Artificial intelligence-based live-line maintenance intelligent task allocation method and system

Abstract

The application discloses an intelligent task allocation method and system for live maintenance based on artificial intelligence, and relates to the technical field of power system operation and maintenance. The method comprises the following steps: performing rule analysis on the state rollback path of the live maintenance task under the condition of execution failure or execution interruption, and calculating the corresponding operation reversibility index; obtaining historical operation data and current operation state data of the operation personnel, and calculating the initial value of the risk credit of the operation personnel; introducing time decay rules and task consumption rules, and calculating the available risk credit limit of the operation personnel; performing double-threshold allocation access determination, performing matching reasoning on the live maintenance task and the operation personnel, and generating a task allocation result; and updating the available risk credit limit of the corresponding operation personnel according to the task execution result, which is used for subsequent live maintenance task allocation. The application significantly improves the safety and reliability of task allocation, the execution stability, and the adaptive optimization capability of the overall operation of the system.

Description

Technical Field

[0001] This application relates to the field of power system operation and maintenance technology, and in particular to an intelligent task allocation method and system for live-line maintenance based on artificial intelligence. Background Technology

[0002] With the continuous expansion of the power grid and the increasing demands for power supply reliability, live-line maintenance, as an important technical means of maintaining and repairing defects in power transmission and distribution equipment under uninterrupted power conditions, has been widely applied in the operation and maintenance of transmission and transformation lines, substations, and distribution networks. Live-line maintenance can effectively reduce the number of power outages and minimize the impact on user power supply. However, it also features complex operating environments, high operational risks, and strong irreversibility. Once operational errors or sudden anomalies occur, they can easily lead to equipment damage, personal injury, or even large-scale power outages. Therefore, extremely high requirements are placed on the safety, rationality, and scientific nature of task allocation.

[0003] In related technologies, existing methods mostly focus on the static labeling of task difficulty or risk level, lacking systematic analysis of whether tasks have the ability to be rolled back after execution failure or interruption. In reality, the consequences of failure vary significantly among different live-line maintenance tasks, and current dispatching systems generally cannot distinguish these differences in the controllability of failure consequences. At the same time, the modeling of operator capabilities relies heavily on static indicators such as qualification level, years of service, and historical scores, neglecting dynamic state factors such as continuous working hours, recent task load, and cumulative risk exposure. This can easily lead to the over-concentration of dispatching a small number of experienced personnel during high-load operation, thereby amplifying the safety hazards caused by fatigued work and continuous high-risk work at the system level, which requires improvement. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent task allocation method and system for live-line maintenance based on artificial intelligence, so as to solve the problems mentioned in the background art.

[0005] Firstly, the intelligent task allocation method for live-line maintenance based on artificial intelligence provided in this application adopts the following technical solution: Obtain information on live-line maintenance tasks to be assigned, perform regular analysis on the state rollback path of live-line maintenance tasks in the event of execution failure or interruption, extract the reversibility influencing factors of live-line maintenance tasks, and calculate the corresponding operation reversibility index. Obtain historical and current work status data of personnel involved in live-line maintenance tasks, and calculate the initial risk credit value of the personnel. Based on the initial risk credit value, a time decay rule and a task consumption rule are introduced to calculate the available risk credit limit for the operator in the current allocation cycle. The reversibility index of the operation and the available risk credit limit are used as joint judgment inputs to perform a dual threshold allocation admission judgment. Under the condition of satisfying the dual threshold allocation admission judgment, matching reasoning is performed on the live maintenance task and the operator to generate the task allocation result. After the live-line maintenance task is completed, the available risk credit limit of the corresponding operator is updated based on the task execution results for subsequent live-line maintenance task allocation.

[0006] Preferably, the steps of obtaining information on live-line maintenance tasks to be assigned, performing regular analysis on the state rollback paths of live-line maintenance tasks in the event of execution failure or interruption, extracting the reversibility influencing factors of live-line maintenance tasks, and calculating the corresponding job reversibility index are as follows: Obtain information on live-line maintenance tasks to be assigned, including the work object, work method, and work step sequence; For the live-line maintenance task, based on the sequence of its operation steps, the system state changes in the case of execution failure or interruption of each operation step are enumerated and analyzed to generate the corresponding state rollback path. The state rollback path is subject to rule-based determination, which includes whether the rollback process can be completed under power outage conditions, whether the rollback operation introduces new power-related risks, and whether the rollback process requires additional scheduling resources, to obtain the rule-based determination result. Based on the rule-based judgment results, the reversibility influencing factors of live-line maintenance tasks are extracted, and the corresponding work reversibility index is calculated.

[0007] Preferably, the step of extracting the reversibility influencing factor of the live-line maintenance task based on the rule-based judgment result and calculating the corresponding operation reversibility index is as follows: Based on the rule-based determination results, obtain state determination information related to state rollback, and extract the reversibility influence factor of the live-line maintenance task from the state determination information. The reversibility influencing factors include uninterrupted power outage reversibility factors, power-on risk introduction factors, and resource rescheduling factors. The reversibility influencing factors are quantified and normalized. The reversibility influence factors after quantification and normalization are fused and calculated according to the preset combination rules to obtain the corresponding operation reversibility index. The operation reversibility index is used to characterize the overall rollback capability of live-line maintenance tasks in the event of execution failure or interruption.

[0008] Preferably, the step of obtaining historical work data and current work status data of personnel involved in live-line maintenance tasks, and calculating the initial risk credit value of the personnel, specifically includes: Obtain historical work data and current work status data of the personnel involved in the live-line maintenance task, and perform structured processing on the historical work data and current work status data to form a personnel status feature dataset; Based on the historical operation data, the historical task success rate, frequency of abnormal and irregular operations, and number of high-risk task executions are extracted from the operator status feature dataset, and historical performance indicators are calculated comprehensively. Based on the current work status data, the current continuous work duration, the number of tasks undertaken in the current allocation cycle, and the matching degree between the current task type and historical experience are extracted from the work personnel status feature dataset, and a status correction index is calculated in a comprehensive manner. The historical stability evaluation index and the state correction index are input together and combined according to the preset fusion rules to obtain the initial risk credit value of the corresponding operator, which is used to characterize the operator's basic risk-bearing capacity at the beginning of the current allocation cycle.

