Power enterprise human resource intelligent portrait modeling method

Abstract

The invention discloses a power enterprise human resource intelligent portrait modeling method, which relates to the technical field of power human resources, and comprises the following steps: uniformly abstracting multi-source data such as a human resource system, a work order system, a job ticket system, safety supervision, training examination and the like into a power business event; constructing a power operation and maintenance knowledge graph covering employees, teams, equipment, courts and task types, and forming a multi-time-scale employee portrait and a team portrait for distinguishing slow variables and fast variables on the graph; secondly, a task scene portrait is generated in combination with work order information and power grid working conditions, risk budget constraints based on the portrait are introduced, and scheduling and dispatching matching is completed on the premise that qualification and regulations are met; and continuously correcting the portrait and risk budget by using the task result event, and periodically re-estimating the event weight, so that the portrait, the scheduling and the result form a closed loop, thereby improving the overall controllability of the high-risk operation and the refinement and transparency level of the scheduling decision.

Description

Technical Field

[0001] This invention relates to the field of power human resources technology, specifically to a method for intelligent profiling and modeling human resources in power enterprises. Background Technology

[0002] In the power system, large and medium-sized power enterprises are responsible for the construction, operation, and maintenance of wide-area power grids, while grassroots power supply stations and operation and maintenance teams are responsible for planned maintenance, fault repair, and emergency power supply. These tasks are high-risk, requiring strict adherence to safety regulations and rapid deployment of personnel in extreme weather or large-scale power outages. Although enterprises have established internal information systems for human resources, work orders, work permits, safety supervision, training, and power grid dispatching, it remains challenging to conduct refined and dynamic management of frontline personnel's capabilities, fatigue levels, and safety risks.

[0003] Currently, human resources systems primarily maintain basic employee information, job titles, qualification certificates, performance records, and training records. Some organizations have developed employee profiles and digital HR functions to statistically display employee structure and performance evaluations. However, these profiles are mostly based on static master data and annual / quarterly indicators, with low update frequency, making it difficult to reflect changes in employee capabilities and fatigue accumulation during recent emergency repairs, consecutive night shifts, and high-risk operations. Furthermore, data is isolated between human resources systems and production business systems such as work orders, work permits, safety supervision, and training. Related business events are difficult to uniformly collect and model at the employee and team levels, making existing profiles inadequate to reflect the true state of personnel in specific power supply scenarios.

[0004] In terms of scheduling and work assignment, the current common practice is to manually compile schedules based on shift rules and experience, or to set shift templates and time constraints in general scheduling software. Personnel selection for high-risk tasks mainly relies on the subjective judgment of shift supervisors, team leaders, and dispatchers. Even when some systems incorporate qualification information, it often remains at the level of simple filtering, rarely utilizing multi-source data such as work order history, safety violations and accidents, training results, and power grid condition predictions in a unified manner. There is a lack of technical pathways to continuously feed task execution results and safety indicators into personnel evaluation and scheduling strategies. Under these circumstances, managers find it difficult to promptly grasp the intensity of high-risk operations and risk exposure levels of personnel within a scheduling cycle, and also struggle to coordinate the manpower needs of key equipment, key areas, and important users. This easily leads to improper personnel matching and resource imbalance, increasing safety hazards and affecting the quality of power supply services.

[0005] Therefore, the current technical problem to be solved is that, given the complex operating scenarios and strict safety requirements of power companies, existing technologies are unable to reflect the capabilities, fatigue, and safety status related to production and business events at the employee and team level in a timely manner. They are also unable to provide a unified and continuously updated computational basis for scheduling, work assignment, and training decisions based on comprehensive power grid operating conditions and safety risk constraints. Summary of the Invention

[0006] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a method for intelligent profiling and modeling human resources in power enterprises. This method includes constructing a power operation and maintenance knowledge graph covering employees, work teams, equipment, transformer substations, and task types, and creating multi-timescale employee and work team profiles on the graph that distinguish between slow and fast variables. Secondly, it combines work order information with power grid operating conditions to generate task scenario profiles, introduces risk budget constraints based on these profiles, and completes scheduling and work assignment matching under the premise of meeting qualifications and procedures. Thirdly, it continuously corrects the profiles and risk budgets using task outcome events, periodically reassessing event weights to create a closed loop between profiles, scheduling, and outcomes. This method solves the technical problems in the background technology.

[0007] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: A method for intelligent profiling of human resources in power enterprises includes collecting employee data and business data from human resources and production business systems, abstracting the data into power business events that record event types, participating employees and work objects, writing power business events into a power operation and maintenance knowledge graph and establishing employee profiles and work group profiles. Acquire operation and maintenance tasks and power grid operating data, combine operation and maintenance tasks with power grid operating data to form task scenario profiles, set risk budgets based on employee profiles and work team profiles, and under the risk budgets, match task scenario profiles with employee profiles and work team profiles to generate scheduling and work assignment plans. For completed tasks, the task results are collected, and the task results are abstracted into task result feedback events and written into the power operation and maintenance knowledge graph. The employee profile and team profile are updated according to the task result feedback events, and the event weight and risk budget are periodically adjusted to update the shift scheduling and work assignment.

[0008] Furthermore, when data related to employees, work teams, power supply stations, equipment, transformer areas, and lines are abstracted into power business events, each power business event includes event type, participating employees, participating work teams, work objects, occurrence time, duration, work ticket level, voltage level, work environment information, and initial risk level, and is organized into a power business event sequence in chronological order.

[0009] Furthermore, the power operation and maintenance knowledge graph includes employees, work teams, power supply stations, equipment, transformer substations, lines, task types, and accident nodes. Among them, employees and work teams are connected through subordinate relationship edges, work teams are connected to equipment, transformer substations, and lines through responsibility relationship edges, employees and work teams are connected to task types through participation relationship edges, and accidents are connected to task types and employees through association relationship edges. Each relationship edge stores time information, risk information, and work environment information.

[0010] Furthermore, when generating employee profiles, education level, qualification level, length of service, and past performance are used as static features. The number of operations, operation type distribution, and violation records related to employees in the power business event sequence are statistically analyzed in multiple time windows as dynamic features. In the dynamic features, slow variables reflecting long-term capability accumulation and fast variables reflecting recent operation status are distinguished. Corresponding long-term profiles and recent profiles are maintained in the power operation and maintenance knowledge graph.

