Search

CN-121032448-B - Human resource allocation method and device for elevator maintenance, electronic equipment, medium and program product

CN121032448BCN 121032448 BCN121032448 BCN 121032448BCN-121032448-B

Abstract

The invention belongs to the technical field of data processing, and particularly discloses a human resource allocation method, a device, electronic equipment, a medium and a program product for elevator maintenance, wherein the method comprises the steps of constructing and updating a task-personnel dependency graph comprising task nodes, personnel nodes, time sequence edges, cooperation edges and the like by detecting personnel temporary lottery events and recording key information; the method comprises the steps of determining an affected task set by spreading delay from an event source node, evaluating task impact weight based on parameters such as urgency, default cost and the like, extracting a high-weight task, taking an event source as a center, extracting a local optimization domain by combining a multi-dimensional threshold value, freezing domain distribution, inputting intra-domain information into a large model to generate a candidate strategy, and finally screening an optimal scheme based on constraint conditions and an objective function. The invention realizes the efficient human resource reallocation under the local minimum disturbance, reduces delay cost and default risk, and improves the human resource utilization rate.

Inventors

  • HAN FANG
  • Bian Shouguo
  • YANG XUJUN
  • XU NA
  • HE JUN

Assignees

  • 中海物业管理有限公司
  • 海纳万商物业管理有限公司

Dates

Publication Date
20260512
Application Date
20251030

Claims (7)