[0009] Preferably, the step of calculating the available risk credit limit for operators in the current allocation period based on the initial risk credit value, by introducing time decay rules and task consumption rules, specifically includes: Obtain the start time and current time of the current task allocation cycle, calculate the cumulative working time of the operator in the current task allocation cycle, extract the current continuous working time, and generate a time decay weight that represents the degree of time consumption based on the cumulative working time and the current continuous working time. The system collects data on tasks executed by operators within the current allocation cycle, obtains the risk level parameters and complexity level parameters corresponding to the executed tasks, and generates task consumption weights that characterize the task load based on the number of executed tasks, risk level parameters, and complexity level parameters. The time decay weight and task consumption weight are weighted and fused to generate a joint credit consumption parameter; Extract the initial risk credit value, and deduct the initial risk credit value according to the joint credit consumption parameter to obtain the available risk credit limit for the operator in the current task allocation cycle.

[0010] Preferably, the step of extracting the initial risk credit value and deducting it according to the joint credit consumption parameters to obtain the available risk credit limit for the operator in the current task allocation cycle is as follows: Extract the initial risk credit value, and determine the deduction method to perform deduction calculation on the initial risk credit value according to the preset consumption range in which the joint credit consumption parameter is located. The deduction method includes a proportional deduction strategy and a segmented deduction strategy. When using a proportional deduction strategy, the joint credit consumption parameter is mapped to a deduction ratio, and a proportional deduction operation is performed on the initial risk credit value according to the deduction ratio to obtain a preliminary credit limit result. When adopting a segmented deduction strategy, multiple joint credit consumption parameter ranges are preset, the range to which the joint credit consumption parameter belongs is determined, and the segmented deduction operation is performed on the initial risk credit value according to the deduction rule matched by the range to obtain the preliminary credit limit result. Boundary constraints are applied to the preliminary credit limit result to ensure that it is not less than zero and not greater than the initial risk credit value, thereby obtaining the available risk credit limit for the operator within the current task allocation cycle.

[0011] Preferably, the steps of using the work reversibility index and available risk credit limit as joint judgment inputs, performing dual-threshold allocation admission judgment, and performing matching reasoning on the live-line maintenance task and the operator to generate task allocation results under the condition of satisfying the dual-threshold allocation admission judgment are as follows: Obtain the data of tasks to be executed within the current allocation period. The data of tasks to be executed includes the risk level parameter and the complexity level parameter of the tasks to be executed. Configure the corresponding reversibility access threshold and credit access threshold according to the data of tasks to be executed. The reversibility index of the operation and the available risk credit limit are compared with the reversibility access threshold and the credit access threshold, respectively. A joint determination of the two thresholds is performed to generate an access determination result on whether or not to allow entry into the task matching stage. When the admission determination result indicates that entry into the task matching stage is permitted, the job requirement parameters of the task to be executed and the execution capability parameters of the candidate workers are obtained. Matching reasoning is performed on the job requirement parameters and execution capability parameters to generate task allocation results and execute them.

[0012] Preferably, after the live-line maintenance task is completed, the available risk credit limit of the corresponding operator is updated based on the task execution result for subsequent live-line maintenance task allocation. The specific steps are as follows: Obtain the execution result information of the live-line maintenance task. The execution result information includes task completion index, abnormal impact index, risk deviation index and efficiency deviation index. Construct corresponding task execution evaluation tags based on the execution result information. A multi-level task execution evaluation system is preset, and the evaluation level to which the execution result of the current live-line maintenance task belongs is determined based on the task execution evaluation label. Establish a mapping relationship between the evaluation level and the credit adjustment strategy, and determine the credit adjustment strategy that matches the operator. The credit adjustment strategy includes a reward-based replenishment strategy, a neutral maintenance strategy, and a penalty-based deduction strategy. Based on the credit adjustment strategy, the available risk credit limit within the current allocation period is increased or decreased to update the available risk credit limit for operators, which is then used for subsequent live-line maintenance task allocation.

[0013] Secondly, the intelligent task allocation system for live-line maintenance based on artificial intelligence provided in this application adopts the following technical solution: An AI-based intelligent task allocation system for live-line maintenance includes: The reversibility assessment module obtains information on the live-line maintenance tasks to be assigned, performs regular analysis on the state rollback path of the live-line maintenance tasks in the event of execution failure or interruption, extracts the reversibility influencing factors of the live-line maintenance tasks, and calculates the corresponding operation reversibility index. The risk credit modeling module acquires historical and current work status data of personnel involved in live-line maintenance tasks and calculates the initial risk credit value of the personnel. The credit consumption calculation module, based on the initial risk credit value, introduces time decay rules and task consumption rules to calculate the available risk credit limit for operators in the current allocation cycle; The matching decision module uses the work reversibility index and available risk credit limit as joint judgment inputs, performs dual threshold allocation admission judgment, and performs matching reasoning on the live maintenance task and the operator under the condition of satisfying the dual threshold allocation admission judgment to generate task allocation results. The credit feedback update module updates the available risk credit limit of the corresponding operator based on the task execution results after the live-line maintenance task is completed, for use in the subsequent allocation of live-line maintenance tasks.

[0014] In summary, this application includes at least one of the following beneficial technical effects: 1. By modeling the state rollback path after a failed or interrupted live-line maintenance task, key factors are extracted and quantified into an operation reversibility index. This enables the engineering controllability classification of task risks, breaking through the traditional coarse-grained classification model based on task type and voltage level. It distinguishes between high-risk reversible and irreversible operations from the perspectives of controllability, rollback, and remediation, providing a precise risk scale for task allocation. Integrating historical operation data with the current state, a comprehensive model is built for dimensions such as personnel stability and compliance, generating an initial value for quantified risk credit. This allows for an objective assessment of personnel risk tolerance, avoiding subjective assignment based on seniority. Introducing time decay and task consumption rules transforms static risk credit into dynamic, consumable, and recoverable resources. The system dynamically calculates the available risk credit limit for the current period, achieving temporal adjustment and system balance of personnel risk tolerance, avoiding potential risks such as concentrated risk accumulation. Using the operation reversibility index and available risk credit limit as joint inputs, a dual-threshold admission mechanism is used to construct a hard constraint coupling relationship between task risk and personnel tolerance. This balances task completion efficiency while ensuring safety, improving the safety and rationality of scheduling strategies. After the task is executed, the available risk credit limit for personnel is updated based on the feedback of the results, forming an adaptive loop of behavior-consumption-feedback-correction, continuously correcting the risk profile of personnel, strengthening positive incentives and risk suppression, and improving the long-term safety, stability and self-learning ability of the scheduling system.

