can be perceived by subsequent algorithms.

[0047] Subsequently, starting from the node corresponding to the target monitoring parameter specified in the operation and maintenance task instruction, a directed multi-step propagation search is performed in the dynamic correlation network. The search depth (number of hops) can be preset according to the task complexity (e.g., 2-3 hops). During the search process, the system accumulates the weight of each edge on the traversed path and sets a dynamic threshold. Only those nodes that start from the target node and accumulate weights exceeding this dynamic threshold through the path are initially listed as "candidate correlation nodes". This dynamic threshold is not a fixed value, but is fine-tuned according to the current overall load factor of the power grid or the system stability index—the heavier the load or the more unstable the system, the lower the threshold may be to capture more potential correlations and improve early warning sensitivity.

[0048] To further ensure the real-time effectiveness of correlations, a real-time change rate filtering mechanism is introduced. For each candidate correlation node found, the system calculates the rate of change of the monitoring parameter it represents within the current sliding time window (e.g., the last 10 sampling periods). Simultaneously, the rate of change of the target monitoring parameter is calculated within the same time window. Through correlation analysis, only candidate nodes whose rate of change is significantly positively correlated (co-directional change) or significantly negatively correlated (inversely correlated) with the rate of change of the target parameter are retained. This step is crucial; it filters out parameters that, although correlated in historical faults, do not exhibit synergistic or antagonistic changes in the current real-time operating scenario, thus ensuring the "immediacy" and "scenario specificity" of the discovered correlations.

[0049] Finally, the remaining parameter nodes after the multi-step propagation search and real-time change filtering are formally identified as the relevant monitoring parameters required for this task. These parameters, together with the target parameters, constitute a "parameter observation set" with close internal connections in the current operating state, laying a solid foundation for subsequently developing an efficient and accurate data upload strategy.

[0050] In summary, the correlation monitoring parameter identification method provided in this step represents a significant technological advancement and yields better results compared to traditional identification methods based on fixed rules or simple statistical correlations.

[0051] This approach achieves a leap from "static experience" to "dynamic perception": by constructing a correlation network that integrates historical fault frequencies and temporal relationships, and then superimposing real-time state activation and change rate filtering on this network, this method not only inherits the wisdom of historical operation and maintenance experience but also keenly captures and adapts to the rapidly changing real-time operating status of the power grid. The correlations it identifies are no longer static and universal but dynamic and highly adapted to the current scenario, greatly improving the accuracy and timeliness of correlation discovery.

[0052] The depth and interpretability of association discovery are enhanced: Employing a multi-step propagation search, it can discover parameters that are indirectly related to the target parameter but may be equally important (e.g., connected to the target parameter through an intermediate parameter), overcoming the limitation of traditional methods that typically only discover directly co-occurring parameters, thus achieving deeper association mining. Simultaneously, edge-weight-based path search and rate-of-change-based filtering ensure that each identified association parameter has a traceable "reason" (strong historical co-occurrence + real-time collaborative change), improving the interpretability and credibility of the entire process.

[0053] This step provides high-quality input for subsequent accurate decision-making: the set of related monitoring parameters output by this step is a high-confidence group of relevant parameters that has been verified by both historical and real-time data. Using this as a basis for selecting operational strategies ensures that the final data upload action is not an isolated collection of single parameters, but rather a systematic acquisition of a contextual and related "data story." This directly enhances the overall value and diagnostic potential of subsequent uploaded data, enabling the main station system to perform fault analysis or status assessment based on more complete information, thus providing stronger support for operational decisions. Simultaneously, filtering out irrelevant parameters effectively avoids wasting communication and storage resources, improving resource utilization efficiency.

[0054] Step S103: In the preset operation and maintenance strategy library, select candidate operation and maintenance strategies that simultaneously cover the at least one target monitoring parameter and the associated monitoring parameters corresponding to the at least one target monitoring parameter, and use them as a candidate operation and maintenance strategy set.

[0055] In this step, a first parameter set is constructed for the at least one target monitoring parameter, and a second parameter set is constructed for the associated monitoring parameters corresponding to the at least one target monitoring parameter;

[0056] Establish a strategy matching score mechanism, which includes the first dimension of the proportion of operation and maintenance strategies covering the first set of monitored parameters;

[0057] The operation and maintenance strategy covers the second dimension, which is the proportion of the parameters monitored in the second set of parameters.

[0058] The third dimension is the number of correlation pairs between parameters belonging to the first and second parameter sets among the monitoring parameters covered by the operation and maintenance strategy.

[0059] According to the task type specified in the operation and maintenance task instruction, weights are assigned to the first dimension, the second dimension and the third dimension, and the comprehensive matching score of each operation and maintenance strategy in the operation and maintenance strategy library is calculated according to the strategy matching degree scoring mechanism and the corresponding weights.

[0060] Set a dynamic matching threshold, which is dynamically adjusted based on the available upload bandwidth and computing resources of the current power distribution terminal;

[0061] Operation and maintenance strategies with a comprehensive matching score exceeding the dynamic matching threshold are selected to form a candidate operation and maintenance strategy set.

[0062] In one specific embodiment, a multi-dimensional, adaptive operation and maintenance strategy screening process is executed. Its core objective is to accurately identify, from a large pre-set strategy library, those strategies that best meet the current requirements for collaborative observation of "target-correlation" parameters, forming a high-quality candidate set. The specific implementation method is as follows:

[0063] First, the system transforms the output of step S102 into two distinct sets of filtering criteria: a first parameter set (containing all target monitoring parameters specified by the user) and a second parameter set (containing all associated monitoring parameters automatically identified by the system that correspond to the target parameters). These two sets define the core data scope that needs to be focused on in this data upload task.

[0064] Subsequently, the system constructs a policy matching score mechanism, which evaluates each operation and maintenance policy in the policy library from three dimensions:

[0065] First Dimension: Target Coverage. Calculate the percentage of overlap between the list of monitored parameters covered by the strategy and the first parameter set (target parameter set). For example, if the first parameter set has 3 target parameters, and a strategy covers 2 of them, its first dimension score is 2 / 3. This dimension ensures that the selected strategies directly respond to the user's explicit needs.

[0066] The second dimension: Correlation Coverage. This calculates the percentage of overlap between the list of monitored parameters covered by the strategy and the second parameter set (correlation parameter set). This dimension evaluates the strategy's ability to capture potentially important contextual information intelligently discovered by the system.

[0067] The third dimension: Bridging of Correlation Pairs. This is a more insightful dimension. It counts how many pairs of parameters covered by the strategy come from both the first and second parameter sets, respectively. For example, a strategy that covers both the target parameter "Phase A voltage" (from set one) and the correlated parameter "line temperature" (from set two) constitutes a "correlation pair." The score for this dimension is the number or weighted sum of such "correlation pairs." This dimension directly measures the effectiveness of the strategy in "connecting" the user's explicit goals with the correlated context discovered by the system.