[0011] Furthermore, when compiling and forming team profiles and scenario personnel experience profiles, the employee and team affiliation relationship in the power operation and maintenance knowledge graph is used as the basis. The employee profiles belonging to the same team are aggregated into team profiles in terms of capability and safety dimensions. And along the association path between employees and task types, equipment, transformer areas and lines, the employees' work experience in the task is accumulated and mapped to equipment, transformer areas and line nodes to form scenario personnel experience profiles.

[0012] Furthermore, when generating task scenario profiles based on power grid operating conditions, the work order type, work ticket level, voltage level, work object, and planned execution time interval are obtained for each scheduling task. Load forecasts, planned maintenance, and weather warning information for the corresponding area are obtained from the power grid dispatching and operation system. The task attributes and power grid operating condition information are associated and stored as the capacity requirement field and scenario risk field in the task scenario profile.

[0013] Furthermore, when setting risk budgets for employees, based on the safety dimensions in the employee profile, violation and accident records, and the company's safety procedures, the company sets the acceptable high-risk work hours and number of high-risk tasks for each employee within a scheduling cycle. When generating scheduling and assignment plans, the allocation of high-risk tasks is constrained so that the employee's high-risk work hours and number of high-risk tasks within that scheduling cycle do not exceed their risk budget.

[0014] Furthermore, when determining the scheduling and assignment plan based on each profile and risk budget, for each scheduling task, the matching degree between the candidate employee and team and the task scenario profile is evaluated based on the employee profile, team profile, and scenario personnel experience profile. Candidates who meet the qualification conditions are screened, and candidates with insufficient risk budget are removed from the screening results. The employees and teams to perform the scheduling tasks are determined according to the matching degree.

[0015] Furthermore, when collecting task execution results and safety incidents, the task completion status is obtained from the work order management system, violation and accident records are obtained from the safety supervision system, and complaint and evaluation records related to the task are obtained from the customer service system. After the above records are matched with the employees and work groups in the scheduling and dispatching plan, they are abstracted into new power business events, and the capability dimension, safety dimension and risk budget in the relevant employee profiles are updated incrementally accordingly.

[0016] Furthermore, when calculating safety indicators based on the updated power business events, power business events are grouped according to task scenario profile characteristics within a preset statistical period. The number of accidents and violations in each group are counted separately, and the statistical results are compared with the enterprise's safety objectives. Based on the comparison results, the weight parameters of power business events in employee profile calculation and the risk budget-related parameters in scheduling and dispatching schemes are adjusted.

[0017] (III) Beneficial Effects This invention provides a method for intelligent profiling and modeling human resources in power enterprises, which has the following beneficial effects: By abstracting human resources and production business data into power business events and writing them into the power operation and maintenance knowledge graph, the associations of employees, teams and equipment, scenarios and task types are carried in the same structure. This breaks down the isolation between personnel data and production data, and enables subsequent profiling and scheduling decisions to be based on a unified business chain.

[0018] By constructing employee and team profiles based on the power operation and maintenance knowledge graph, which include static features as well as slow and fast variables, basic information, task experience, safety behavior, recent high-risk operations, night shifts, and violations are uniformly expressed, comprehensively reflecting capabilities and risk status, and providing a refined basis for personnel selection for high-risk operations, emergency repairs, and training tasks.

[0019] By summarizing work records along the associated paths of employees, work teams, task types, equipment, transformer areas, and lines, scenario-based personnel experience profiles are formed. This allows equipment, transformer areas, and line nodes to record maintenance experience. When generating task scenario profiles and matching them with employee and work team profiles, personnel capabilities and scenario experience are comprehensively considered. This unifies the information foundation of people and power grid scenarios, reducing improper personnel use due to unfamiliarity with scenarios in complex operations.

[0020] By incorporating power grid operating condition information into task scenario profiles and setting risk budgets for high-risk operations based on employee and work group profiles, the system constrains qualifications, scenario risks, and risk budgets when generating scheduling and dispatch plans. This allows high-risk tasks to be allocated among employees and work groups according to their profile status, preventing high-risk operations from being concentrated on a few individuals, balancing key tasks with overall risks, and achieving a match between people, tasks, and power grid scenarios. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the intelligent profiling modeling method for human resources in power enterprises according to the present invention. Detailed Implementation

[0022] 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. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Please see Figure 1 This invention provides a method for intelligent profiling and modeling of human resources in power enterprises, including: Step 1: In the daily operation of power companies, human resource management systems, work order management systems, work permit systems, safety supervision systems, and training and examination systems record basic personnel information, work order processing procedures, safety permit circulation, violations and accidents, training and examinations, etc.

[0024] This data was initially collected and stored in a scattered manner across different systems, with inconsistent primary keys, time fields, and object identifiers. This meant that subsequent analysis often had to be performed within a single system, making it impossible to reconstruct the complete workflow of an operation along the chain of employee-workgroup-power station-equipment-transformer area-line-task type. If this scattered data format continued, subsequent employee profiling and scheduling decisions would remain at the level of static ledgers, failing to reflect the temporal and scenario characteristics of actual maintenance activities. Therefore, it was necessary to first unify, abstract, and structure the multi-source data.

[0025] First, using power business events as the core intermediary, records from different business systems are merged into a unified event structure. In this structure, each business event must be associated with specific personnel, work teams, and power grid scenario objects. Above this event layer, a power operation and maintenance knowledge graph is built, connecting employee nodes, work team nodes, power supply station nodes, equipment nodes, transformer area nodes, line nodes, task type nodes, and accident nodes with relationships. This allows for querying at any time along the graph structure who, where, what they did, and which equipment and scenarios they were related to.

[0026] First, the data dictionaries of each business system are uniformly sorted out, and the employee number field in the human resources management system is mapped one by one with the responsible person field in the work order management system, the operator field in the work ticket system, the violator field in the safety supervision system, and the examinee field in the training and examination system to establish a unique employee identifier. A unified identifier is established for objects such as work teams, power supply stations, equipment, transformer areas, and lines. Then, taking a single work activity as the basic unit, key fields such as start time, end time, work object, participants, and work environment are extracted from work order flow records, work ticket issuance and termination records, safety supervision and inspection records, and training and examination records. These fields are encapsulated into a power business event, and then classified according to event type into on-site maintenance events, fault repair events, patrol events, violation events, accident events, and training and examination events.