[0015] 2. By structuring the description of assigned live-line maintenance tasks, the work objects, methods, and sequence of steps are clearly defined. Textual plans or manual experience are transformed into a process-oriented work sequence model that the system can parse. This allows the system to understand the work content from a process perspective, avoiding the bias of safety assessments based solely on coarse-grained risk judgments based on task categories. For the sequence of work steps, the potential system state changes caused by execution failures or interruptions are enumerated and analyzed, generating corresponding state rollback paths. Accident handling and emergency recovery paths are pre-computed, providing a direct and verifiable analytical basis for evaluating task risks from the dimensions of recoverability and controllability, thus improving the engineering authenticity and completeness of risk assessments. Engineering constraint rules are introduced to rule-basedly determine key factors such as the uninterrupted power-on feasibility, new live-line risks, and additional resource requirements for each state rollback path. Results are output in a unified format, transforming the rollback feasibility judgments based on expert experience into consistent assessment rules that the system can automatically execute, ensuring the stability, reproducibility, and auditability of risk assessment results. Based on the judgment results, key factors affecting the difficulty of task recovery and the degree of risk diffusion are extracted and comprehensively calculated into a quantitative operation reversibility index. This enables the system to rank and classify the controllability of different task failure consequences, providing a unified calculation scale for subsequent matching with the risk credit capabilities of operators, and enhancing the refinement of scheduling strategies and the controllability of safety boundaries.

[0016] 3. Obtain data on tasks to be executed this week. Based on task risk and complexity levels, dynamically configure reversibility and credit thresholds to construct a tiered control mechanism with high thresholds for high-risk and low thresholds for low-risk tasks, giving the scheduling system task adaptive capabilities for its safety constraint strategies. Compare the job reversibility index and available risk credit limit with the corresponding thresholds, and perform a dual-threshold joint judgment to generate an admission result, avoiding one-dimensional, biased decision-making. By establishing a dual insurance mechanism of task-side reversibility and personnel-side risk tolerance, the probability of accidents escalating due to incorrect allocation is significantly reduced from a systemic perspective. If the admission judgment allows entry into the matching stage, obtain the job requirement parameters of the task to be executed and the execution capability parameters of the candidate personnel, and generate and execute the allocation result through matching reasoning. This hierarchical decision-making architecture prioritizes safety, then pursues optimal efficiency and suitability, preventing blind pursuit of efficiency that exceeds safety boundaries and avoiding excessive conservatism that wastes intelligent optimization space within the safety range, thus achieving a balance between operational safety and job efficiency. Attached Figure Description

[0017] Figure 1 This is a schematic diagram illustrating the specific steps of an embodiment of the intelligent task allocation method for live-line maintenance based on artificial intelligence according to the present invention.

[0018] Figure 2 This is a schematic diagram of the module connections of an embodiment of the intelligent task allocation system for live-line maintenance based on artificial intelligence of the present invention. Detailed Implementation

[0019] The following examples and... Figures 1-2 The present invention will be described in further detail, but the embodiments of the present invention are not limited thereto.

[0020] This invention discloses an intelligent task allocation method for live-line maintenance based on artificial intelligence, which specifically includes the following steps: Step S1: Obtain information on the live-line maintenance tasks to be assigned, perform regular analysis on the state rollback path of the live-line maintenance task in the case of execution failure or interruption, extract the reversibility influencing factor of the live-line maintenance task, and calculate the corresponding operation reversibility index. Step S2: Obtain historical work data and current work status data of the personnel involved in the live-line maintenance task, and calculate the initial risk credit value of the personnel. Step S3: Based on the initial risk credit value, introduce the time decay rule and the task consumption rule to calculate the available risk credit limit for the operator in the current allocation cycle; Step S4: Using the reversibility index of the operation and the available risk credit limit as joint judgment inputs, perform dual threshold allocation access judgment. Under the condition of satisfying the dual threshold allocation access judgment, perform matching reasoning on the live maintenance task and the operator to generate task allocation results. Step S5: After the live-line maintenance task is completed, the available risk credit limit of the corresponding operator is updated according to the task execution result for subsequent live-line maintenance task allocation.

[0021] In practical applications, by performing rule-based modeling and analysis of the state rollback paths in the event of failure or interruption of assigned live-line maintenance tasks, key factors affecting the recoverability of the work are extracted and quantified into a work reversibility index. This enables a structured description of the severity and controllability of the risk consequences of different maintenance tasks. Instead of simply classifying risks by task type or voltage level in a coarse-grained manner, the system categorizes task risks based on whether they are controllable, rollbackable, and remediable after an error. This allows the system to distinguish between high-risk but reversible and high-risk and irreversible work types, providing a risk scale basis for subsequent task allocation strategies. By integrating historical work behavior data with current work status data, a comprehensive model is performed on the stability, standardization, correlation of risk events, and current state adaptability of the workers. This forms a unified and quantified initial risk credit value, enabling an objective quantitative assessment of personnel risk tolerance and avoiding subjective assignment based solely on seniority, years of service, or manual experience. Based on the initial risk credit value, time decay rules and task consumption rules are introduced to dynamically calculate the available risk credit limit for operators within the current allocation cycle. This transforms risk credit from a static capability value into a dynamic resource that is consumable, recoverable, and periodically adjustable. This allows operators' allocation eligibility to automatically shrink after continuous high-risk operations and gradually recover after long-term stable performance, thus avoiding hidden safety hazards such as concentrated risk accumulation and amplified fatigue work. This achieves time-series adjustment and system-level balance control of personnel risk-bearing capacity. Using the job reversibility index and available risk credit limit as joint inputs, a dual-threshold admission mechanism first determines eligibility, then performs task-person matching reasoning to generate allocation results. This constructs a hard-constraint coupling relationship between task risk attributes and personnel risk-bearing capacity, preventing systemic accident combinations where highly irreversible tasks are assigned to personnel with insufficient risk margins. Simultaneously, it allows high-risk but highly reversible tasks to be undertaken by high-credit personnel under controlled conditions, thus achieving a structured balance between safety and task completion efficiency, significantly improving the controllability of the safety boundary and the engineering rationality of the scheduling strategy. After a task is completed, the available risk credit limit of the operator is updated based on the execution result (such as successful completion, abnormal interruption, risk triggering, etc.) and used for subsequent task allocation. This makes the system no longer a one-time assessment and static scoring, but a continuous adaptive adjustment loop of behavior-consumption-feedback-correction. As a result, the risk profile of the personnel can be continuously corrected and converged with the actual operation performance. This not only strengthens the positive incentive constraint of standardized operation, but also forms a long-term inhibitory effect on high-risk behavior, which significantly improves the long-term safety stability and self-learning ability of the entire scheduling system.

