Multi-agent based construction site task scheduling method and system

Abstract

The present application relates to the technical field of task scheduling, in particular to a construction site task scheduling method and system based on multiple agents, comprising the following steps: obtaining task start time and resource arrival time, matching task demand to set priority, judging task conflict to generate section structure table, extracting completed data to activate task to generate trigger chain, constructing path segment to establish task continuity, judging section adjacency to reconstruct scheduling to generate parallel control results. In the present application, the task selection is guided by cross judgment of task start time and resource arrival state, the priority information is dynamically given according to the operation type demand, the intelligent identification of task trigger condition is realized, the mutual exclusion conflict relationship is timely marked by overlapping comparison of regional task time interval, the scheduling behavior is prompted to have pre-identification ability for resource conflict, and the scheduling continuity and resource consistency in multiple agent task collaborative execution are enhanced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of task scheduling, in particular to a construction site task scheduling method and system based on multi-agent. BACKGROUND

[0002] The technical field of task scheduling mainly involves the optimal allocation and reasonable arrangement of limited resources in the time dimension, and its core matters include task priority sorting, resource allocation strategy, time dependency management and multi-objective coordinated control. Task scheduling technology is widely used in manufacturing management, transportation deployment, project construction and information systems, etc. Through the establishment of task queues, the definition of dependency relationships and the optimization of task execution order, the maximum resource utilization and the minimum time cost are realized. Task scheduling in construction management is an important branch of this field, involving personnel arrangement, equipment deployment, material distribution and work progress planning, with the characteristics of high real-time, strong collaboration and complex process dependency. Among them, the traditional construction site task scheduling method refers to manually preparing construction schedules and using basic project management tools to arrange various tasks on the construction site. The technical matters it addresses are how to reasonably coordinate task order, resource allocation and progress connection in the environment of multi-skilled, asynchronous operation and multi-site cross-construction. According to the construction drawings and project contract requirements, the traditional method determines the start and end time of each process based on experience and judgment, allocates worker, machinery and material usage time in a calendar manner through table tools or general scheduling software, and manually adjusts the scheduling order by referring to weather forecasts, material arrival records and other information to complete the construction organization tasks of each stage.

[0003] The existing technology relies on manual setting of progress plans and general scheduling tools in task scheduling, lacks a dynamic matching mechanism based on task attributes and resource states, and cannot real-time perceive the space-time conflicts between tasks in multi-skilled cross-construction, resulting in static lag problems in the scheduling process. The manual adjustment method is not timely in responding to sudden resource changes and task delays, and the task execution order depends on subjective experience, lacking a mechanism based on task state feedback. In the scene of complex process dependency and frequent resource flow in the construction site, it is easy to cause work rhythm dislocation, resource repeated occupation and other phenomena, resulting in decreased scheduling efficiency and progress control deviation. SUMMARY

[0004] The purpose of the present application is to solve the shortcomings in the prior art, and a construction site task scheduling method and system based on multi-agent are proposed.

[0005] In order to achieve the above purpose, the present application adopts the following technical scheme, a construction site task scheduling method based on multi-agent, comprising the following steps:

[0006] S1: Obtain the starting time range of the task to be executed by the agent in the construction area, the resource arrival time and the unit operation efficiency record, judge the time period coverage of the task starting time and the resource arrival time, match the unit operation efficiency record and the task type demand, and form a multi-agent task execution priority group;

[0007] S2: Based on the multi-agent task execution priority group, call the predicted execution time of the job and the job space number, compare the overlapping intervals of the start time and the end time of the tasks in the same area, judge whether concurrent or cross conflict is generated in the same scheduling period, and generate a construction site scheduling behavior section structure table;

[0008] S3: Call the completed task number in the construction site scheduling behavior section structure table, correspondingly extract the operation content quantity and positioning record time, judge the passed task as an activated task, and generate a scheduling task trigger state chain;

[0009] S4: According to the startable task number in the scheduling task trigger state chain, extract the construction area label, operation type label and rhythm control identifier, combine the three labels to build a path segment, and compare the number of consecutive items of the path segment, and generate a task content logical connection graph.

[0010] As a further scheme of the present application, the multi-agent task execution priority group includes the task starting time interval, the resource arrival determination result, and the task type corresponding priority, the construction site scheduling behavior section structure table includes the mutual exclusion identifier, the task number mapping relationship, and the operation space section number, the scheduling task trigger state chain includes the activated task number, the next task starting mark, and the task state mapping chain, and the task content logical connection graph includes the path segment sequence, the content continuity relationship between tasks, and the operation rhythm control identifier.