[0068] To ensure the scoring aligns with specific task objectives, the system introduces task-type-driven weighting. Maintenance task instructions specify the task type, such as "precise fault location," "health status inspection," or "power quality analysis." Based on preset mapping rules, different weight coefficients are assigned to the three dimensions. For example, for the "precise fault location" task, "correlation pair bridging degree" (the third dimension) might be given a higher weight because the correlation between the root cause of the fault and the apparent parameters is crucial; while for "health status inspection," the three dimensions might be considered more evenly.

[0069] Based on the above mechanism, the system calculates a comprehensive matching score for each strategy in the strategy library (the formula is: score of each dimension × sum of corresponding weights). To cope with the dynamically changing operating environment, the system sets a dynamic matching threshold. This threshold is not a fixed value, but is adjusted in real time according to the current available upload bandwidth and local computing resource margin of the power distribution terminal. When network bandwidth is tight or the terminal computing load is high, the system automatically raises the threshold, allowing only a very small number of elite strategies with the highest matching degree to enter the candidate set, in order to save resources; when resources are sufficient, the threshold is appropriately lowered to include more potentially useful strategies for subsequent optimization selection.

[0070] Finally, the system traverses the policy library and formally includes all operation and maintenance policies with a comprehensive matching score exceeding the current dynamic matching threshold into the candidate operation and maintenance policy set. The policies in this set are the best subset that achieves the best balance between "target parameter observation" and "related context acquisition" under the current task objectives and resource constraints.

[0071] In summary, the operation and maintenance strategy selection method provided in this step completely changes the traditional strategy selection mode based on keyword matching or fixed classification, realizing a leap from "simple inclusion" to "intelligent matching", and bringing significant technical benefits:

[0072] This method achieves multi-dimensional and precise quantification and task alignment for strategy selection: by designing three dimensions—target coverage, association coverage, and association pair bridging degree—and introducing task type weights, it transforms the fuzzy "strategy relevance" into a calculable "matching score." This shifts the selection process from qualitative to quantitative, ensuring that the selected strategies not only include user-specified parameters but also effectively bridge the deep associations discovered by the system with the user's goals. This results in the final uploaded data packet having inherent logical consistency and problem-oriented focus, greatly enhancing the data's analytical value and decision support capabilities.

[0073] A resource-aware adaptive filtering threshold is constructed: a major innovation of this method is the introduction of a dynamic matching threshold based on real-time available upload bandwidth and computing resources. This imbues the policy filtering system with "environmental awareness," enabling it to flexibly adjust the filtering stringency according to the sufficiency of communication and computing resources. When resources are bottlenecked, it automatically focuses on core policies, ensuring uninterrupted critical data transmission; when resources are plentiful, it expands the selection pool, providing more high-quality options for subsequent optimization and scheduling. This mechanism significantly enhances the system's robustness and resource utilization efficiency in complex and fluctuating network environments, avoiding resource waste or critical data omissions caused by fixed thresholds.

[0074] This step lays a high-quality decision-making foundation for subsequent optimization and scheduling: the "candidate operation and maintenance strategy set" produced in this step is a pool of high-quality strategies that has undergone precise quantitative evaluation and initial screening based on resource constraints. Compared to blindly selecting from the entire database, this set is more controllable in size and of higher quality, allowing subsequent optimization operations such as cluster partitioning, sorting, and scheduling to be performed within a better solution space. This significantly reduces subsequent computational complexity and improves the overall efficiency of the final generated upload execution sequence. In short, this step, through high-quality initial screening, provides "high-quality raw materials" for the entire data upload pipeline and is a key prerequisite for ultimately achieving efficient and intelligent upload scheduling.

[0075] Step S104: Based on the correlation strength between all monitoring parameters covered by each candidate operation and maintenance strategy in the candidate operation and maintenance strategy set, the candidate strategy set is divided into multiple strategy clusters, each strategy cluster corresponds to a correlation strength interval, and each strategy cluster is stored in a different queue to be executed.

[0076] In this step, for each candidate operation and maintenance strategy, all monitoring parameters covered by the candidate operation and maintenance strategy are obtained;

[0077] Based on the historical failure mode database, calculate the conditional anomaly probability between each pair of monitoring parameters covered by the candidate operation and maintenance strategy, that is, the probability that the other monitoring parameter will also be abnormal if one of the monitoring parameters is abnormal.

[0078] Based on the calculated conditional anomaly probability matrix, the candidate operation and maintenance strategy set is divided into multiple strategy clusters using the spectral clustering algorithm;

[0079] For each strategy cluster, calculate the average conditional anomaly probability among the coverage monitoring parameters of all candidate operation and maintenance strategies within the strategy cluster, and use it as the correlation strength characterization value of the strategy cluster.

[0080] Based on the distribution of association strength characterization values, the strategy clusters are mapped to a preset discrete association strength interval.

[0081] In one specific embodiment, an unsupervised learning process based on the inherent correlation patterns of data is executed. Its core objective is to intelligently group the candidate operation and maintenance strategies selected in step S103, clustering strategies with similar "internal correlation characteristics" together to lay the foundation for subsequent differentiated scheduling. The specific implementation method is as follows:

[0082] First, for each strategy in the candidate operation and maintenance strategy set, the system extracts all the monitoring parameters it covers and forms a parameter list. For example, strategy A may cover three parameters: "A-phase voltage", "B-phase current", and "equipment temperature".

[0083] Subsequently, the system invokes the historical failure mode database to quantitatively analyze the symbiotic relationships between these parameters. The core calculation metric is the conditional anomaly probability. For any two distinct parameters X and Y in the parameter list covered by the strategy, the system calculates the probability P(Y anomalous|X anomalous). This probability is derived from historical data statistics: the proportion of events in which Y is also recorded as anomalous among all events where X is recorded as anomalous. By traversing all parameter pairs, a conditional anomaly probability matrix (or list) can be generated for each strategy. This matrix characterizes the pairwise correlation between parameters within the small group involved in this strategy under failure states.

[0084] Next, the system uses the conditional anomaly probability matrix of all candidate strategies as features and employs the Spectral Clustering algorithm to divide the entire candidate strategy set into multiple strategy clusters. Spectral Clustering is suitable for discovering natural groupings of data within complex structures. In this application, we treat the association probability feature of each strategy as a point in a high-dimensional space. The algorithm constructs a similarity matrix (e.g., based on the similarity between the association probability distributions of two strategies) and performs feature decomposition, ultimately assigning them to different clusters. Strategies within the same cluster cover parameter groups with highly similar internal association patterns (e.g., both belonging to the "strong electrical quantity association" pattern or the "weak electrical-temperature association" pattern), while strategies in different clusters exhibit significantly different internal association patterns.

