Power system human resource allocation method and system based on multi-agent cooperation

Abstract

A power system human resource allocation method based on multi-agent cooperation comprises the steps that S1, a human resource task collection module is constructed, and newly-added human tasks of a power system are collected in real time; s2, quantitatively decomposing each human task into a plurality of sub-tasks, calculating the complexity of each sub-task, and constructing a human task balance optimization model; s3, constructing an intelligent agent role allocation model according to the sub-task complexity, the intelligent agent capability matrix and the real-time load of the intelligent agent; and S4, performing iterative operation through a multi-objective optimization algorithm, solving the human task balance optimization model and the agent role allocation model, allocating the subtasks to the agents to obtain a task allocation matrix and estimated completion time, and executing the subtasks by the agents according to the allocation matrix to realize allocation of the human tasks. The design not only considers the balance of resource consumption and processing time of human tasks, but also allocates based on allocation efficiency and task completion efficiency.

Description

Technical Field

[0001] The present invention relates to the fields of intelligent systems, artificial intelligence, and human resource management, and particularly to a method and system for human resource allocation in a power system based on multi-agent collaboration. The system realizes the automatic decomposition and execution of complex human resource allocation tasks through an autonomous agent collaboration module, and can adapt to the multi-level and cross-departmental allocation requirements of power enterprises. The task decomposition, agent role allocation, and collaboration mechanism of the system significantly improve the allocation efficiency and are applicable to the intelligent allocation of multi-tasks in multi-level organizational management. Background Art

[0002] In the field of human resource management, traditional allocation methods usually involve a large amount of manual intervention. Especially in large multi-level organizations such as power enterprises, the demand differences for different positions are large, and the allocation difficulty is high. In the daily operation of an enterprise, human resource allocation not only involves cross-departmental collaboration but also requires reasonable allocation of resources according to the complexity of different tasks, which poses a great challenge to traditional human resource management. With the increasing diversification and complexity of business requirements, enterprises increasingly need intelligent systems that can automatically process multi-level and multi-task allocations to achieve efficient person-position matching and position management.

[0003] Current intelligent human resource allocation systems usually have difficulty handling complex allocation scenarios with multiple human tasks and multiple roles, and lack a flexible and efficient mechanism for decomposing human tasks and allocating roles. Most traditional systems are based on a single-agent rule engine. When encountering multi-human task and multi-department collaboration requirements, there are often situations of backlog of human tasks or uneven resource allocation, and they cannot quickly respond to changing business needs. Therefore, there is an urgent need for an innovative method that can automatically decompose complex human tasks, allocate roles through the collaboration of multiple agents, and record historical data through a human task memory module to dynamically adjust the human task allocation of each agent, thereby effectively solving the efficiency bottleneck problem of existing allocation systems.

[0004] The human resource allocation system based on multi-agent collaboration proposed by the present invention can automatically complete the decomposition and execution of human tasks in multi-level allocation requirements through the role collaboration of multiple autonomous agents, realize flexible resource allocation and dynamic scheduling, adapt to the multi-level and cross-departmental allocation requirements of power enterprises, and has strong practicability and applicability. Summary of the Invention

[0005] The object of the present invention is to overcome the problems existing in the prior art, and provide a method and system for human resource allocation in a power system based on multi-agent collaboration. The core of the system lies in the autonomous agent collaboration module, which decomposes complex human tasks, assigns roles, and records and manages historical data through multi-agent collaboration. Specifically, each agent independently undertakes different sub-human tasks according to the different attributes of human tasks, job requirements, and resource situations. The system can record the historical data during the execution of human tasks through the memory module, and dynamically adjust the agent role assignment according to the actual situation to achieve the automation of multi-human task collaboration and configuration management.

[0006] To achieve the above object, the technical solution of the present invention is:

[0007] In the first aspect, the present invention provides a method for human resource allocation in a power system based on multi-agent collaboration, including the following steps:

[0008] S1. Construct a human resource task collection module to collect newly added human tasks in the power system in real time;

[0009] S2. Quantitatively decompose each human task into multiple sub-tasks, calculate the complexity of each sub-task, and construct a human task balance optimization model;

[0010] S3. Construct an agent role assignment model according to the sub-task complexity, agent ability matrix, and real-time load of the agent;

[0011] S4. Perform iterative operations through a multi-objective optimization algorithm to solve the human task balance optimization model and the agent role assignment model, assign the sub-tasks to the agents, obtain a task assignment matrix and an estimated completion time, and the agents execute the sub-tasks according to the assignment matrix to achieve the assignment of human tasks.

[0012] In the above S2, using natural language processing technology, extract key features from the description of human tasks, decompose the task types, task time limits, job requirements, and task priorities of different sub-tasks, and at the same time extract the dependency relationship information between sub-tasks, where the dependency relationship includes parallel and sequential relationships; convert the unstructured job description into structured data, and perform noise removal and semantic unification;

[0013] Perform semantic analysis and weight assignment on the key features related to different sub-tasks, and calculate the complexity of each sub-task respectively:

[0014]

[0015] Among them: C m represents the complexity of sub-task m, w n is the weight factor of the nth feature, f nis the nth eigenvalue of task m;

[0016] Construct a balance optimization model for human tasks. The balance optimization function F for each human task is as follows:

[0017]

[0018] where m t is the completion status symbol of subtask m at time t. When m t = 1, it means that subtask m is not completed at time t. When m t = 0, it means that subtask m is completed at time t; R mk represents the resource consumption of subtask m assigned to agent k. M is the total number of subtasks in the personnel assignment task, and T mk is the processing time for agent k to complete subtask m. K is the total number of agents executing the personnel assignment task, and α is the balance coefficient;

[0019] Based on the task types, task time limits, position requirements, and subtask relationships of different subtasks, construct the constraints of the balance optimization function.