[0011] As a further scheme of the present application, the obtaining step of the multi-agent task execution priority group is specifically:

[0012] S111: Obtain the starting time range of the task to be executed by the agent in the construction area, the resource arrival time and the unit operation efficiency record, judge the interval coverage of the resource arrival time and the task starting time range of each task, select the task whose task starting time range is completely covered by the resource arrival time, and generate a resource coverage task interval;

[0013] S112: Call the resource coverage task interval, extract the corresponding task type demand and construction site task record, and filter the tasks whose task type matches the construction site task record by comparing the consistency of the task identifier and the task type field, and generate a matched task type task set;

[0014] S113: Based on the matching task type task set, the job types corresponding to multiple tasks are extracted, the tasks are divided into groups according to the job types, the resource input, unit job efficiency, task required period and job hindering factor of the tasks in the groups are extracted, the job execution priority of the tasks in each job type group is calculated, the tasks in the job type are sorted in descending order of numerical value, and a multi-agent task execution priority group is obtained.

[0015] As a further scheme of the present application, the obtaining step of the construction site scheduling behavior section structure table is specifically:

[0016] S211: Based on the multi-agent task execution priority group, the predicted execution time of the job and the job space number are called, the task combination in the same area is screened, the start time and end time of the task are obtained, the time overlap interval is compared, whether a cross conflict is generated in the same scheduling period is judged, and a cross conflict task combination is obtained.

[0017] S212: According to the cross conflict task combination, the task combination generating cross conflict in the same scheduling period is identified, the corresponding job space number is extracted, classified and registered according to the section range, and a conflict job space section number list is obtained.

[0018] S213: The conflict job space section number list is called, the job space range and execution time between tasks in the conflict combination are compared, the space overlap length, task priority value are extracted, and the space cross density is combined to calculate the mutual exclusion conflict identification value between task pairs, set the mutual exclusion identification, and establish the construction site scheduling behavior section structure table.

[0019] As a further scheme of the present application, the obtaining step of the scheduling task trigger state chain is specifically:

[0020] S311: The completed task number in the construction site scheduling behavior section structure table is called, the job completion flag, operation content quantity and positioning record time are extracted, the three types of data are matched and numbered respectively, and the numbering dimension is reconstructed to form a job state data combination item;

[0021] S312: According to the multiple numbered contents in the job state data combination item, it is judged that the job completion flag is an identification value, and the operation content quantity and task number group value are subjected to structure offset processing, the single task combination response quantity is calculated, the task number of the single task combination response quantity greater than zero is recorded, and an activated task number item is obtained.

[0022] S313: The activated task number item is called, the corresponding subsequent task number is searched, the subsequent task is allocated with a start identification field, and a mapping path relationship between the activated task number and the subsequent task is established, and a scheduling task trigger state chain is generated.

[0023] As a further scheme of the present application, the task content logical connection map obtaining step is specifically:

[0024] S411: According to the task number in the scheduling task trigger state chain, the corresponding construction area label, operation type label and rhythm control identifier are extracted, a corresponding label item set is established using the task number and the three label contents, and the three label contents are spliced in the order of the task number to generate a path segment sequence set;

[0025] S412: The path segment sequence set is called, the order position of the label item in the sequence is extracted for the continuous task path segment, consistency judgment between adjacent segments is performed, path segment combinations with associated features of the label content are identified, the label comparison relationship in the path segment group is obtained, and a label consistency comparison result is generated;

[0026] S413: According to the label consistency comparison result, in combination with the task number order information, it is judged whether the label combination features between adjacent task path segments have a continuous relationship, a one-way connection path is constructed for the task numbers with a continuous relationship, the task nodes are organized according to the path structure, and a task content logical connection map is generated.

[0027] As a further scheme of the present application, the method further comprises a step S5:

[0028] S5: The task number in the task content logical connection map is called, and the task pair forming a connection relationship is judged for the adjacentness of the section number, the task group that meets the judgment condition is entered into scheduling reconstruction, the operation rhythm identifier is compared with the real-time section operation rhythm, and the synchronization group is verified and merged into the same section to generate a construction site task parallel control result;

[0029] The construction site task parallel control result includes a section synchronization operation group, a scheduling rhythm consistency verification result and a scheduling reconstruction section set.

[0030] As a further scheme of the present application, the task content logical connection map obtaining step is specifically:

[0031] S511: Based on the task number in the task content logical connection map, each pair of task numbers forming a connection relationship and the section number are called, and according to the order of the numbers, whether the section numbers of adjacent tasks form a continuous relationship is compared, the adjacentness of the task pair execution section number forming a connection relationship is judged, and a group of adjacent task pair numbers is generated;

[0032] S512: According to the adjacent task pair number group, the work rhythm identifier corresponding to each task pair and the work rhythm value of the real-time section are extracted, the consistency between the task work rhythm identifier and the real-time section work rhythm value is compared, the work rhythm synchronized task combination is screened, and a synchronized task combination list is obtained.

[0033] S513: The segment number and task number set belonging to the synchronized task combination list are called, the tasks in the same combination are merged into the corresponding section, the task number control relationship under the corresponding section is identified, the construction site task parallel control result is generated.

