A method and system for processing power equipment operation tasks

Abstract

This invention discloses a method and system for processing power equipment operation tasks, applied in the field of operation task processing technology. The method includes: performing attention analysis on the current monitoring data of all modes of the target power equipment based on modal association operations to obtain the current multimodal dependency matrix; obtaining the current equipment status data based on the current multimodal dependency matrix and multimodal regression operations; obtaining the current operation data based on the feedback signal of the current equipment status data and the equipment operation relationship graph; constructing equipment composite operation constraints based on the equipment node data, personnel node data, and node relationship data of the operation relationship graph; generating operation task instructions based on the current status data and the current operation data; and responding to the operation task instructions based on the equipment composite operation constraints to realize the operation task processing of the target power equipment. This invention improves the accuracy of power equipment operation task processing.

Description

Technical Field

[0001] This invention relates to the field of task processing technology, and in particular to a method and system for processing tasks of power equipment. Background Technology

[0002] In existing equipment maintenance work schemes, the processing of tasks for power equipment typically involves independently analyzing various monitoring data of the equipment to determine the operating status of each individual monitoring dimension, while relying on human experience to schedule personnel and assign tasks. This approach has several drawbacks: First, it fails to explore the inherent dependencies between different monitoring modalities, making it impossible to fully utilize the correlation information of multimodal data, resulting in low accuracy of the obtained equipment status data and difficulty in detecting potential equipment anomalies; second, it lacks relevant operational constraints and relies entirely on human experience for task assignment, which easily leads to low task matching, thereby reducing the accuracy of equipment maintenance. Summary of the Invention

[0003] This invention provides a method and system for processing power equipment operation tasks to solve the technical problem of low accuracy in existing equipment operation and maintenance.

[0004] To address the aforementioned technical problems, embodiments of the present invention provide a method for processing power equipment operation tasks, including: Based on the attention classification model, the modal association operation performs attention analysis on the current monitoring data of all modes of the target power equipment to obtain the current multimodal dependency matrix of the target power equipment. Based on the current multimodal dependency matrix and the multimodal regression operation of the attention classification model, the current state data of the target power equipment is obtained; Based on the feedback signal of the current status data and the operation relationship map corresponding to the target power equipment, the current operation data of the target power equipment is obtained; Based on the equipment node data, personnel node data, and node relationship data of the operation relationship graph, a composite operation constraint for the target power equipment is constructed. Based on the current status data and the current job data, generate job task instructions; Based on the composite operation constraints, the system responds to the operation task instructions to realize the operation task processing of the target power equipment according to the response results of the operation task instructions.

[0005] As one preferred embodiment, the modal association operation based on the attention classification model performs attention analysis on the current monitoring data of all modes of the target power equipment to obtain the multimodal dependency matrix of the target power equipment, including: Cross-attention analysis is performed on the current monitoring feature vectors of all modes of the target power equipment to obtain the interdependence strength between any two modes; Based on the interdependence strength of any two modes, construct the multimodal dependency matrix of the target power equipment.

[0006] As one preferred embodiment, obtaining the current state data of the target power equipment based on the current multimodal dependency matrix and the multimodal regression operation of the attention classification model includes: Obtain the target multimodal dependency matrix of the target power equipment, wherein the target multimodal dependency matrix is ​​constructed based on the data of the previous monitoring period of the current monitoring data; Based on the multimodal regression operation of the attention classification model, temporal context processing is performed on the current multimodal dependency matrix and the target multimodal dependency matrix to obtain the multidimensional state scalar of the target power equipment; Based on the analysis results of the multidimensional state scalar, the current state data of the target power equipment is obtained.

[0007] As one preferred embodiment, the step of obtaining the current operating data of the target power equipment based on the feedback signal of the current status data and the operating relationship map corresponding to the target power equipment includes: Acquire the equipment attribute data of the target power equipment, the personnel operation data corresponding to the target power equipment, and the personnel relationship data of the target power equipment; Construct the target device node corresponding to the target power equipment, and determine the device node data of the target device node based on the device attribute data of the target power equipment; Construct the target personnel node corresponding to the target power equipment, and determine the personnel node data of the target personnel node based on the personnel operation data corresponding to the target power equipment; Based on the personnel relationship data of the target power equipment, determine the node relationship data between the target equipment node and the target personnel node; Based on the device node data of the target device node, the personnel node data of the target personnel node, and the node relationship data, an operation relationship graph corresponding to the target power equipment is constructed.

[0008] As one preferred embodiment, obtaining the current operating data of the target power equipment based on the feedback signal of the current status data and the operating relationship map corresponding to the target power equipment includes: The current status data is sent to all user terminals corresponding to the target power equipment; When a task confirmation signal corresponding to the current status data is received, the target terminal that fed back the task confirmation signal among all the user terminals is determined; Based on the task confirmation signal, the target terminal, and the corresponding operation relationship map of the target power equipment, the current operation data of the target power equipment is obtained.

[0009] As one preferred embodiment, obtaining the current operation data of the target power equipment based on the task confirmation signal and the operation relationship map corresponding to the target power equipment includes: Based on the task confirmation signal, the equipment attribute data, and the personnel operation data, a current operation strategy for the target power equipment is constructed. Based on the analysis results of the current operation strategy, the current operation data of the target power equipment is obtained.

