A power equipment operation and maintenance scheduling system and method based on multi-source state awareness

Abstract

This application provides a power equipment operation and maintenance scheduling system and method based on multi-source state awareness. It constructs a state transition graph between power equipment by leveraging state transition features and state association features among various state nodes, thereby obtaining confidence constraint features during power equipment operation. It acquires equipment operation data and extracts health scores and topological dependencies between each power equipment from this data. Based on the health scores and topological dependencies, it performs preventative identification of potential fault modes and operation and maintenance demand sequences for the power equipment, resulting in a preventative operation and maintenance task sequence for multi-equipment collaboration. The scheduling priority of operation and maintenance tasks in the power equipment is adjusted based on the confidence constraint features and the preventative operation and maintenance task sequence. Based on this scheme, predictive collaborative scheduling based on state graphs and health scores can be achieved.

Description

Technical Field

[0001] This application relates to the field of power operation and maintenance technology, and more specifically, to a power equipment operation and maintenance scheduling system and method based on multi-source state perception. Background Technology

[0002] Power equipment is the general term for all mechanical and electrical devices used to produce, convert, transmit, distribute, control, and protect electrical energy in the various stages of power generation, transmission, transformation, distribution, and consumption. It is the material basis for ensuring the safe, stable, and economical operation of the power system. Power equipment includes generators, transformers, circuit breakers, disconnect switches, power cables, relay protection devices, and various monitoring and control equipment.

[0003] Traditional power equipment operation and maintenance (O&M) models primarily rely on pre-set cycle preventative testing, periodic inspections, and reactive emergency repairs after faults occur. Their decision-making logic depends on operator experience and static procedures in equipment manuals. At the data application level, this model typically sets fixed thresholds for real-time data from a single sensor or monitoring point to trigger alarms when these thresholds are exceeded. The signals processed are discrete and lagging single-point information, lacking correlation analysis of the continuous evolution trend of the equipment's overall state. This response mechanism, based on isolated events and fragmented data, cannot proactively identify the gradual degradation process within equipment, and is even less able to predict systemic cascading risks caused by electrical coupling and functional dependencies between equipment. This leads to O&M work often being in a passive, isolated handling mode, resulting in low resource allocation efficiency and an inability to effectively prevent the occurrence and spread of complex faults. Therefore, how to achieve predictive collaborative scheduling based on state maps and health scores to improve the resource utilization rate of power grid O&M has become a challenge for the industry. Summary of the Invention

[0004] This application provides a power equipment operation and maintenance scheduling system and method based on multi-source state perception, which can realize predictive collaborative scheduling based on state map and health score, thereby improving the resource utilization rate of power grid operation and maintenance.

[0005] Firstly, this application provides a power equipment operation and maintenance scheduling method based on multi-source state awareness, including:

[0006] Collect multi-source status sensing data of power equipment during operation;

[0007] State transition features and state association features between various state nodes in power equipment are extracted from the multi-source state perception data. A state transition map between power equipment is constructed using the state transition features and the state association features, thereby obtaining the confidence constraint features in the operation of power equipment.

[0008] The system acquires equipment operation data of power equipment, extracts the health score of each power equipment and the topological dependency relationship between power equipment from the equipment operation data, and performs preventive identification of potential failure modes and operation and maintenance needs of power equipment based on each health score and the topological dependency relationship, thereby obtaining a preventive operation and maintenance task sequence for multi-equipment collaboration.

[0009] The scheduling priority of maintenance tasks in power equipment is adjusted based on the confidence constraint features and the preventive maintenance task sequence.

[0010] In some embodiments, extracting state transition features and state association features between various state nodes in a power device from the multi-source state sensing data specifically includes:

[0011] The multi-source state perception data is subjected to time-series alignment and standardization to obtain a multi-dimensional state sequence under a unified timestamp;

[0012] The health status of power equipment at different operating stages is identified from the multi-dimensional state sequence, thereby obtaining multiple state nodes;

[0013] Calculate the conditional transition probability and association strength between any two state nodes to obtain the state transition characteristics and state association characteristics between each state node in the power equipment.

[0014] In some embodiments, constructing a state transition map between power devices using the state transition features and the state association features, and then obtaining the confidence constraint features during the operation of the power devices, specifically includes:

[0015] An initial state transition graph is constructed using the state nodes of power equipment as graph nodes, the state transition features as the weights of directed edges, and the state association features as the attributes of edges.

[0016] Topological analysis and community detection are performed on the initial state transition graph to identify healthy state clusters with coupling relationships within and between each power device.

[0017] By statistically analyzing the transition stability index and confidence interval of each healthy cluster, the confidence constraint characteristics of power equipment operation are obtained.

[0018] In some embodiments, extracting the health scores of each power device and the topological dependencies between power devices from the device operation data specifically includes:

[0019] A health baseline model for equipment is established by using historical operation and maintenance records. The performance indicators in the equipment operation data are compared with the health baseline model to obtain the operating deviation of each power device.

[0020] The health score of each power device is determined based on the operational deviation and the dwell time of the corresponding power device at the current state node.

[0021] Based on the power grid wiring diagram, power supply path, and load dependency in the equipment operation data, the physical and electrical connection relationships between power equipment are analyzed for functional dependency, and then mapped to the topological dependency relationships between power equipment.

[0022] In some embodiments, the potential failure modes and maintenance requirements of power equipment are preventively identified based on each health score and the topological dependency, resulting in a preventive maintenance task sequence for multi-device collaboration, specifically including:

[0023] Based on various health scores, all electrical equipment is classified into health levels, and electrical equipment with health scores below a preset threshold is identified as target monitoring objects;

[0024] By performing dependency path backtracking analysis on the target monitoring object through the topological dependency relationship, upstream devices and downstream devices sharing the load that have a direct impact on the operating status of the target monitoring object are identified, and a set of associated devices is obtained.

[0025] Based on the state transition map, the evolution trend of the state of each device in the associated device set is used to predict the concurrent fault mode and the fault propagation sequence.