[0022] The steps for obtaining information on live-line maintenance tasks to be assigned, performing regular analysis on the state rollback paths of live-line maintenance tasks in the event of execution failure or interruption, extracting the reversibility influencing factors of live-line maintenance tasks, and calculating the corresponding job reversibility index are as follows: Step S11: Obtain the live-line maintenance task information to be assigned. The live-line maintenance task information includes the work object, work method and work step sequence. Step S12: For the live-line maintenance task, based on the order of its operation steps, enumerate and analyze the system state changes in the case of execution failure or interruption of each operation step, and generate the corresponding state rollback path. Step S13: Perform rule-based determination on the state rollback path. The rule-based determination includes whether the rollback process can be completed under power outage conditions, whether the rollback operation introduces new power-related risks, and whether the rollback process requires additional scheduling resources, and obtain the rule-based determination result. Step S14: Based on the rule-based judgment results, extract the reversibility influence factor of the live-line maintenance task and calculate the corresponding operation reversibility index.

[0023] In practical applications, by structuring the assigned live-line maintenance tasks, the work objects, methods, and sequence of steps are clearly identified. This transforms the maintenance tasks, originally described using textual plans or manual experience, into a process-oriented work sequence model that the system can parse. This allows the system to understand the work content from a process perspective, rather than solely from the task name or type, avoiding safety assessment biases caused by coarse-grained risk judgments based only on task categories. For the sequence of work steps in live-line maintenance tasks, the system enumerates and analyzes the potential system state changes that may occur when each step fails or is interrupted, generating corresponding state rollback paths. Accident handling and emergency recovery paths are pre-computed, ensuring the system focuses not only on how to do it but also on what to do if things go wrong. This provides a direct and verifiable analytical basis for subsequent risk assessment from the dimensions of recoverability and controllability, significantly improving the engineering authenticity and completeness of the risk assessment. By introducing a set of engineering constraint rules, key factors such as whether each state rollback path supports uninterrupted power supply, whether it introduces new live-line risks, and whether additional scheduling resources are required are systematically determined, resulting in a unified set of judgment results. This transforms the rollback feasibility judgment, which previously relied heavily on expert experience, into a consistent assessment rule that can be automatically executed by the system. This avoids inconsistent judgments from different dispatchers or experts on the same rollback plan, thus improving the stability, reproducibility, and auditability of risk assessment results from a fundamental mechanism. Based on the rule-based judgment results, key factors affecting the difficulty of task recovery and the degree of risk diffusion are further extracted and comprehensively calculated into a unified quantitative operational reversibility index. This enables the system to rank and classify different live-line maintenance tasks in terms of the controllability of failure consequences, providing a unified calculation scale for matching subsequent risk creditworthiness with the operators, significantly enhancing the refinement of scheduling strategies and the controllability of safety boundaries.

[0024] Based on the rule-based judgment results, the steps for extracting the reversibility influencing factors of live-line maintenance tasks and calculating the corresponding work reversibility index are as follows: Step S141: Based on the regularized determination result, obtain state determination information related to state rollback, and extract the reversibility influence factor of the live-line maintenance task from the state determination information; Step S142, the reversibility influencing factors include uninterrupted power-off rollback factors, power-on risk introduction factors, and resource rescheduling factors, and the reversibility influencing factors are quantified and normalized. Step S143: The reversibility influence factors after quantification and normalization are fused and calculated according to the preset combination rules to obtain the corresponding operation reversibility index. The operation reversibility index is used to characterize the overall rollback capability of the live-line maintenance task in the event of execution failure or interruption.

[0025] In practical applications, based on the rule-based judgment results, further state judgment information related to state rollback is obtained, and reversibility influencing factors for live-line maintenance tasks are extracted. The completed qualitative or semi-qualitative rule-based judgment results are transformed into a structured set of influencing factors that can participate in subsequent numerical calculations. This avoids merely remaining at the binary judgment level of whether rollback is possible, but further reveals which factors influence rollback capability and from which dimensions each factor constitutes a constraint. This lays a clear and scalable parameter foundation for constructing a multi-factor comprehensive model of reversibility index. By quantifying and normalizing influencing factors with different physical meanings and dimensions, such as uninterrupted rollback factors, live-line risk introduction factors, and resource rescheduling factors, the incomparability of different factors in terms of dimensions, value ranges, and scales is eliminated. On the other hand, it provides a mathematically operable basis for subsequent fusion calculations according to unified rules, upgrading the reversibility assessment process from empirical judgment to a standardized calculation process that is repeatable, adjustable, and verifiable. According to the preset combination rules, the reversibility influencing factors after quantitative normalization are fused and calculated to generate an operation reversibility index to characterize the overall reversibility of the task. This enables the system to uniformly sort, classify and control the failure consequences of different live-line maintenance tasks in terms of controllability, providing a quantitative basis for subsequent joint matching with the risk credit ability of operators, and significantly improving the safety controllability and refinement of task allocation strategies.