[0010] As a preferred embodiment, the construction of composite operational constraints for the target power equipment based on the equipment node data, personnel node data, and node relationship data of the operational relationship graph includes: Construct a vector of required resources for the device corresponding to the device node data, a vector of provided resources corresponding to the personnel node data, and a vector of matching degree corresponding to the node relationship data; The resource vector required by the device and the resource vector provided are analyzed to obtain the operation cost constraint data. Based on the analysis results of the matching degree vector, the job collaboration constraint data is obtained; Based on the operation cost constraint data and the operation coordination constraint data, a composite operation constraint for the target power equipment is constructed.

[0011] As one preferred embodiment, generating the job task instruction based on the current status data and the current job data includes: The current status data is parsed to obtain the task description data of the target power equipment; Based on the parsing results of the current operation data, the operation description data of the target power equipment is determined; Feature extraction is performed on the task description data and the job description data to obtain job task features used to generate job task instructions.

[0012] As one preferred embodiment, the step of responding to the task instruction based on the composite task constraint, and realizing the task processing of the target power equipment according to the response result of the task instruction, includes: The predicted response result is obtained based on the task instruction; Based on the prediction adaptation results of the composite task constraints and the predicted response results, the processing strategy for the task instructions is determined. The processing strategy is used to respond to the task instruction, so as to realize the task processing of the target power equipment according to the response result of the task instruction.

[0013] Another embodiment of the present invention provides a power equipment operation task processing system, comprising: The attention analysis module is used for modal association operations based on the attention classification model to perform attention analysis on the current monitoring data of all modes of the target power equipment to obtain the current multimodal dependency matrix of the target power equipment. The current state data determination module is used to obtain the current state data of the target power equipment based on the current multimodal dependency matrix and the multimodal regression operation of the attention classification model. The current operation data determination module is used to obtain the current operation data of the target power equipment based on the feedback signal of the current status data and the operation relationship map corresponding to the target power equipment; The composite operation constraint construction module is used to construct composite operation constraints for the target power equipment based on the equipment node data, personnel node data, and node relationship data of the operation relationship graph. The task instruction generation module is used to generate task instructions based on the current status data and the current task data; The task processing module is used to respond to the task instructions based on the composite task constraints, so as to realize the task processing of the target power equipment according to the response result of the task instructions. Attached Figure Description

[0014] Figure 1 This is one of the flowcharts illustrating the power equipment operation task processing method provided by the present invention; Figure 2 This is the second flowchart of the power equipment operation task processing method provided by the present invention; Figure 3 This is a schematic diagram of the power equipment operation task processing system provided by the present invention.

[0015] Figure label: Among them, 301 is the attention analysis module; 302 is the current state data determination module; 303 is the current task data determination module; 304 is the compound task constraint construction module; 305 is the task instruction generation module; and 306 is the task processing module. Detailed Implementation

[0016] 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, and not all embodiments. The purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0017] In the description of this application, the terms "first," "second," "third," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined with "first," "second," "third," etc., may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0018] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; they can refer to mechanical connections or electrical connections; they can refer to direct connections or indirect connections through an intermediate medium; and they can refer to the internal communication between two components. The terms "vertical," "horizontal," "left," "right," "upper," "lower," and similar expressions used herein are for illustrative purposes only and do not indicate or imply that the device or component referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as limiting the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0019] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this specification is for the purpose of describing specific embodiments only and is not intended to limit the invention. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0020] See Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the power equipment operation task processing method provided by the present invention, as shown below. Figure 1 As shown, this embodiment includes steps 100 to 600, and the specific steps are as follows: Step 100: Based on the attention classification model, perform attention analysis on the current monitoring data of all modes of the target power equipment to obtain the current multimodal dependency matrix of the target power equipment; The attention classification model in this embodiment is a deep learning model capable of simultaneously processing multiple modal monitoring data. This model can analyze the impact of the correlations between different modal data on the state of power equipment. The attention classification model in this embodiment can be a classification model based on cross-attention or a classification model based on self-attention. The modal association operation in this embodiment is used to perform correlation matching on monitoring data of different modalities, and to uncover the interdependencies between different modalities. The current monitoring data in this embodiment is multi-dimensional monitoring data obtained by monitoring the target power equipment within the current monitoring period, which may include infrared thermal imaging data, vibration sensor data, sound monitoring data, and video surveillance data, etc. The multimodal dependency matrix in this embodiment is a matrix used to characterize the interdependence strength between different modal monitoring data. The rows and columns of the multimodal dependency matrix correspond to different monitoring modalities, and the values ​​of the matrix elements can be used to represent the dependency strength between two corresponding modalities.

[0021] In one feasible implementation, the target power equipment is a 110kV oil-immersed transformer, whose current monitoring data includes three modes: body temperature sensing data, dissolved gas concentration data in the oil, and mechanical vibration sensing data. First, the monitoring data for these three modes are mapped to feature vectors of a uniform dimension. Then, a cross-attention mechanism is used to calculate the attention weight between any two modal feature vectors, thereby obtaining the interdependence strength between any two modes. For example, the dependency strength between the temperature mode and the gas mode is 0.82; the dependency strength between the temperature mode and the vibration mode is 0.65; and the dependency strength between the gas mode and the vibration mode is 0.58. Then, a 3×3 current multimodal dependency matrix is ​​constructed based on these intermodal dependency strengths.

[0022] In another feasible implementation, the target power equipment is a 10kV vacuum circuit breaker, whose current monitoring data includes four modes: opening and closing coil current data, contact temperature data, mechanical travel data, and partial discharge data. First, the monitoring data for these four modes are standardized to obtain standardized feature vectors. Then, a self-attention mechanism is used to perform global attention analysis on the feature vectors of all modes to obtain the dependency strength between any two modes. For example, the dependency strength between coil current and mechanical travel is 0.91; the dependency strength between contact temperature and partial discharge is 0.47. Then, a 4×4 current multimodal dependency matrix is ​​constructed based on these dependency strengths.

