CN-121981463-A - Construction site dynamic job scheduling method based on multi-agent cooperation

CN121981463ACN 121981463 ACN121981463 ACN 121981463ACN-121981463-A

Abstract

The application relates to the technical field of group intelligence, in particular to a multi-agent cooperation-based construction site dynamic job scheduling method, which comprises the steps of constructing a task state matrix; in the iteration of an optimization algorithm, the positions of population individuals are decoded into task priorities, the expected execution time of each task node is obtained by combining a task state matrix, the process coupling degree for representing the process connection compactness is calculated, the process coupling degree of the population individuals in the current iteration is counted, the evolution stagnation risk is calculated by combining the standard deviation of the process coupling degree in the initial iteration, a target strategy is selected from a Cauchy variation strategy and a Gaussian variation strategy to update the population individuals, and the global optimal position when the iteration is ended is converted into a scheduling instruction set. By the technical scheme, invalid waiting time among working procedures can be compressed, and comprehensive working efficiency of a construction site is improved.

Inventors

TAO FENG
LI JUAN
XIE BO
XIE XINGBING
ZOU HONG

Assignees

湖北荆力电力集团有限公司

Dates

Publication Date: 20260505
Application Date: 20260120

Claims (10)

1. The method for dispatching the dynamic job of the construction site based on the cooperation of multiple agents is characterized by comprising the steps of obtaining real-time positions, task states and environment data of each agent of the construction site to obtain multi-source original data; smoothing and aligning the multisource original data to construct a task state matrix reflecting real-time positions and task progress of the intelligent agents; In the iteration of the optimization algorithm, decoding the positions of population individuals into task priorities, and carrying out event simulation by combining the task state matrix to obtain the estimated execution time of each task node; calculating the procedure coupling degree for representing the procedure connection tightness degree by using the predicted execution time, the task priority and the preset procedure standard conversion time; Counting the process coupling degree of population individuals in the current iteration, and calculating an evolution stagnation risk for representing the evolution capability of the population by combining the standard deviation of the process coupling degree in the initial iteration; selecting a target strategy from a cauchy variation strategy and a gaussian variation strategy based on the evolution stagnation risk to update the positions of population individuals; And converting the global optimal position when the iteration is terminated into a scheduling instruction set containing the operation schedule and the path planning of each agent, and issuing and executing the scheduling instruction set through an industrial network.
2. The multi-agent collaboration-based construction site dynamic job scheduling method of claim 1, wherein the smoothing and aligning the multi-source raw data comprises: smoothing the positioning data in the multi-source original data by using a Kalman filtering algorithm; And performing time alignment on each dimension data in the multi-source original data by using a linear interpolation method based on a preset job scheduling period.
3. The multi-agent collaboration-based construction site dynamic job scheduling method of claim 1, wherein calculating the process coupling degree comprises: acquiring a difference value between the estimated subsequent task start time and the estimated precursor task end time of each pair of coupling procedures in the estimated execution time; calculating the deviation value of the difference value and the process standard conversion time; and weighting the inverse of the deviation value based on the task priority to obtain the process coupling degree.
4. The multi-agent collaboration-based construction site dynamic job scheduling method of claim 1, wherein calculating the risk of evolutionary arrest comprises: obtaining an average value and a maximum value of the process coupling degree of population individuals in the current iteration, and calculating the ratio of the average value to the maximum value; Multiplying the ratio by a correction term reflecting the population distribution state to obtain the evolution stagnation risk.
5. The multi-agent collaboration-based construction site dynamic job scheduling method of claim 4, wherein correcting the ratio using standard deviation comprises: And calculating the ratio of the standard deviation of the process coupling degree in the current iteration to the population diversity threshold, wherein the correction term and the ratio are in negative correlation.
6. The multi-agent cooperation based construction site dynamic job scheduling method according to claim 1, wherein the optimization algorithm is a dung beetle optimization algorithm.
7. The multi-agent collaboration-based construction site dynamic job scheduling method of claim 1, wherein selecting the target strategy from the cauchy variation strategy and the gaussian variation strategy comprises: Generating a random number in the current iteration, and comparing the random number with the evolution stagnation risk; in response to the random number being less than the risk of evolutionary arrest, performing the cauchy variation strategy; the gaussian variation strategy is performed in response to the random number not being less than the risk of evolutionary arrest.
8. The multi-agent collaboration-based construction site dynamic job scheduling method of claim 1, wherein decoding the locations of the population individuals into task priorities comprises: The method comprises the steps of obtaining position vectors of population individuals, respectively determining the numerical values of each dimension in the position vectors as priority weights of corresponding tasks to be scheduled, and sequencing all the tasks to be scheduled according to the order of the priority weights from large to small to obtain the task priorities reflecting the task execution sequence.
9. The method for dynamically scheduling the job on the foundation construction site based on the multi-agent cooperation according to claim 8, wherein the step of converting the global optimal position into the scheduling instruction set comprises the steps of sequentially distributing tasks to the earliest idle agent by utilizing a greedy distribution strategy according to the task priority to obtain a preliminary job scheme, and carrying out path planning on the preliminary job scheme by combining the real-time positions of the agents to obtain the scheduling instruction set.
10. The multi-agent collaboration-based on-site dynamic job scheduling method as set forth in claim 1, wherein the scheduling method employing a rolling time domain optimization strategy comprises updating the task state matrix at a preset frequency, and restarting the optimization algorithm in response to job environment changes or task state anomalies to generate an updated set of scheduling instructions.