[0027] After abstraction, all power business events are sorted according to their start time. Time series of business events are maintained separately for the same employee and the same work group. Through operations such as time interval calculation and task duration calculation, temporal characteristics such as event density and the span of continuous high-intensity tasks within a certain time window can be obtained during subsequent profiling and modeling. It should be noted that the specific length of the time window is not limited in this process; instead, the event timestamps are preserved as is, allowing subsequent algorithms to flexibly set the observation window according to different business needs.

[0028] An exponential weighting function is introduced to evaluate the degree of event temporal concentration. For example, the event temporal weight for a particular employee at a given observation time is expressed as: Wherein, time difference: the time interval between the observation time and the occurrence time of the i-th business event; attenuation coefficient: the event time attenuation coefficient, which is a real number greater than 0; event weight: the time weight of the i-th business event at the observation time, which ranges from 0 to 1.

[0029] This function allows for differentiated weighting of events at different time intervals in subsequent profile calculations. When used, it transforms scattered records into unified power business events without altering the existing business system structure, forming event time series organized by employee and work group. This provides a foundation for subsequently distinguishing between recent and long-term behaviors over time.

[0030] Furthermore, based on the abstraction of power business events, a power operation and maintenance knowledge graph is constructed. Specifically, employees, work teams, power supply stations, equipment, transformer areas, lines, task types, and accidents are treated as different types of nodes. During initialization, an organizational structure tree and equipment ledger are imported to form a basic node set.

[0031] For each power business event, iterate through the information recorded in the event, such as the participating employees, their work groups, the equipment and transformer areas involved, the lines to which the work is carried out, and the task type. Establish relationship edges between the corresponding nodes. For example, the employee belongs to a work group, the work group belongs to a power supply station, the work group is responsible for the transformer area, the type of task the employee participates in, the equipment affected by the task, and the accident-related tasks. Attributes such as the event occurrence time, work ticket level, voltage level, work environment hazard level, and initial risk level are added to the edges.

[0032] In the process of constructing a knowledge graph, a set of relationship strength coefficients is defined for each type of relationship edge to characterize the cumulative impact of similar events on the closeness of node relationships.

[0033] For example, regarding the relationship of an employee's participation in a certain task type, the product of the event weight function and the task importance coefficient is accumulated in the node relationship weight. This accumulation of multiple events reflects the employee's depth of experience in that task type. The knowledge graph's storage structure can use an adjacency list or a graph database to ensure efficient traversal along the employee-task type-device-station-line path.

[0034] When used, data originally scattered across multiple systems in power companies is integrated into a unified structure based on nodes and relationships. This ensures that every power business event falls onto a specific path in the knowledge graph, achieving unified association between employees, work teams, power supply stations, equipment, transformer areas, and lines. This provides structural support for subsequent multi-entity profiling. At the same time, the explicit recording of relationship strength and time attributes in the knowledge graph allows for the rapid reconstruction of the corresponding operation and maintenance behavior chain from different query perspectives, thereby improving the transparency and interpretability of the entire modeling scheme.

[0035] After completing the abstraction of power business events and the construction of the knowledge graph, simply accumulating or averaging events still cannot effectively distinguish between an employee's long-term capabilities and recent performance, nor can it provide a comprehensive profile suitable for scheduling at the team and scenario levels. Power field operations include extreme scenarios such as high-risk work and high-load emergency repairs. An employee may perform well over a long period, but within a certain time window, due to consecutive night shifts or intensive high-risk work, their actual safety status may significantly decline.

[0036] Therefore, it is necessary to introduce a dynamic profiling mechanism with multiple time scales on the basis of a unified knowledge graph, and to incorporate time factors and scene factors into the profiling calculation.

[0037] Using knowledge graphs as a framework, profiles combining static and dynamic features are constructed for employees, work teams, and power grid scenarios. Static features reflect long-term stable attributes such as education, qualifications, length of service, and past performance. Dynamic features are aggregated along the timeline and scenario paths in the knowledge graph and further broken down into slow and fast variables. Slow variables focus on long-term capability accumulation and long-term safe behavior, while fast variables focus on recent work intensity, fatigue, and risk exposure.

[0038] Then, the employee-level profiles are aggregated from bottom to top along the knowledge graph into team profiles and power supply station profiles, and then propagated from top to bottom to scene nodes along the path of employee-task type-equipment-transformer area-line, forming personnel experience profiles of equipment, transformer area and line, thereby establishing a unified profile system among multiple subjects and multiple levels.

[0039] First, a set of static characteristics is defined for each employee in the knowledge graph, including education type, major, type and level of qualification certificates obtained, years of service, years in the current position, and performance appraisal results over the years. These static characteristics come directly from the human resource management system and are stored in the employee node attributes after being uniformly coded.

[0040] Based on this, and considering the characteristics reflecting long-term capabilities and long-term safety behaviors, a relatively long observation period is selected from the power business event sequence, such as one or several years. Within this period, events related to the employee, such as maintenance, emergency repairs, inspections, accidents, and violations, are categorized into different task types, equipment categories, and voltage levels according to paths in a knowledge graph. The number of events, the proportion of event types, and the frequency of accidents and violations under each category are cumulatively calculated. Events occurring earlier are appropriately attenuated using the aforementioned weighting function, ensuring that events closer to the observation time still have some influence in the slow variables, but the overall change is relatively gradual. This yields a slow variable vector reflecting the employee's long-term capabilities and long-term safety behaviors.

[0041] To avoid slow variables being overly sensitive or overly insensitive, the decay coefficient in the weighting function can be set within a preset range and adjusted according to enterprise management requirements: a smaller decay coefficient emphasizes long-term accumulation, while a larger coefficient places more emphasis on performance in recent years. This technique, through the fusion of static features and slow variable features, enables employee profiles to comprehensively reflect basic conditions, task experience, and safety records in the long-term dimension.

[0042] When used, managers can intuitively understand the ability level and safety reliability of employees in different task types and equipment categories through the slow variable section, providing a long-term basis for determining whether they can assume fixed responsibility areas. At the same time, the corresponding feature dimensions are also easy to aggregate in subsequent team profiles and scenario profiles.