[0026] Based on the concurrent failure modes and the failure propagation sequence, a preventive operation and maintenance task sequence is determined that is optimized in both time and space.

[0027] In some embodiments, the feedback adjustment of the scheduling priority of maintenance tasks in power equipment based on the confidence constraint features and the preventive maintenance task sequence specifically includes:

[0028] For each maintenance task in the preventive maintenance task sequence, the scheduling priority of the maintenance task in the power equipment is quantified and scored according to the equipment health, the number of dependent equipment and the urgency of the predicted fault window, so as to obtain the preliminary priority of the maintenance task.

[0029] The confidence constraint features are used to perform confidence verification on the preliminary priority to obtain the scheduling priority of the operation and maintenance task, and then the scheduling priority of each operation and maintenance task is obtained.

[0030] In some embodiments, the multi-source state perception data includes equipment operating parameters, environmental monitoring indicators, and historical operation and maintenance records.

[0031] Secondly, this application provides a power equipment operation and maintenance scheduling system based on multi-source state awareness, comprising:

[0032] The data acquisition module is used to collect multi-source status sensing data of power equipment during operation;

[0033] The processing module is used to extract state transition features and state association features between various state nodes in the power equipment from the multi-source state perception data, construct a state transition map between the power equipment through the state transition features and the state association features, and then obtain the confidence constraint features in the operation of the power equipment.

[0034] The processing module is also used to acquire equipment operation data of power equipment, and then extract the health score of each power equipment and the topological dependency relationship between power equipment from the equipment operation data. Based on the health score and the topological dependency relationship, the module performs preventive identification of potential fault modes and operation and maintenance requirements of power equipment, and obtains a preventive operation and maintenance task sequence for multi-device collaboration.

[0035] The execution module is used to adjust the scheduling priority of maintenance tasks in power equipment based on the confidence constraint features and the preventive maintenance task sequence.

[0036] Thirdly, this application provides a computer device, which includes a memory and a processor. The memory is used to store a computer program, and the processor is used to call and run the computer program from the memory, so that the computer device executes the above-described power equipment operation and maintenance scheduling method based on multi-source state awareness.

[0037] Fourthly, this application provides a computer-readable storage medium storing instructions or code that, when executed on a computer, cause the computer to implement the aforementioned power equipment operation and maintenance scheduling method based on multi-source state awareness.

[0038] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects:

[0039] This application provides a power equipment operation and maintenance scheduling system and method based on multi-source state awareness. The system collects multi-source state awareness data of power equipment during operation; extracts state transition features and state association features between various state nodes in the power equipment from the multi-source state awareness data; constructs a state transition map between power equipment using the state transition features and state association features, thereby obtaining confidence constraint features during power equipment operation; acquires equipment operation data of the power equipment; extracts health scores of each power equipment and topological dependencies between power equipment from the equipment operation data; performs preventative identification of potential fault modes and operation and maintenance demand sequences of the power equipment based on the health scores and the topological dependencies, obtaining a preventative operation and maintenance task sequence for multi-equipment collaboration; and adjusts the scheduling priority of operation and maintenance tasks in the power equipment based on the confidence constraint features and the preventative operation and maintenance task sequence.

[0040] Therefore, in this application, the scheduling priority of maintenance tasks in power equipment is adjusted based on the confidence constraint features and the preventive maintenance task sequence. First, determining the confidence constraint features yields a quantitative confidence boundary for the equipment state evolution pattern. By performing stability analysis and probability interval estimation on the state transition map, the state transition paths, originally based on historical statistics and potentially uncertain, are transformed into constraints with clear confidence level assessments. Quality control and risk calibration are embedded within the data-driven prediction model, enabling automatic identification and priority adoption of fault warning paths with clear evolution patterns and reliable prediction results. This is equivalent to introducing a confidence filter in scheduling decisions, effectively filtering out low-reliability warnings caused by data noise, short-term fluctuations, or small samples. This ensures that the forecasting basis upon which the scheduling instructions are based is solid and reliable, avoiding the excessive consumption of operation and maintenance resources on responding to warnings with high uncertainty and the possibility of false alarms. This allows limited manpower, material resources, and time windows to be more accurately invested in truly high-risk, high-certainty preventive operations, improving the accuracy of resource allocation. Then, by determining the sequence of preventive operation and maintenance tasks, a collaborative intervention plan that takes into account both the health degradation of individual devices and the coupling relationship of system topology can be obtained. This achieves a leap from single-point warning to a scheduling strategy of joint prevention and control of systemic risks. It not only identifies risk points based on the health score of individual devices, but also performs dependency path backtracking and impact range analysis through topological dependencies, expanding isolated warning devices into a set of related devices, and predicting potential concurrent failure modes and propagation sequences within this set based on state transition graphs. Its effect lies in breaking the isolated handling mode in traditional operation and maintenance, prompting scheduling decisions to shift from passively responding to single equipment failures to proactively planning collaborative maintenance of equipment clusters. The generated operation and maintenance task sequence considers the order of fault propagation in the time dimension and coordinates geographical location and electrical connections in the spatial dimension. This allows a single power outage maintenance or the deployment of a working group to simultaneously resolve multiple potential faults with strong coupling relationships, greatly optimizing the batch arrangement and path planning of operation and maintenance tasks. It significantly reduces repetitive work and resource waiting losses caused by repeated power outages, cross-operations, or fault chain reactions, and compresses the total amount and duration of unnecessary operation and maintenance activities as a whole, improving the comprehensive utilization efficiency of resources in the time and spatial dimensions. In summary, based on the above scheme, predictive collaborative scheduling based on state graphs and health scores can be realized, thereby improving the resource utilization rate of power grid operation and maintenance. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 This is an exemplary flowchart of a power equipment operation and maintenance scheduling method based on multi-source state awareness, as shown in some embodiments of this application.

[0043] Figure 2 This is a flowchart illustrating the determination of a preventative maintenance task sequence according to some embodiments of this application;

[0044] Figure 3 This is a schematic diagram of the structure of a power equipment operation and maintenance scheduling system based on multi-source state awareness, as shown in some embodiments of this application.