[0034] The multi-agent-based construction site task scheduling system is used to execute the multi-agent-based construction site task scheduling method described above, and the system comprises:

[0035] The task triggering module obtains the starting time range of the agent task to be executed in the construction area, the resource access time and the unit work efficiency record, judges the overlapping interval of the task starting time and the resource access time, defines the task matched with the unit work efficiency record as a triggerable task, sets the task execution priority, generates a multi-agent task execution priority group;

[0036] The conflict detection module calls the task starting time, work space number and expected execution time in the multi-agent task execution priority group, compares the start and end time of the tasks in the same area, marks the task combination and the corresponding space section number that cause the intersection conflict, and establishes a construction site scheduling behavior section structure table;

[0037] The state judgment module calls the work completion flag, operation content quantity and positioning record time of the construction site scheduling behavior section structure table, marks the tasks that meet the conditions as active state, and generates a scheduling task triggering state chain;

[0038] The path construction module extracts the construction area label, work type label and rhythm control identifier according to the startable task number in the scheduling task triggering state chain, compares the number of continuous items of the path segment, and obtains a task content logical connection graph;

[0039] The parallel control module calls the task pair section number in the task content logical connection graph, judges the adjacency of the space section, compares the rhythm control identifier and the real-time work rhythm, merges the rhythm synchronized tasks into the same section, and generates a construction site task parallel control result.

[0040] Compared with the prior art, the advantages and positive effects of the present application are that:

[0041] In the application, the task screening is guided by the cross judgment of the task starting time and the resource in-place state, the priority information is dynamically given by combining the job type demand, the intelligent identification of the task trigger condition is realized, the mutual exclusion conflict relationship is marked in time by the overlap comparison of the regional task time interval, the pre-identification ability of the resource conflict is promoted to the scheduling behavior, the real-time response mechanism is given to the subsequent task by the activation condition reflected by the task state data, the path continuous identification relationship is established in the task sequence to effectively support the coherent promotion of the rhythm control, the job group fusion is realized by the adjacency and synchronism judgment of the space section, the precise matching of the job rhythm consistency is promoted, the synchronization of the logical relationship between tasks and the spatial deployment is ensured to achieve, and the scheduling continuity and the resource consistency in the multi-agent task collaborative execution are enhanced. BRIEF DESCRIPTION OF DRAWINGS

[0042] Figure 1 It is a workflow schematic diagram of the application;

[0043] Figure 2 It is a flowchart of the acquisition of the multi-agent task execution priority group in the application;

[0044] Figure 3 It is a flowchart of the acquisition of the construction site scheduling behavior section structure table in the application;

[0045] Figure 4 It is a flowchart of the acquisition of the scheduling task trigger state chain in the application;

[0046] Figure 5 It is a flowchart of the acquisition of the task content logic connection map in the application;

[0047] Figure 6 It is a flowchart of the acquisition of the construction site task parallel control result in the application. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical scheme and advantages of the application more clear and obvious, the application is further described in detail below by combining with the drawings and examples. It should be understood that the specific examples described here are only used to explain the application, and are not used to limit the application.

[0049] In the description of the present application, it should be understood that the terms "length", "width", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation of the present application. In addition, in the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise explicitly and specifically limited.

[0050] Please refer to Figure 1 The present application provides a technical solution, a construction site task scheduling method based on multiple agents, comprising the following steps:

[0051] S1: Obtain the start time range of the tasks to be executed by the agents in the construction area, the resource arrival time and the unit work efficiency record, perform time period coverage judgment on the task start time and the resource arrival time, match the unit work efficiency record with the task type demand, set the tasks that meet the conditions as triggerable tasks, set execution priorities for the triggerable tasks according to the work types, and form a multiple-agent task execution priority group;

[0052] S2: Based on the multiple-agent task execution priority group, call the predicted execution time of the work and the work space number, compare the start time and end time of the tasks in the same area, judge whether concurrent or cross conflict occurs in the same scheduling period, set mutual exclusion identifiers for the task combinations that exist in the cross, register the corresponding task numbers and work space section numbers, and generate a construction site scheduling behavior section structure table;

[0053] S3: Call the completed task numbers in the construction site scheduling behavior section structure table, correspondingly extract the work completion flag, the number of operation contents and the positioning record time, perform state judgment and value comparison on the three data, set the tasks that pass the judgment as active tasks, mark the next sequence tasks as startable, generate a scheduling task trigger state chain;

[0054] S4: According to the startable task numbers in the scheduling task trigger state chain, extract the construction area label, the work type label and the rhythm control identifier, combine the three labels to build a path segment, compare the number of consecutive items of the path segment, establish a content continuity identifier relationship between the tasks, and generate a task content logical connection map;