[0023] It should be noted that the dependency strength between modes can also be determined through the message passing mechanism of graph neural networks. First, the multimodal monitoring data of the target power equipment is constructed into a graph, where graph nodes represent different modes and edges represent potential correlations between modes. Then, the graph neural network calculates the attention weights between nodes through the message passing mechanism. For example, when the electrical parameter mode shows abnormal current fluctuations and the vibration mode detects abnormal vibration frequencies, the connection weights between the nodes of these two modes are strengthened.

[0024] The current multimodal dependency matrix in this embodiment can reflect the correlation characteristics between multimodalities.

[0025] Step 200: Based on the current multimodal dependency matrix and the multimodal regression operation of the attention classification model, obtain the current state data of the target power equipment; The multimodal regression operation in this embodiment is a function within the attention classification model used to perform regression analysis on the features of the multimodal dependency matrix to output the quantification results of the device state. The current state data in this embodiment is quantified data characterizing the current operational health status of the target power equipment, and may include information such as the equipment's health score, abnormal state type, and abnormality degree.

[0026] In one feasible implementation, the 3×3 multimodal dependency matrix of the transformer obtained in the above embodiment is input into the multimodal regression layer of the attention classification model. The multimodal regression layer adopts a fully connected network structure to map the features of the current multimodal dependency matrix and finally outputs the current state data of the transformer, in which the health score is 72, the abnormal state is overheating, and the abnormality level is a medium warning.

[0027] In another feasible implementation, the 4×4 multimodal dependency matrix of the circuit breaker obtained above is input into the multimodal regression layer of the attention classification model. The multimodal regression layer adopts a lightweight CNN network structure to extract the local and global features of the dependency matrix and finally outputs the current state data of the circuit breaker, in which the health score is 85, there is no obvious abnormal state, and the operating state is normal.

[0028] Step 300: Based on the feedback signal of the current status data and the operation relationship map corresponding to the target power equipment, obtain the current operation data of the target power equipment; In this embodiment, the feedback signal is a personnel response signal. The job relationship graph in this embodiment is a knowledge graph used to characterize the association between the target power equipment and the operators, including equipment nodes, personnel nodes, and the associations between nodes. The current job data in this embodiment is job-related data used to characterize the current state of the target power equipment, and may include information such as candidate operators, resources required for the job, and estimated job duration.

[0029] In one feasible implementation, for a transformer in a moderate overheating warning state, the status data is sent to the corresponding personnel terminal, and a feedback signal is obtained from the personnel terminal. For example, the status data is sent to the terminal of the person handling the equipment problem. When multiple terminals are involved, the personnel need to actively provide feedback through their terminals. The terminal feedback indicates that the corresponding personnel can handle the equipment problem. The operation relationship graph includes equipment attributes, personnel attributes, and the operation association between personnel and equipment. Through graph matching, the current operation data of the target power equipment is obtained. Among them, the candidate operators are those skilled in transformer oil handling, the required resources for the operation are an oil chromatograph and an infrared thermometer, and the estimated operation time is 3 hours.

[0030] In another feasible implementation, the feedback signal refers to the content returned by maintenance personnel after receiving and viewing equipment status data. The feedback signal can carry data such as whether a job is accepted, the available time for the job, and the resources required for the job. In specific implementation, the generated current status data is pushed to relevant maintenance personnel via a mobile terminal. The job relationship graph in this embodiment includes power equipment nodes and personnel nodes, each node containing its corresponding attribute data; there is an association between power equipment nodes and personnel nodes, which can be the power equipment managed by the personnel. First, the list of maintenance personnel corresponding to the target power equipment is queried from the job relationship graph. Then, based on the personnel's skill level, workload, and geographical location, the scope of personnel who should be given priority to receive status data is determined. After receiving the status data, maintenance personnel can send a task confirmation signal via the terminal, including information such as the estimated arrival time, required tools and equipment, and personnel configuration requirements. After collecting the feedback signal, the job suitability of each maintenance personnel is calculated by combining it with the node relationship data in the job relationship graph. For example, for high-risk transformer faults, maintenance personnel with high-voltage equipment maintenance qualifications and high historical job scores are given priority. The current job data in this embodiment includes information such as the job supervisor, job time window, job resource configuration, and safety measure requirements.

[0031] Step 400: Based on the equipment node data, personnel node data, and node relationship data of the operation relationship graph, construct the composite operation constraints of the target power equipment; In this embodiment, the equipment node data represents the attributes of equipment in the job relationship graph, and may include information such as equipment type, maintenance requirements, and required resources. The personnel node data represents the attributes of workers in the job relationship graph, and may include information such as skill level, work experience, and available time. The node relationship data represents the association between equipment nodes and personnel nodes in the job relationship graph, and may include information such as the historical job matching degree between personnel and equipment, and collaborative work experience among personnel. The composite job constraints in this embodiment are multi-dimensional constraints used to limit the execution of job tasks, and may include job cost constraints, job risk constraints, and job collaboration constraints.

[0032] In one feasible implementation, for the transformer's operating scenario, equipment node data is extracted from the operation relationship graph to obtain the required maintenance resources and operating hours for the equipment; personnel node data is extracted to obtain the personnel's man-hour costs and skill levels; and node relationship data is extracted to obtain the historical operation matching degree between personnel and the equipment. Based on this data, composite operation constraints are constructed, where the operation cost constraint can be that the total cost of the current operation does not exceed a certain threshold, and the operation coordination constraint can be that the matching degree between the operator's skill level and the equipment maintenance requirements is not lower than a certain threshold.