[0085] To quantify the overall association characteristics of each cluster, the system calculates an association strength characterization value for each policy cluster. This is calculated by averaging the conditional anomaly probabilities of all parameter pairs covered by all policies within the cluster. This average value reflects the "average strength" of the association pattern represented by the cluster.

[0086] Finally, based on the distribution of the association strength characteristics of all policy clusters (e.g., through quantile analysis), the system maps them to several preset discrete association strength intervals, such as ["high association strength interval", "medium association strength interval", "low association strength interval"]. Each policy cluster is assigned to a unique interval. The system creates an independent queue for execution for each association strength interval and stores the policies from all policy clusters belonging to that interval into the corresponding queue for subsequent processing. Thus, the originally flattened set of candidate policies is organized into multiple hierarchical queues with different inherent association characteristics.

[0087] In summary, the strategy clustering and queuing methods provided in this step, by deeply exploring the implicit correlation patterns between monitoring parameters within the strategies, achieve intelligent organization and management of operation and maintenance strategies, resulting in significant technical improvements.

[0088] This approach represents a cognitive upgrade from "isolated strategies" to "patterned strategy clusters": traditional methods treat each operational strategy as an independent entity. This method, by calculating the probability of conditional anomalies and using spectral clustering, reveals the similarity in "internal parameter association patterns" between strategies, thus organizing them into "strategy clusters" with shared characteristics. This partitioning method, based on the inherent structure of the data, is far superior to classification based on shallow attributes such as strategy name and creation time, making the management, understanding, and subsequent scheduling of massive strategies more scientific and orderly. It provides operations personnel with a higher-level, patterned view.

[0089] This method provides a precise and reasonable basis for differentiated priority scheduling: by calculating the "association strength characterization value" of each policy cluster and mapping it to discrete strength intervals, it assigns a clear and quantifiable priority label to each queue to be executed. A queue with "high association strength" means that the data combinations that its policies focus on have shown stronger collaborative anomalies in historical faults, and its data often has higher joint value for fault diagnosis. This provides a direct, objective, and business-interpretable basis for determining the queue execution order in step S106, transforming scheduling decisions from rough judgments based on experience to refined and automated decisions based on data association strength, ensuring that high-value data combinations are prioritized.

[0090] This significantly improves the efficiency and effectiveness of subsequent scheduling algorithms: pre-dividing strategies into different queues essentially decomposes the complex global sorting problem into two relatively simple sub-problems: "inter-queue sorting" and "intra-queue sorting." This "divide and conquer" strategy greatly reduces the computational complexity of the scheduling algorithm. Furthermore, since strategies within the same queue have similar association patterns, competition and comparison are more comparable when performing dynamic weight-based internal sorting, resulting in a more reasonable final sorting result. This hierarchical processing structure provides an optimized data structure foundation for building an efficient and accurate intelligent scheduling system, serving as a crucial intelligent bridge connecting strategy selection and final scheduling execution.

[0091] Step S105: Obtain the dynamic weight coefficients of the target monitoring parameter and each of the associated monitoring parameters, and sort the candidate operation and maintenance strategies in the same queue to be executed within the cluster according to the dynamic weight coefficients.

[0092] In this step, an inherent benchmark weight is formed based on the frequency with which the monitoring parameters are marked as key causes or core characterization parameters in the historical failure mode database. The monitoring parameters are target monitoring parameters or associated monitoring parameters.

[0093] The deviation between the statistical characteristics of the monitoring parameters within the current time window and the characteristics of the historical normal operation interval is calculated and used as a weight adjustment factor for the real-time operation layer.

[0094] Based on the task objectives explicitly stated in the operation and maintenance task instructions, the corresponding target preference coefficients are matched from the preset mapping rule base;

[0095] The inherent baseline weight, real-time operation layer weight adjustment factor, and target preference coefficient are fused together to calculate and output the dynamic weight coefficient of each monitoring parameter.

[0096] The intra-cluster sorting rules include:

[0097] Calculate the weight evaluation value of each candidate operation and maintenance strategy, wherein the weight evaluation value is the sum of the dynamic weight coefficients corresponding to all monitoring parameters covered by the candidate operation and maintenance strategy;

[0098] Candidate operation and maintenance strategies whose weight evaluation values ​​exceed the preset threshold are marked as high-weight candidate operation and maintenance strategies and stored in the first priority sub-queue; the remaining candidate operation and maintenance strategies are marked as regular candidate operation and maintenance strategies and stored in the second priority sub-queue.

[0099] Within the first priority sub-queue and the second priority sub-queue, a secondary sorting is performed according to the weight evaluation value from high to low.

[0100] In one specific embodiment, an intelligent computing process involving multi-source information fusion is executed. Its core objective is to assign a dynamically changing "value" quantification index—a dynamic weight coefficient—to each monitoring parameter, based on which the grouped operation and maintenance strategies are finely ranked. The specific implementation method is as follows:

[0101] First, the system calculates an inherent baseline weight for each monitored parameter. This weight, based on historical experience, reflects the parameter's essential importance in fault diagnosis. Specifically, it iterates through every root cause analysis record in the historical fault mode database; when a parameter is marked as a "critical trigger" or "core characteristic parameter" by operations and maintenance experts, its count is incremented. Finally, the inherent baseline weight for each parameter is proportional to the frequency of its marking and is normalized. This ensures that important parameters repeatedly validated in long-term operations and maintenance practice receive higher base scores.

[0102] Secondly, the system calculates a real-time operational layer weight adjustment factor to capture the urgency of the current operational status. For each parameter, the system obtains its numerical sequence within the current sliding time window (e.g., the past 30 minutes). The statistical characteristics of this sequence (e.g., mean, variance, trend slope) are calculated and compared with a "healthy benchmark interval" learned from historical normal operation data. Deviation is calculated using a composite function that comprehensively considers the extent to which the current value exceeds the benchmark interval, the degree of increased volatility, and the abnormal steepness of the trend. The greater the deviation, the higher the generated real-time adjustment factor value, indicating that the parameter is currently performing abnormally and requires more attention.

[0103] Third, the system introduces target preference coefficients to align the weighting system with the specific objectives of this maintenance operation. The system parses maintenance task instructions (such as "locating transient faults" and "assessing equipment aging") and matches the corresponding "parameter importance preference table" from a preset mapping rule base. For example, the "locating transient faults" task may assign higher preference coefficients to high time resolution parameters such as "transient waveform data" and "switch position signal"; the "assessing equipment aging" task may increase the preference coefficients for parameters such as "long-term operating temperature trend" and "cumulative number of actions".