[0045] Figure 4 This is a schematic diagram of the structure of a computer device that implements a power equipment operation and maintenance scheduling method based on multi-source state awareness, according to some embodiments of this application. Detailed Implementation

[0046] To better understand the technical solution of this application, the technical solution of this application will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0047] refer to Figure 1 The figure is an exemplary flowchart of a power equipment operation and maintenance scheduling method based on multi-source state awareness, according to some embodiments of this application. The power equipment operation and maintenance scheduling method based on multi-source state awareness mainly includes the following steps:

[0048] In step 101, multi-source state sensing data of the power equipment during operation are collected.

[0049] It should be noted that, in this application, the source state perception data includes equipment operating parameters, environmental monitoring indicators, and historical operation and maintenance records. Equipment operating parameters are real-time or periodic measurement data that characterize the electrical, mechanical, and thermodynamic performance of the power equipment itself; environmental monitoring indicators are external parameter data that describe the influence of the micro-environmental conditions in which the power equipment is located on its operating state; and historical operation and maintenance records are archival data used to record the process and results of past maintenance, repair, testing, and abnormal events of the equipment.

[0050] In practice, various sensors and monitoring terminals deployed on the power equipment and its operating environment periodically or triggeredly collect equipment operating parameters, such as current, voltage, temperature, vibration, and partial discharge signals. Simultaneously, environmental monitoring indicators, such as ambient temperature, humidity, air pressure, and pollution levels, are also collected. Furthermore, historical maintenance records throughout the equipment's entire lifecycle are retrieved from the production management system and asset database, including but not limited to maintenance reports, preventative test data, defect records, and handling results. After standardizing the format and aligning the timestamps of the equipment operating parameters, environmental monitoring indicators, and historical maintenance records, the multi-dimensional data sequence under a unified time reference is used as multi-source state perception data.

[0051] In step 102, state transition features and state association features between various state nodes in the power equipment are extracted from the multi-source state perception data. A state transition map between the power equipment is constructed using the state transition features and the state association features, thereby obtaining the confidence constraint features in the operation of the power equipment.

[0052] In some embodiments, extracting state transition features and state association features between various state nodes in a power device from the multi-source state sensing data can be achieved using the following steps:

[0053] The multi-source state perception data is subjected to time-series alignment and standardization to obtain a multi-dimensional state sequence under a unified timestamp;

[0054] The health status of power equipment at different operating stages is identified from the multi-dimensional state sequence, thereby obtaining multiple state nodes;

[0055] Calculate the conditional transition probability and association strength between any two state nodes to obtain the state transition characteristics and state association characteristics between each state node in the power equipment.

[0056] It should be noted that, in this application, the multi-dimensional state sequence is a set of equipment state observation data composed of multiple state-aware parameters arranged chronologically within a unified time frame; the health state is a state category reflecting the overall performance of power equipment under a specified operating stage; the state node is a discretized and symbolic identifier used to represent a specified health state in an abstract model; the conditional transition probability is a probability value that quantifies the likelihood of the power equipment transitioning to another specified state node under given conditions at the current state node; the correlation strength is a measure of the closeness of the interdependence or common change relationship between two different state nodes in terms of their occurrence conditions or performance characteristics; the state transition feature is a feature quantity used to characterize the evolution trend of the health state of power equipment over time; and the state correlation feature is a feature quantity used to characterize the intrinsic relationship between different health states.

[0057] In specific implementation, firstly, multi-source state sensing data from different sensors, monitoring systems, and databases are received. These multi-source data have different acquisition periods and initial times. A time base unification algorithm is used, for example, using the system's master clock as a reference, to attach a time tag accurate to milliseconds or seconds to each data record. For data with different acquisition frequencies, interpolation methods are used to supplement low-frequency data to a unified, higher time resolution. Parameters with significant numerical differences (such as voltage values ​​in kilovolts and temperature in degrees Celsius) are standardized, for example, using a min-max normalization method to map all parameter values ​​to the [0,1] interval, eliminating the influence of dimensions. All processed parameter observations at each unified time point are combined into a multi-dimensional vector, and these vectors are arranged in chronological order. The ordered set of vectors is used as a multi-dimensional state sequence. Then, based on the aforementioned multi-dimensional state sequence, unsupervised clustering analysis is used to automatically identify the typical health status of the equipment. Specifically, representative historical time periods are extracted from the sequence, and clustering algorithms (such as k-means clustering) are used to group the multi-dimensional vectors at these time points. The algorithm iterates based on the similarity of vectors in multidimensional space (e.g., Euclidean distance), grouping observations with high similarity into the same cluster. Data points within each cluster exhibit concentration and stability across various dimensions, representing a similar operating condition for the equipment, i.e., a healthy state. For example, the clustering results may form three main clusters: one cluster corresponding to a "healthy state" where all parameters are within the normal range; one cluster corresponding to a "pay attention state" where some parameters (e.g., temperature) are slightly elevated but still within acceptable limits; and one cluster corresponding to a "warning state" where key parameters are close to or exceed alarm thresholds. Each identified cluster is defined as an abstract state node, and the set of state nodes and their cluster centers (representing typical parameter combinations for that state) are considered as multiple state nodes. Finally, based on the multidimensional state sequence and the identified state nodes, a label sequence of equipment state changes over time is constructed, where each time point corresponds to its respective state node label. The state transitions between all adjacent time points in the state label sequence are statistically analyzed. For any two state nodes A and B, the conditional transition probability is calculated as follows: Count the number of times the state changes to B in the next time step given that the current state is A; then divide this number by the total number of times the current state is A. The resulting ratio is the conditional transition probability of transitioning from state node A to state node B. The pairwise conditional transition probabilities of all state nodes constitute the state transition probability matrix, which is the state transition feature. Simultaneously, to uncover non-temporal correlations, the correlation strength between any two state nodes is calculated. One approach is to calculate the correlation between the time periods when the device is in state node A and state node B over a long time scale, either in terms of occurrence time or environmental conditions. For example, calculate the correlation coefficient of the parameter vectors representing these two states throughout the entire sequence.The absolute value of this correlation coefficient represents the strength of the association between state nodes A and B. The correlation strength values ​​between all pairs of state nodes are aggregated, and the aggregated result is used as the state association feature.