[0043] Furthermore, to reflect employees' recent status and contextualized experience, a short observation window with a relatively short time span is selected, such as the most recent few days or weeks. Within this time window, power business events are unfolded along the knowledge graph starting from the employee node. For recent events such as nighttime emergency repairs, high-risk work tickets, continuous long-term work, and recent violations or emergencies, the event duration, start time, risk level, and work environment are extracted from the event sequence. Indicators such as the employee's time spent in high-risk operations, the longest continuous work duration, the number of nighttime operations, and the frequency of recent violations within this observation window are calculated and combined into a fast variable feature vector to characterize the current fatigue state and short-term safety risks.

[0044] After calculating fast variables at the employee level, the static, slow, and fast variable characteristics of all employees within the same work group are aggregated according to preset rules to form a work group profile. For example, by taking the maximum and average values ​​of key capability dimensions, the overall capability and core capability of the work group are characterized; by taking the average and extreme values ​​of fast variables, the overall fatigue level of the work group and the state of the most fatigued member are characterized. Subsequently, for each piece of equipment, transformer area, and line node, along the path in the knowledge graph from the transformer area the work group is responsible for, the task the employee participates in, to the equipment the task affects, the profiles of employees and work groups involved in the operation of that scenario are aggregated to form a scenario personnel experience profile, which is used to measure the level of maintenance support that the scenario has received historically.

[0045] When in use, employee dynamic variables, team profiles, and scenario personnel experience profiles are linked together in the same knowledge graph, so that at any point in time, it is possible to query the current fatigue and risk status of an employee, the overall risk load of a team, and the historical support level of a certain equipment or distribution area at the personnel level.

[0046] It provides a multi-subject dynamic profile that can be directly accessed for subsequent scheduling and work assignment. By using scenario-based personnel experience profiles, the power grid scenario is no longer an abstract geographical or electrical object, but a scenario with memory that carries specific human operation and maintenance history, thus laying the foundation for further risk identification and task matching at the scenario level.

[0047] Step 2: In the production command center and power supply station dispatch room of the power company, the on-duty personnel need to face a large number of task requests of different natures every day. These include planned maintenance tasks that are defined several weeks in advance, emergency repair tasks for sudden faults, routine inspection tasks, and routine on-duty tasks.

[0048] These tasks differ significantly in terms of voltage level, work order level, equipment involved, scope of impact, and completion deadline. However, in traditional scheduling practices, they are often roughly distinguished using labels such as maintenance, emergency repair, and patrol, and personnel are assigned based on human experience. With the establishment of multi-agent profiles in step one, if the tasks themselves are not meticulously characterized, it will be impossible to effectively connect task requirements with employee profiles, team profiles, and scenario-based personnel experience profiles.

[0049] Therefore, it is necessary to first transform the task into a task scenario profile that can be directly compared with the profile, and to incorporate information on load, fault risk and weather conditions from the power grid operating condition forecast, so that the task is no longer an abstract work order, but a specific object with both capability requirements and risk requirements.

[0050] Furthermore, the approach unfolds from top to bottom, focusing on clarifying tasks and understanding scenarios: On the one hand, basic attributes of tasks and power grid operating conditions are extracted from the work order management system and power grid dispatching and operation system to construct task scenario profiles, which reflect what the task needs to do, where it needs to be done, and in what operating environment it needs to be done; On the other hand, the task scenario profiles are directly referenced from the equipment nodes, transformer area nodes, and line nodes already established in step one, so that the tasks are naturally embedded in the power operation and maintenance knowledge graph structure, and can be integrated into the subsequent scheduling decision logic and employee profiles and team profiles for calculation.

[0051] First, the field content in the work order management system is read one by one for different types of tasks, such as planned maintenance tasks, fault repair tasks, inspection tasks, shifts, and emergency drills, so as to make the task itself clear.

[0052] Specifically, for each task record, the work order type field, work ticket level field, voltage level field, work object equipment number field, substation number field, line number field, planned start time field, planned end time field, adjustable time interval field, supply guarantee level field, and estimated construction period field are parsed. Based on the task type, the required number of personnel and role division are determined, such as whether a team leader is needed or whether personnel with special operation certificates are needed.

[0053] After parsing the basic attributes, the required capabilities for the task are mapped to the capability dimensions of the employee and team profiles from Step 1, based on the task type and the work object. For example, if the work object equipment number points to a 10kV distribution transformer and the work ticket level is a high-level work ticket, then the capability requirements for this task should include dimensions such as 10kV equipment operation experience, high-level work ticket execution experience, and transformer body maintenance experience. If the supply guarantee level field indicates that the task involves important users or major events, then a supply guarantee task experience requirement dimension is added to the task scenario profile to prioritize matching employees or teams with a good track record in similar supply guarantee tasks.

[0054] At this point, a task scenario profile has a clear structure at the capability level, which can be matched one-to-one with the corresponding capability dimensions in the employee profile and team profile, thus laying the foundation for subsequent matching calculations.

[0055] With the basic attributes of the task already clear, we further inject power grid operating condition information to characterize the scenario risks of the task under the power grid operating conditions.

[0056] The specific approach involves reading load forecasts, planned maintenance plans, equipment health status assessments, lists of important users, weather forecasts, and extreme weather warnings from the power grid dispatching and operation system for the next scheduling cycle. This information is then aggregated by region, transformer area, and line, so that each task can find its corresponding operating scenario.

[0057] Furthermore, based on three aspects—load level, failure history, and weather impact—a scenario risk score is constructed for each task's scenario. A linear weighted approach can be used to synthesize multiple risk factors into a single scalar risk value, for example: Among them, scenario risk scoring : The overall risk level of the scenario, a non-negative real number; Load risk factor: Represents the ratio of the load level of the scenario to the rated capacity in the current scheduling cycle, which can be obtained by normalizing the load forecast results, and its value is a real number between 0 and 1; Calculated based on the ratio of load forecast results to equipment rated capacity. Specifically, the maximum load rate can be set. Then, it is linearly normalized to:

[0058] The load factor, within the specified range, is obtained by dividing the load forecast from the dispatch system by the rated capacity.

[0059] Failure Risk Factor: Represents the failure occurrence rate of this scenario over a historical period. For example, it is obtained by normalizing the number of failures within a certain time window, and its value is a real number between 0 and 1; possible values ​​are: Based on statistical results such as the number of failures and the average failure interval within a certain historical window, the data is mapped to an interval through linear mapping or piecewise functions. For example, let the number of historical failures be... Maximum number of tolerable faults ; Weather risk factor: Represents the degree of impact from extreme weather, and can be mapped to a real number between 0 and 1 based on the weather warning level; mapping is performed through a rule table according to the meteorological warning level. For example, no warning is mapped to... Generally, early warnings are mapped as Major early warning is mapped to .