[0104] Next is the fusion computation, whose core logic is to dynamically weigh historical experience, real-time status, and task objectives. The system employs a weighted summation model, but its key innovation lies in the dynamic nature of the weight allocation itself. The system sets up a state judgment mechanism: if the real-time deviation of a parameter exceeds a preset "significant anomaly" threshold, then in the current fusion computation for that parameter, the real-time runtime layer weight adjustment factor is assigned a very high fusion weight (e.g., 70%), while the weights of the inherent benchmark weight and the target preference coefficient are correspondingly reduced. This ensures that when a parameter exhibits significant anomalies, its dynamic weight can immediately and sensitively reflect this emergency situation. If the parameter operates smoothly, the fusion weight tilts towards the inherent benchmark and target preference, reflecting the planning and purposefulness under steady-state conditions. Through this mechanism, a dynamic weight coefficient is output for each parameter.

[0105] Finally, within each queue to be executed, the system performs intra-cluster sorting:

[0106] Calculate the strategy weight evaluation value: For each candidate operation and maintenance strategy in the queue, sum the dynamic weight coefficients of all monitoring parameters it covers to obtain the "weight evaluation value" of the strategy. This value intuitively reflects the comprehensive data value that can be obtained by executing the strategy.

[0107] Sub-queues are created by setting a dynamic threshold (which can be determined based on the overall evaluation value distribution of the queue or the currently available resources). Policies with evaluation values ​​exceeding this threshold are marked as high-weight candidate maintenance policies and placed in the first priority sub-queue dedicated to that queue; the remaining policies are marked as regular candidate maintenance policies and placed in the second priority sub-queue. This achieves initial hierarchical classification within the queues.

[0108] Secondary sorting within sub-queues: Within the first and second priority sub-queues, strategies are sorted in descending order of their "weight evaluation value." At this point, each strategy in the queue to be executed has obtained a precise, dynamically value-based ranking position.

[0109] In summary, the dynamic weight calculation and intra-cluster ranking method provided in this step, by creatively integrating historical experience, real-time status, and task objectives from three perspectives, and designing a sensitive dynamic fusion mechanism and a two-level ranking structure, achieves precise value assessment of operation and maintenance strategies and refined scheduling preparation, bringing about groundbreaking technical results:

[0110] This method represents a fundamental shift in weight calculation from "static and fixed" to "dynamic and context-aware": In traditional methods, parameter weights are often pre-configured statically. This method introduces a real-time runtime layer weight adjustment factor and a target preference coefficient, and designs dynamic fusion rules based on real-time abnormal states. This makes the weight of each parameter a variable that evolves in real-time with the system's operating conditions and maintenance intentions. This ensures that when a fault emerges or develops, the weight of relevant parameters can "automatically" surge, driving the system to prioritize the collection of the most critical data, greatly improving the system's sensitivity and targeted handling of sudden abnormal events.

[0111] A precise strategy evaluation system integrating multi-dimensional value judgments has been constructed: by organically combining the "inherent benchmark weight" reflecting long-term reliability, the "real-time adjustment factor" reflecting current urgency, and the "goal preference coefficient" reflecting current work focus, this method no longer relies solely on a single aspect for strategy evaluation. A strategy that covers historically important parameters, exhibits abnormal current performance, and is highly relevant to the task objective will have a prominent "weight evaluation value." This multi-dimensional integrated evaluation makes the strategy ranking results more comprehensive and the decision-making more scientific, enabling the selection of the data collection actions with the highest "cost-effectiveness" in a specific spatiotemporal context.

[0112] A two-level sorting mechanism provides a structured optimal preparation for efficient scheduling: the two-level structure of "high-weight / regular sub-queue partitioning" plus "secondary sorting within sub-queues" implemented within each queue is an ingenious preprocessing step. It clearly "tiers" and "fully sorts" the value distribution of strategies within the queues. This greatly facilitates the subsequent scheduling execution module: the scheduling algorithm can quickly locate the "head" high-value strategies in each queue and obtain a complete priority list when a queue needs to be explored in depth. This structure significantly reduces the algorithmic complexity of generating the final global scheduling sequence and ensures that, at any scheduling depth, the selected strategy is locally optimal under the current conditions, thereby maximizing the overall efficiency of data uploading.

[0113] Step S106: Determine the execution order of the queues to be executed based on the relative strength intervals of each queue, and select a predetermined number of operation and maintenance strategies from the queues to be executed in sequence to generate the final upload execution sequence, and control the power distribution terminal to upload data in sequence.

[0114] In this step, a dynamic scheduling mechanism based on sliding time windows is adopted to divide the available upload time window into multiple consecutive scheduling time slots;

[0115] Calculate the initial priority score of each queue to be executed based on the association strength representation value of the policy cluster corresponding to each queue to be executed;

[0116] The real-time status indicators of each queue to be executed are obtained, and the initial priority scores of each queue are dynamically adjusted according to the real-time status indicators to obtain the target priority scores of each queue to be executed. The real-time status indicators include the proportion of high-weight candidate operation and maintenance strategies in the queue to be executed, the average estimated execution time of candidate operation and maintenance strategies in the queue to be executed, and the sum of the real-time anomaly degree of the monitoring parameters covered by the candidate operation and maintenance strategies in the queue to be executed.

[0117] At the beginning of each scheduling slot, all queues to be executed are sorted according to the dynamically adjusted target priority score;

[0118] In descending order of priority, a preset number of operation and maintenance policies are selected from each queue to be executed and added to the upload execution sequence. The selection of the preset number of operation and maintenance policies follows these principles:

[0119] The number of candidate operation and maintenance strategies selected from each queue to be executed is proportional to the priority score of the queue to be executed.

[0120] Ensure that each queue to be executed is selected at least once within a scheduling cycle, unless the queue to be executed is empty;

[0121] The total number of policies selected within a single scheduling slot shall not exceed the maximum number of policies that the scheduling slot can accommodate for execution;

[0122] The selected operation and maintenance strategies are arranged according to the priority order of their respective queues and the sorting order within each queue to form the execution sequence fragments of the current scheduling slot;

[0123] The execution sequence segments of multiple consecutive scheduling time slots are combined to form the final upload execution sequence.

[0124] In one specific embodiment, the final resource scheduling and sequence synthesis are performed by employing a dynamic priority polling scheduling mechanism based on a sliding time window. This mechanism transforms the intelligently grouped and sorted maintenance strategy queue into a highly efficient sequence of upload instructions that can be executed sequentially by the power distribution terminal. The specific implementation method is as follows:

[0125] First, the system initiates dynamic scheduling based on a sliding time window. According to the communication plan negotiated between the master station system and the distribution terminals, or the real-time available communication time period, the system divides a large, continuous available upload time window (e.g., the next 10 minutes) into multiple shorter, continuous scheduling time slots (e.g., 30 seconds per slot). Each time slot is the basic unit for scheduling decisions and execution. The window "slides" over time, ensuring that scheduling always targets the nearest future operable time period.