[0058] In some embodiments, constructing a state transition map between power devices using the state transition features and the state association features, and then obtaining the confidence constraint features during the operation of the power devices, can be achieved through the following steps:

[0059] An initial state transition graph is constructed using the state nodes of power equipment as graph nodes, the state transition features as the weights of directed edges, and the state association features as the attributes of edges.

[0060] Topological analysis and community detection are performed on the initial state transition graph to identify healthy state clusters with coupling relationships within and between each power device.

[0061] By statistically analyzing the transition stability index and confidence interval of each healthy cluster, the confidence constraint characteristics of power equipment operation are obtained.

[0062] It should be noted that, in this application, a graph node is an abstract element used to represent a basic entity or state in a graphical network model; the weight of a directed edge is a numerical value used in a graphical network model to quantify the strength or probability of a directed connection from one node to another; the attributes of an edge are additional information, besides the weight, used in a graphical network model to describe and supplement the connection relationship (edge) between two nodes; the initial state transition graph is a network graph model that has not undergone in-depth analysis and processing, used for preliminary visualization and structured representation of all health states (state nodes) of power equipment and their mutual transitions and relationships; and the health state cluster is used to refer to the state transition graph. In the spectrum, a subgraph or set of nodes identified by a community detection algorithm consists of multiple closely related state nodes in terms of state evolution or feature behavior; the transition stability index is a quantitative measure used to measure the volatility or predictability of the transition relationships between state nodes within a healthy state cluster; the confidence interval is used statistically to express a numerical range in which the estimated value of an unknown parameter (such as the true value of the transition probability) may fall, and this range is associated with a specified confidence level; the confidence constraint feature is a set of features used to provide reliability boundaries and constraints for prediction and decision-making based on the state transition graph, specifically composed of the stability index and probability confidence interval of each healthy state cluster.

[0063] In practice, firstly, each identified state node is mapped to a graph node. Based on the state transition features (i.e., the state transition probability matrix), connections are established between any two graph nodes. If the conditional transition probability from state node A to state node B is greater than zero, a directed edge is created from graph node A to graph node B, and the specific value of the conditional transition probability is assigned as the weight of this directed edge. Simultaneously, the association strength value between state nodes A and B, extracted from the state association feature matrix, is added as an attribute to the directed edge. By traversing all state node pairs and creating nodes and directed edges according to the above rules, a complete network structure containing all nodes, weighted directed edges, and edge attributes is ultimately formed as the initial state transition graph. Then, topological analysis is performed on the initial state transition graph to calculate basic indicators such as the in-degree and out-degree of each graph node, and to identify critical paths and loops in the graph. Finally, a community detection algorithm, such as the Louvain algorithm based on modularity optimization, is applied to partition the graph. The algorithm iteratively groups nodes, aiming to maximize the ratio between the connection density (sum of edge weights) within a community and the connection sparsity between communities. During the calculation, not only the weights of directed edges are considered, but the association strength of edges may also be taken into account. After the algorithm completes, the graph will be divided into several node groups. Within each node group, the state nodes exhibit high state transition probabilities or strong associations, indicating strong coupling in evolution or characteristics. These states may originate from different degradation stages of the same device, or from similar states affected by common factors (such as the same environmental stress) from different devices. Each node group identified by the algorithm is defined as a healthy state cluster. Finally, statistical analysis is performed on each healthy state cluster; the variance of the transition probabilities of all directed edges within the cluster is calculated. The smaller the variance, the stronger the transition relationship between states. The more stable and predictable the data, the more likely the reciprocal (or normalized) of the variance is used as the transition stability index for the cluster. The confidence interval for the transition probability of a critical transition path within the cluster (e.g., from "attention state" to "warning state") is calculated. Specifically, based on the number of successful transitions and the total number of transition attempts observed in historical data, a binomial distribution or Bayesian estimation method is used to calculate the upper and lower bounds where the true value of the transition probability might fall at a set confidence level (e.g., 95%). These two boundary values ​​constitute the confidence interval for the path's transition probability. The above confidence interval calculation is performed for all critical paths within a cluster. The set of transition stability indices for all healthy clusters, and the set of confidence intervals for the transition probabilities of critical paths within each cluster, are merged and organized into a structured feature data set. This dataset serves as the confidence constraint feature for power equipment operation.

[0064] In step 103, the equipment operation data of the power equipment is obtained, and then the health score of each power equipment and the topological dependency relationship between the power equipment are extracted from the equipment operation data. Based on the health score and the topological dependency relationship, the potential failure modes and operation and maintenance requirements of the power equipment are preventively identified, and a preventive operation and maintenance task sequence for multi-equipment collaboration is obtained.

[0065] In some embodiments, acquiring equipment operation data of power equipment can be achieved by: periodically or event-triggeredly reading the latest operating parameters of the power equipment by calling the application programming interfaces of real-time monitoring systems, data acquisition and monitoring control systems, and various online monitoring devices deployed on the equipment itself. These operating parameters include, but are not limited to, electrical quantities (such as three-phase current, voltage, power, and power factor), mechanical quantities (such as vibration amplitude, noise decibels, and oil level), thermal quantities (such as winding hot spot temperature, casing temperature, and ambient temperature), and specified monitoring signals (such as partial discharge pulse count and gas content). Simultaneously, the current on / off status of the equipment, protection signals, control commands, and other switch quantity information are also acquired. All heterogeneous data from different data sources reflecting the real-time operating conditions of the equipment are aggregated to form a data snapshot with a timestamp close to the current moment, covering multi-dimensional operating characteristics of the equipment. The latest data snapshot, after basic verification (such as removing obvious outliers and determining whether communication is interrupted), is used as the equipment operation data. It should be noted that in this application, the equipment operation data is a set of the latest observation and measurement data used to reflect the current and recent actual working status, performance parameters, and operating environment of the power equipment.