[0060] Load risk weight Fault risk weights and weather risk weights: both are non-negative real numbers used to characterize the relative importance of the three types of risk factors in the overall risk, and their sum is 1. , , Normalization can be achieved using simple linear normalization or piecewise functions, for example, normalizing the load factor below 80% to... 80% to 95% correspond Over 95% correspond This falls under rule configuration and can be customized.

[0061] In a typical application scenario, ahead of the summer heatwave, a power supply station used this technology to generate scenario risk scores for all transformer substations within its jurisdiction. For a main line carrying multiple commercial loads, given that the predicted load for the coming week is close to its rated capacity, there have been multiple records of equipment failures due to high temperatures in the past year, and the weather forecast includes both high temperature and thunderstorm warnings, the load risk factor corresponding to the tasks on this line is as follows: After comprehensive calculation, the scenario risk score is significantly higher than that of ordinary power distribution areas due to the combined effects of fault risk factors and weather risk factors. On-duty personnel can directly see these tasks marked as high-risk on the task list, which will lead to more careful scheduling of shifts and personnel.

[0062] When in use, the task scenario profile includes an additional quantitative indicator to reflect the risks of power grid operation; this indicator can be used in conjunction with the long-term capabilities and recent status in the employee profile, so that scheduling and assignment are truly based on the three-element relationship of people-task-power grid scenario.

[0063] Furthermore, even with task scenario profiles and scenario risk scores already established, relying solely on manual review of these profiles and risk scores to assign personnel can easily lead to situations where high-risk tasks are concentrated on a small number of key personnel, some work teams are in a high-risk state for extended periods, and some employees do not receive opportunities for training.

[0064] Meanwhile, power companies' safety regulations typically specify requirements for continuous working hours, high-risk work intervals, and observation periods after violations. However, these requirements are often only stated in the policy documents and are not translated into constraints that can be directly used in scheduling and work assignment calculations. Therefore, it is necessary to go a step further in step two, extracting key information from employee profiles, team profiles, and task scenario profiles into risk budgets and safety margins. Based on this, a multi-objective scheduling and work assignment decision-making logic should be constructed to ensure that each task allocation not only complies with policy requirements but also takes into account safety, efficiency, and fairness.

[0065] Therefore, firstly, for employees and work teams, an acceptable high-risk work budget is set based on historical high-risk work records and current fatigue status, forming the concepts of risk budget and safety margin; then, when calculating shifts and assignments, the utilization of the risk budget is used as a hard constraint, and scenario risk scores, ability matching degree, and fatigue status are uniformly incorporated into the task matching score calculation; finally, a comprehensive objective function is constructed at the global level, and the optimal value of the comprehensive objective function is obtained by solving the shift plan, thereby balancing timely task completion and reasonable personnel allocation within the safety boundary.

[0066] First, based on the slow and fast variables of employees obtained in step one, identify the high-risk operations that each employee participates in within a certain time window.

[0067] For each employee, the cumulative time spent on high-risk tasks during the observation period is recorded as the cumulative time spent on high-risk tasks. At the same time, based on the employee's qualifications, length of service, past accidents and violations, and safety training, a permissible budget duration for high-risk operations is set for the employee. The cumulative duration of high-risk operations can be obtained by summarizing the duration of events marked as high-risk in step one, while the total individual risk budget can be set by the safety management department according to the nature of the job and individual circumstances.

[0068] Based on this, the risk budget utilization rate is calculated for each employee to evaluate the extent to which the employee utilizes risk resources within the current period, and can be defined as: Among them, employee risk budget utilization rate: a dimensionless ratio, ranging from 0 to positive infinity, when the cumulative time of high-risk operations is less than the total individual risk budget. hour, It is between 0 and 1, and when it is equal to 1, Equal to 1, when Exceeding hour, A value greater than 1 indicates that the risk budget has been overdrawn; Cumulative duration of high-risk operations: a non-negative real number, representing the total duration of all high-risk operations performed by the employee during the observation period; Total personal risk budget: A positive real number, representing the total amount of high-risk work hours that the employee can tolerate during the observation period without exceeding the safety red line.

[0069] For employees or work groups Its total risk budget can be recorded as An employee's individual allowed budget duration can be viewed as an allocation value for the current period. This is recorded at the individual level as the allowed budget duration. This is equivalent to the total amount of the main risk budget within the current cycle.

[0070] For team risk budgeting, the sums of all team members' data or weighted data can be added together to obtain the team's cumulative high-risk operation time and the team's allowed high-risk budget time. Then, the team's risk budget utilization rate can be calculated in the same way. In this way, we can control an individual employee's overexertion at the individual level, and control the overall risk exposure at the team level, avoiding the concentration of most high-risk tasks in the same team.

[0071] When in use: When scheduling and calculating work assignments, as long as the risk budget utilization rate is limited to no more than the preset threshold, it can be ensured that no employee or work group is subjected to high-risk work tasks that exceed their safety tolerance. At the same time, for employees or work groups with low risk budget utilization rates, an appropriate amount of high-difficulty tasks can be consciously assigned to achieve balanced training under the premise of safety.

[0072] After the risk budgets for employees and work teams have been quantified, the capability requirements, scenario risk scores, employee fatigue status, and risk budget utilization rates in the task scenario profiles need to be uniformly incorporated into the matching score calculation so that the system can provide an evidence-based score on whether an employee is suitable for undertaking a certain task.

[0073] Therefore, a task matching score can be constructed, which comprehensively considers the degree of ability matching, scenario risk, and employee fatigue level, for example: Among them, task matching score: represents the score of the first task matching task. The overall suitability of each candidate entity (which can be a single employee or a work group) for undertaking the i-th task is represented by a real number. Capability matching degree: A real number between 0 and 1, used to represent the degree of conformity between the candidate subject and the task scenario profile in terms of capability dimension. It can be calculated by the similarity of the corresponding dimensions of the employee profile, team profile, and task scenario profile; the employee capability profile can be represented as a vector. Representing task capability requirements as a vector The similarity is obtained by normalizing the dot product or cosine similarity with . ; Scenario Risk Score The aforementioned scenario risk score is a non-negative real number; employee fatigue coefficient. : Non-negative real number, used to reflect the cumulative situation of high-intensity work and nighttime tasks in the recent fast variables of the candidate subject; Among them: employee fatigue coefficient : Range of values The higher the value, the higher the recent level of fatigue.