[0020] In step S3, agent role assignment: The initial assignment of agent roles is based on its ability matrix E = [E ij . The matrix element E ij represents the efficiency of agent i in executing task type j. E ij is obtained by weighted calculation based on the computing performance parameters of agent i and the historical completion data of different task types;

[0021] Construct an agent role assignment model based on the agent role assignment objective function S:

[0022]

[0023] where: X ij represents the decision symbol for whether task j is assigned to agent i. X ij ∈{0,1}. When X ij = 1, it means that task j is assigned to agent i. I is the total number of callable agents in the system, and J is the sum of the subtasks to be completed in the system; E ij represents the efficiency of agent i in executing task j, and L i is the current load of agent i, and λ is the load balance coefficient;

[0024] The real-time load of agent i is defined as:

[0025]

[0026] where: T iDenote the set of subtasks currently being executed by agent i; C j is the complexity of subtask j;

[0027] Based on the task types that the agent can complete, the computing performance parameters, and the ability matrix E, construct the task assignment constraints for the agent.

[0028] In S4, tasks with higher priorities will be executed by the agent first. During the process of the agent executing subtasks according to the assignment matrix, there is a situation of task migration. When the load difference |L i -L i′ | between any two agents i and i' exceeds the preset threshold ΔL, that is, it satisfies: |L i -L i′ | > ΔL, then there is a task migration between agents i and i'. Select subtasks that meet the migration conditions from the task set of the high-load agent and migrate them to the task set of the low-load agent, and then update the task assignment matrix and the load parameters of the agent.

[0029] In S4, the agent executes subtasks according to the assignment matrix. After the subtask is completed, determine whether the completion status of the subtask is successful or failed. For subtasks that fail to execute, check the task log and the execution history of the agent to determine the reason for the failure:

[0030] If the load of the actual resource consumption of the subtask is greater than the complexity of the calculated subtask, it is determined that the reason for the failure is insufficient resources. Then, assign this subtask to an agent with higher processing capabilities and increase the priority of the subtask;

[0031] If there is a logical conflict in a single subtask resulting in the failure of the subtask execution, it is determined that the reason for the failure is high complexity. Then, further divide this subtask into multiple subtasks, increase the task processing priority, and then re-assign them to multiple agents for execution;

[0032] If the number of times a subtask fails to execute exceeds the specified threshold, start an error reporting mechanism, generate an alarm, and recommend manual intervention.

[0033] In S4, the agent executes subtasks according to the assignment matrix. After the subtask is completed, record the subtask log: the agent number that completed the subtask, the human task number to which the subtask belongs, the subtask type, the key features of the subtask, the operation duration of the subtask, the actual resource consumption of the subtask operation, the completion status of the subtask; record the above subtask log information as the historical completion data of the corresponding agent.

[0034] In the method for allocating human resources in the power system, local devices and agents are preferentially used to perform the human resource allocation task. When the resource utilization rate of local devices is greater than the set threshold, a resource collaborative call mechanism is started, and the subtasks that meet the collaborative conditions locally are transferred to other departments with a resource utilization rate less than the set threshold for collaborative processing, and the task processing logs are recorded.

[0035] In a second aspect, the present invention provides a power system human resource allocation system based on multi-agent collaboration. The system is used to execute the foregoing power system human resource allocation method based on multi-agent collaboration, and specifically includes: a human resource task collection module, a task decomposition and balance optimization module, an agent role assignment module, and a solution operation module;

[0036] The human resource task collection module: is used to collect newly added human tasks in the power system in real time;

[0037] The task decomposition and balance optimization module: is used to quantitatively decompose each human task into multiple subtasks, calculate the complexity of each subtask, and construct a human task balance optimization model;

[0038] The agent role assignment module: is used to construct an agent role assignment model according to the subtask complexity, the agent ability matrix, and the real-time load of the agent;

[0039] The solution operation module: performs iterative operations through a multi-objective optimization algorithm, solves the human task balance optimization model and the agent role assignment model, allocates the subtasks to the agents, obtains a task allocation matrix and an estimated completion time, and the agents execute the subtasks according to the allocation matrix to realize the allocation of human tasks.

[0040] In a third aspect, the present invention provides a power system human resource allocation device based on multi-agent collaboration, including a memory and a processor. The memory is used to store computer program code and transmit the computer program code to the processor;

[0041] The processor is used to execute the foregoing power system human resource allocation method based on the instructions in the computer program code.

[0042] In a fourth aspect, the present invention provides a computer program product, including a computer program, and the computer program is executed by the processor to perform the foregoing power system human resource allocation method based on multi-agent collaboration.