[0033] Step 500: Generate a task instruction based on the current status data and the current job data; In this embodiment, the work task instruction is structured data carrying specific work information. Its information format can be: Personnel - Time - Resources - Problem with Equipment. In practice, the current status data and current work data obtained above are merged. The current status data provides a detailed description of the equipment fault, including fault type, fault location, and fault severity; the current work data provides a specific arrangement for work execution, including personnel, work time, and work resources. Based on this information, standardized work task instructions are automatically generated. The work task instruction in this embodiment can include fields such as work task ID, equipment information, fault description, work requirements, safety measures, resource list, time schedule, and personnel assignment. For example, for a transformer bushing oil leakage fault, an instruction containing specific work steps such as "replacing the bushing seal ring, preparing insulating oil, setting up a safety fence, and testing insulation resistance" is generated. The instruction generation process considers work standards and safety procedures to ensure that the instruction content complies with power industry regulations. This embodiment can also implement hierarchical management of instructions, setting different approval processes according to the severity of the fault to ensure that high-risk operations are fully reviewed.

[0034] In another feasible implementation, job task instructions can be generated using a template-based approach, combined with artificial intelligence technology for personalized customization. When generating specific instructions, first, the most suitable template is selected, and then adjustments are made based on the specific conditions of the equipment and the capabilities of the personnel. For example, for experienced senior technicians, the instructions will include more complex fault diagnosis steps; for novice personnel, the instructions will include more detailed operating instructions and safety reminders. By intelligently generating job task instructions, equipment status assessment results can be transformed into executable work plans, ensuring the standardization and normalization of maintenance operations, improving work efficiency and quality, and reducing the risk of human error.

[0035] Step 600: Based on the composite operation constraints, respond to the operation task instruction to realize the operation task processing of the target power equipment according to the response result of the operation task instruction.

[0036] In this embodiment, the composite task constraints are used to determine the feasibility of a task or to filter multiple generated tasks. Specifically, upon receiving a task instruction, constraint verification is performed first. The verification process can include three levels: the first level verifies basic constraints (such as personnel qualifications and safety requirements); if not met, the task is rejected directly. The second level verifies resource constraints (such as tool availability and material inventory); if not met, a resource allocation process is triggered. The third level verifies optimization constraints (such as cost-effectiveness and time efficiency), used to select the optimal solution from multiple feasible options. The response results in this embodiment can include three types: task acceptance, task adjustment, or task rejection. When all constraints are met, the task is accepted and a work plan is generated; when some non-critical constraints are not met, the plan is adjusted (such as changing personnel and adjusting time); when critical constraints are not met, the task is rejected and alternative work suggestions are provided. The response results are communicated to relevant personnel through multiple channels, and the task relationship graph and equipment status information are automatically updated. For accepted tasks, the work progress is continuously tracked, and work process data is collected, which can be used for subsequent constraint model optimization and personnel capability assessment.

[0037] This invention analyzes multimodal monitoring data using an attention classification model, which can accurately uncover the dependencies between different monitoring dimensions, accurately quantify the current operating status of the equipment, and, combined with a job relationship graph, quickly match suitable job resources and personnel, construct multi-dimensional job constraints, and finally generate and execute job task instructions that conform to the constraints, thereby improving the accuracy of power equipment job task processing.

[0038] In another embodiment of the power equipment operation task processing method provided by the present invention, the above steps specifically include: Step 110: Perform cross-attention analysis on the current monitoring feature vectors of all modes of the target power equipment to obtain the interdependence strength between any two modes; Step 120: Construct the multimodal dependency matrix of the target power equipment based on the interdependence strength of any two modes.

[0039] In this embodiment, the current monitoring feature vector refers to the numerical feature representation extracted from the original monitoring data, which may include image modal features, sound modal features, and vibration modal time-frequency features, etc. The cross-attention analysis in this embodiment refers to calculating the attention weights between feature vectors of different modalities. This process can employ the multi-head attention mechanism in the Transformer architecture. In specific implementation, firstly, the original monitoring data of each modality is converted into feature vectors, and then the converted feature vectors are input into the cross-attention layer. The cross-attention layer obtains the attention weights between modalities by calculating the similarity between the query vector, key vector, and value vector. For example, when the image modality detects abnormal heating on the device surface, and the sound modality detects abnormal discharge sound, the cross-attention mechanism automatically strengthens the weights between these two modalities, determining that there is a causal relationship between them. In this embodiment, the mutual dependence strength is a value in the range [0,1], representing the degree of correlation between the two modal data in the device status assessment; the larger the value, the stronger the dependence. In this embodiment, the multimodal dependency matrix can be a symmetric matrix with diagonal elements of 1 (representing complete dependence of each mode on itself) and off-diagonal elements representing the dependence strength between different modes.

[0040] In practice, the cross-attention analysis process can include the following steps: First, generate corresponding query, key, and value vectors for each modality feature vector; then, calculate the attention score between all modality pairs; next, normalize the attention score using the softmax function; finally, weighted aggregate values ​​to obtain the attention output. Residual connections and layer normalization can also be introduced to improve the model's training stability and expressive power. To handle features at different scales, a multi-scale attention mechanism can be employed to calculate modality dependencies at different granularities.