[0026] The steps for obtaining historical and current work status data of personnel involved in live-line maintenance tasks, and calculating the initial risk credit value of these personnel, are as follows: Step S21: Obtain historical work data and current work status data of the workers involved in the live-line maintenance task, and perform structured processing on the historical work data and current work status data to form a worker status feature dataset. Step S22: Based on the historical operation data, extract the historical task success rate, frequency of abnormal and irregular operations, and number of high-risk task executions from the operator status feature dataset, and calculate the historical performance indicators in a comprehensive manner. Step S23: Based on the current work status data, extract the current continuous work duration, the number of tasks undertaken in the current allocation cycle, and the matching degree between the current task type and historical experience from the work personnel status feature dataset, and comprehensively calculate the status correction index. Step S24: The historical stability evaluation index and the state correction index are jointly input and combined and calculated according to the preset fusion rules to obtain the initial risk credit value of the corresponding operator, which is used to characterize the operator's basic risk-bearing capacity at the beginning of the current allocation cycle.

[0027] In practical applications, by acquiring historical and current work status data of personnel involved in live-line maintenance tasks and performing structured processing to form a personnel status characteristic dataset, this approach avoids relying solely on single-dimensional or single-moment data for static assessment of personnel capabilities. It provides comprehensive and scalable data support for subsequent comprehensive characterization of personnel risk tolerance from both historical reliability and current status perspectives. Based on historical work data, indicators such as historical task success rate, frequency of abnormal and irregular operations, and number of high-risk task executions are extracted from the personnel status characteristic dataset. These are then used to calculate historical performance indicators, characterizing the historical reliability of personnel from the perspective of long-term behavioral stability and safety records. This transforms the previous reliance on management experience to determine an individual's reliability into quantifiable, comparable, and algorithmic decision-making-compatible numerical indicators, thereby introducing an objective and traceable historical evidence basis for risk credit assessment. Based on current operational status data, the system extracts the current continuous working duration, the number of tasks undertaken within the current allocation cycle, and the matching degree between the current task type and historical experience from the operator status feature dataset. A status correction index is then calculated to dynamically characterize the operator's immediate workload and status adaptability within the current time window. This avoids static overestimation or underestimation of personnel based solely on historical performance. Instead, it incorporates realistic factors such as fatigue, load level, and experience matching degree to contextually correct the risk credit assessment results, making the assessment results closer to the actual operational safety state. The historical performance index and the status correction index are jointly input and combined according to preset fusion rules to generate an initial risk credit value representing the operator's basic risk-bearing capacity at the start of the current allocation cycle. This provides a unified initial benchmark for subsequently introducing time decay and task consumption mechanisms, upgrading personnel risk assessment from experience-based grading to a credit limit model that dynamically evolves with time and tasks, significantly improving the safety and precision of task allocation decisions.

[0028] Based on the initial risk credit value, and introducing time decay and task consumption rules, the steps for calculating the available risk credit limit for operators in the current allocation period are as follows: Step S31: Obtain the start time and current time of the current task allocation cycle, calculate the cumulative working time of the operator in the current task allocation cycle, extract the current continuous working time, and generate a time decay weight representing the degree of time consumption based on the cumulative working time and the current continuous working time. Step S32: Statistically analyze the data of tasks executed by operators in the current allocation cycle, obtain the risk level parameters and complexity level parameters corresponding to the executed tasks, and generate task consumption weights that characterize the task load based on the number of executed tasks, risk level parameters and complexity level parameters. Step S33: Perform weighted fusion processing on the time decay weight and task consumption weight to generate joint credit consumption parameters; Step S34: Extract the initial risk credit value, and perform a deduction operation on the initial risk credit value according to the joint credit consumption parameters to obtain the available risk credit limit for the operator in the current task allocation cycle.

[0029] In practical applications, the start and current times of the current task allocation cycle are obtained. The cumulative working time of the operators within the allocation cycle is calculated, and the current continuous working time is extracted. Then, a time decay weight is generated based on the cumulative and continuous working times. This characterizes the energy consumption and state decline trend of operators due to long or continuous work from a time perspective, transforming the previously difficult-to-quantify cumulative fatigue effect into a weight parameter that can be used in calculations. This prevents the system from ignoring the hidden risks brought by continuous work during task allocation and, from a mechanism perspective, inhibits the excessive concentration of scheduling for the same personnel. By statistically analyzing the number of tasks executed by operators within the current allocation cycle and obtaining the corresponding risk level and complexity level parameters, task consumption weights are generated. This characterizes the risk-bearing capacity already consumed by operators within the current cycle from two dimensions: task load intensity and risk exposure degree. This avoids using only the number of tasks as the basis for workload judgment, but introduces task risk level and complexity factors to differentiate the consumption effects of different tasks, making the risk credit consumption model more consistent with the actual engineering risk distribution characteristics. By weighting and fusing the time decay weight and the task consumption weight, a joint credit consumption parameter is generated. This avoids biases caused by single-dimensional evaluation, enabling the system to comprehensively reflect the cumulative impact of longer operation times and heavier tasks on personnel's risk tolerance. This provides a unified and adjustable consumption scale for subsequent deductions of initial risk credit values. By extracting the initial risk credit value of operators and deducting it according to the joint credit consumption parameter, the available risk credit limit for the current allocation period is obtained. This transforms the personnel's original risk tolerance capacity into a real-time available risk limit under current time and task load conditions. Risk credit is no longer a static personnel attribute but a dynamic constraint resource that continuously changes with time and task execution. This provides a directly comparable and controllable quantitative basis for subsequent task access judgments and allocation decisions.

[0030] The steps for extracting the initial risk credit value and deducting it according to the joint credit consumption parameters to obtain the available risk credit limit for the operator within the current task allocation cycle are as follows: Step S341: Extract the initial risk credit value, and determine the deduction method to perform deduction calculation on the initial risk credit value according to the preset consumption range where the joint credit consumption parameter is located. The deduction method includes a proportional deduction strategy and a segmented deduction strategy. Step S342: When adopting the proportional deduction strategy, the joint credit consumption parameter is mapped to the deduction ratio, and the initial risk credit value is deducted proportionally according to the deduction ratio to obtain the preliminary credit limit result. Step S343: When adopting the segmented deduction strategy, multiple joint credit consumption parameter ranges are preset, the range to which the joint credit consumption parameter belongs is determined, and the segmented deduction operation is performed on the initial risk credit value according to the deduction rule matched by the range to obtain the preliminary credit limit result. Step S344: Perform boundary constraint processing on the preliminary credit limit result to ensure that it is not less than zero and not greater than the initial risk credit value, thereby obtaining the available risk credit limit for the operator in the current task allocation cycle.