[0043] Compared with the prior art, the beneficial effects of the present invention are:

[0044] 1. In the human task balance optimization model of a power system human resource allocation method based on multi-agent collaboration in the present invention, in combination with the actual requirements of power grid human resource management, a set of decomposition methods for complex tasks is designed. Through task characteristic analysis, the overall task is decomposed into multiple subtasks, and further hierarchical management is carried out according to priority, complexity, and resource requirements; the agents are divided into core agents and auxiliary agents, which respectively undertake the roles of task execution and task support. To adapt to the dynamic changes in task requirements, a role adjustment mechanism is introduced into the system.

[0045] 2. In the agent role assignment model of a power system human resource allocation method based on multi-agent collaboration in the present invention, specific roles are assigned to each agent to quickly respond to different job requirements. Each agent executes in parallel in the subtasks according to the role division, improving the efficiency of the configuration task. When solving the model, the balance of resource consumption and processing time of human tasks is considered simultaneously, and a task assignment model for agents is constructed based on assignment efficiency and task completion efficiency.

[0046] 3. A power system human resource allocation system based on multi-agent collaboration in the present invention includes a human resource task collection module, a task decomposition and balance optimization module, an agent role assignment module, and a solution operation module; this system is used to implement the steps of the power system human resource allocation method based on multi-agent collaboration provided in any of the above technical solutions. Therefore, this system also includes all the beneficial effects of the power system human resource allocation method based on multi-agent collaboration provided in any of the above technical solutions, which will not be elaborated here.

[0047] 4. A power system human resource allocation device based on multi-agent collaboration in the present invention includes a processor and a memory. The memory is used to store computer program code and transmit the computer program code to the processor, and the processor is used to execute the power system human resource allocation method based on multi-agent collaboration provided in any of the above technical solutions according to the instructions in the computer program code. Therefore, this device also includes all the beneficial effects of the power system human resource allocation method based on multi-agent collaboration provided in any of the above technical solutions, which will not be elaborated here.

[0048] 5. A computer program product in the present invention, when the computer program is executed by a processor, implements the steps of the power system human resource allocation method based on multi-agent collaboration provided in any of the above technical solutions. Therefore, this computer program product also includes all the beneficial effects of the power system human resource allocation method based on multi-agent collaboration provided in any of the above technical solutions, which will not be elaborated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 is the flowchart of the method of the present invention.

[0050] Figure 2 This is the system module diagram of the present invention.

[0051] Figure 3 This is the schematic diagram of the device of the present invention. Detailed implementation manners

[0052] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific implementation manners.

[0053] Example 1:

[0054] Refer to Figure 1 , a method for human resource allocation in a power system based on multi-agent collaboration, including the following steps:

[0055] S1. Construct a human resource task collection module to collect newly added human tasks in the power system in real time;

[0056] S2. Quantitatively decompose each human task into multiple subtasks, calculate the complexity of each subtask, and construct a human task balance optimization model;

[0057] In S2, using natural language processing technology, extract key features from the description of the human task, decompose the task types, task time limits, position requirements, and task priorities of different subtasks, and at the same time extract the dependency relationship information between subtasks, where the dependency relationship includes parallel and sequential relationships; convert the unstructured position description into structured data, and perform noise removal and semantic unification;

[0058] Perform semantic analysis and weight assignment on the key features related to different subtasks, and calculate the complexity of each subtask respectively:

[0059]

[0060] Among them: C m represents the complexity of subtask m, w n is the weight factor of the nth feature, and f n is the nth feature value of task m;

[0061] Construct a human task balance optimization model, and the balance optimization function F of each human task is:

[0062]

[0063] Among them, m t is the completion status symbol of subtask m at time t. When m t = 1, it means that subtask m is not completed at time t. When m t = 0, it means that subtask m is completed at time t; R mkIndicates the resource consumption of subtask m assigned to agent k. M is the total number of subtasks in the personnel assignment task, and T mk is the processing time for agent k to complete subtask m. K is the total number of agents executing the personnel assignment task, and α is the balance coefficient;

[0064] Based on the task type, task time limit, position requirements, and subtask relationships of different subtasks, construct the constraints of the balance optimization function. It should be emphasized here that the constraints of different tasks will vary. Some tasks focus more on time, some on knowledge position requirements, and some tasks need to meet all aspects simultaneously. Therefore, the task content and relevant standard materials are intelligently extracted.

[0065] Human task feature extraction and complexity analysis: Based on the professional characteristics of grid position requirements, extract features such as the skill requirements, priorities, and resource constraints of human tasks, and provide a basis for human task allocation through a complexity quantification model.

[0066] Hierarchical task decomposition: Divide the task into core tasks and auxiliary tasks. The core tasks are responsible for the agents directly completing the tasks, and the auxiliary tasks are handed over to the auxiliary agents to assist in completion.

[0067] Task parallel processing: Agents achieve the maximization of configuration efficiency by parallelly decomposing subtasks while ensuring resource utilization.