[0041] This embodiment constructs a multimodal dependency matrix through cross-attention analysis, which can accurately capture the complex correlation between different modal data, providing richer feature information for subsequent equipment status assessment, and improving the accuracy and robustness of fault diagnosis, especially in the case of inconsistent multimodal data or missing partial modal data.

[0042] In another embodiment of the power equipment operation task processing method provided by the present invention, the above steps specifically include: Step 210: Obtain the target multimodal dependency matrix of the target power equipment. The target multimodal dependency matrix is ​​constructed based on the data of the previous monitoring period of the current monitoring data. Step 220: Based on the multimodal regression operation of the attention classification model, perform temporal context processing on the current multimodal dependency matrix and the target multimodal dependency matrix to obtain the multidimensional state scalar of the target power equipment; Step 230: Obtain the current state data of the target power equipment based on the analysis results of the multidimensional state scalar.

[0043] In this embodiment, the target multimodal dependency matrix refers to the historical reference matrix, used to capture the temporal change characteristics of equipment status. A sliding time window is maintained to store the multimodal dependency matrices for the most recent N monitoring periods. The positive integer N can be dynamically adjusted according to the equipment type and monitoring frequency; for critical equipment, N takes a larger value, while for general equipment, N takes a smaller value. The temporal context processing in this embodiment refers to analyzing the evolutionary relationship between the current state and historical states. This process can employ recurrent neural network structures such as LSTM or GRU. In specific implementation, the current multimodal dependency matrix and the target multimodal dependency matrix are input into the temporal processing module in chronological order. This temporal processing module generates a multidimensional state scalar by calculating the state change rate, trend, and pattern. In this embodiment, the multidimensional state scalar is a vector that can contain multiple dimensions of state information, such as equipment health index, failure probability, remaining lifespan, and risk level. Each dimension corresponds to a specific evaluation index.

[0044] The temporal context processing can include three sub-processes: differential analysis, trend prediction, and anomaly detection. Differential analysis calculates the difference between the current matrix and the historical matrix, quantifying the degree of state change; trend prediction predicts future state evolution based on historical change patterns; and anomaly detection identifies state changes that significantly deviate from the normal evolution pattern. The parsing of multidimensional state scalars can employ a combination of rule engines and machine learning. The rule engine maps scalar values ​​to specific device state descriptions based on preset thresholds and logical relationships; the machine learning model optimizes the accuracy of the mapping rules based on historical data and expert annotations. For example, when the health index is below 0.6 and the failure probability is above 0.8, the device is determined to be in a severe failure state and requires immediate action.

[0045] This embodiment, by introducing temporal context processing, can capture the dynamic evolution characteristics of the device state, not only assess the current state but also predict future trends, providing more comprehensive information support for maintenance decisions. It effectively avoids misjudgments and omissions that may be caused by static assessment methods, and improves the foresight and accuracy of state assessment.

[0046] See Figure 2 , Figure 2 This is a flowchart illustrating another embodiment of the power equipment operation task processing method provided by the present invention, as shown below. Figure 2 As shown, this embodiment includes steps one through four, and each step is detailed below: Step 10: Obtain the equipment attribute data of the target power equipment, the personnel operation data corresponding to the target power equipment, and the personnel relationship data of the target power equipment; Step 20: Construct the target device node corresponding to the target power equipment, and determine the device node data of the target device node based on the device attribute data of the target power equipment; Step 30: Construct the target personnel node corresponding to the target power equipment, and determine the personnel node data of the target personnel node based on the personnel operation data corresponding to the target power equipment; Step 40: Based on the personnel relationship data of the target power equipment, determine the node relationship data between the target equipment node and the target personnel node; Step 50: Based on the device node data of the target device node, the personnel node data of the target personnel node, and the node relationship data, construct the operation relationship graph corresponding to the target power equipment.

[0047] The equipment attribute data in this embodiment may include basic information such as equipment type, rated parameters, installation location, service life, and historical fault records. The personnel operation data in this embodiment may include personnel skill certificates, work experience descriptions, historical operation records, current work content, and geographical location. The personnel relationship data in this embodiment may include information such as the responsibility relationship between personnel and equipment, the collaborative relationship between personnel, and the professional field association of personnel. In specific implementation, firstly, equipment attribute data of the target power equipment is obtained from the equipment management database, relevant personnel operation data is obtained from the personnel management system, and personnel relationship data is mined from historical operation records. Then, graph nodes are constructed. Equipment nodes include attributes such as equipment ID, equipment type, and fault complexity; personnel nodes include attributes such as personnel ID, skill level, work content, and location coordinates. Node relationship data is represented by relationship edges, including various relationship types such as "responsible," "collaborative," and "professionally related." Each relationship edge has a weight value, indicating the strength of the relationship.

[0048] The construction of the job relationship graph can adopt an incremental update strategy. During initial construction, basic nodes and relationships are established; during operation, node attributes and relationship weights are dynamically adjusted based on actual job data. For example, when a person successfully handles a specific type of equipment failure, the relationship weight between that person and that type of equipment can be increased; when a person collaborates with another person multiple times to complete complex tasks, the collaboration relationship weight between the two can be increased.

[0049] This embodiment constructs a job relationship graph, integrating equipment, personnel, and relationship data into a unified knowledge representation, providing structured decision support for subsequent job task allocation, improving the matching accuracy between personnel and equipment, and the rationality of job task allocation.