[0126] Secondly, the system calculates a target priority score for each queue to be executed, which serves as the core basis for scheduling. The calculation process consists of two steps:

[0127] Calculate the initial priority: Based on the association strength characterization value determined in step S104 for each queue, it is converted into an initial priority score through a preset linear or nonlinear mapping function (such as the Sigmoid function). This ensures that queues with higher association strength receive higher base priorities.

[0128] Dynamic Adjustment: The system acquires three real-time status indicators for each queue: (a) the proportion of "high-weight candidate maintenance strategies" in the queue (reflecting the density of high-value strategies within the queue); (b) the average estimated execution time of all strategies in the queue (predicted based on historical execution time data, reflecting execution cost); and (c) the sum of real-time anomalies of the monitoring parameters covered by all strategies in the queue (obtained from the weight calculation in step S105, reflecting the current urgency of the data involved in the queue). The system dynamically adjusts the initial priority score using these three indicators based on a preset adjustment algorithm (such as weighted summation or a more complex rule model). For example, a high-association-strength queue, if it simultaneously contains a high proportion of high-weight strategies and its parameters are currently significantly abnormal, will have its priority further increased significantly; conversely, if its strategy's estimated execution time is too long, its priority may be moderately suppressed when resources are scarce. The adjusted result is the final target priority score.

[0129] At the beginning of each scheduling time slot, the system sorts all non-empty queues in descending order based on their latest target priority scores, thus determining the access order of each queue within the time slot.

[0130] Next, the system enters the sequential selection and quota allocation phase. The system processes each queue sequentially according to the aforementioned sorting results. A policy selection quota for the current queue is allocated within the current time slot. The quota calculation is proportional to the queue's target priority score and follows the core principle: ensuring that each non-empty queue gets at least one selection opportunity within a complete scheduling cycle (i.e., traversing all queues once). This prevents low-priority queues from being "starved." Simultaneously, the system maintains a counter for the total number of policies selected within a time slot, ensuring that this number does not exceed the maximum number of policy executions a single time slot can accommodate (determined by the communication rate and the average data volume of the policy).

[0131] For the current queue, the system starts from the head of its first priority sub-queue (i.e., the strategy with the highest weight evaluation value) and selects strategies sequentially according to the internal order of the sub-queue. Each time a strategy is selected, the quota is decremented by 1, and the total number of selected strategies is incremented by 1. When all strategies in the first sub-queue are selected or the quota reaches 0, if there is still quota available, selection continues sequentially from the head of the second priority sub-queue. This process continues until the queue's quota is exhausted, all strategies in the queue are selected, or the total number of selected strategies reaches the time slot limit.

[0132] When a scheduling time slot ends, the system will arrange the operation and maintenance strategies selected from all queues in this time slot according to the above rules, in the order in which they were selected, to form an execution sequence segment.

[0133] Finally, the system concatenates the execution sequence fragments of all consecutive time slots generated within the current sliding time window, chronologically, to form the final upload execution sequence corresponding to that window. This sequence is then sent to the distribution terminal, controlling it to strictly execute the data encapsulation and upload operations defined by each maintenance strategy in the sequence. When the sequence of a window is completed or the time expires, the system slides to the next time window and repeats the above scheduling process.

[0134] In summary, the dynamic scheduling and sequence generation method provided in this step, through innovative sliding window polling, multi-factor dynamic priority adjustment, and proportional quota allocation mechanism, successfully solves the core challenge of efficient and fair scheduling of heterogeneous policy queues under resource constraints, achieving the following significant technical effects:

[0135] A remarkable balance between value prioritization and system efficiency has been achieved: the core of the scheduling mechanism—dynamic target priority scoring—creatively integrates static correlation strength with dynamic three-dimensional real-time status (value density, execution cost, and urgency). This allows scheduling decisions to not only be based on the long-term potential value of the strategy (correlation strength) but also to respond sensitively to instantaneous changes in the real-time system's state. For example, when a queue with moderate correlation strength but containing a large number of real-time anomaly parameters appears, its priority will be dynamically and significantly increased to ensure that urgent data is not delayed. At the same time, quota allocation proportional to priority scores ensures that high-value queues receive more resources, while the fairness guarantee of "at least once per cycle" prevents low-correlation but necessary data from being completely ignored, thus achieving a precise balance between maximizing the value of data uploads and optimizing system resource utilization globally.

[0136] A robust scheduling framework combining real-time responsiveness and predictability was constructed: Sliding time window-based scheduling discretizes the continuous scheduling problem into periodic decision points, enabling the system to make optimized decisions for finite future time periods. This approach allows for rapid response to priority changes (recalculated at the start of each time slot) while avoiding the computational complexity and uncertainty associated with global infinite look-ahead. Combined with a two-level sorting structure (inter-queue sorting and intra-queue sub-queue sorting), the scheduler can determine the next strategy to execute within constant time, ensuring extremely low latency and high real-time performance of scheduling decisions, meeting the power system's requirement for rapid response. Furthermore, the entire scheduling process has clear and highly deterministic rules, making the uploading behavior predictable and analyzable for operations and maintenance personnel, thus improving the system's controllability.

[0137] A high-quality execution sequence tightly coupled with business logic and resource constraints was generated: the final uploaded execution sequence is the concentrated embodiment and ultimate implementation of the results of all the aforementioned intelligent processing steps (association identification, strategy filtering, dynamic weighting, and intra-cluster sorting). It is not a simple list, but an optimized solution deeply embedded in business logic (reflected by priority and quota) and strictly constrained by resource boundaries (time slot capacity, execution upper limit). This sequence ensures that the distribution terminal executes the most important and appropriate set of data upload actions within the current communication window, taking into account data association value, real-time urgency, task objectives, and available resources. This fundamentally improves the data utility and information density of a single communication interaction, providing the highest quality data raw materials for the backend master station system to perform accurate status analysis and fault diagnosis, achieving a systematic improvement in the efficiency of the entire link from data acquisition to decision support.

[0138] In summary, the method of this application, by constructing a dynamic correlation network and performing multi-step propagation search and real-time change filtering, intelligently identifies the set of monitoring parameters strongly correlated with the target parameters, significantly improving the correlation and contextual integrity of data collection. Furthermore, it adopts a multi-dimensional scoring mechanism that integrates target coverage, correlation coverage, and correlation pair bridging degree, and introduces a resource-aware dynamic matching threshold to accurately select a set of candidate strategies with high fit from the strategy library. Through a spectral clustering algorithm based on the probability of historical condition anomalies, the strategies are divided into strategy clusters with different intrinsic correlation strengths and queued, laying the foundation for differentiated scheduling. By integrating inherent baseline weights, real-time runtime layer weight adjustment factors, and task target preference coefficients, dynamic weights are calculated to drive the generation of strategy priorities that can sensitively reflect the real-time status of the system and operational intentions. Finally, a polling scheduling mechanism based on a sliding time window and incorporating multi-factor dynamic priority adjustment is adopted to generate an upload execution sequence that tightly couples business logic and communication resource constraints. This method achieves full-link intelligence from data association perception and intelligent policy selection to resource optimization and scheduling, effectively improving the value density of operation and maintenance data uploads, resource utilization efficiency, and system response agility to abnormal events.