[0074] Recent night shifts Within a defined observation window (e.g., the last 7 days), the number of night shift tasks undertaken by this entity is obtained by statistically analyzing the task time periods of business events, and can be normalized to... For example, use the number of times / maximum number of night shifts.

[0075] Cumulative duration of high-risk operations The total duration of high-risk tasks within the same observation window can be further divided by a preset maximum value. Normalization to Frequency of violations The number of violations occurring within the observation window can be divided by the maximum allowed number of violations to obtain the internal value. Weight : Non-negative real number, summing to one, used to reflect the relative contribution of night shifts, continuous high-risk operations and frequency of violations to fatigue, and can be configured by the management department based on experience.

[0076] Ability weighting coefficient The scenario risk penalty coefficient and fatigue penalty coefficient are both non-negative real numbers, used to adjust the influence of the three factors on the overall score, and can be configured by the administrator according to security policies and business needs.

[0077] When scheduling and assigning tasks, for each task, the system calculates the corresponding task matching score from the set of candidate employees or work groups. Furthermore, the risk budget utilization rate is used as a constraint, and only candidate entities with a risk budget utilization rate lower than a preset threshold are selected to participate in the scoring and ranking.

[0078] After obtaining the matching scores between all tasks and candidate subjects, a comprehensive objective function is further constructed, which weights and combines the objectives of safety, efficiency, and fairness into a scalar objective that can be solved, for example: The overall objective function is the degree of merit of the entire scheduling plan, which is a real number. Safety Target: A non-negative real number reflecting the safety performance of the plan. It can be constructed based on the degree to which high-risk tasks are assigned to high-capacity, low-fatigue subjects, and whether the risk budget utilization rate remains within the safe range; the sum of penalties when higher-risk tasks are undertaken by higher- or larger subjects. Efficiency Target: A non-negative real number reflecting the tightness of task completion time and schedule; the sum of the absolute values ​​of the differences between the completion time and the planned completion time for all tasks. Fairness Target: A non-negative real number reflecting the balance of task allocation among different employees and work groups; the sum of the absolute deviations of each employee's workload from the overall average workload; safety weighting coefficient. The efficiency weight coefficient and the fairness weight coefficient are all non-negative real numbers, representing the importance of the three types of objectives for the enterprise in this scheduling cycle. The sum of the three is 1, which facilitates subsequent comparison of schemes under different configurations.

[0079] When in use, the ability matching degree, scenario risk score, employee fatigue coefficient and risk budget utilization rate work together to affect the task matching score, and influence the final scheduling results through a comprehensive objective function. This ensures that high-risk tasks are prioritized for personnel with sufficient long-term experience and good recent performance, which can reduce the possibility of personal injury and equipment accidents. Under the premise of meeting safety constraints, it enables a more reasonable allocation of different teams and employees between high-difficulty tasks and ordinary tasks, which is conducive to the long-term development of talent pipelines.

[0080] Step 3: After scheduling and assignment are completed in Step 2, the tasks are officially executed on-site by specific work teams and employees. During on-site execution, whether the task is completed on time, whether there are overtime or rework, whether violations or accidents occur, and whether the customer is satisfied, all directly reflect whether the employee profiles, work team profiles, task scenario profiles, and comprehensive matching scores set in Steps 1 and 2 are reasonable.

[0081] If these execution results are not systematically collected and organized, but only remain in paper records or scattered written minutes, then the profiling and scheduling strategies will remain at the initial assumption level for a long time, and will be unable to be adjusted according to new work patterns, equipment problems and changes in personnel status.

[0082] Therefore, it is necessary to design a task-level result feedback path that is triggered immediately after each task is completed, to structure the key facts in the task execution process and write them back to the power business event sequence, and to accurately map them to employee nodes, team nodes and scenario nodes in the knowledge graph.

[0083] Fields related to task results are extracted from the work order system, safety supervision system, and customer service system, and these are uniformly abstracted into task result feedback events. Then, in the power operation and maintenance knowledge graph, the feedback events are hierarchically attached to the event lists of the corresponding nodes along the path of task type node - equipment node - transformer area node - line node - employee node - team node. Finally, based on the nature of the task result, the slow and fast variables related to the task in the employee profile, team profile, and scenario personnel experience profile are adjusted incrementally with a limited range to achieve the correction of capability and safety dimensions.

[0084] For each completed task, the system retrieves information from the work order system, including task completion time, whether it exceeded the time limit, whether it required rework, actual operation time, whether the originally planned equipment was replaced, and whether there was a temporary power outage adjustment. It also retrieves information from the safety supervision system, including whether any violations occurred during the task execution, the type of violation, whether any accidents or near misses were triggered, accident level records, and on-site safety inspection records. Finally, it retrieves information from the customer service system, including user complaints related to the task, power outage information releases, and customer satisfaction evaluations after the power outage ended.

[0085] The aforementioned data sources exist in different field formats. They are organized into task result feedback events to form a unified structure. The main fields include completion status identifier, time deviation identifier, safety event identifier, violation category identifier, accident level identifier, customer evaluation identifier, and on-site anomaly description field.

[0086] Next, based on the task scenario profile formed in step two, obtain the task type, equipment, transformer substation, line, affiliated power supply station, and participating employees and work teams information. In the power operation and maintenance knowledge graph, find the corresponding task type node and equipment node, and attach the task result feedback event as a new time-stamped event to the relationship edge between the task type node and the equipment node; then, along the path of equipment node-transformer substation node-line node-power supply station node, append the summary information of the event to the time-stamped sequence of each layer of scenario nodes; at the same time, add an item to the event list of employee node and work team node, so that any participating employee, any work team, and any scenario node will have the execution record of the task explicitly included in their historical event sequence.

[0087] After the task result feedback event has been successfully linked, the task results are used to make limited incremental adjustments to the capability and safety dimensions of the relevant employee and team profiles.