  1. 1. The human resource allocation method for elevator maintenance is characterized by comprising the following steps of: Step S1, when a temporary personnel lottery event is detected, recording event triggering time and affected personnel identification; Step S2, a task-personnel dependency graph is constructed, wherein the dependency graph comprises a plurality of nodes, a plurality of edges and corresponding attributes thereof, the nodes comprise task nodes and personnel nodes, the edges comprise time sequence edges, cooperation edges, resource edges and supervision edges, the temporary lottery event is based on and the available time window and related edge attributes of affected personnel are updated, the task nodes further comprise task node attributes, the task node attributes comprise skill requirements, time windows, reference duration and service level protocol costs, the personnel nodes further comprise personnel node attributes, the personnel node attributes comprise skill sets, available time windows and positions, the time sequence edges represent time dependence of continuous tasks of the same personnel, the cooperation edges represent participation relations of common tasks of the plurality of persons, the resource edges represent association of tasks and shared resources, and the supervision edges represent supervision relations of the tasks and supervision personnel; Step S3, starting from an event source node, propagating delay along the dependency graph, and calculating to obtain an affected task set, wherein the event source node is a temporary lottery person node; S4, evaluating impact weights of the affected tasks based on influence parameters to obtain the first k tasks with highest weights, wherein k is more than 0, and the influence parameters comprise urgency of the tasks, default cost, customer criticality and site sensitivity; step S5, taking an event source node as a center, extracting a local optimization domain of the dependency graph based on a multi-dimensional threshold, and freezing task and personnel distribution outside the optimization domain, wherein the multi-dimensional threshold comprises a graph distance threshold, a travel time threshold and a time window overlapping degree threshold; step S6, inputting the node related information in the optimized domain into a large model to generate candidate strategies for human resource allocation, wherein the node related information at least comprises task information, personnel information and travel time matrix abstract, the large model is a GPT-based model and comprises an input analysis layer and a strategy generation layer, The strategy generation layer is used for generating a plurality of candidate strategies based on exchange operators, insertion operators, translation operators and collaborative task splitting, wherein the exchange operators are used for exchanging tasks of personnel based on skill matching and time window compatibility, the insertion operators are used for inserting high-weight tasks based on available time window neutral positions of the personnel, the translation operators are used for translating time windows of low-weight tasks, and the collaborative task splitting is used for splitting the tasks of multi-person collaboration into serial sub-stages; step S7, screening an optimal scheme from the candidate strategies based on constraint conditions and objective functions, wherein the method specifically comprises the following steps: Traversing all the candidate strategies, wherein the strategies conforming to the constraint conditions are used as effective candidate sets; Calculating an objective function for each strategy in the effective candidate set, wherein the minimum function value is an optimal scheme; wherein the constraint conditions include skill matching constraint, time window constraint, resource conflict constraint and travel time constraint; The objective function f=w 1 × DelayCost + w 2 × Disturbance, where DelayCost is the sum of products of impact delay weights and actual delay durations of all tasks, disturbance is an index for measuring the change amount of the candidate set strategy to the original plan, and w 1 and w 2 are preset weight coefficients.
  2. 2. The human resource allocation method for elevator maintenance according to claim 1, wherein, The computing obtains an affected task set, comprising: initializing a set, and incorporating initial task nodes associated with event source nodes into the set; Starting from the initial task node, traversing the associated nodes along different types of edges in the dependency graph, judging whether the associated nodes cannot execute the task according to an original plan due to the delay of the preceding task, and if the task meets the condition, incorporating the task into a set; and stopping propagation when all the associated nodes are traversed and no new task accords with the joining condition.
  3. 3. The human resource allocation method for elevator maintenance according to claim 1, wherein, The extracting the local optimization domain of the dependency graph based on the multidimensional threshold comprises the following steps: screening out candidate task sets, candidate person sets and associated edges which meet the conditions from global tasks and persons based on the graph distance threshold, the travel time threshold and the time window overlapping degree threshold to form a preliminary local area; and incorporating the top k tasks with the highest weights into the local area to form the local optimization domain.
  4. 4. A manpower resources allotment device for elevator maintenance, characterized by comprising: The system comprises a lottery event detection module, a lottery event detection module and a lottery event detection module, wherein the lottery event detection module is used for recording event triggering time and affected personnel identification when a personnel temporary lottery event is detected; The system comprises a dependency graph construction and updating module, a task-personnel dependency graph, a monitoring module and a task-personnel dependency graph, wherein the dependency graph comprises a plurality of nodes, a plurality of edges and corresponding attributes thereof, the nodes comprise task nodes and personnel nodes, the edges comprise time sequence edges, cooperation edges, resource edges and monitoring edges, the time sequence edges are based on temporary lottery events and update available time windows and related edge attributes of affected personnel; The affected task set determining module is used for starting from an event source node, propagating delay along the dependency graph, and calculating to obtain an affected task set, wherein the event source node is a temporary lottery person node; The impact weight calculation module is used for evaluating the impact weight of the affected task based on the influence parameters to obtain the first k tasks with the highest weight, wherein k is more than 0, and the influence parameters comprise the urgency of the task, the default cost, the client criticality and the site sensitivity; The optimization domain determining module is used for extracting a local optimization domain of the dependency graph based on a multi-dimensional threshold by taking an event source node as a center, and freezing task and personnel distribution outside the optimization domain, wherein the multi-dimensional threshold comprises a graph distance threshold, a travel time threshold and a time window overlapping degree threshold; The candidate strategy generation module is used for inputting the node related information in the optimization domain into a large model to generate candidate strategies for human resource allocation, wherein the node related information at least comprises task information, personnel information and travel time matrix abstract, the large model is a GPT-based model and comprises an input analysis layer and a strategy generation layer, The strategy generation layer is used for generating a plurality of candidate strategies based on exchange operators, insertion operators, translation operators and collaborative task splitting, wherein the exchange operators are used for exchanging tasks of personnel based on skill matching and time window compatibility, the insertion operators are used for inserting high-weight tasks based on available time window neutral positions of the personnel, the translation operators are used for translating time windows of low-weight tasks, and the collaborative task splitting is used for splitting the tasks of multi-person collaboration into serial sub-stages; the optimal scheme selecting module is used for selecting an optimal scheme from the candidate strategies based on constraint conditions and objective functions, and specifically comprises the following steps: Traversing all the candidate strategies, wherein the strategies conforming to the constraint conditions are used as effective candidate sets; Calculating an objective function for each strategy in the effective candidate set, wherein the minimum function value is an optimal scheme; wherein the constraint conditions include skill matching constraint, time window constraint, resource conflict constraint and travel time constraint; The objective function f=w 1 × DelayCost + w 2 × Disturbance, where DelayCost is the sum of products of impact delay weights and actual delay durations of all tasks, disturbance is an index for measuring the change amount of the candidate set strategy to the original plan, and w 1 and w 2 are preset weight coefficients.
  5. 5. An electronic device, the electronic device comprising: at least one processor, and a memory communicatively coupled to the processor, wherein, The memory stores instructions executable by the processor to enable the processor to perform the method of any one of claims 1-3.
  6. 6. A computer readable medium having stored thereon computer program instructions executable by a processor to implement the method of any of claims 1-3.
  7. 7. A computer program product comprising a computer program which, when executed by a processor, implements the method of any of claims 1-3.