[0139] Please see Figure 2 The diagram shows a structural block diagram of a power distribution terminal operation and maintenance data upload system according to this application.

[0140] like Figure 2 As shown, the operation and maintenance data upload system 200 includes an acquisition module 210, an identification module 220, a selection module 230, a division module 240, a sorting module 250, and a generation module 260.

[0141] The system includes: an acquisition module 210, configured to acquire a real-time operating status data set of the power distribution terminal and a preset maintenance task instruction, wherein the maintenance task instruction contains at least one target monitoring parameter; an identification module 220, configured to identify at least one associated monitoring parameter that has a potential correlation with the at least one target monitoring parameter based on the real-time operating status data set and a preset historical fault mode database; a selection module 230, configured to select candidate maintenance strategies from a preset maintenance strategy library that simultaneously cover the at least one target monitoring parameter and the associated monitoring parameters corresponding to the at least one target monitoring parameter, as a candidate maintenance strategy set; a division module 240, configured to divide the candidate maintenance strategy set into multiple strategy clusters based on the correlation strength between all monitoring parameters covered by each candidate maintenance strategy in the candidate maintenance strategy set, with each strategy cluster corresponding to a correlation strength interval, and storing each strategy cluster in a different execution queue; and a sorting module 250, configured to acquire the target... The system identifies the dynamic weight coefficients of the monitoring parameters and the associated monitoring parameters, and sorts the candidate operation and maintenance strategies in the same queue for execution within a cluster based on these dynamic weight coefficients. The cluster sorting rules include: calculating the weight evaluation value of each candidate operation and maintenance strategy, where the weight evaluation value is the sum of the dynamic weight coefficients of all monitoring parameters covered by the candidate operation and maintenance strategy; marking candidate operation and maintenance strategies with weight evaluation values ​​exceeding a preset threshold as high-weight candidate operation and maintenance strategies and storing them in a first priority sub-queue, and marking the remaining candidate operation and maintenance strategies as regular candidate operation and maintenance strategies and storing them in a second priority sub-queue; performing secondary sorting within the first priority sub-queue and the second priority sub-queue according to the weight evaluation values ​​from high to low; and configuring the generation module 260 to determine the execution order of the queues for execution based on the relative strength intervals of the associated parameters, and to sequentially select a predetermined number of operation and maintenance strategies from the queues for execution to generate the final upload execution sequence, thereby controlling the power distribution terminal to upload data sequentially.

[0142] It should be understood that Figure 2 The modules and references described in the document Figure 1 The steps described in the text correspond to those in the method described above. Therefore, the operations, features, and corresponding technical effects described above also apply to the method described in the text. Figure 2 The various modules in the document will not be described in detail here.

[0143] In other embodiments, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the operation and maintenance data uploading method for the power distribution terminal in any of the above method embodiments.

[0144] In one embodiment, the computer-readable storage medium of the present invention stores computer-executable instructions, which are configured as follows:

[0145] Acquire a set of real-time operating status data of the power distribution terminal and preset operation and maintenance task instructions, wherein the operation and maintenance task instructions contain at least one target monitoring parameter;

[0146] Based on the real-time operating status data set and the preset historical fault mode database, at least one associated monitoring parameter that has a potential correlation with the at least one target monitoring parameter is identified.

[0147] In the preset operation and maintenance strategy library, select candidate operation and maintenance strategies that simultaneously cover the at least one target monitoring parameter and the associated monitoring parameters corresponding to the at least one target monitoring parameter, and use them as a candidate operation and maintenance strategy set.

[0148] Based on the correlation strength between all monitoring parameters covered by each candidate operation and maintenance strategy in the candidate operation and maintenance strategy set, the candidate strategy set is divided into multiple strategy clusters, each strategy cluster corresponds to a correlation strength interval, and each strategy cluster is stored in a different queue to be executed.

[0149] Obtain the dynamic weight coefficients of the target monitoring parameters and each of the associated monitoring parameters, and sort the candidate operation and maintenance strategies in the same queue for execution within a cluster according to the dynamic weight coefficients. The intra-cluster sorting rules include:

[0150] Calculate the weight evaluation value of each candidate operation and maintenance strategy, wherein the weight evaluation value is the sum of the dynamic weight coefficients corresponding to all monitoring parameters covered by the candidate operation and maintenance strategy;

[0151] Candidate operation and maintenance strategies whose weight evaluation values ​​exceed the preset threshold are marked as high-weight candidate operation and maintenance strategies and stored in the first priority sub-queue; the remaining candidate operation and maintenance strategies are marked as regular candidate operation and maintenance strategies and stored in the second priority sub-queue.

[0152] Within the first priority sub-queue and the second priority sub-queue, a secondary sorting is performed according to the weight evaluation value from high to low;

[0153] Based on the correlation strength intervals corresponding to each queue to be executed, the execution order of the queues to be executed is determined, and a predetermined number of operation and maintenance strategies are selected from the queues to be executed in sequence to generate the final upload execution sequence, and the power distribution terminal is controlled to upload data in sequence.

[0154] Computer-readable storage media may include a stored program area and a stored data area, wherein the stored program area may store an operating system and an application program required for at least one function; the stored data area may store data created based on the use of the power distribution terminal's operation and maintenance data upload system, etc. Furthermore, the computer-readable storage medium may include high-speed random access memory, and may also include memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, the computer-readable storage medium may optionally include memory remotely configured relative to a processor, and these remote memories can be connected to the power distribution terminal's operation and maintenance data upload system via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0155] Figure 3 This is a schematic diagram of the structure of the electronic device provided in the embodiment of the present invention, such as... Figure 3 As shown, the device includes a processor 310 and a memory 320. The electronic device may also include an input device 330 and an output device 340. The processor 310, memory 320, input device 330, and output device 340 can be connected via a bus or other means. Figure 3 Taking a bus connection as an example, the memory 320 is the computer-readable storage medium described above. The processor 310 executes various server functions and data processing by running non-volatile software programs, instructions, and modules stored in the memory 320, thereby realizing the power distribution terminal operation and maintenance data upload method described in the above embodiment. The input device 330 can receive input digital or character information and generate key signal inputs related to user settings and function control of the power distribution terminal operation and maintenance data upload system. The output device 340 may include a display screen or other display device.