[0031] In practical applications, by extracting the initial risk credit value of operators and based on the preset consumption range where the joint credit consumption parameters lie, the system automatically selects either a proportional deduction strategy or a segmented deduction strategy to perform deduction calculations on the initial risk credit value. This allows the credit consumption mechanism to adopt deduction methods with different sensitivities and penalty intensities according to different consumption intensity ranges, avoiding the use of the same linear deduction model for minor and severe consumption. This achieves a non-linear constraint effect of smooth adjustment for small consumption and rapid tightening for large consumption, thereby improving the stability and security of the risk control strategy. When selecting a proportional deduction strategy, the joint credit consumption parameters are mapped to the corresponding deduction ratio. The initial risk credit value is then proportionally deducted according to this ratio to generate a preliminary credit limit result. This ensures that the credit limit changes steadily and gradually with consumption changes, avoiding drastic jumps in allocation decisions caused by small fluctuations in status, and improving the robustness and controllability of the scheduling system within the normal operating range. When selecting a segmented deduction strategy, multiple joint credit consumption parameter ranges are preset, and corresponding deduction rules are matched according to the range to which the joint credit consumption parameters belong. Segmented deduction calculations are then performed on the initial risk credit value to generate a preliminary credit limit result. When credit consumption reaches a high or dangerous range, the deduction intensity is rapidly amplified to achieve forced convergence and hard limitation of high-risk states. This allows the system to quickly compress the available task space when personnel conditions significantly deteriorate or risk exposure accumulates excessively, thereby preventing further risk accumulation. By applying boundary constraints of no less than zero and no greater than the initial risk credit value to the preliminary credit limit result, the available risk credit limit for operators within the current task allocation cycle is generated. This prevents unreasonable situations such as negative credit limits or reverse gains due to abnormal parameters, extreme working conditions, or numerical fluctuations, ensuring the stability, interpretability, and engineering feasibility of the risk credit model at the system level.

[0032] The steps of using the reversibility index of the operation and the available risk credit limit as joint inputs to perform a dual-threshold allocation admission judgment, and performing matching reasoning on the live-line maintenance task and the operator to generate the task allocation result under the condition that the dual-threshold allocation admission judgment is met, are as follows: Step S41: Obtain the data of tasks to be executed in the current allocation period. The data of tasks to be executed includes the risk level parameter and the complexity level parameter of the tasks to be executed. Configure the corresponding reversibility access threshold and credit access threshold according to the data of tasks to be executed. Step S42: Compare the reversibility index of the operation and the available risk credit limit with the reversibility access threshold and the credit access threshold respectively, perform a joint determination of the two thresholds, and generate an access determination result of whether or not to allow entry into the task matching stage. Step S43: When the admission determination result is that the task matching stage is allowed, the job requirement parameters of the task to be executed and the execution capability parameters of the candidate workers are obtained. Matching reasoning is performed on the job requirement parameters and execution capability parameters to generate task allocation results and execute them.

[0033] In practical applications, by acquiring the data of tasks to be executed within the current allocation cycle and dynamically configuring corresponding reversibility and credit access thresholds for different tasks based on their risk level and complexity level parameters, a hierarchical control mechanism is implemented, with high thresholds for high-risk tasks and low thresholds for low-risk tasks. This enables the scheduling system's safety constraint strategy to have task-adaptive capabilities. By comparing the job reversibility index and available risk credit limit with the corresponding reversibility and credit access thresholds, a joint determination of the two thresholds is performed, generating an access determination result for whether to allow entry into the task matching stage. This avoids making one-sided decisions based solely on a single dimension (such as only considering task difficulty or only considering personnel capabilities), establishing a dual insurance mechanism of task-side reversibility and personnel-side risk-bearing capacity from an institutional perspective, significantly reducing the probability of accidents escalating due to incorrect allocation. When the admission judgment result allows entry into the task matching stage, the operation requirement parameters of the task to be executed and the execution capability parameters of the candidate operators are obtained. The matching reasoning of the two is performed to generate and execute the task allocation result. Under the premise of passing the safety gate control, the qualified candidates are intelligently optimized and allocated in a way that is guided by efficiency and suitability. This achieves a hierarchical decision-making architecture that prioritizes safety and then optimizes efficiency. The system will not break the safety boundary due to blindly pursuing efficiency, nor will it abandon the intelligent optimization space within the safety range due to excessive conservatism, thus balancing operational safety and work efficiency.

[0034] After the live-line maintenance task is completed, the available risk credit limit of the corresponding operator is updated based on the task execution results for subsequent live-line maintenance task allocation. The specific steps are as follows: Step S51: Obtain the execution result information of the live-line maintenance task. The execution result information includes task completion index, abnormal impact index, risk deviation index and efficiency deviation index. Construct corresponding task execution evaluation tags based on the execution result information. Step S52: Preset multi-level task execution evaluation levels, and determine the evaluation level to which the execution result of the current live-line maintenance task belongs based on the task execution evaluation label; Step S53: Establish the mapping relationship between the evaluation level and the credit adjustment strategy, and determine the credit adjustment strategy that matches the operator. The credit adjustment strategy includes a reward-based replenishment strategy, a neutral maintenance strategy, and a penalty-based deduction strategy. Step S54: Based on the credit adjustment strategy, perform an increase or decrease operation on the available risk credit limit in the current allocation cycle to update the available risk credit limit of the operators for subsequent live-line maintenance task allocation.