[0050] In another embodiment of the power equipment operation task processing method provided by the present invention, the above steps specifically include: Step 310: Send the current status data to all user terminals corresponding to the target power equipment; Step 320: When the task confirmation signal corresponding to the current status data is received, determine the target terminal among all the user terminals that fed back the task confirmation signal; Step 330: Based on the task confirmation signal, the target terminal, and the corresponding operation relationship map of the target power equipment, obtain the current operation data of the target power equipment.

[0051] In this embodiment, the user terminal can be a mobile device used by maintenance personnel. The task confirmation signal in this embodiment is the maintenance personnel's response information to the task, including acceptance, rejection, estimated arrival time, required resources, and personnel configuration. In specific implementation, firstly, based on the task relationship map, all relevant maintenance personnel corresponding to the target power equipment are identified. Then, current status data is sent to these personnel's user terminals via a message push service. The push content includes a description of the equipment fault, urgency level, location information, and preliminary handling suggestions. A response time threshold (e.g., 30 minutes) is set, and all task confirmation signals are collected within the threshold. The specific responsible personnel are determined by the target terminal that provides the task confirmation signal.

[0052] The processing of task confirmation signals can employ a priority ranking algorithm. A comprehensive score is calculated for each confirmation signal based on factors such as node relationship weights in the job relationship graph, personnel historical job scores, and current workload. For example, for high-risk equipment failures, personnel with relevant qualifications, high historical scores, and low current workload are prioritized. This embodiment also supports a multi-person collaboration mode; when a single person cannot meet the job requirements, multiple personnel can be selected to form a work team. The job data generation process considers the time commitment and resource requirements in the task confirmation signals, combined with the urgency of the equipment status, to generate the optimal job task.

[0053] This embodiment achieves rapid response and reasonable allocation of maintenance tasks by intelligently distributing and collecting task confirmation signals, ensuring that critical faults can be handled in a timely manner. At the same time, it avoids the response delay and resource waste that may occur in traditional manual dispatching, thereby improving the timeliness and resource utilization efficiency of power equipment maintenance.

[0054] In another embodiment of the power equipment operation task processing method provided by the present invention, the above steps specifically include: Step 331: Based on the task confirmation signal, the equipment attribute data, and the personnel operation data, construct the current operation strategy for the target power equipment; Step 332: Based on the analysis results of the current operation strategy, obtain the current operation data of the target power equipment.

[0055] The current work strategy in this embodiment is a comprehensive work planning scheme, including multiple dimensions such as personnel allocation, time scheduling, resource allocation, and safety measures. In specific implementation, firstly, key information from task confirmation signals is integrated, such as personnel availability time, skill matching, and resource requirements; then, by combining equipment attribute data and personnel work data, a multi-objective optimization model is constructed. This multi-objective optimization model simultaneously considers multiple objectives such as work quality, time efficiency, cost control, and risk avoidance to find the optimal work strategy. For example, for urgent faults, time efficiency is prioritized; for complex faults, work quality is prioritized.

[0056] The current operational strategy can be constructed using a hierarchical planning approach. The first layer determines the personnel composition, selecting the main supervisor and support staff. The second layer defines the operational time window, considering equipment downtime and personnel availability. The third layer determines the resource allocation plan, including tools, materials, and vehicles. The fourth layer determines safety measures, developing corresponding protection plans based on risk assessment results. The strategy parsing process transforms these planning results into specific operational data, including work instructions, resource lists, schedules, and safety procedures. It also supports dynamic strategy adjustments, enabling rapid replanning in case of unforeseen circumstances during operations.

[0057] This embodiment transforms scattered task confirmation information into a systematic work plan by constructing a current work strategy, ensuring the scientific nature and feasibility of maintenance work, effectively balancing various constraints and optimization objectives, and improving the overall quality and efficiency of power equipment maintenance work.

[0058] In another embodiment of the power equipment operation task processing method provided by the present invention, the above steps specifically include: Step 410: Construct the resource vector required by the device corresponding to the device node data, the resource provided vector corresponding to the personnel node data, and the matching degree vector corresponding to the node relationship data; Step 420: Analyze the resource vector required by the equipment and the resource vector provided to obtain the operation cost constraint data; Step 430: Based on the parsing results of the matching degree vector, obtain the job collaboration constraint data; Step 440: Based on the operation cost constraint data and the operation coordination constraint data, construct the composite operation constraint for the target power equipment.

[0059] In this embodiment, the resource requirement vector represents the various resource demands for equipment maintenance, including human resources, material resources, time resources, and space resources. The resource provision vector in this embodiment refers to the availability of various resources, including available personnel, inventory materials, available time windows, and available workspace. The matching degree vector in this embodiment refers to the degree of compatibility between personnel and equipment, and between personnel themselves, including dimensions such as skill matching degree, collaboration matching degree, and experience matching degree. In practical implementation, relevant data is extracted from the work relationship graph to construct three vectors. Specific constraint data is generated through vector operations and rule judgments.

[0060] The calculation of operational cost constraints includes both direct and indirect costs. Direct costs include labor costs, material costs, and equipment usage costs; indirect costs include downtime losses, safety risk costs, and quality risk costs. A cost function model is used to map resource vectors to specific cost values. The calculation of operational collaboration constraints includes indicators such as personnel collaboration degree, skill complementarity, and communication efficiency. By analyzing historical operational data, a collaborative effect prediction model is established. The construction of composite operational constraints adopts a constraint hierarchy, classifying cost constraints and collaboration constraints according to their importance to form a constraint satisfaction priority.

[0061] This embodiment constructs composite job constraints using a vectorization method, accurately quantifies various constraint conditions, and improves the operability of job task allocation.