[0156] The aforementioned electronic device can execute the method provided in the embodiments of the present invention, and has the corresponding functional modules and beneficial effects for executing the method. Technical details not described in detail in this embodiment can be found in the method provided in the embodiments of the present invention.

[0157] In one implementation, the above-described electronic device is applied to a power distribution terminal operation and maintenance data upload system for a client, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to:

[0158] Acquire a set of real-time operating status data of the power distribution terminal and preset operation and maintenance task instructions, wherein the operation and maintenance task instructions contain at least one target monitoring parameter;

[0159] Based on the real-time operating status data set and the preset historical fault mode database, at least one associated monitoring parameter that has a potential correlation with the at least one target monitoring parameter is identified.

[0160] In the preset operation and maintenance strategy library, select candidate operation and maintenance strategies that simultaneously cover the at least one target monitoring parameter and the associated monitoring parameters corresponding to the at least one target monitoring parameter, and use them as a candidate operation and maintenance strategy set.

[0161] Based on the correlation strength between all monitoring parameters covered by each candidate operation and maintenance strategy in the candidate operation and maintenance strategy set, the candidate strategy set is divided into multiple strategy clusters, each strategy cluster corresponds to a correlation strength interval, and each strategy cluster is stored in a different queue to be executed.

[0162] Obtain the dynamic weight coefficients of the target monitoring parameters and each of the associated monitoring parameters, and sort the candidate operation and maintenance strategies in the same queue for execution within a cluster according to the dynamic weight coefficients. The intra-cluster sorting rules include:

[0163] Calculate the weight evaluation value of each candidate operation and maintenance strategy, wherein the weight evaluation value is the sum of the dynamic weight coefficients corresponding to all monitoring parameters covered by the candidate operation and maintenance strategy;

[0164] Candidate operation and maintenance strategies whose weight evaluation values ​​exceed the preset threshold are marked as high-weight candidate operation and maintenance strategies and stored in the first priority sub-queue; the remaining candidate operation and maintenance strategies are marked as regular candidate operation and maintenance strategies and stored in the second priority sub-queue.

[0165] Within the first priority sub-queue and the second priority sub-queue, a secondary sorting is performed according to the weight evaluation value from high to low;

[0166] Based on the correlation strength intervals corresponding to each queue to be executed, the execution order of the queues to be executed is determined, and a predetermined number of operation and maintenance strategies are selected from the queues to be executed in sequence to generate the final upload execution sequence, and the power distribution terminal is controlled to upload data in sequence.

[0167] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence 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 ROM / RAM, magnetic disk, 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 various embodiments or some parts of embodiments.

[0168] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for uploading operation and maintenance data of a power distribution terminal, characterized in that, include: Acquire a set of real-time operating status data of the power distribution terminal and preset operation and maintenance task instructions, wherein the operation and maintenance task instructions contain at least one target monitoring parameter; Based on the real-time operating status data set and the preset historical fault mode database, at least one associated monitoring parameter that has a potential correlation with the at least one target monitoring parameter is identified. In the preset operation and maintenance strategy library, select candidate operation and maintenance strategies that simultaneously cover the at least one target monitoring parameter and the associated monitoring parameters corresponding to the at least one target monitoring parameter, and use them as a candidate operation and maintenance strategy set. Based on the correlation strength between all monitoring parameters covered by each candidate operation and maintenance strategy in the candidate operation and maintenance strategy set, the candidate strategy set is divided into multiple strategy clusters, each strategy cluster corresponds to a correlation strength interval, and each strategy cluster is stored in a different queue to be executed. Obtain the dynamic weight coefficients of the target monitoring parameters and each of the associated monitoring parameters, and sort the candidate operation and maintenance strategies in the same queue for execution within a cluster according to the dynamic weight coefficients. The intra-cluster sorting rules include: Calculate the weight evaluation value of each candidate operation and maintenance strategy, wherein the weight evaluation value is the sum of the dynamic weight coefficients corresponding to all monitoring parameters covered by the candidate operation and maintenance strategy; Candidate operation and maintenance strategies whose weight evaluation values ​​exceed the preset threshold are marked as high-weight candidate operation and maintenance strategies and stored in the first priority sub-queue; the remaining candidate operation and maintenance strategies are marked as regular candidate operation and maintenance strategies and stored in the second priority sub-queue. Within the first priority sub-queue and the second priority sub-queue, a secondary sorting is performed according to the weight evaluation value from high to low; Based on the correlation strength intervals corresponding to each queue to be executed, the execution order of the queues to be executed is determined, and a predetermined number of operation and maintenance strategies are selected from the queues to be executed in sequence to generate the final upload execution sequence, and the power distribution terminal is controlled to upload data in sequence.

2. The method for uploading operation and maintenance data of a power distribution terminal according to claim 1, characterized in that, The at least one associated monitoring parameter identified based on the real-time operating status data set and the preset historical fault mode database includes: Construct a dynamic correlation network, where the nodes of the dynamic correlation network represent different types of monitoring parameters, and the weights of the edges of the dynamic correlation network are determined based on the frequency and temporal relationship of simultaneous abnormal states in the historical fault mode database. The real-time operating status data set is mapped to the dynamic association network to activate parameter nodes that are currently in an abnormal state or close to the threshold state. Starting from the node corresponding to the at least one target monitoring parameter, perform a multi-step propagation search in the dynamic association network to identify other parameter nodes that are within a specified number of hops and are connected to the target node via a path whose edge weight exceeds a dynamic threshold. Based on the real-time change rate of parameters in the real-time operating status data set, the searched related nodes are filtered: only those parameter nodes whose change rate is significantly positively or negatively correlated with the change rate of the target monitoring parameter within the current time window are retained. The monitoring parameters represented by the remaining parameter nodes after filtering are identified as associated monitoring parameters that have a potential correlation with the target monitoring parameter.

3. The method for uploading operation and maintenance data of a power distribution terminal according to claim 1, characterized in that, The step of selecting candidate operation and maintenance strategies from a preset operation and maintenance strategy library that simultaneously covers the at least one target monitoring parameter and the associated monitoring parameters corresponding to the at least one target monitoring parameter includes: A first parameter set is constructed for the at least one target monitoring parameter, and a second parameter set is constructed for the associated monitoring parameters corresponding to the at least one target monitoring parameter; Establish a strategy matching score mechanism, which includes the first dimension of the proportion of operation and maintenance strategies covering the first set of monitored parameters; The operation and maintenance strategy covers the second dimension, which is the proportion of the parameters monitored in the second set of parameters. The third dimension is the number of correlation pairs between parameters belonging to the first and second parameter sets among the monitoring parameters covered by the operation and maintenance strategy. According to the task type specified in the operation and maintenance task instruction, weights are assigned to the first dimension, the second dimension and the third dimension, and the comprehensive matching score of each operation and maintenance strategy in the operation and maintenance strategy library is calculated according to the strategy matching degree scoring mechanism and the corresponding weights. Set a dynamic matching threshold, which is dynamically adjusted based on the available upload bandwidth and computing resources of the current power distribution terminal; Operation and maintenance strategies with a comprehensive matching score exceeding the dynamic matching threshold are selected to form a candidate operation and maintenance strategy set.