[0066] In some embodiments, extracting the health scores of each power device and the topological dependencies between power devices from the device operation data can be achieved using the following steps:

[0067] A health baseline model for equipment is established by using historical operation and maintenance records. The performance indicators in the equipment operation data are compared with the health baseline model to obtain the operating deviation of each power device.

[0068] The health score of each power device is determined based on the operational deviation and the dwell time of the corresponding power device at the current state node.

[0069] Based on the power grid wiring diagram, power supply path, and load dependency in the equipment operation data, the physical and electrical connection relationships between power equipment are analyzed for functional dependency, and then mapped to the topological dependency relationships between power equipment.

[0070] It should be noted that in this application, the equipment health benchmark model is a reference standard or mathematical model describing the range that various key performance indicators of power equipment should fall within under known healthy or normal operating conditions; performance indicators are core quantifiable parameters selected from equipment operating data that effectively reflect the specified functions and health status of the equipment; operating deviation is a single numerical value used to quantify the degree to which the current actual performance indicator value of the power equipment deviates from the reference value defined by its corresponding health benchmark model; the current state node is used to represent the current health state category of the power equipment as determined based on the latest multi-dimensional state sequence; and the dwell time is used to represent... The total time that a power device has remained in its current state from the moment it enters the current state node until the current assessment time is considered. The health score is a single quantitative score used to comprehensively characterize the overall health level of a single power device at the current moment; generally, the lower the score, the higher the health risk. The physical and electrical connection relationship describes the direct connection structure between power devices formed by actual physical media such as conductors, lines, and buses, which can transmit electrical energy or signals. The topology dependency relationship is a matrix-based or graph-based description used to quantitatively express the degree of mutual influence between power devices due to functional dependencies under a specified power grid topology and operating mode.

[0071] In practice, the process begins with mining historical maintenance records to identify all historical time periods marked as "equipment healthy," "accepted after major overhaul," or "preventive test passed," indicating a good state. Multi-dimensional state sequences of stable equipment operation within these time periods are extracted. For each selected performance indicator, its typical value range under this health condition is statistically analyzed, such as calculating its mean and standard deviation. For indicators under different load rates or environmental conditions, segmented benchmarks or multi-dimensional benchmark surfaces can be established. The set of reference values ​​or allowable fluctuation ranges for each type of equipment's performance indicators serves as the equipment health benchmark model. Current equipment operating data is read, and the current measured values ​​of each performance indicator corresponding to the benchmark model are extracted. The measured value of each indicator is compared with its reference value in the benchmark model, specifically by calculating the absolute value of the relative deviation or the standardized Mahalanobis distance. The deviation calculation results of each performance indicator are then combined into an overall metric, for example, using a weighted average method, and this comprehensive metric is taken as the operating deviation of the power equipment.

[0072] Then, in practical implementation, the current state node of each power device is determined. The cumulative time since the device entered that state node is obtained as the dwell time. The health score calculation comprehensively considers two factors: operational deviation and dwell time. A base health score is set, for example, a maximum score of 100 points. The greater the operational deviation, the further the current state deviates from the health benchmark, and more points should be deducted from the base score. At the same time, for certain specified deteriorating state nodes, the longer the device stays in them, the deeper the cumulative damage or deterioration may be, so corresponding points also need to be deducted based on the dwell time. Specific deduction rules can be defined through functional relationships, for example: Health Score = Base Score - a × Operational Deviation - b × Dwell Time Coefficient in Specified State, where a and b are weighting coefficients determined based on expert experience or historical data fitting. This calculation process generates a specific value for each device, which is used as the health score for that power device. Finally, based on the real-time grid wiring diagram, power flow distribution data, and load allocation information obtained from the device operation data or the grid energy management system, the direct physical connection relationships between all devices are analyzed from the wiring diagram, such as the connection between transformers and buses, and the connection between circuit breakers and lines, to construct the physical connection topology of the power grid. Based on the current power supply path (i.e., power flow direction and path) and load allocation, functional dependencies are analyzed. For example, the power supply of a distribution line depends on the transformers and buses of its upstream substation; the power supply of an important load depends on all switching equipment and lines on its power supply link. This dependency includes direct power supply dependency, protection coordination dependency, and the intensity of electrical influence determined by electrical distance. This functional dependency is quantified, for example, by defining a dependency coefficient between one device and another. This coefficient can be calculated comprehensively based on factors such as the fault impact range, load importance level, and the tightness of electrical coupling. By iterating through all device pairs and calculating their dependency coefficients, a square matrix is ​​formed where rows and columns are devices and elements are dependency coefficients. This dependency coefficient matrix is ​​used as the topological dependency between power devices.

[0073] In some embodiments, based on each health score and the topological dependencies, potential failure modes and maintenance requirements of power equipment are preventively identified to obtain a preventive maintenance task sequence for multi-device collaboration, as referenced. Figure 2 The diagram is a flowchart illustrating the determination of a preventative maintenance task sequence in some embodiments of this application. In this embodiment, the determination of the preventative maintenance task sequence can be achieved using the following steps:

[0074] In step 1031, all electrical equipment is classified into health levels based on each health score, and electrical equipment with a health score lower than a preset threshold is identified as the target monitoring object.

[0075] In step 1032, the target monitoring object is subjected to dependency path backtracking analysis through the topological dependency relationship to identify upstream devices and downstream devices sharing the load that have a direct impact on the operating status of the target monitoring object, thereby obtaining a set of associated devices;

[0076] In step 1033, based on the state transition map, the evolution trend of the state of each device in the associated device set is predicted to obtain concurrent fault modes and fault propagation timelines.

[0077] In step 1034, a preventive operation and maintenance task sequence that is optimized in time and space is determined based on the concurrent failure mode and the failure propagation timing.