[0062] In another embodiment of the power equipment operation task processing method provided by the present invention, the above steps specifically include: Step 510: Parse the current state data to obtain the task description data of the target power equipment; Step 520: Based on the parsing results of the current operation data, determine the operation description data of the target power equipment; Step 530: Extract features from the task description data and the job description data to obtain job task features used to generate job task instructions.

[0063] In this embodiment, the task description data characterizes the specific features and handling requirements of equipment failures, including information such as failure type, failure location, scope of impact, urgency, and safety risks. The job description data in this embodiment characterizes the specific arrangements and requirements for job execution, including information such as personnel allocation, time schedule, resource requirements, operating procedures, and quality standards.

[0064] Natural language processing (NLP) techniques are used to parse the current state data and extract key task elements; a rule engine is used to parse the current job data and extract key job elements. The feature extraction process in this embodiment includes three sub-processes: text feature extraction, numerical feature extraction, and relational feature extraction. Methods such as TF-IDF, word embedding, and statistical analysis can be used to convert the raw data into structured feature vectors.

[0065] In practical implementation, the construction of task features can adopt a multimodal fusion strategy. Text features include fault description keywords, operation step keywords, and safety precaution keywords; numerical features include estimated task duration, resource requirements, and risk level scores; relational features include the weights of personnel-equipment relationships and equipment-tool relationships. A feature selection algorithm is used to filter the most discriminative features for generating task instructions. The instruction generation process combines template filling and sequence generation to ensure the completeness and readability of the instruction content.

[0066] This embodiment generates job task instructions through feature extraction, which can transform complex raw data into concise and clear job instructions, improve the automation and accuracy of instruction generation, provide maintenance personnel with operable job guidance, and improve the efficiency and quality of job execution.

[0067] In another embodiment of the power equipment operation task processing method provided by the present invention, the above steps specifically include: Step 610: Obtain the predicted response result based on the job task instruction; Step 620: Determine the processing strategy for the task instruction based on the prediction adaptation result of the composite task constraint and the prediction response result; Step 630: Respond to the operation task instruction based on the processing strategy, so as to realize the operation task processing of the target power equipment according to the response result of the operation task instruction.

[0068] In this embodiment, the predicted response result characterizes the expected effect of executing the task instruction, including predicted indicators across multiple dimensions such as expected completion time, expected cost, expected risk, and expected quality. The predicted adaptation result in this embodiment characterizes the degree of matching between the composite task constraints and the predicted response result, and a comprehensive score is obtained by calculating the constraint satisfaction and target achievement.

[0069] In practice, the various outcomes of instruction execution are first predicted. Then, the predicted results are compared and analyzed with the constraints of the complex task to calculate the fit score. The processing strategies in this embodiment include various types such as acceptance, adjustment, rejection, and postponement strategies, determined based on the fit score and the severity of constraint violations.

[0070] The predicted response can be calculated using Monte Carlo simulation, estimating various possible execution outcomes through multiple random samplings. The predicted adaptation result can be calculated using multi-criteria decision analysis, assigning weights to different constraints and optimization objectives to calculate a comprehensive adaptation score. The processing strategy can be determined using a rule engine, selecting the optimal strategy based on preset thresholds and logical relationships. For example, when the adaptation score is higher than 0.8, an acceptance strategy is adopted; when the score is between 0.6 and 0.8, an adjustment strategy is adopted; and when the score is lower than 0.6, a rejection or postponement strategy is adopted. This embodiment also supports dynamic optimization of the strategy, continuously adjusting the prediction model and decision rules based on actual execution results.

[0071] This embodiment responds to work task instructions through predictive response and adaptation analysis, assesses the feasibility and effectiveness of task execution in advance, avoids unreasonable task allocation, and ensures that each work task can be executed under optimal constraints. This improves the success rate and resource utilization efficiency of power equipment maintenance operations, while reducing operational risks and costs.

[0072] The power equipment operation task processing system provided by the present invention is described below. The power equipment operation task processing system described below can be referred to in correspondence with the power equipment operation task processing method described above.

[0073] Please refer to Figure 3 The present invention also provides a power equipment operation task processing system, comprising: Attention analysis module 301 is used for modal association operation based on attention classification model to perform attention analysis on the current monitoring data of all modes of the target power equipment to obtain the current multimodal dependency matrix of the target power equipment; The current state data determination module 302 is used to obtain the current state data of the target power equipment based on the current multimodal dependency matrix and the multimodal regression operation of the attention classification model. The current operation data determination module 303 is used to obtain the current operation data of the target power equipment based on the feedback signal of the current status data and the operation relationship map corresponding to the target power equipment; The composite operation constraint construction module 304 is used to construct composite operation constraints for the target power equipment based on the equipment node data, personnel node data, and node relationship data of the operation relationship graph. The task instruction generation module 305 is used to generate task instructions based on the current status data and the current task data; The task processing module 306 is used to respond to the task instruction based on the composite task constraint, so as to realize the task processing of the target power equipment according to the response result of the task instruction.

[0074] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for processing power equipment operation tasks, characterized in that, include: Based on the attention classification model, the modal association operation performs attention analysis on the current monitoring data of all modes of the target power equipment to obtain the current multimodal dependency matrix of the target power equipment. Based on the current multimodal dependency matrix and the multimodal regression operation of the attention classification model, the current state data of the target power equipment is obtained; Based on the feedback signal of the current status data and the operation relationship map corresponding to the target power equipment, the current operation data of the target power equipment is obtained; Based on the equipment node data, personnel node data, and node relationship data of the operation relationship graph, a composite operation constraint for the target power equipment is constructed. Based on the current status data and the current job data, generate job task instructions; Based on the composite operation constraints, the system responds to the operation task instructions to realize the operation task processing of the target power equipment according to the response results of the operation task instructions.