[0035] In practical applications, by acquiring the execution results information of live-line maintenance tasks and comprehensively analyzing task completion indicators, anomaly impact indicators, risk deviation indicators, and efficiency deviation indicators, corresponding task execution evaluation labels are constructed. This provides a unified and standardized input carrier for subsequent grading and judgment of work quality and selection of credit adjustment strategies, making the execution result evaluation process comparable, reusable, and automatable. By pre-setting multi-level task execution evaluation grades and determining the evaluation grade to which the current task execution result belongs based on the task execution evaluation labels, continuous or multi-dimensional execution quality performance is mapped to discrete and controllable grade ranges. This avoids unstable or overly sensitive credit adjustments based directly on original indicators. Instead, a buffer layer is introduced through a grade segmentation mechanism to improve the stability and controllability of the credit evolution process, while also facilitating rule configuration and strategy interpretation by management personnel. By establishing a mapping relationship between evaluation levels and credit adjustment strategies, and accordingly determining reward-based replenishment strategies, neutral maintenance strategies, or penalty-based deduction strategies that match the operators, a positive and negative feedback adjustment mechanism is constructed. This mechanism restores credit for good performance and tightens permissions for poor performance, enabling the system to evolve by self-reinforcing safe behaviors and suppressing risky behaviors. Based on the credit adjustment strategy, the available risk credit limit within the current allocation cycle is increased or decreased to update the operator's risk credit status. This directly feeds back the execution result of a single task to the operator's eligibility and scope for subsequent task allocation, creating a closed-loop evolutionary structure of evaluation-consumption-execution-feedback-re-evaluation for the risk credit model. Consequently, the scheduling system is no longer a static rule system, but an adaptive safety scheduling mechanism that can continuously self-calibrate based on the long-term performance of personnel.

[0036] The AI-based intelligent task allocation system for live-line maintenance, by applying the AI-based intelligent task allocation method for live-line maintenance described above, includes: The reversibility assessment module obtains information on the live-line maintenance tasks to be assigned, performs regular analysis on the state rollback path of the live-line maintenance tasks in the event of execution failure or interruption, extracts the reversibility influencing factors of the live-line maintenance tasks, and calculates the corresponding operation reversibility index. The risk credit modeling module acquires historical and current work status data of personnel involved in live-line maintenance tasks and calculates the initial risk credit value of the personnel. The credit consumption calculation module, based on the initial risk credit value, introduces time decay rules and task consumption rules to calculate the available risk credit limit for operators in the current allocation cycle; The matching decision module uses the work reversibility index and available risk credit limit as joint judgment inputs, performs dual threshold allocation admission judgment, and performs matching reasoning on the live maintenance task and the operator under the condition of satisfying the dual threshold allocation admission judgment to generate task allocation results. The credit feedback update module updates the available risk credit limit of the corresponding operator based on the task execution results after the live-line maintenance task is completed, for use in the subsequent allocation of live-line maintenance tasks.

[0037] In practical applications, the reversibility assessment module transforms the reversibility of live-line maintenance tasks under abnormal conditions from experience-based judgment into a quantifiable operational reversibility index, providing a quantitative criterion for task risk controllability. The risk credit modeling module integrates personnel's historical work performance and current work status to construct an initial risk credit value model reflecting the basic risk-bearing capacity of the personnel. The credit consumption calculation module transforms personnel risk credit from a static initial value into an available risk credit limit that dynamically decays over time and with task load. The judgment and matching decision-making module performs admission judgment and intelligent matching reasoning under the dual constraints of task controllability and personnel capacity, generating safe and controllable task allocation results. The credit feedback update module feeds back the task execution results to the personnel risk credit status, achieving closed-loop evolution and adaptive adjustment of risk credit.

[0038] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. An intelligent task allocation method for live-line maintenance based on artificial intelligence, characterized in that, Includes the following steps: Obtain information on live-line maintenance tasks to be assigned, perform regular analysis on the state rollback path of live-line maintenance tasks in the event of execution failure or interruption, extract the reversibility influencing factors of live-line maintenance tasks, and calculate the corresponding operation reversibility index. Obtain historical and current work status data of personnel involved in live-line maintenance tasks, and calculate the initial risk credit value of the personnel. Based on the initial risk credit value, a time decay rule and a task consumption rule are introduced to calculate the available risk credit limit for the operator in the current allocation cycle. The reversibility index of the operation and the available risk credit limit are used as joint judgment inputs to perform a dual threshold allocation admission judgment. Under the condition of satisfying the dual threshold allocation admission judgment, matching reasoning is performed on the live maintenance task and the operator to generate the task allocation result. After the live-line maintenance task is completed, the available risk credit limit of the corresponding operator is updated based on the task execution results for subsequent live-line maintenance task allocation.

2. The intelligent task allocation method for live-line maintenance based on artificial intelligence according to claim 1, characterized in that, The steps of obtaining information on live-line maintenance tasks to be assigned, performing regular analysis on the state rollback paths of live-line maintenance tasks in the event of execution failure or interruption, extracting the reversibility influencing factors of live-line maintenance tasks, and calculating the corresponding job reversibility index are as follows: Obtain information on live-line maintenance tasks to be assigned, including the work object, work method, and work step sequence; For the live-line maintenance task, based on the sequence of its operation steps, the system state changes in the case of execution failure or interruption of each operation step are enumerated and analyzed to generate the corresponding state rollback path. The state rollback path is subject to rule-based determination, which includes whether the rollback process can be completed under power outage conditions, whether the rollback operation introduces new power-related risks, and whether the rollback process requires additional scheduling resources, to obtain the rule-based determination result. Based on the rule-based judgment results, the reversibility influencing factors of live-line maintenance tasks are extracted, and the corresponding work reversibility index is calculated.

3. The intelligent task allocation method for live-line maintenance based on artificial intelligence according to claim 2, characterized in that, The step of extracting the reversibility influencing factor of the live-line maintenance task based on the rule-based judgment result and calculating the corresponding operation reversibility index is as follows: Based on the rule-based determination results, obtain state determination information related to state rollback, and extract the reversibility influence factor of the live-line maintenance task from the state determination information. The reversibility influencing factors include uninterrupted power outage reversibility factors, power-on risk introduction factors, and resource rescheduling factors. The reversibility influencing factors are quantified and normalized. The reversibility influence factors after quantification and normalization are fused and calculated according to the preset combination rules to obtain the corresponding operation reversibility index. The operation reversibility index is used to characterize the overall rollback capability of live-line maintenance tasks in the event of execution failure or interruption.