[0078] It should be noted that, in this application, the target monitoring object refers to an individual power equipment selected through health level classification that currently has a substandard health level or a high risk of deterioration and requires key analysis and treatment; upstream equipment refers to power equipment located on the power supply side of the target monitoring object in the power supply path or functional logic, providing power or control signal support to the target monitoring object; downstream equipment refers to power equipment located on the load side of the target monitoring object in the power supply path or functional logic, obtaining power from the target monitoring object or being directly affected by its operating status; the associated equipment set includes the target monitoring object itself, as well as those connected through dependent paths. Analysis identifies a group of devices, including upstream and downstream devices, that are strongly correlated with their operating status; concurrent failure modes describe a combination of multiple device failures that may occur simultaneously or in successive short periods due to common causes, cascading effects, or strong coupling relationships within a set of related devices; failure propagation sequence describes the estimated order and time interval in which a failure, starting from one device, triggers the next related device failure in a concurrent failure mode; preventative maintenance task sequence is an ordered list of plans used to guide the sequential, spatiotemporally coordinated inspection, maintenance, or repair work on a group of related power devices within a specified future time period.

[0079] In practice, the first step is to obtain the health scores of all power equipment. Several consecutive score ranges are defined, each corresponding to a health level. For example, scores between 90 and 100 are classified as "healthy," between 70 and 90 as "attention," and below 70 as "warning." Each device is then assigned to its corresponding health level based on its score. A specific preset threshold, such as 75, is set based on this classification. The system automatically iterates through all devices, filtering out those with health scores below the preset threshold. These filtered devices, due to their low scores, indicate that their current state has significantly deviated from the health benchmark or is in an unfavorable trend, and therefore need to be prioritized for further in-depth analysis. The list of all devices with scores below the preset threshold is then used as target monitoring objects. Next, the list of target monitoring objects and the topological dependencies between power equipment are read. For each target monitoring object in the list, such as a low-scoring outgoing circuit breaker, dependency path backtracking analysis is performed. The analysis process is divided into two directions: First, tracing back upstream, based on topological dependencies, identifying upstream equipment that directly supplies power to the circuit breaker, such as busbars and main transformers supplying power to the busbars, until tracing back to the power source point, and identifying these devices as upstream equipment that directly affects the target; Second, probing downstream, identifying lines directly supplied by the circuit breaker and the loads carried by those lines, identifying those load devices with high importance or strong coupling protection relationships with the circuit breaker as downstream devices sharing the load, merging and deduplicating the target monitoring object itself, all identified key upstream devices, and key downstream devices to form a device list, and using this device list as the set of associated devices.

[0080] Then, in the specific implementation, the current state node of each device in the associated device set is obtained; the state transition graph is consulted, and for each device, the possible future states it might evolve to from its current state node along directed edge weights (transition probabilities) are analyzed. Particular attention is paid to transition paths that ultimately point to a "fault" or "critical warning" state. By analyzing the magnitude of the transition probabilities and the required time characteristics on these paths, the possible fault types and their time windows for each device are predicted. Next, the entire set is comprehensively examined: if the predicted fault time windows of multiple devices in the set overlap, or if the predicted fault pattern of one device significantly increases the probability of occurrence of another device through topological dependencies, then these spatiotemporally correlated predicted fault combinations are identified as a concurrent fault mode. Simultaneously, based on the direction and strength of dependencies between devices and the time estimation of state transition paths, the estimated sequence and delay time of the impact of a device failure propagating to other devices along the dependency chain are inferred, forming a fault propagation sequence. These predicted fault combinations and propagation sequence relationships are used as concurrent fault modes and fault propagation sequences. Finally, using the predicted concurrent fault modes and fault propagation sequences as core inputs, overall planning of operation and maintenance tasks is carried out. To address each concurrent fault mode, the devices requiring intervention are transformed into specific preventive operation and maintenance tasks, such as "performing insulation testing on device A" or "replacing worn parts of device B". Based on the fault propagation sequence, the logical sequence of these tasks is determined, prioritizing the source devices that may trigger cascading failures. On the basis of the logical sequence, further consideration is given to the coordinated optimization of time and space: in terms of time, tasks for devices in the same substation or adjacent geographical locations are arranged on the same working day as much as possible to reduce personnel travel time; the estimated time of each task is evaluated, and the task schedule is reasonably arranged in combination with constraints such as power grid outage windows. Spatially, optimize the patrol or operation routes of the working group; arrange all the identified tasks according to the optimized execution time, location, and sequence to form a detailed, ordered list with timestamps and location information, and use this ordered task list as a sequence of preventive maintenance tasks.

[0081] In step 104, the scheduling priority of maintenance tasks in power equipment is adjusted based on the confidence constraint features and the preventive maintenance task sequence.

[0082] In some embodiments, the feedback adjustment of the scheduling priority of maintenance tasks in power equipment based on the confidence constraint characteristics and the preventive maintenance task sequence can be achieved by the following steps:

[0083] For each maintenance task in the preventive maintenance task sequence, the scheduling priority of the maintenance task in the power equipment is quantified and scored according to the equipment health, the number of dependent equipment and the urgency of the predicted fault window, so as to obtain the preliminary priority of the maintenance task.

[0084] The confidence constraint features are used to perform confidence verification on the preliminary priority to obtain the scheduling priority of the operation and maintenance task, and then the scheduling priority of each operation and maintenance task is obtained.

[0085] It should be noted that, in this application, "device health" refers to the health score of the specified power equipment targeted by the maintenance task at the time of task generation; "number of dependent devices" refers to the number of other power equipment whose operating status directly affects the power equipment targeted by the maintenance task and which need to maintain normal function in the topology dependency relationship; "predicted fault window urgency" is used to quantify the proximity or time urgency of the time window when the device may fail, predicted based on the state transition graph, to the current time; "preliminary priority" is used to represent the preliminary ranking value of the execution order of maintenance tasks after considering only the direct factors such as the device's own state, fault urgency, and scope of impact, without verification by other global constraints; and "scheduling priority" is the ranking value used to guide the execution order of maintenance tasks in the actual allocation of maintenance resources.