2. The power equipment operation task processing method as described in claim 1, characterized in that, The modal association operation based on the attention classification model performs attention analysis on the current monitoring data of all modes of the target power equipment to obtain the multimodal dependency matrix of the target power equipment, including: Cross-attention analysis is performed on the current monitoring feature vectors of all modes of the target power equipment to obtain the interdependence strength between any two modes; Based on the interdependence strength of any two modes, construct the multimodal dependency matrix of the target power equipment.

3. The power equipment operation task processing method as described in claim 2, characterized in that, The process of obtaining the current state data of the target power equipment based on the current multimodal dependency matrix and the multimodal regression operation of the attention classification model includes: Obtain the target multimodal dependency matrix of the target power equipment, wherein the target multimodal dependency matrix is ​​constructed based on the data of the previous monitoring period of the current monitoring data; Based on the multimodal regression operation of the attention classification model, temporal context processing is performed on the current multimodal dependency matrix and the target multimodal dependency matrix to obtain the multidimensional state scalar of the target power equipment; Based on the analysis results of the multidimensional state scalar, the current state data of the target power equipment is obtained.

4. The power equipment operation task processing method as described in claim 1, characterized in that, Before obtaining the current operation data of the target power equipment based on the feedback signal of the current status data and the operation relationship map corresponding to the target power equipment, the following steps are included: Acquire the equipment attribute data of the target power equipment, the personnel operation data corresponding to the target power equipment, and the personnel relationship data of the target power equipment; Construct the target device node corresponding to the target power equipment, and determine the device node data of the target device node based on the device attribute data of the target power equipment; Construct the target personnel node corresponding to the target power equipment, and determine the personnel node data of the target personnel node based on the personnel operation data corresponding to the target power equipment; Based on the personnel relationship data of the target power equipment, determine the node relationship data between the target equipment node and the target personnel node; Based on the device node data of the target device node, the personnel node data of the target personnel node, and the node relationship data, an operation relationship graph corresponding to the target power equipment is constructed.

5. The power equipment operation task processing method as described in claim 4, characterized in that, The step of obtaining the current operation data of the target power equipment based on the feedback signal of the current status data and the operation relationship map corresponding to the target power equipment includes: The current status data is sent to all user terminals corresponding to the target power equipment; When a task confirmation signal corresponding to the current status data is received, the target terminal that fed back the task confirmation signal among all the user terminals is determined; Based on the task confirmation signal, the target terminal, and the corresponding operation relationship map of the target power equipment, the current operation data of the target power equipment is obtained.

6. The power equipment operation task processing method as described in claim 5, characterized in that, The process of obtaining the current operation data of the target power equipment based on the task confirmation signal and the operation relationship map corresponding to the target power equipment includes: Based on the task confirmation signal, the equipment attribute data, and the personnel operation data, a current operation strategy for the target power equipment is constructed. Based on the analysis results of the current operation strategy, the current operation data of the target power equipment is obtained.

7. The power equipment operation task processing method as described in claim 1, characterized in that, The composite operational constraints for the target power equipment, constructed based on the equipment node data, personnel node data, and node relationship data of the operational relationship graph, include: Construct a vector of required resources for the device corresponding to the device node data, a vector of provided resources corresponding to the personnel node data, and a vector of matching degree corresponding to the node relationship data; The resource vector required by the device and the resource vector provided are analyzed to obtain the operation cost constraint data. Based on the analysis results of the matching degree vector, the job collaboration constraint data is obtained; Based on the operation cost constraint data and the operation coordination constraint data, a composite operation constraint for the target power equipment is constructed.

8. The power equipment operation task processing method as described in claim 1, characterized in that, The step of generating a job task instruction based on the current status data and the current job data includes: The current status data is parsed to obtain the task description data of the target power equipment; Based on the parsing results of the current operation data, the operation description data of the target power equipment is determined; Feature extraction is performed on the task description data and the job description data to obtain job task features used to generate job task instructions.

9. The power equipment operation task processing method as described in claim 1, characterized in that, The step of responding to the task instruction based on the composite task constraint, and realizing the task processing of the target power equipment according to the response result of the task instruction, includes: The predicted response result is obtained based on the task instruction; Based on the prediction adaptation results of the composite task constraints and the predicted response results, the processing strategy for the task instructions is determined. The processing strategy is used to respond to the task instruction, so as to realize the task processing of the target power equipment according to the response result of the task instruction.

10. A power equipment operation task processing system, characterized in that, include: The attention analysis module is used for modal association operations based on the attention classification model to perform attention analysis on the current monitoring data of all modes of the target power equipment to obtain the current multimodal dependency matrix of the target power equipment. The current state data determination module is used to obtain the current state data of the target power equipment based on the current multimodal dependency matrix and the multimodal regression operation of the attention classification model. The current operation data determination module is used to obtain the current operation data of the target power equipment based on the feedback signal of the current status data and the operation relationship map corresponding to the target power equipment; The composite operation constraint construction module is used to construct composite operation constraints for the target power equipment based on the equipment node data, personnel node data, and node relationship data of the operation relationship graph. The task instruction generation module is used to generate task instructions based on the current status data and the current task data; The task processing module is used to respond to the task instructions based on the composite task constraints, so as to realize the task processing of the target power equipment according to the response result of the task instructions.

Images