[0055] S5: call the task number in the task content logical connection graph, make adjacency judgment on the section number of the task pair forming the connection relationship, and the task group meeting the judgment condition enters the scheduling reconstruction, compare the operation rhythm identifier with the real-time section operation rhythm, verify the synchronization group, and then merge into the same section to generate the construction site task parallel control result;

[0056] The multi-agent task execution priority group includes a task start time interval, a resource arrival determination result, and a task type corresponding priority. The construction site scheduling behavior section structure table includes a mutual exclusion identifier, a task number mapping relationship, and an operation space section number. The scheduling task trigger state chain includes an activated task number, a next task start marker, and a task state mapping chain. The task content logical connection graph includes a path segment sequence, a content continuity relationship between tasks, and an operation rhythm control identifier. The construction site task parallel control result includes a section synchronization operation group, a scheduling rhythm consistency verification result, and a scheduling reconstruction section set.

[0057] Please refer to Figure 2 The acquisition step of the multi-agent task execution priority group is specifically:

[0058] S111: Obtain the start time range of the agent to be executed task in the construction area, the resource arrival time, and the unit operation efficiency record. Interval coverage judgment is performed on the resource arrival time of each task and the task start time range. Tasks whose task start time range are completely covered by the resource arrival time are selected to generate a resource coverage task interval.

[0059] The planned start time of the task is set as the upper and lower bounds of the interval, where the upper bound is set as the task plan start time delayed by 5 minutes, and the lower bound is set as the plan start time advanced by 5 minutes. The obtained resource arrival time is recorded in minutes. The task time record and resource arrival data are read in sequence. The matching operation is performed with the task number as the key. The resource arrival time of each task is compared with the boundaries of its start time interval. If the resource arrival time is 15:25 and the start time interval of the task is 15:20 to 15:30, the comparison result meets the resource arrival time within the start interval range, and it is determined that the coverage relationship is established. If it does not meet, the resource is excluded. Such judgment is performed on the resource time point. Tasks that meet the "resource time point completely surrounded by the task time interval" are selected and marked as executable tasks and are included in the resource coverage task interval list. When processing batch tasks, a batch judgment mechanism is used for vector operation. Matrix A is a two-dimensional array of task time intervals, and matrix B is a resource time array. The interval inclusion judgment between the elements is realized to quickly select and generate the resource coverage task interval.

[0060] S112: Call the resource coverage task range, extract the corresponding task type requirements and construction site task records, compare the consistency between the task identifier and the task type field, filter the tasks whose task types match the construction site task records, and generate a set of matching task types.

[0061] The task type field is extracted item by item and matched with the task type identifier in the construction site task record. First, the task types are uniformly converted into strings of uniform length. The type similarity is extracted by comparing the strings character by character. Let the task type be "concrete pouring" and the construction site task record be "concrete construction". Their codes are 11001 and 11010 respectively. The Hamming distance between the two is taken as 1, which is less than the set threshold of 2. The task is judged to be of the same type and included in the matching result. In this process, the task types are compared one by one to build a similarity matrix. The judgment is made according to the set matching threshold. The threshold of 2 is based on empirical data and the field error tolerance range. The setting process is detailed in Table 1. The optimal threshold range is evaluated by comparing the results of 5 construction site task types with system task types.

[0062] Table 1: Task Type Similarity Threshold Setting Table

[0063]

[0064] As shown in Table 1, when the Hamming distance does not exceed 2, the matching rate remains above 85%, which meets the requirements for determining the type of engineering task. Based on the selected matching results, a set of matching task types is formed.

[0065] S113: Based on the task set matching task types, extract the job types corresponding to multiple tasks, divide the tasks into groups according to job types, and extract the resource input, unit work efficiency, required work period, and job obstacle factors for each group, using the following formula:

[0066] ,

[0067] Calculate the task execution priority within each task type group, sort the tasks within the task type in descending order of their numerical values, and obtain the multi-agent task execution priority group.

[0068] in, For the first In the type of assignment, the first The task execution priority, Representing the In the type of assignment, the first Resource input for the task Representing the In the type of assignment, the first The unit resource output efficiency of the task. The first The first The job hindering factor of the first The first The standard resource requirement of the first The first The total number of tasks;

[0069] The formula calculation logic is as follows: by constructing a fractional structure, the task resource input, resource efficiency, task standard demand and construction hindering factor are comprehensively considered. The absolute value of the difference between the actual output and the standard demand of resources in each task is calculated in the numerator part, which is used to reflect the deviation between the task resource allocation and the actual need. Then the square value of the resource input of this task is added to the square value of the unit output, and divided by the unit output value, to construct the adjustment effect of resource intensity on work efficiency. The sum of the two items constitutes the numerator of the priority index. The denominator is the linear sum of the resource input and output rate of each task, which is used as a normalization processing factor to control the scale difference brought by the different number of tasks. The output is the task priority value in the unified dimension. The whole formula evaluates the resource supply and demand difference, work efficiency and task intensity at the same time, and gets a sortable numerical result.