Description

Construction site dynamic job scheduling method based on multi-agent cooperation Technical Field The application relates to the technical field of parameter control, in particular to a multi-agent cooperation-based construction site dynamic job scheduling method. Background In large-scale infrastructure construction engineering, the collaborative work of a plurality of construction machines such as an excavator, a dump truck, a bulldozer and the like is involved, the construction machines are regarded as intelligent bodies with sensing and decision making capabilities, and the dynamic allocation and path planning of complex construction tasks through a collaborative mechanism are key to improving the construction efficiency. In the face of abrupt changes in construction environments and strict logic dependencies between construction procedures (e.g., excavation-loading-transportation-rolling), there is a need for a real-time scheduling scheme that ensures orderly advancement of large-scale capital construction projects. In order to solve the complex combination optimization problem, a dung beetle optimization algorithm and other meta-heuristic algorithms are often selected for solving. The dung beetle optimizing algorithm builds a mathematical optimizing model by simulating biological behaviors such as rolling balls, dancing, foraging and the like of the dung beetles, carries out iterative search on task queues and mechanical distribution of a construction site, and searches a scheduling sequence with the shortest total construction period in a complex optimizing space by utilizing the biological behavior model so as to provide a preliminary configuration scheme of task distribution and mechanical dispatching for the construction site. However, the method has obvious defects in the application process, and because the algorithm searching mode is relatively fixed and the self-adaptive adjustment mechanism aiming at the coupling characteristic of the working procedure of the basic building is lacked, the algorithm is extremely easy to be trapped into local optimum, further the working procedure connection among the intelligent agents in the actually generated scheduling scheme is extremely loose, a large amount of recessive waiting time exists in the construction site, and a working instruction set with tight and feasible working procedure connection is difficult to be quickly generated in the dynamically changed construction environment. Disclosure of Invention In order to solve the technical problem that a working procedure instruction set which is tightly connected and feasible is difficult to quickly generate in a dynamically-changed construction environment, the application provides a multi-agent cooperation-based construction site dynamic working scheduling method, which can compress invalid waiting time among working procedures and improve comprehensive working efficiency of a construction site. The application provides a multi-agent collaborative construction site dynamic job scheduling method, which comprises the steps of obtaining real-time positions, task states and environment data of agents in a construction site to obtain multi-source original data, conducting smoothing and alignment processing on the multi-source original data to construct a task state matrix reflecting real-time positions and task progress of the agents, decoding positions of individuals in a population into task priorities in optimization algorithm iteration, conducting event simulation by combining the task state matrix to obtain expected execution time of each task node, calculating process coupling degree for representing process connection tightness by utilizing the expected execution time, the task priorities and preset process standard conversion time, calculating evolution stagnation risks for representing population evolution capacity by means of calculating the process coupling degree of the individuals in the current iteration and combining standard deviation of the process coupling degree in initial iteration, selecting target strategies from Cauchy variation strategies and Gaussian variation strategies based on the evolution stagnation risks to update positions of the individuals, converting the global optimal positions of the individuals in population termination into operation time tables and paths containing the individuals in the iteration termination, and sending out industrial planning instructions by means of an industrial network planning and performing network planning. By combining the real-time position, the task state and the procedure coupling evaluation, a dynamic scheduling closed loop is established, and the scheduling decision can be flexibly adjusted along with the real-time change of the progress of each station of the construction site. Preferably, the smoothing and aligning processing of the multi-source original data comprises smoothing positioning data in the multi-source original data by us