4. The method for uploading operation and maintenance data of a power distribution terminal according to claim 1, characterized in that, The step of dividing the candidate operation and maintenance strategy set into multiple strategy clusters based on the correlation strength among all monitoring parameters covered by each candidate operation and maintenance strategy in the candidate operation and maintenance strategy set, with each strategy cluster corresponding to a correlation strength interval, includes: For each candidate operation and maintenance strategy, obtain all monitoring parameters covered by the candidate operation and maintenance strategy; Based on the historical failure mode database, calculate the conditional anomaly probability between each pair of monitoring parameters covered by the candidate operation and maintenance strategy, that is, the probability that the other monitoring parameter will also be abnormal if one of the monitoring parameters is abnormal. Based on the calculated conditional anomaly probability matrix, the candidate operation and maintenance strategy set is divided into multiple strategy clusters using the spectral clustering algorithm; For each strategy cluster, calculate the average conditional anomaly probability among the coverage monitoring parameters of all candidate operation and maintenance strategies within the strategy cluster, and use it as the correlation strength characterization value of the strategy cluster. Based on the distribution of association strength characterization values, the strategy clusters are mapped to a preset discrete association strength interval.

5. The method for uploading operation and maintenance data of a power distribution terminal according to claim 1, characterized in that, The process of obtaining the dynamic weighting coefficients of the target monitoring parameters and each of the associated monitoring parameters includes: Based on the frequency of monitoring parameters being marked as key causes or core characterization parameters in the historical failure mode database, an inherent benchmark weight is formed, wherein the monitoring parameters are target monitoring parameters or associated monitoring parameters. The deviation between the statistical characteristics of the monitoring parameters within the current time window and the characteristics of the historical normal operation interval is calculated and used as a weight adjustment factor for the real-time operation layer. Based on the task objectives explicitly stated in the operation and maintenance task instructions, the corresponding target preference coefficients are matched from the preset mapping rule base; The inherent baseline weight, real-time operation layer weight adjustment factor, and target preference coefficient are fused together to calculate and output the dynamic weight coefficient of each monitoring parameter.

6. The method for uploading operation and maintenance data of a power distribution terminal according to claim 1, characterized in that, The process of determining the execution order of the queues based on the correlation strength intervals corresponding to each queue, and then sequentially selecting a predetermined number of operation and maintenance strategies from the queues to generate the final upload execution sequence includes: A dynamic scheduling mechanism based on sliding time windows is adopted to divide the available upload time window into multiple consecutive scheduling time slots; Calculate the initial priority score of each queue to be executed based on the association strength representation value of the policy cluster corresponding to each queue to be executed; The real-time status indicators of each queue to be executed are obtained, and the initial priority scores of each queue are dynamically adjusted according to the real-time status indicators to obtain the target priority scores of each queue to be executed. The real-time status indicators include the proportion of high-weight candidate operation and maintenance strategies in the queue to be executed, the average estimated execution time of candidate operation and maintenance strategies in the queue to be executed, and the sum of the real-time anomaly degree of the monitoring parameters covered by the candidate operation and maintenance strategies in the queue to be executed. At the beginning of each scheduling slot, all queues to be executed are sorted according to the dynamically adjusted target priority score; In descending order of priority, a preset number of operation and maintenance policies are selected from each queue to be executed and added to the upload execution sequence. The selection of the preset number of operation and maintenance policies follows these principles: The number of candidate operation and maintenance strategies selected from each queue to be executed is proportional to the priority score of the queue to be executed. Ensure that each queue to be executed is selected at least once within a scheduling cycle, unless the queue to be executed is empty; The total number of policies selected within a single scheduling slot shall not exceed the maximum number of policies that the scheduling slot can accommodate for execution; The selected operation and maintenance strategies are arranged according to the priority order of their respective queues and the sorting order within each queue to form the execution sequence fragments of the current scheduling slot; The execution sequence segments of multiple consecutive scheduling time slots are combined to form the final upload execution sequence.

7. A power distribution terminal operation and maintenance data uploading system, characterized in that, include: The acquisition module is configured to acquire a set of real-time operating status data of the power distribution terminal and a preset operation and maintenance task instruction, wherein the operation and maintenance task instruction contains at least one target monitoring parameter. The identification module is configured to identify at least one associated monitoring parameter that has a potential correlation with the at least one target monitoring parameter based on the real-time operating status data set and a preset historical fault mode database. The selection module is configured to select candidate operation and maintenance strategies from a preset operation and maintenance strategy library that simultaneously cover the at least one target monitoring parameter and the associated monitoring parameters corresponding to the at least one target monitoring parameter, as a candidate operation and maintenance strategy set. The partitioning module is configured to divide the candidate operation and maintenance strategy set into multiple strategy clusters based on the correlation strength between all monitoring parameters covered by each candidate operation and maintenance strategy in the candidate operation and maintenance strategy set. Each strategy cluster corresponds to a correlation strength interval, and each strategy cluster is stored in a different queue to be executed. The sorting module is configured to obtain the dynamic weight coefficients of the target monitoring parameters and each of the associated monitoring parameters, and to perform intra-cluster sorting of each candidate operation and maintenance strategy in the same queue to be executed according to the dynamic weight coefficients. The intra-cluster sorting rules include: Calculate the weight evaluation value of each candidate operation and maintenance strategy, wherein the weight evaluation value is the sum of the dynamic weight coefficients corresponding to all monitoring parameters covered by the candidate operation and maintenance strategy; Candidate operation and maintenance strategies whose weight evaluation values ​​exceed the preset threshold are marked as high-weight candidate operation and maintenance strategies and stored in the first priority sub-queue; the remaining candidate operation and maintenance strategies are marked as regular candidate operation and maintenance strategies and stored in the second priority sub-queue. Within the first priority sub-queue and the second priority sub-queue, a secondary sorting is performed according to the weight evaluation value from high to low; The generation module is configured to determine the execution order of the queues to be executed based on the relative strength of the association intervals corresponding to each queue, and to select a predetermined number of operation and maintenance strategies from the queues in order to generate the final upload execution sequence, thereby controlling the power distribution terminal to upload data in sequence.

8. An electronic device, characterized in that, include: At least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by a processor, it implements the method described in any one of claims 1 to 6.

Images