[0070] The job execution priority is a numerical index for measuring the degree to which the task should be arranged for execution under the current resource configuration. The higher the value, the more the task should be prioritized in the conditions of resource efficiency, demand matching and capacity conflict.

[0071] Parameter meaning and calculation process:

[0072] The resource input required by the task is the number of man-hours;

[0073] The unit resource output rate is square meters per hour;

[0074] The resource hindering factor reflects the resistance influence of site congestion, construction difficulty, etc. The value range is 0.1 to 1.5;

[0075] The standard resource requirement of the task is the number of man-hours;

[0076] Suppose there are 3 tasks in the template job type, and their parameters are shown in Table 2:

[0077] Table 2: Task parameter table of template job type

[0078]

[0079] As shown in Table 2, the above formula is as follows:

[0080] For T1 task:

[0081] ;

[0082] ;

[0083] The sum is 65;

[0084] For T2 task:

[0085] ;

[0086] ;

[0087] The sum is 138.5;

[0088] For T3 task:

[0089] ;

[0090] ;

[0091] The sum is 81.8;

[0092] The sum of the numerator is:

[0093] ;

[0094] The denominator is:

[0095] ;

[0096] Substitute the formula for calculation:

[0097] ;

[0098] The result shows that the average priority of task execution in the template operation type is 5.49, which can be used for subsequent task sorting process;

[0099] The advantage of the formula is that by superimposing the deviation between resources and expectations through the composite expression of resources and output rate, the rationality of task resource allocation can be effectively evaluated, and the goal of dynamic priority sorting in the overall system under the multi-task scenario is achieved.

[0100] Please refer to Figure 3 , the acquisition step of the site scheduling behavior section structure table is specifically:

[0101] S211: Based on the multi-agent task execution priority group, the estimated execution time of the job and the job space number are called to filter the task combination in the same area, the start time and end time of the task are obtained, the time overlap interval is compared, whether there is cross conflict in the same scheduling cycle is judged, and the cross conflict task combination is obtained;

[0102] The estimated execution time parameter of each task and the job space number to which it belongs are called, the tasks are classified into corresponding job areas according to the space number, for the task combination in the same area, the planned start time and end time data of each task are extracted one by one, and a task execution time overlap judgment mechanism is constructed by pairwise comparison. The time range of task A is 08:00 to 09:00, the time range of task B is 08:30 to 09:30, the overlapping part of the time interval is 08:30 to 09:00, and the overlapping length is 30 minutes. The execution judgment action is to extract the start and end time of tasks A and B and calculate the length of the overlapping interval by taking the maximum start time and the minimum end time of the two. If the result is greater than 0, it means there is a conflict, and if it is 0 or negative, it is determined that there is no intersection. Repeat this process to process the task pair combination in the same area task group, and generate task pairs with execution time conflicts in the same scheduling cycle. In actual scenarios, if high-altitude operation and concrete pouring tasks are scheduled at the same time in the same area, conflicts will occur due to equipment and personnel exclusivity, which need to be included in the subsequent conflict identification judgment process to obtain the cross conflict task combination.

[0103] S212: According to the cross conflict task combination, identify the task combination that produces cross conflict in the same scheduling cycle, extract the corresponding job space number, classify and register by section range, and get the conflict job space section number list;

[0104] Identify the job space number of each task group, use the number to locate the conflict task area, call the preset job space code field in the task parameter table, and match the task ID and space number by key value to identify the location index of the task in the space number list. Classify and summarize the tasks involved in the conflict according to the space range dimension, and use the space code prefix uniform method for classification identification. For example, the space number is set to "X101", "X102", etc. "X10" is used as a common prefix for space section grouping. Extract the number of tasks in each section, task distribution density, etc. in ascending order of space number to get the conflict job space section number list.

[0105] S213: Call the conflict job space section number list, compare the job space range and execution time between the conflict combination, extract the space overlap length, task priority value, and combine the space intersection density, using the formula:

[0106] ;

[0107] The mutual exclusion conflict identification value between the computing task pairs is set, the mutual exclusion identification is set, and the construction site scheduling behavior section structure table is established;

[0108] Among them, The mutual exclusion conflict identification value of the task pair and , represent the conflict frequency density of task and task in a unit operation section within a unit time, represent the crossing density value of task in the space section th region, , respectively represent the priority value of task and task ;

[0109] Formula calculation logic: adopt fraction structure, the numerator part is based on the unit time conflict frequency between the task pairs , superimposed with the sum of the crossing density of task a in each space section, the interference degree of task a is reflected by the cross-sectional summary of the crossing density, and the denominator is the sum of the priority values of task a and b, which is used to normalize the dependence degree of the conflict index on the scheduling priority of the task itself. The whole formula integrates the conflict frequency and the spatial density in a weighted manner, outputs the standardized mutual exclusion conflict identification value, and the operation structure has a linear addition and proportional control double structure, which ensures that the conflict severity can be quantitatively expressed;