[0068] S3. Construct an agent role assignment model based on subtask complexity, agent ability matrix, and the real-time load of agents;

[0069] In S3, agent role assignment: The initial assignment of agent roles is based on their ability matrix E = [E ij , and the matrix element E ij represents the efficiency of agent i in executing task type j. E ij is obtained by weighted calculation based on the computing performance parameters of agent i and the historical completion data of different task types;

[0070] Construct an agent role assignment model based on the agent role assignment objective function S:

[0071]

[0072] Where: X ij represents the decision symbol for whether task j is assigned to agent i. X ij ∈{0,1}. When X ij = 1, it means that task j is assigned to agent i. I is the total number of callable agents in the system, and J is the sum of the subtasks to be completed in the system; E ij represents the efficiency of agent i in executing task j, and L iis the load of the current agent i, and λ is the load balancing coefficient;

[0073] The real-time load of agent i is defined as:

[0074]

[0075] where: T i represents the set of subtasks assigned to agent i that are being executed; C j is the complexity of subtask j;

[0076] Based on the task types that the agent can complete, the computing performance parameters, and the ability matrix E, construct the task assignment constraints for the agent. For an agent, its ability to process different types of data is different, and through long-term operation training, the ability matrix of the agent will change. Therefore, it is necessary to regularly update and evaluate the ability matrix and level of the agent. The data for the update and evaluation comes from the historical task logs of the agent.

[0077] According to the task priority, the ability matrix of the agent, and the resource status, dynamically generate a task assignment plan. The core agent focuses on the core tasks with high complexity, and the auxiliary agents provide support such as computing resources and information sharing.

[0078] S4. Through iterative operations using a multi-objective optimization algorithm, solve the human task balance optimization model and the agent role assignment model, allocate subtasks to agents, obtain the task assignment matrix and the estimated completion time, and the agents execute the subtasks according to the assignment matrix to achieve the assignment of human tasks. Furthermore, achieve human resource allocation.

[0079] In the above S4, tasks with high priority will be preferentially executed by the agent. During the process of the agent executing subtasks according to the assignment matrix, there is a situation of task migration. When the load difference |L i -L i′ | between any two agents i and i' exceeds the preset threshold ΔL, that is, when |L i -L i′ | > ΔL, then there is a task migration between agents i and i'. Select subtasks that meet the migration conditions from the task set of the high-load agent and migrate them to the task set of the low-load agent, and then update the task assignment matrix and the load parameters of the agent.

[0080] The migration condition for the subtask is that the subtask complexity and task type meet the task assignment constraints of the low-load agent;

[0081] For the calculation of the migration amount, for subtasks that meet the migration conditions of the subtasks, select the subtask with the lowest complexity among them and migrate them one by one until the load difference is less than the preset threshold ΔL, at which time the migration is completed.

[0082] The value ΔL can be set based on the following methods:

[0083] Dynamic calculation: Adjust dynamically according to the mean and standard deviation of historical load data.

[0084] Fixed value: Set a fixed value through expert experience or business requirements.

[0085] Iterative process: If the load is still unbalanced after migration, repeat the above steps until the load difference is less than the threshold ΔL or the maximum number of iterations is reached.

[0086] Dynamic load balancing: The system automatically adjusts task allocation by monitoring the task progress of each agent in real time to ensure load balance among agents, avoiding resource idleness or overload. The dynamic adjustment mechanism significantly reduces the task backlog of high-load agents and improves the overall resource utilization rate.

[0087] Role collaboration optimization: Through the feedback mechanism between the auxiliary agent and the core agent, the system can efficiently handle task complexity and emergencies.

[0088] System adaptability improvement: The system can flexibly adjust the allocation scheme according to changes in task requirements and is applicable to human resource allocation in complex scenarios such as the power industry.

[0089] In S4, the agent executes subtasks according to the allocation matrix. After the subtask is completed, determine whether the completion status of the subtask is successful or failed. For the subtask that fails to execute, judge the failure reason by checking the task log and the execution history of the agent:

[0090] If the load of the actual resource consumption of the subtask is greater than the complexity of the calculated subtask, it is determined that the failure reason is insufficient resources, and then the subtask is assigned to an agent with higher processing capacity, and the priority of the subtask is increased;

[0091] If a logical conflict exists in a single subtask resulting in the failure of the subtask execution, it is determined that the failure reason is high complexity, and then the subtask is further divided into multiple subtasks, the task processing priority is increased, and then it is reassigned to multiple agents for execution;

[0092] If the number of times the subtask fails to execute exceeds the specified threshold, start the error reporting mechanism, generate an alarm and suggest manual intervention.

[0093] Relationship management between subtasks: The task memory module supports modeling and management of the relationships between subtasks to ensure the smooth execution of complex tasks. The processing methods of the two main relationships are as follows:

[0094] Parallel Sub - tasks: For independent parallel sub - tasks, the system distributes them to multiple agents in parallel according to the task characteristics, while recording the execution order and completion status of the tasks to ensure the efficient use of resources.

[0095] Sequential Sub - tasks: For sub - tasks with dependencies, the system maintains a task dependency graph to record the dependency relationships between tasks.

[0096] For example:

[0097] G=(V,F)

[0098] Where: V is the set of human tasks; F is the dependency relationship between tasks (a directed edge represents the constraint from a pre - task to a post - task).

[0099] The system ensures that subsequent tasks are executed only after their prerequisite tasks are successfully completed according to the topological sorting in the dependency graph.

[0100] Historical Data Analysis

[0101] The historical data analysis module provides a basis for task allocation and agent optimization by extracting the behavior patterns of agents. The analysis process is as follows:

[0102]

[0103] Where: P i represents the task completion rate of agent i; T i is the set of tasks executed by agent i; S j is the completion status of task j, 1 represents success, and 0 represents failure.