[0088] Specifically, for each task, a task result quality score is first calculated based on its execution results. This score comprehensively reflects the completion status, safety status, and service evaluation. For example, it can be defined as: Among them, the task result quality score The overall quality score of the indexing task after completion, with a value between 0 and 1; weighting coefficient. , , : A non-negative real number used to balance the weights of completion performance, safety performance, and service performance in the quality score, and usually satisfies a sum of 1; Finished product These are non-negative real numbers, which can be mapped to values ​​between 0 and 1 based on factors such as whether the work was completed on schedule, whether rework occurred, or whether the power outage area was temporarily expanded; safety performance. These are non-negative real numbers, which can be mapped to values ​​between 0 and 1 based on factors such as whether a violation occurred, the severity of the violation, and the level of an accident or near miss; service performance It is a non-negative real number that can be mapped to a value between 0 and 1 based on factors such as whether user complaints are generated or customer feedback ratings are received.

[0089] After obtaining the task result quality score Subsequently, for each employee and each work group involved in the task, the direction and intensity of the task's influence on the corresponding capability and safety dimensions are determined based on their role in the task (such as team leader, main operator, and assistant) and the weight of each capability dimension in the task capability requirement vector.

[0090] If the task result quality score If the score is high and there are no violations or accidents, a small positive accumulation will be made in the ability dimension related to the task type and equipment type in the slow variable of the employee profile, and the stable operation performance related characteristics will be slightly improved in the safety behavior dimension. If the task result quality score If the safety level is low, especially if a violation or accident occurs, a negative correction will be made to the safety behavior dimension of the corresponding employee, and the safety dimension of the relevant work group profile will be adjusted accordingly based on the division of responsibility for the violation.

[0091] Furthermore, for high-risk tasks, step two has already factored in the risk budget consumption based on risk intensity and operation duration. After the task results are reported, if the task is successfully completed under strict safety measures and the task result quality score is [not specified], [further details will be provided]. Higher, allowing for a larger total risk budget. Alternatively, the subsequent estimate of risk intensity may be slightly increased; conversely, if a dangerous situation or accident occurs during the task, the total risk budget allocated to relevant personnel and teams in future cycles may be correspondingly decreased. The performance of each task is transformed into visible changes in the performance profile, allowing capability accumulation and safety performance to be systematically recorded and reflected along a timeline. For a high-risk task... The risk intensity can be defined as the comprehensive risk score of the task. For linear or piecewise function mappings within a certain interval, the job duration is the actual execution time of the task, and the risk budget consumption can be regarded as the accumulation of the time.

[0092] Real-time task-level profile updates can ensure that every work record is reflected in the profiles of individuals and work teams. However, the characteristics of safe production and manpower dispatch in power companies are both long-term and phased. For example, annual safety indicators, seasonal load characteristics, equipment aging trends, and personnel turnover will affect accident occurrence patterns and task distribution patterns over a long time scale.

[0093] Relying solely on task-level adjustments is insufficient to capture these long-term trends, and the total risk budget... Comprehensive risk assessment of the task Furthermore, the weighting parameters in the comprehensive matching score will remain at the initial setting level for a long time, making it difficult to reflect the overall management goals of the enterprise and changes in the external environment. Therefore, this step needs to conduct statistical analysis and strategy reassessment on the relationship between task results, profile status, and scheduling results over a longer time window, and reflect the results as joint parameter corrections, so that the model has an opportunity to review and self-adjust from a global perspective in each cycle.

[0094] At the end of a pre-set period (e.g., a month or a quarter), the quality of the task results for all tasks within that period is scored. Comprehensive risk assessment of the task Sample the fast and slow variables of the profile status of employees and work teams when they are performing tasks, and construct a result-status comparison sample. Comprehensive Risk Score of the Task The scenario risk score of the task's location can be used directly. For more precise calculations, multiply by the inherent risk level factor of the task (e.g., the coefficient corresponding to the work order level). The inherent risk coefficient of the operation is determined by the operation ticket level, voltage level, etc., through a rule table.

[0095] Next, these samples are aggregated and analyzed to calculate statistical indicators such as accident rate, violation rate, and rework rate under different combinations of profile statuses and task scenarios. Furthermore, these statistical indicators are compared with the company's preset safety, efficiency, and equilibrium goals to construct a strategy deviation metric. Finally, the strategy deviation metric is used to assign event impact weights and total risk budget. The weighting coefficients in the comprehensive matching score are jointly adjusted to form the parameter set used in the new cycle.

[0096] Furthermore, to analyze long-term trends in safety performance, all high-risk task samples within the period are selected, and the number of tasks in which accidents or serious violations occur is recorded as the accident sample number. The total number of all high-risk tasks within the period is recorded as the total number of high-risk tasks. The proportion of observed security events can then be expressed as: Among them, the proportion of observed security incidents : Dimensionless proportion, ranging from 0 to 1, representing the average frequency of accidents or serious violations occurring for each high-risk task within the current period; number of accident samples. : A non-negative integer representing the number of accidents or serious violations recorded in the current period; total number of high-risk tasks. : A positive integer representing the total number of tasks marked as high-risk during the current period.

[0097] Furthermore, corporate management typically sets an acceptable target percentage for security incidents in their annual work objectives. This ratio can be determined based on historical levels and regulatory requirements. Therefore, the security policy deviation can be defined as: Among them, security policy deviation : Real number, which can take the values ​​of negative, zero, or positive.

[0098] In practical applications, separate safety policy deviations can be constructed for tasks at different voltage levels, different work categories, or different regions to adjust the weighting of event impacts and the total risk budget. Perform layered correction.

[0099] When used, safety performance is converted into a set of deviation signals that can be directly applied to parameter adjustments, allowing the company's long-term safety goals to enter the model in a clear numerical form, providing a basis for the next step of joint parameter correction.

[0100] Based on security policy deviations In addition, there is a deviation analysis of efficiency and equilibrium objectives, an assessment of the impact weight of events on the profile, and a total risk budget. Joint revisions will be made.

[0101] Taking event impact weighting as an example, in step one, the employee capability and safety dimensions were cumulatively calculated based on event type and scenario. The contribution of different types of events is controlled by a set of event weight coefficients. If, within a certain period, it is observed that the accident rate is significantly higher in a specific task type or scenario than in other task types, the event weights related to that task type and scenario can be amplified, making the negative impact of future events of the same type on the profile more significant.