[0110] The mutual exclusion conflict identification value is a numerical index for quantifying the degree of overlap and mutual interference of two tasks occurring simultaneously in space and time. This value considers the unit time conflict frequency, spatial crossing density, and the execution priority of each task. The larger the identification value, the stronger the repulsion between tasks, and the simultaneous execution should be avoided in scheduling arrangement;

[0111] Parameter meaning and calculation process:

[0112] is the conflict frequency density (unit: times / hour) of and in the same section within a unit time;

[0113] represents the crossing density value of task in the space section ;

[0114] , respectively represent the priority value of task The priority value of the task pair T105 and T109 is 1.00, and the priority value of the task pair T106 and T110 is 1.12. The higher the value, the stronger the conflict, and the task pair needs to be avoided in subsequent scheduling. The priority value of the task pair T105 and T109 is 1.00, and the priority value of the task pair T106 and T110 is 1.12. The higher the value, the stronger the conflict, and the task pair needs to be avoided in subsequent scheduling.

[0115] In the specific operation, the task The conflict frequency of the task in the same space segment is 4 times / hour, and the task has 3 cross density values under this space segment, which are 1.2, 1.0, and 0.8, and the priority of the task and is 3.0 and 4.0 respectively, then the operation process is as follows:

[0116] Calculate the numerator part:

[0117] ;

[0118] Calculate the denominator part:

[0119] ;

[0120] Substitute the formula to calculate:

[0121] ;

[0122] The result shows that the conflict recognition value of the task pair is 1.00, which provides a numerical basis for further scheduling and judgment.

[0123] Table 3: Conflict task pair parameter example table

[0124]

[0125] As shown in Table 3, the conflict recognition value of the task pair T105 and T109 is 1.00, and the conflict recognition value of the task pair T106 and T110 is 1.12. The higher the value, the stronger the conflict, and needs to be avoided in subsequent scheduling.

[0126] Please refer to Figure 4 , the acquisition steps of the scheduling task trigger state chain are as follows:

[0127] S311: Call the completed task number in the construction site scheduling behavior section structure table, corresponding to the extraction of job completion flag, operation content quantity and positioning record time, match the three types of data with the number and unify the number dimension for reconstruction, form the job state data combination item;

[0128] The task number field is retrieved, and the task state field in the job completion flag table is key value corresponding, the corresponding task flag value of the task with completion status marked as "1" is extracted, and the operation content number record table and positioning record table data corresponding to the task are called to extract the total number of operation items involved in each task and the operation space position information, and a three-dimensional task state array is constructed; in the construction process, the three data dimensions are aligned and reconstructed according to the task number, and are respectively set as dimension A: task completion flag set, dimension B: operation content number set, and dimension C: job space positioning set, to form a triple If the task T201 completion status is "1", the operation content is 4 items, and the operation space is "X105", the triple is recorded as (T201 1 4 X105), and the task data is executed to perform such triple assembly action to form a job state data combination item.

[0129] S312: According to the multiple number contents in the job state data combination item, it is judged that the job completion flag is the identification value, and the operation content number and task number group value are structure offset processed, and the formula is adopted:

[0130] ;

[0131] The single task combination response quantity is calculated, the task number of the single task combination response quantity greater than zero is recorded, and the activated task number item is obtained;

[0132] Wherein, represents the single task combination response quantity of the first task, represents the job completion flag of the first task, represents the operation content number of the first task, represents the median value of the operation content number in the same number task group of the first task, represents the corresponding operation node number of the first task in the job stage, represents the scheduling response interference factor of the task number in the corresponding work area, represents the product item of the operation content number and the matching task number in the task number task, represents the number of reverse waiting times in the task execution of the task number ;

[0133] Formula calculation logic: combined with three hierarchical substructures, the completion flag As a control switch item, the difference between the number of operation items and the number of operation items and interference factors The interaction effect term is used to reflect the cumulative effect of the actual completed task on local disturbances; the denominator uses the weighted number of nodes. Used for normalization, so that the impact of the difference in the number of operation terms on the response is balanced by the node complexity; square root term Used to convert area to linear space so that the task load area does not expand by the square of the area; the last term is the linear priority level number. The advantage of the formula is that by combining task completion, operational complexity, spatial factors and execution level, it can achieve accurate response screening before the task chain is triggered.

[0134] The single-task combination response is a quantitative indicator that measures the activation and execution of a task in the current job state. It comprehensively reflects multiple factors such as its operational complexity, node density, spatial interference, and priority. The larger the value, the more significant the impact of the task on the current scheduling structure and the stronger the execution triggering conditions.