[0104] The analysis results are used for:

[0105] Performance Evaluation: Identify efficient agents and preferentially allocate tasks.

[0106] Bottleneck Optimization: For agents with low completion rates, optimize their task allocation strategies or ability models.

[0107] Prediction and Early Warning: According to the trend of historical data, predict potential failure risks and adjust resource allocation in advance.

[0108] Through the task memory module and historical data management, the system can comprehensively record and dynamically optimize the task execution process, not only improving resource utilization, but also effectively reducing duplicate work and resource waste caused by task failures. This module is especially suitable for complex human resource allocation scenarios with multiple levels and multiple tasks.

[0109] Data Recording and Analysis: Record key parameters during the task execution process, including task completion time, resource consumption, and success rate.

[0110] Failed task handling: For failed tasks, the system automatically analyzes the reasons and reallocates the tasks to ensure the stability and continuity of task execution.

[0111] Knowledge graph construction: Structurally store historical data and dynamically call it in subsequent task assignments.

[0112] In S4, the agent executes subtasks according to the allocation matrix. After the subtasks are completed, record the subtask logs: the agent number that completed the subtask, the human task number to which the subtask belongs, the subtask type, the key features of the subtask, the operation duration of the subtask, the resource consumption of the actual operation of the subtask, and the completion status of the subtask; record the above subtask log information as the historical completion data of the corresponding agent.

[0113] In the power system human resource allocation method, preferentially use local devices and agents for human resource allocation tasks. When the resource utilization rate of local devices is greater than the set threshold, start the resource collaborative call mechanism, transfer local subtasks that meet the collaborative conditions to other departments with a resource utilization rate less than the set threshold for collaborative processing, and record their task processing logs.

[0114] Collaborative conditions for tasks: Subtasks with a task complexity and priority both lower than the set task complexity threshold and priority threshold; at the same time, the subtasks need to meet the task allocation constraints of the receiving department.

[0115] Job function specialization: Based on the multi-level organizational structure of the power grid, task allocation fully considers the differences in the functions of each position to ensure that resource allocation meets business requirements.

[0116] Auxiliary agent enhancement: The auxiliary agent not only supports the decomposition of human tasks, but also provides data support for the execution of the core agent through knowledge sharing and resource optimization.

[0117] Cross-departmental task collaboration: By introducing a task relay mechanism between agents, the system achieves seamless connection in multi-department and multi-position task allocation, improving the overall efficiency.

[0118] The cross-departmental collaboration mechanism solves the problems of resource allocation and agent task coordination between multiple departments through resource sharing and agent task feedback mechanisms, aiming to improve the overall resource utilization rate of the system and the agent task execution efficiency. This mechanism can not only optimize resource allocation, but also optimize the agent task allocation algorithm in real time through the feedback and dynamic adjustment mechanism, forming a linkage relationship with the previous agent task allocation model.

[0119] Cross-departmental resource sharing

[0120] Cross-departmental resource sharing performs dynamic resource allocation based on the global resource pool and task demand model to avoid the phenomenon of resource idleness or uneven allocation.

[0121] Construction of the global resource pool: The global resource pool G contains the available resources of each department (such as human resources, computing power, and task tools), and is defined by the following formula:

[0122]

[0123] Where: R p represents the set of available resources of the p-th department. Task requirement modeling: The task requirements of each department are represented by a task requirement matrix D, and the element d pj reflects the resource requirements of department p for task j.

[0124] Resource allocation objective function: The system achieves optimal resource allocation by minimizing the following resource allocation cost function:

[0125]

[0126] Where: r pj represents the amount of resources allocated to task j in department p, P is the total number of departments, and J’ represents the total amount of tasks allocated to department p; w j is the weight factor of task j (measuring its importance and urgency).

[0127] Dynamic resource adjustment: According to the task requirement priorities and resource utilization rates of the agents, dynamically adjust the resource allocation strategy to give priority to meeting the needs of high-weight agent tasks and avoid resource waste.

[0128] Task allocation linkage mechanism

[0129] To achieve the linkage between resource allocation and task allocation, this mechanism establishes a feedback loop through resource status monitoring and task allocation strategy optimization:

[0130] Resource status monitoring: The system monitors the resource utilization of agents in each department in real time, records the idle resources R idle and overloaded resources R overload , and evaluates the resource utilization rate through the following formula:

[0131]

[0132] Where: U p is the resource utilization rate of department p; T p represents the set of tasks allocated to department p; r pj is the resource requirement of task j.

[0133] Linked task adjustment: When the resource status is unbalanced (such as when resources are idle in a certain department), the system triggers the following linked task adjustment mechanism:

[0134] Task migration: Migrate low-priority tasks from overloaded departments to departments with idle resources.

[0135] Sub-task decomposition: Further decompose high-complexity tasks and reallocate sub-tasks in departments with idle resources.

[0136] Task allocation optimization formula: The linkage mechanism re-adjusts task allocation through the following optimization formula:

[0137]

[0138] Where: M pj represents the efficiency of department p in executing task j; X pj represents the flag indicating whether task j is allocated to department p, and X pj = 1 indicates that task j is allocated to department p.