[0086] In practical implementation, firstly, the preventive maintenance task sequence is read, and each maintenance task is traversed. For each task, the health score of the target power equipment, the number of downstream dependent equipment calculated based on the topology dependency matrix, and the start time of the predicted fault time window for the equipment obtained from the fault prediction results are obtained. The urgency of the predicted fault window is calculated, for example, it can be represented by the reciprocal of the difference between the fault window start time and the current time (in hours or days). The smaller the difference, the higher the urgency score. A priority base score is set, and different weight coefficients are assigned to the three indicators: equipment health (taking the reciprocal or negative mapping, i.e., the lower the health, the higher the score), the number of dependent equipment, and the urgency of the predicted fault window. Through a linear or non-linear weighting function, the converted values ​​of these three indicators are combined with the weights for comprehensive calculation. For example: preliminary priority score = base score + w1 * (health coefficient) + w2 * (number of dependent equipment) + w3 * (urgency score), where w1, w2, and w3 are weights. For each task in the sequence, calculate such a value and use the calculated value of all tasks as the initial priority of the operation and maintenance task; then, obtain the current state node of the power equipment corresponding to each operation and maintenance task, as well as the state transition path on which the fault is predicted; query the confidence constraint features to find the transition stability index of the health state cluster where the equipment is located, and the transition probability confidence interval corresponding to the predicted transition path. Verification and judgment are performed: If the transfer stability index of the cluster where the device is located is lower than the preset stability threshold, or the transfer probability confidence interval of the predicted path is too wide, it indicates that the reliability of the predicted path is low. The core scoring criterion of the urgency of the fault window predicted based on this path has a large uncertainty. In this case, the system will adjust the initial priority score of the task downward, for example, by multiplying it by a reliability discount factor less than 1, in order to reduce the risk of resource misallocation due to inaccurate prediction. Conversely, if the stability index is high and the confidence interval is narrow, the initial priority score can be maintained or slightly increased. After the above-mentioned confidence-based verification and adjustment of the initial priority of all tasks, a new set of scores reflecting the reliability of the prediction is generated. This set of final adjusted scores is used as the scheduling priority of the operation and maintenance tasks.

[0087] Furthermore, in another aspect of this application, in some embodiments, this application provides a power equipment operation and maintenance scheduling system based on multi-source state awareness, referencing... Figure 3 The figure is a schematic diagram of the structure of a power equipment operation and maintenance scheduling system based on multi-source state awareness, according to some embodiments of this application. The power equipment operation and maintenance scheduling system based on multi-source state awareness includes: a data acquisition module 201, a processing module 202, and an execution module 203, which are described below:

[0088] The acquisition module 201 in this application is mainly used to acquire multi-source state sensing data of power equipment during operation.

[0089] Processing module 202 in this application is used to extract state transition features and state association features between various state nodes in the power equipment from the multi-source state perception data, construct a state transition map between the power equipment through the state transition features and the state association features, and then obtain the confidence constraint features in the operation of the power equipment.

[0090] It should be noted that the processing module 202 is also used to acquire the equipment operation data of the power equipment, and then extract the health score of each power equipment and the topological dependency relationship between the power equipment from the equipment operation data. Based on the health score and the topological dependency relationship, the potential failure modes and operation and maintenance requirements of the power equipment are preventively identified to obtain a preventive operation and maintenance task sequence for multi-device collaboration.

[0091] The execution module 203 in this application is mainly used to adjust the scheduling priority of maintenance tasks in power equipment based on the confidence constraint features and the preventive maintenance task sequence.

[0092] The foregoing has detailed examples of a power equipment operation and maintenance scheduling system and method based on multi-source state awareness provided in the embodiments of this application. It is understood that the corresponding apparatus, in order to achieve the above functions, includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should readily recognize that, in conjunction with the units and algorithm steps of the various examples described in the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed by hardware or by computer software driving hardware depends on the specified application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specified application, but such implementation should not be considered beyond the scope of this application.

[0093] In some embodiments, this application also provides a computer device, the computer device including a memory and a processor, the memory for storing a computer program, and the processor for calling and running the computer program from the memory, so that the computer device executes the above-described power equipment operation and maintenance scheduling method based on multi-source state awareness.

[0094] In some embodiments, reference Figure 4The dashed lines in the figure indicate that the unit or module is optional. This figure is a structural schematic diagram of a computer device implementing a power equipment operation and maintenance scheduling method based on multi-source state awareness, according to an embodiment of this application. The power equipment operation and maintenance scheduling method based on multi-source state awareness described in the above embodiments can be achieved through… Figure 4 The computer device shown is used to implement this, and the computer device includes at least one processor 301, a memory 302 and at least one communication unit 305. The computer device may be a terminal device, a server or a chip.

[0095] Processor 301 can be a general-purpose processor or a special-purpose processor. For example, processor 301 can be a central processing unit (CPU), which can be used to control computer devices, execute software programs, and process data from software programs. The computer device may also include a communication unit 305 for inputting (receiving) and outputting (transmitting) signals.

[0096] For example, the computer device may be a chip, and the communication unit 305 may be the input and / or output circuit of the chip, or the communication unit 305 may be the communication interface of the chip, which may be a component of a terminal device, network device or other device.

[0097] For example, the computer device may be a terminal device or a server, and the communication unit 305 may be a transceiver of the terminal device or the server, or the communication unit 305 may be a transceiver circuit of the terminal device or the server.

[0098] The computer device may include one or more memories 302 storing a program 304. The program 304 can be executed by a processor 301 to generate instructions 303, causing the processor 301 to execute the method described in the above method embodiments according to the instructions 303. Optionally, the memory 302 may also store data (such as a target audit model). Optionally, the processor 301 may also read data stored in the memory 302, which may be stored at the same storage address as the program 304, or it may be stored at a different storage address than the program 304.