[0102] For the total risk budget It can be based on security policy deviations Apply differentiated adjustments to employees with different profile statuses. For example, for employees with good long-term safety performance and low accident correlation in slow variables, if their overall safety performance in this period is better than the target (i.e., safety strategy deviation), further adjustments may be made. (Negative), which can slightly increase its total risk budget. This allows for future opportunities to undertake more challenging, high-risk tasks in the next cycle; for employees with a high correlation to recent accidents or a high frequency of violations, their total risk budget can be reduced individually, even if the overall safety strategy is not significantly flawed. This will enable it to primarily participate in low-risk tasks in the next cycle.

[0103] When using this, the event impact weight and the total risk budget should be considered. Instead of static configuration, the system undergoes gradual fine-tuning based on periodic policy reassessment. This allows the system's security sensitivity to dynamically change with actual incidents and violations, while maintaining controllable adjustment ranges to prevent drastic fluctuations caused by a single extreme event. The joint adjustment of event weights and risk budgets ensures that subsequent stages, such as profile building in step one and personnel scheduling in step two, reflect new security tendencies and management preferences, achieving a complete closed loop from task results to policy deviations, parameter updates, profiles, and scheduling.

[0104] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0105] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0106] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0107] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0108] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for intelligent profiling and modeling human resources in power enterprises, characterized in that: comprising, Collecting employee data and business data in the human resource and production business system, abstracting the data as power business events of record event types, participating employees and work objects, writing the power business events into a power operation and maintenance knowledge graph and establishing employee portraits and team portraits; Obtaining operation and maintenance tasks and power grid working condition data, combining the operation and maintenance tasks with the power grid working condition to form a task scene portrait, setting a risk budget according to the employee portraits and team portraits, matching the task scene portrait with the employee portraits and team portraits under the risk budget to generate a scheduling and dispatching scheme; Collecting task results for completed tasks, abstracting the task results as task result feedback events and writing them into the power operation and maintenance knowledge graph, updating the employee portraits and team portraits according to the task result feedback events, and periodically adjusting event weights and risk budgets to update the scheduling and dispatching according to the employee portraits and team portraits.

2. The power enterprise human resource intelligent portrait modeling method according to claim 1, wherein: when abstracting data related to employees, teams, power supply stations, equipment, transformer areas and lines as power business events, each power business event includes an event type, participating employees, participating teams, work objects, a time of occurrence, a duration, a work ticket level, a voltage level, work environment information and an initial risk level, and is organized as a power business event sequence in chronological order.

3. The power enterprise human resource intelligent portrait modeling method according to claim 2, wherein: the power operation and maintenance knowledge graph includes employees, teams, power supply stations, equipment, transformer areas, lines, task types and accident nodes, wherein employees and teams are connected by membership relationship edges, teams and equipment, transformer areas and lines are connected by responsible relationship edges, employees and teams are connected by participating relationship edges, accidents are connected to task types and employees by association relationship edges, and time information, risk information and work environment information are stored on each relationship edge.

4. The power enterprise human resource intelligent portrait modeling method according to claim 3, wherein: when generating an employee portrait, education, qualification level, length of service and past performance are taken as static features, and the number of work times, work type distribution and violation records of the employee related in the power business event sequence are taken as dynamic features in multiple time windows, and slow variables reflecting long-term ability accumulation and fast variables reflecting recent work state are distinguished in the dynamic features, and corresponding long-term portraits and recent portraits are maintained in the power operation and maintenance knowledge graph.

5. The power enterprise human resource intelligent portrait modeling method according to claim 4, wherein: when summarizing to form a team portrait and a scene personnel experience portrait, the employee portraits of employees belonging to the same team are aggregated in the ability dimension and the safety dimension to form a team portrait based on the membership relationship of employees and teams in the power operation and maintenance knowledge graph, and the work experience of employees on tasks is accumulated and mapped to equipment, transformer area and line nodes along the association path of employees and task types, equipment, transformer area and lines to form a scene personnel experience portrait.

6. The power enterprise human resource intelligent portrait modeling method according to claim 5, wherein: ​ ​ ​ ​ When generating the task scene portrait based on the grid operating conditions, the work order type, work order level, voltage level, work object, and planned execution time interval of each scheduling task are obtained, and the load prediction, planned maintenance, and weather warning information of the corresponding area are obtained from the grid dispatching and operation system. The task attributes and grid operating condition information are associated and stored as the ability demand field and scene risk field in the task scene portrait.

7. The power enterprise human resource intelligent portrait modeling method of claim 6, wherein: When setting the risk budget for the employees, the length of high-risk work and the number of high-risk tasks that each employee can bear within a scheduling cycle are set according to the safety dimension in the employee portrait, the violation and accident records, and the enterprise safety regulations. When generating the scheduling and dispatching scheme, the high-risk task assignment constraint is set so that the length of high-risk work and the number of high-risk tasks of the employee within the scheduling cycle do not exceed the risk budget.

8. The power enterprise human resource intelligent portrait modeling method of claim 7, wherein: When determining the scheduling and dispatching scheme according to the portraits and risk budgets, for each scheduling task, the matching degree of the candidate employees and teams with the task scene portrait is evaluated based on the employee portrait, team portrait, and scene personnel experience portrait, the candidate objects that meet the qualification conditions are selected, the candidate objects that do not meet the risk budget are excluded from the selection results, and the employees and teams that execute the scheduling tasks are determined according to the matching degree.

9. The power enterprise human resource intelligent portrait modeling method of claim 8, wherein: When collecting the task execution results and safety events, the task completion status is obtained from the work order management system, the violation and accident records are obtained from the safety supervision system, and the complaints and evaluation records related to the tasks are obtained from the customer service system. The above records are abstracted into new power business events corresponding to the employees and teams in the scheduling and dispatching scheme, and the ability dimension and safety dimension in the related employee portrait and the risk budget are incrementally updated based on the power business events.

10. The power enterprise human resource intelligent portrait modeling method of claim 9, wherein: When statistical safety indicators are based on updated power business events, the power business events are grouped according to the task scene portrait characteristics within a preset statistical period, the number of accidents and the number of violations in each group are counted, and the statistical results are compared with the enterprise safety target. The weight parameters in the employee portrait calculation and the parameters related to the risk budget in the scheduling and dispatching scheme are adjusted according to the comparison results.

Images