4. The intelligent task allocation method for live-line maintenance based on artificial intelligence according to claim 1, characterized in that, The step of obtaining historical work data and current work status data of personnel participating in live-line maintenance tasks, and calculating the initial risk credit value of the personnel, specifically includes: Obtain historical work data and current work status data of the personnel involved in the live-line maintenance task, and perform structured processing on the historical work data and current work status data to form a personnel status feature dataset; Based on the historical operation data, the historical task success rate, frequency of abnormal and irregular operations, and number of high-risk task executions are extracted from the operator status feature dataset, and historical performance indicators are calculated comprehensively. Based on the current work status data, the current continuous work duration, the number of tasks undertaken in the current allocation cycle, and the matching degree between the current task type and historical experience are extracted from the work personnel status feature dataset, and a status correction index is calculated in a comprehensive manner. The historical stability evaluation index and the state correction index are input together and combined according to the preset fusion rules to obtain the initial risk credit value of the corresponding operator, which is used to characterize the operator's basic risk-bearing capacity at the beginning of the current allocation cycle.

5. The intelligent task allocation method for live-line maintenance based on artificial intelligence according to claim 4, characterized in that, The step of calculating the available risk credit limit for operators in the current allocation cycle based on the initial risk credit value, by introducing time decay rules and task consumption rules, is as follows: Obtain the start time and current time of the current task allocation cycle, calculate the cumulative working time of the operator in the current task allocation cycle, extract the current continuous working time, and generate a time decay weight that represents the degree of time consumption based on the cumulative working time and the current continuous working time. The system collects data on tasks executed by operators within the current allocation cycle, obtains the risk level parameters and complexity level parameters corresponding to the executed tasks, and generates task consumption weights that characterize the task load based on the number of executed tasks, risk level parameters, and complexity level parameters. The time decay weight and task consumption weight are weighted and fused to generate a joint credit consumption parameter; Extract the initial risk credit value, and deduct the initial risk credit value according to the joint credit consumption parameter to obtain the available risk credit limit for the operator in the current task allocation cycle.

6. The intelligent task allocation method for live-line maintenance based on artificial intelligence according to claim 5, characterized in that, The step of extracting the initial risk credit value and deducting it according to the joint credit consumption parameters to obtain the available risk credit limit for the operator in the current task allocation cycle is as follows: Extract the initial risk credit value, and determine the deduction method to perform deduction calculation on the initial risk credit value according to the preset consumption range in which the joint credit consumption parameter is located. The deduction method includes a proportional deduction strategy and a segmented deduction strategy. When using a proportional deduction strategy, the joint credit consumption parameter is mapped to a deduction ratio, and a proportional deduction operation is performed on the initial risk credit value according to the deduction ratio to obtain a preliminary credit limit result. When adopting a segmented deduction strategy, multiple joint credit consumption parameter ranges are preset, the range to which the joint credit consumption parameter belongs is determined, and the segmented deduction operation is performed on the initial risk credit value according to the deduction rule matched by the range to obtain the preliminary credit limit result. Boundary constraints are applied to the preliminary credit limit result to ensure that it is not less than zero and not greater than the initial risk credit value, thereby obtaining the available risk credit limit for the operator within the current task allocation cycle.

7. The intelligent task allocation method for live-line maintenance based on artificial intelligence according to claim 1, characterized in that, The steps of using the work reversibility index and available risk credit limit as joint inputs to perform a dual-threshold allocation admission judgment, and performing matching reasoning on the live-line maintenance task and the operator to generate the task allocation result under the condition that the dual-threshold allocation admission judgment is met, are as follows: Obtain the data of tasks to be executed within the current allocation period. The data of tasks to be executed includes the risk level parameter and the complexity level parameter of the tasks to be executed. Configure the corresponding reversibility access threshold and credit access threshold according to the data of tasks to be executed. The reversibility index of the operation and the available risk credit limit are compared with the reversibility access threshold and the credit access threshold, respectively. A joint determination of the two thresholds is performed to generate an access determination result on whether or not to allow entry into the task matching stage. When the admission determination result indicates that entry into the task matching stage is permitted, the job requirement parameters of the task to be executed and the execution capability parameters of the candidate workers are obtained. Matching reasoning is performed on the job requirement parameters and execution capability parameters to generate task allocation results and execute them.

8. The intelligent task allocation method for live-line maintenance based on artificial intelligence according to claim 1, characterized in that, The step of updating the available risk credit limit of the corresponding operator based on the task execution result after the completion of the live-line maintenance task, for use in the subsequent allocation of live-line maintenance tasks, is as follows: Obtain the execution result information of the live-line maintenance task. The execution result information includes task completion index, abnormal impact index, risk deviation index and efficiency deviation index. Construct corresponding task execution evaluation tags based on the execution result information. A multi-level task execution evaluation system is preset, and the evaluation level to which the execution result of the current live-line maintenance task belongs is determined based on the task execution evaluation label. Establish a mapping relationship between the evaluation level and the credit adjustment strategy, and determine the credit adjustment strategy that matches the operator. The credit adjustment strategy includes a reward-based replenishment strategy, a neutral maintenance strategy, and a penalty-based deduction strategy. Based on the credit adjustment strategy, the available risk credit limit within the current allocation period is increased or decreased to update the available risk credit limit for operators, which is then used for subsequent live-line maintenance task allocation.

9. An intelligent task allocation system for live-line maintenance based on artificial intelligence, characterized in that, The application of the AI-based intelligent task allocation method for live-line maintenance as described in any one of claims 1-8 includes: The reversibility assessment module obtains information on the live-line maintenance tasks to be assigned, performs regular analysis on the state rollback path of the live-line maintenance tasks in the event of execution failure or interruption, extracts the reversibility influencing factors of the live-line maintenance tasks, and calculates the corresponding operation reversibility index. The risk credit modeling module acquires historical and current work status data of personnel involved in live-line maintenance tasks and calculates the initial risk credit value of the personnel. The credit consumption calculation module, based on the initial risk credit value, introduces time decay rules and task consumption rules to calculate the available risk credit limit for operators in the current allocation cycle; The matching decision module uses the work reversibility index and available risk credit limit as joint judgment inputs, performs dual threshold allocation admission judgment, and performs matching reasoning on the live maintenance task and the operator under the condition of satisfying the dual threshold allocation admission judgment to generate task allocation results. The credit feedback update module updates the available risk credit limit of the corresponding operator based on the task execution results after the live-line maintenance task is completed, for use in the subsequent allocation of live-line maintenance tasks.

Images