[0099] The processor 301 and memory 302 can be configured separately or integrated together, for example, integrated on the system on chip (SOC) of the terminal device.

[0100] It should be understood that each step of the above method embodiment can be completed by hardware logic circuits or software instructions in the processor 301. The processor 301 can be a CPU, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, such as discrete gate, transistor logic devices, or discrete hardware components.

[0101] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0102] For example, in some embodiments, this application also provides a computer-readable storage medium storing instructions or code that, when executed on a computer, cause the computer to implement the above-described power equipment operation and maintenance scheduling method based on multi-source state awareness.

[0103] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0104] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A power equipment operation and maintenance scheduling method based on multi-source state perception, characterized in that, Includes the following steps: Collect multi-source status sensing data of power equipment during operation; State transition features and state association features between various state nodes in power equipment are extracted from the multi-source state perception data. A state transition map between power equipment is constructed using the state transition features and the state association features, thereby obtaining the confidence constraint features in the operation of power equipment. The system acquires equipment operation data of power equipment, extracts the health score of each power equipment and the topological dependency relationship between power equipment from the equipment operation data, and performs preventive identification of potential failure modes and operation and maintenance needs of power equipment based on each health score and the topological dependency relationship, thereby obtaining a preventive operation and maintenance task sequence for multi-equipment collaboration. The scheduling priority of maintenance tasks in power equipment is adjusted based on the confidence constraint features and the preventive maintenance task sequence.

2. The method as described in claim 1, characterized in that, Extracting state transition features and state association features between various state nodes in power equipment from the multi-source state perception data specifically includes: The multi-source state perception data is subjected to time-series alignment and standardization to obtain a multi-dimensional state sequence under a unified timestamp; The health status of power equipment at different operating stages is identified from the multi-dimensional state sequence, thereby obtaining multiple state nodes; Calculate the conditional transition probability and association strength between any two state nodes to obtain the state transition characteristics and state association characteristics between each state node in the power equipment.

3. The method as described in claim 1, characterized in that, By constructing a state transition map between power equipment using the state transition features and the state association features, the confidence constraint features during the operation of power equipment are obtained, specifically including: An initial state transition graph is constructed using the state nodes of power equipment as graph nodes, the state transition features as the weights of directed edges, and the state association features as the attributes of edges. Topological analysis and community detection are performed on the initial state transition graph to identify healthy state clusters with coupling relationships within and between each power device. By statistically analyzing the transition stability index and confidence interval of each healthy cluster, the confidence constraint characteristics of power equipment operation are obtained.

4. The method as described in claim 1, characterized in that, Extracting the health scores of each power device and the topological dependencies between power devices from the equipment operation data specifically includes: A health baseline model for equipment is established by using historical operation and maintenance records. The performance indicators in the equipment operation data are compared with the health baseline model to obtain the operating deviation of each power device. The health score of each power device is determined based on the operational deviation and the dwell time of the corresponding power device at the current state node. Based on the power grid wiring diagram, power supply path, and load dependency in the equipment operation data, the physical and electrical connection relationships between power equipment are analyzed for functional dependency, and then mapped to the topological dependency relationships between power equipment.

5. The method as described in claim 1, characterized in that, Based on the health scores and the aforementioned topological dependencies, the potential failure modes and maintenance requirements of power equipment are preventively identified, resulting in a preventative maintenance task sequence for multi-device collaboration, specifically including: Based on various health scores, all electrical equipment is classified into health levels, and electrical equipment with health scores below a preset threshold is identified as target monitoring objects; By performing dependency path backtracking analysis on the target monitoring object through the topological dependency relationship, upstream devices and downstream devices sharing the load that have a direct impact on the operating status of the target monitoring object are identified, and a set of associated devices is obtained. Based on the state transition map, the evolution trend of the state of each device in the associated device set is used to predict the concurrent fault mode and the fault propagation sequence. Based on the concurrent failure modes and the failure propagation sequence, a preventive operation and maintenance task sequence is determined that is optimized in both time and space.

6. The method as described in claim 1, characterized in that, The feedback adjustment of the scheduling priority of maintenance tasks in power equipment based on the confidence constraint features and the preventive maintenance task sequence specifically includes: For each maintenance task in the preventive maintenance task sequence, the scheduling priority of the maintenance task in the power equipment is quantified and scored according to the equipment health, the number of dependent equipment and the urgency of the predicted fault window, so as to obtain the preliminary priority of the maintenance task. The confidence constraint features are used to perform confidence verification on the preliminary priority to obtain the scheduling priority of the operation and maintenance task, and then the scheduling priority of each operation and maintenance task is obtained.

7. The method as described in claim 1, characterized in that, The multi-source status perception data includes equipment operating parameters, environmental monitoring indicators, and historical operation and maintenance records.

8. A power equipment operation and maintenance scheduling system based on multi-source state perception, characterized in that, include: The data acquisition module is used to collect multi-source status sensing data of power equipment during operation; The processing module is used to extract state transition features and state association features between various state nodes in the power equipment from the multi-source state perception data, construct a state transition map between the power equipment through the state transition features and the state association features, and then obtain the confidence constraint features in the operation of the power equipment. The processing module is also used to acquire equipment operation data of power equipment, and then extract the health score of each power equipment and the topological dependency relationship between power equipment from the equipment operation data. Based on the health score and the topological dependency relationship, the module performs preventive identification of potential fault modes and operation and maintenance requirements of power equipment, and obtains a preventive operation and maintenance task sequence for multi-device collaboration. The execution module is used to adjust the scheduling priority of maintenance tasks in power equipment based on the confidence constraint features and the preventive maintenance task sequence.

9. A computer device, characterized in that, The computer device includes a memory and a processor. The memory is used to store computer programs, and the processor is used to call and run the computer programs from the memory, so that the computer device executes the power equipment operation and maintenance scheduling method based on multi-source state awareness as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions or code that, when executed on a computer, cause the computer to implement the power equipment operation and maintenance scheduling method based on multi-source state awareness as described in any one of claims 1 to 7.

Images