[0135] Extract the task number field from each combination sequentially. Iterate through the completion status flags of each task to determine if it is in a completed state. If the flag is "1", proceed to the calculation process to extract the number of operations performed on that task. Average number of operations per task with the same task number within the task number group Number of operation nodes in the relevant work phase Interference factors in the work area Operation content parameters area value Task priority numbering ;

[0136] The parameters are defined as follows:

[0137] Let 1 be the task completion flag and 0 be the incomplete flag.

[0138] Taking task T305 as an example, let its completion marker be 1, the number of operation items be 6, the average number of operation items per number group be 4, the number of operation nodes in the operation phase be 2, the interference factor be 1.1, and the operation area be 36m². 2 The priority number is 2. Substituting it into the formula, the calculation is as follows:

[0139] Calculate the difference term and multiply it by the interference factor:

[0140] ;

[0141] Add the normalized node factor:

[0142] ;

[0143] Square root area term:

[0144] ;

[0145] Substitute the formula for calculation:

[0146] ;

[0147] The results show that the single task combination response quantity is 8.733, which is used to activate the task number item to determine whether the activation threshold (set the activation threshold to 5) is met, and the task meets the activation condition;

[0148] Table 4: Unit task combination response quantity calculation example table

[0149]

[0150] As shown in Table 4, tasks T305 and T306 obtain response quantities of 8.733 and 6.000, respectively, both of which are greater than the activation threshold of 5, so both of them will be marked as active tasks.

[0151] S313: Call the active task number item, retrieve the corresponding next task number, assign the start identification field to the next task, and establish a mapping path relationship between the active task number and the next task, and generate a scheduling task trigger state chain;

[0152] Find the successor task item identifier in the task list, obtain the next task number that has a dependency relationship with the active task number item through the relationship mapping table between tasks, and traverse each active task item in turn. The successor task is marked as a triggerable task, and each pair of "active task-next task" relationship is registered in the task mapping table to form a bidirectional retrievable structure in the form of task number pairs. Set the next task of the active task T305 as T309 and T311, construct the relationship pairs (T305 T309), (T305 T311), and record them in the mapping list structure. The subsequent scheduling process triggers the task state chain operation according to this structure to generate a scheduling task trigger state chain.

[0153] Please refer to Figure 5 , the steps of obtaining the task content logical continuation graph are as follows:

[0154] S411: According to the startable task number in the scheduling task trigger state chain, extract the corresponding construction area label, operation type label and rhythm control identifier, use the task number and the three label contents to establish a corresponding label item set, and then splice the three label contents in order of task number to generate a path segment sequence set;

[0155] Read the field content in the task attribute table, extract the construction area label, operation type label and rhythm control identifier corresponding to the task, set the task T001 corresponding area as "Southeast Zone One", operation type as "Steel Arrangement", and rhythm control identifier as "Rhythm Section A", and repeat the extraction operation for task numbers T002, T003, etc. The three types of labels of each task need to be positioned in the structured data table according to the task number. After field reading, the three labels are merged and spliced in order, and T001 is spliced as "Southeast Zone One-Steel Arrangement-Rhythm Section A", and T002 is "Southeast Zone One-Concrete Pouring-Rhythm Section A", forming a label path segment. After all splicing is completed, organize according to the task number sequence to establish a sequence set of path segments, which can be stored in the task path tracking table for subsequent calling. The splicing operation is completed by connecting the label field content with a hyphen. If the T003 label is "Northwest Zone Two, Formwork Installation, Rhythm Section B", the spliced result is "Northwest Zone Two-Formwork Installation-Rhythm Section B", and the path segment sequence set is obtained.

[0156] S412: Call the path segment sequence set, extract the order position of the label item in the sequence for the continuous task path segment, judge the consistency between adjacent segments, identify the path segment combination with correlated label content, obtain the label comparison relationship in the path segment group, and generate the label consistency comparison result.

[0157] Compare the labels of any two adjacent path segments to determine whether there is a correlation between them. Extract adjacent pairs in the order of path segment appearance, set the first and second, the second and third, etc. Then identify the label content in each group of path segments, and compare the construction area label, operation type label and rhythm control identifier in order. If the label content in the two path segments is exactly the same, it is considered to be highly consistent. If two labels are the same, it is considered to be moderately consistent. If only one is the same, it is considered to be low consistency. If they are completely inconsistent, they are considered to have no continuity. Path segment 1 is "Southeast Zone One-Steel Arrangement-Rhythm Section A", path segment 2 is "Southeast Zone One-Formwork Installation-Rhythm Section A", the construction area and rhythm control identifier are the same, and it is determined to be moderately consistent. Continue to compare path segment 2 with path segment 3. If path segment 3 is "Southeast Zone One-Formwork Installation-Rhythm Section B", only the operation type is the same, and it is determined to be low consistency. In actual operation, the comparison level of each group of path segments needs to be marked to the path label comparison for recording, and the label consistency comparison result is generated.