Description

Human resource allocation method and device for elevator maintenance, electronic equipment, medium and program product Technical Field The invention belongs to the technical field of data processing, and particularly relates to a human resource allocation method, a device, electronic equipment, a medium and a program product for elevator maintenance. Background In the existing elevator maintenance work, personnel temporary lottery events frequently occur, and task delay, default risks and customer complaints are easily caused. The traditional human resource allocation method depends on manual experience, lacks of system consideration on the association relation between tasks and personnel, and cannot dynamically evaluate delay propagation paths and priorities, so that resource allocation is uneven, critical tasks are delayed, and overall efficiency is reduced. Meanwhile, tasks relate to multidimensional constraints such as skill matching, time window limitation, cooperative relation, shared resources and the like, and the optimality and feasibility are difficult to consider by manual allocation. Especially in the multi-task and multi-person parallel scene, the existing method lacks the capability of rapidly re-allocating the tasks with high local influence, and is difficult to efficiently generate a feasible allocation scheme in the minimum disturbance range. Disclosure of Invention In view of the above, the present invention provides a human resource allocation method, apparatus, electronic device, medium and computer program product for elevator maintenance, so as to solve the above technical problems. The invention provides a human resource allocation method for elevator maintenance, which comprises the following steps: Step S1, when a temporary personnel lottery event is detected, recording event triggering time and affected personnel identification; Step S2, a task-personnel dependency graph is constructed, wherein the dependency graph comprises a plurality of nodes, a plurality of edges and corresponding attributes thereof, the nodes comprise task nodes and personnel nodes, the edges comprise time sequence edges, cooperation edges, resource edges and supervision edges, and the available time window and related edge attributes of affected personnel are updated based on temporary lottery events; Step S3, starting from an event source node, propagating delay along the dependency graph, and calculating to obtain an affected task set, wherein the event source node is a temporary lottery person node; S4, evaluating impact weights of the affected tasks based on influence parameters to obtain the first k tasks with highest weights, wherein k is more than 0, and the influence parameters comprise urgency of the tasks, default cost, customer criticality and site sensitivity; step S5, taking an event source node as a center, extracting a local optimization domain of the dependency graph based on a multi-dimensional threshold, and freezing task and personnel distribution outside the optimization domain, wherein the multi-dimensional threshold comprises a graph distance threshold, a travel time threshold and a time window overlapping degree threshold; S6, inputting the node related information in the optimized domain into a large model to generate candidate strategies for human resource allocation, wherein the node related information at least comprises task information, personnel information and travel time matrix abstracts; and S7, screening an optimal scheme from the candidate strategies based on the constraint conditions and the objective function. In another aspect of the present application, there is also provided a human resource allocation apparatus for elevator maintenance, including: The system comprises a lottery event detection module, a lottery event detection module and a lottery event detection module, wherein the lottery event detection module is used for recording event triggering time and affected personnel identification when a personnel temporary lottery event is detected; The dependency graph construction and updating module is used for constructing a task-personnel dependency graph, the dependency graph comprises a plurality of nodes, a plurality of edges and corresponding attributes thereof, wherein the nodes comprise task nodes and personnel nodes, the edges comprise time sequence edges, cooperation edges, resource edges and supervision edges, and the available time window and related edge attributes of affected personnel are updated based on temporary lottery events; The affected task set determining module is used for starting from an event source node, propagating delay along the dependency graph, and calculating to obtain an affected task set, wherein the event source node is a temporary lottery person node; The impact weight calculation module is used for evaluating the impact weight of the affected task based on the influence parameters to obtain the first k tasks with the highest weig