[0139] Collaborative management and feedback mechanism

[0140] The collaborative management mechanism further optimizes resource and task allocation through a feedback model:

[0141] Feedback model design: The task completion efficiency is quantified by the following formula:

[0142]

[0143] Where: is the number of tasks completed by department p; is the total task volume of department i; α is the resource idle penalty factor.

[0144] Dynamic feedback and adjustment: According to the feedback results, adjust the resource and task allocation strategies: Prioritize allocating resources to efficient departments; for inefficient departments, re-evaluate the task allocation plan.

[0145] Through the cross-departmental collaboration mechanism, the present invention can:

[0146] Improve resource utilization rate: Avoid resource idleness or overload and ensure the efficient utilization of resources in each department.

[0147] Optimize task allocation efficiency: Achieve the linkage optimization of task allocation and resource allocation and improve the overall execution efficiency.

[0148] Enhance system adaptability: The dynamic adjustment mechanism ensures that the system can cope with complex and changing business requirements, and is particularly applicable to the cross-departmental task collaboration scenario in the power industry.

[0149] Example 2:

[0150] In the human task of human resource allocation in a power enterprise, for different job requirements and complex and changing human task requirements, the system realizes efficient allocation through the following steps:

[0151] For example, for a line maintenance task, first extract key information such as the line name, maintenance scope, maintenance level, and key maintenance time. Based on this key information, along with the specific location, time, and potential order of maintenance, break down the maintenance items into several subtasks, such as candidate screening, skill matching, and interview arrangement. Calculate the complexity of each subtask, and construct a balance optimization function for human tasks for this line maintenance task.

[0152] Agent role assignment: The system assigns specific roles to each agent to quickly respond to different task assignment requirements. Each agent calculates the personnel scheduling and assignment situation in the subtasks according to the role division, obtains the pre - plan arrangement for the subtasks, and distributes the pre - plan arrangement to the corresponding task personnel to guide the execution of the tasks, improving the efficiency of configuring human tasks.

[0153] Task memory recording and backtracking: The memory module records the data during the execution of human tasks, including the completion time and resource consumption of each sub - human task. Through these historical data, the system can dynamically adjust the allocation strategy in subsequent human tasks.

[0154] When the system detects sudden changes in certain job requirements or an increase in the urgency of human tasks, it will adjust the role assignment of the agents to ensure that key job requirements are prioritized.

[0155] The system performs backtracking analysis based on the task historical data recorded in the memory module to optimize the roles and task assignments of each agent, so as to improve the overall configuration effect of the system.

[0156] Regarding the collaborative deployment of tasks: For example, in the event of extreme disaster weather, ensuring power supply becomes the primary responsibility of the power system. At this time, real - time management is required for human tasks such as fault detection and hazard elimination. The number of tasks increases exponentially, resulting in the resource utilization rate of local devices being greater than the set threshold. At this time, start collaborative deployment, call computing resources outside the disaster - affected area to assist in the execution of human resource allocation tasks, so as to achieve effective scheduling as soon as possible.

[0157] Embodiment 3:

[0158] See Figure 2 , a power system human resource allocation system based on multi - agent collaboration. The system is used to execute the aforementioned power system human resource allocation method based on multi - agent collaboration, and specifically includes: a human resource task collection module, a task decomposition and balance optimization module, an agent role assignment module, and a solution operation module;

[0159] The human resource task collection module is used to collect newly added human tasks in the power system in real - time;

[0160] The task decomposition and balance optimization module is used to quantitatively decompose each human task into multiple subtasks, calculate the complexity of each subtask, and construct a human task balance optimization model;

[0161] In the task decomposition and balance optimization module, using natural language processing technology, key features are extracted from the description of the human task, and the task types, task time limits, job requirements, and task priorities of different subtasks are decomposed. At the same time, the dependency relationship information between subtasks is extracted, and the dependency relationship includes parallel and sequential relationships; the unstructured job description is converted into structured data, and noise removal and semantic unification are performed;

[0162] Semantic analysis and weight assignment are performed on the key features related to different subtasks, and the complexity of each subtask is calculated respectively:

[0163]

[0164] Among them: C m represents the complexity of subtask m, w n is the weight factor of the nth feature, and f n is the nth feature value of task m;

[0165] Construct a human task balance optimization model. The balance optimization function F for each human task is:

[0166]

[0167] Among them, m t is the completion status symbol of subtask m at time t. When m t =1, it means that subtask m is not completed at time t. When m t =0, it means that subtask m is completed at time t; R mk represents the resource consumption of subtask m assigned to agent k. M is the total number of subtasks in the personnel assignment task, T mk is the processing time for agent k to complete subtask m. K is the total number of agents executing the personnel assignment task, and α is the balance coefficient;

[0168] Based on the task types, task time limits, job requirements, and subtask relationships of different subtasks, constraints for the balance optimization function are constructed.

[0169] The agent role assignment module is used to construct an agent role assignment model based on the subtask complexity, agent ability matrix, and real-time load of the agent;

[0170] In the agent role assignment module, agent role assignment: The initial assignment of agent roles is based on its ability matrix E = [E ij , and the matrix element E ijRepresents the efficiency of agent i in executing task type j, E ij Obtained by weighted calculation of the computing performance parameters of agent i and the completion data of different historical task types;

[0171] Construct an agent role assignment model based on the agent role assignment objective function S:

[0172]

[0173] Where: X ij Represents the decision symbol for whether task j is assigned to agent i, X ij ∈{0,1}, when X ij =1, it means that task j is assigned to agent i. I is the total number of callable agents in the system, and J is the sum of the subtasks to be completed in the system; E ij Represents the efficiency of agent i in executing task j, L i Is the current load of agent i, and λ is the load balancing coefficient;

[0174] The real-time load of agent i is defined as:

[0175]

[0176] Where: T i Represents the set of subtasks currently assigned to agent i for execution; C j Is the complexity of subtask j;

[0177] Based on the task types that the agent can complete, computing performance parameters, and the ability matrix E, construct the task assignment constraints of the agent.

[0178] Solution operation module: Perform iterative operations through a multi-objective optimization algorithm to solve the human task balance optimization model and the agent role assignment model, assign subtasks to agents, obtain the task assignment matrix and the estimated completion time, and the agents execute subtasks according to the assignment matrix to achieve the assignment of human tasks.

[0179] In the solution operation module, tasks with higher priorities will be preferentially executed by agents. During the process of agents executing subtasks according to the assignment matrix, there is a situation of task migration. When the load difference |L i -L i′ | between any two agents i and i' exceeds the preset threshold ΔL, that is, when |L i -L i′ |>ΔL, then task migration occurs between agents i and i'. Select subtasks that meet the migration conditions from the task set of the high-load agent and migrate them to the task set of the low-load agent, and then update the task assignment matrix and the load parameters of the agents.

[0180] The agent executes subtasks according to the allocation matrix. After the subtasks are completed, it determines whether the completion status of the subtasks is successful or failed. For the subtasks that fail to execute, by checking the task log and the execution history of the agent, it judges the reason for failure:

[0181] If the load of the actual resource consumption of the subtask is greater than the complexity of the calculated subtask, it is determined that the reason for failure is insufficient resources, and then the subtask is assigned to an agent with higher processing capacity, and the priority of the subtask is increased;

[0182] If the execution of a single subtask fails due to logical conflicts, it is determined that the reason for failure is high complexity, and then the subtask is further divided into multiple subtasks, the task processing priority is increased, and then it is reassigned to multiple agents for execution;

[0183] If the number of times a subtask fails to execute exceeds the specified threshold, an error reporting mechanism is started, and an alarm is generated and it is recommended that manual intervention be carried out.

[0184] The agent executes subtasks according to the allocation matrix. After the subtasks are completed, it records the subtask log: the agent number that completes the subtask, the human task number to which the subtask belongs, the subtask type, the key features of the subtask, the operation duration of the subtask, the resource consumption of the actual operation of the subtask, the completion status of the subtask; the above-mentioned subtask log information is recorded as the historical completion data of the corresponding agent.

[0185] In the human resource allocation method of the power system, local devices and agents are preferentially used to perform human resource allocation tasks. When the resource utilization rate of local devices is greater than the set threshold, a resource collaborative call mechanism is started, and the local subtasks that meet the collaborative conditions are transferred to other departments with a resource utilization rate less than the set threshold for collaborative processing, and their processing task logs are recorded.

[0186] Embodiment 4:

[0187] See Figure 3 , a power system human resource allocation device based on multi-agent collaboration, including a memory and a processor. The memory is used to store computer program codes and transmit the computer program codes to the processor;

[0188] The processor is used to execute the aforementioned power system human resource allocation method based on the instructions in the computer program code.

[0189] Embodiment 5:

[0190] A computer program product includes a computer program, and the computer program is executed by the processor to perform the aforementioned power system human resource allocation method based on multi-agent collaboration.

Claims

1. A method for human resource allocation in a power system based on multi-agent collaboration, characterized in that It includes the following steps: S1. Construct a human resource task collection module to collect newly added human tasks in the power system in real time; S2. Quantitatively decompose each human task into multiple subtasks, calculate the complexity of each subtask, and construct a human task balance optimization model; S3. Construct an agent role assignment model based on the subtask complexity, agent capability matrix, and real-time load of the agent; S4. Perform iterative operations through a multi-objective optimization algorithm to solve the human task balance optimization model and the agent role assignment model, allocate the subtasks to the agents, obtain a task assignment matrix and an estimated completion time, and the agents execute the subtasks according to the assignment matrix to achieve the assignment of human tasks.

2. The power system human resource allocation method based on multi-agent collaboration according to claim 1, characterized in that: In S2, using natural language processing technology, extract key features from the description of the human task, decompose the task types, task time limits, position requirements, and task priorities of different subtasks, and at the same time extract the dependency relationship information between subtasks, where the dependency relationship includes parallel and sequential relationships; convert the unstructured position description into structured data, and perform noise removal and semantic unification; Perform semantic analysis and weight assignment on the key features related to different subtasks, and calculate the complexity of each subtask respectively: Where: C m represents the complexity of subtask m, w n is the weight factor of the nth feature, f n is the nth feature value of task m; Construct a human task balance optimization model, and the balance optimization function F of each human task is: where m t is the completion status symbol of subtask m at time t. When m t = 1, it indicates that subtask m is not completed at time t. When m t = 0, it indicates that subtask m is completed at time t; R mk represents the resource consumption of subtask m assigned to agent k. M is the total number of subtasks in the personnel assignment task. T mk is the processing time for agent k to complete subtask m. K is the total number of agents executing the personnel assignment task, and α is the balance coefficient; Based on the task types, task time limits, position requirements, and subtask relationships of different subtasks, construct the constraints of the balance optimization function.

3. The power system human resource allocation method based on multi-agent collaboration according to claim 1, characterized in that: In S3, agent role assignment: The initial assignment of agent roles is based on its ability matrix E = [E ij , where the matrix element E ij represents the efficiency of agent i in executing task type j, and E ij is obtained by weighted calculation of the computing performance parameters of agent i and the historical completion data of different task types; Construct an agent role assignment model based on the agent role assignment objective function S: Where: X ij is a decision symbol indicating whether task j is assigned to agent i, X ij ∈ {0, 1}, when X ij = 1, it means that task j is assigned to agent i, I is the total number of callable agents in the system, and J is the sum of the sub-tasks to be completed in the system; E ij represents the efficiency of agent i in executing task j, L i is the load of the current agent i, and λ is the load balancing coefficient; The real-time load of agent i is defined as: Where: T i represents the set of subtasks being executed by agent i; C j is the complexity of subtask j; Based on the task types that the agent can complete, computing performance parameters, and capability matrix E, construct the task assignment constraints of the agent.

4. The power system human resource allocation method based on multi-agent collaboration according to claim 1, characterized in that: In S4, tasks with higher priorities are preferentially executed by the agent. During the process of the agent executing subtasks according to the allocation matrix, there is a situation of task migration. When the load difference |L i -L i′ | between any two agents i and i' exceeds the preset threshold ΔL, that is, when |L i -L i′ | > ΔL, then task migration occurs between agents i and i'. From the task set of the high-load agent, subtasks that meet the migration conditions are selected and migrated to the task set of the low-load agent, and then the task allocation matrix and the load parameters of the agent are updated.

5. The power system human resource allocation method based on multi-agent collaboration according to claim 1, characterized in that: In S4, the agent executes the subtask according to the assignment matrix. After the subtask is completed, determine whether the completion status of the subtask is successful or failed. For the subtask that fails to execute, check the task log and the execution history of the agent to determine the reason for the failure: If the actual resource consumption load of the subtask is greater than the calculated complexity of the subtask, it is determined that the reason for the failure is insufficient resources, and then the subtask is assigned to an agent with higher processing capacity, and the priority of the subtask is increased; If there is a logical conflict in a single subtask resulting in the failure of the subtask execution, it is determined that the reason for the failure is high complexity, and then the subtask is further divided into multiple subtasks, the task processing priority is increased, and then it is reassigned to multiple agents for execution; If the number of times the subtask fails to execute exceeds the specified threshold, start an error reporting mechanism, generate an alarm and suggest manual intervention.

6. The method for human resource allocation in a power system based on multi-agent collaboration according to claim 1, wherein: In S4, the agent executes subtasks according to the allocation matrix. After the subtasks are completed, the subtask logs are recorded: the agent number that completed the subtask, the human task number to which the subtask belongs, the subtask type, the key features of the subtask, the operation duration of the subtask, the resource consumption of the actual operation of the subtask, and the completion status of the subtask; the above subtask completion information is recorded as the historical completion data of the corresponding agent.

7. The method for human resource allocation in a power system based on multi-agent collaboration according to claim 1, wherein: In the method for human resource allocation in the power system, local devices and agents are preferentially used to perform human resource allocation tasks. When the resource utilization rate of local devices is greater than the set threshold, a resource collaborative call mechanism is started, and the local subtasks that meet the collaborative conditions are transferred to other departments with a resource utilization rate less than the set threshold for collaborative processing, and their processing task logs are recorded.

8. A human resource allocation system for a power system based on multi-agent collaboration, characterized in that, The system is used to execute the method for human resource allocation in a power system based on multi-agent collaboration according to any one of claims 1 to 7, specifically including: a human resource task collection module, a task decomposition and balance optimization module, an agent role allocation module, and a solution operation module; The human resource task collection module is used to collect newly added human tasks in the power system in real time; The task decomposition and balance optimization module is used to quantitatively decompose each human task into multiple subtasks, calculate the complexity of each subtask, and construct a human task balance optimization model; The agent role allocation module is used to construct an agent role allocation model according to the subtask complexity, the agent ability matrix, and the real-time load of the agent; The solution operation module: performs iterative operations through a multi-objective optimization algorithm, solves the human task balance optimization model and the agent role allocation model, allocates the subtasks to the agents, obtains a task allocation matrix and an estimated completion time, and the agents execute the subtasks according to the allocation matrix to achieve the allocation of human tasks.

9. A power system human resource allocation device based on multi-agent collaboration, characterized in that, It includes a memory and a processor. The memory is used to store computer program code and transmit the computer program code to the processor; The processor is used to execute the method for human resource allocation in a power system based on multi-agent collaboration according to any one of claims 1 to 7 according to the instructions in the computer program code.

10. A computer program product, comprising a computer program, characterized in that, The computer program is executed by the processor to perform the method for human resource allocation in a power system based on multi-agent collaboration according to any one of claims 1 to 7.

Images