[0158] S413: According to the label consistency comparison result, combine the task number sequence information, judge whether the label combination characteristics between adjacent task path segments have a continuous relationship, construct a one-way connection path for the task numbers with a continuous relationship, organize the task nodes according to the path structure, and generate a task content logical connection map.

[0159] In combination with the order structure of the original scheduling task chain by task number, it is determined which adjacent task combination features have continuity. The path segment pairs marked as "highly consistent" or "moderately consistent" in the comparison result can be considered to have label combination continuity. If the tasks T002 and T003 are moderately consistent, a one-way connection between their task numbers is established, i.e., T002 is connected to T003. Then, it is checked whether T003 and T004 have the same relationship. If the requirement is met, the connection is continued to form a path sequence structure. The continuous number path, such as T002→T003→T004, is taken as a connection path. The connection path structure that meets the condition is collected. The task number is used to construct a graph structure. The connection relationship between the task numbers can be drawn in the form of a linear directed graph. In the actual construction process, the number index can be used for node mapping. The connection path is organized as an adjacency information list for storage, ensuring that the continuous logical structure between task nodes is clearly displayed in the graph. A task content logical connection graph is generated.

[0160] Please refer to Figure 6 The acquisition steps of the parallel control result of the construction site task are as follows:

[0161] S511: Based on the task number in the task content logical connection graph, the task number and the section number of each pair of tasks forming a connection relationship are called. According to the order of the numbers, the adjacent section number judgment of the task pair forming the connection relationship is performed by comparing whether the section numbers of adjacent tasks form a continuous relationship. The adjacent task pair number group is generated.

[0162] The task pair number list forming the connection relationship in the graph and the corresponding task section number list need to be extracted. The task number is generally a non-repeating integer code, and T1, T2, T3, etc. are set. The section number is a sequential number representing its physical location, such as Z101, Z102, Z103, etc. After extraction, structure recognition operation is performed on each task pair number, i.e., according to the order of the numbers, such as T1→T2, T2→T3, etc., the connection direction between the task pairs is constructed. The section numbers Z a , Z_b of the two tasks in the corresponding task pair are extracted and compared to see if they are continuous sections. The continuity judgment standard of the section number can be set by setting the section number difference ΔZ=Zb-Z aWhether it is an integer multiple of the minimum unit section number interval, the difference can be determined as a logical continuity criterion without setting a specific value, set: if the number interval between Z101 and Z102 is set as ΔZ=1 unit section, Z102 and Z101 are continuous sections, the judgment process can be completed by traversing the task pair set and calling the corresponding section number set to perform logical judgment operations, in the example, if there is a task pair T12→T13, its corresponding section is Z202→Z203, ΔZ=203-202=1 is judged, which satisfies the continuous relationship, then the task pair is determined as an adjacent task pair, the task pair numbers that satisfy the continuous relationship are structurally combined to obtain the adjacent task pair number group.

[0163] S512: According to the adjacent task pair number group, extract the task rhythm identifier corresponding to each group and the task rhythm value in the real-time section, and filter the task combination with synchronous work rhythm by comparing the consistency between the task work rhythm identifier and the real-time section work rhythm value, to obtain the synchronous task combination list;

[0164] Extract the task work rhythm identifier corresponding to each number in each group and the task work rhythm value in the real-time section one by one, the task work rhythm identifier can be the rhythm number of task execution, set J101, J102, etc., and the real-time section work rhythm value can be the standard rhythm identifier currently set or collected in each physical section, set the rhythm value corresponding to section Z101 as J101, and judge whether the task work rhythm identifier and the section work rhythm value are consistent one by one to identify whether the task and the section it is in form a matching relationship in work rhythm, the comparison method used in the judgment operation is character or number comparison, if JTn=JZn, the task rhythm is consistent, otherwise it is inconsistent, the task combination list with synchronous rhythm is constructed by extracting the task combination with consistent relationship, in the example, tasks T15 and T16 are located in section Z205, T15 rhythm identifier is J201, and Z205 real-time rhythm value is also J201, then T15 is a synchronous task, if T16 is J202, it is excluded, only the synchronous rhythm task combination is retained, all adjacent task pair number groups that satisfy the rhythm consistency are sequentially screened out, and the synchronous task combination list is obtained through set classification operation.

[0165] S513: Call the section number and task number set in the synchronous task combination list, merge the tasks in the same combination into the corresponding section to identify the task number control relationship under the corresponding section, and generate the construction site task parallel control result;

[0166] The tasks in the combination need to be merged in the physical section, that is, multiple tasks in the combination are simultaneously included in the parallel control mechanism of the same section, the task number set {T1, T2, …, T n} and the corresponding section number Zx, the task set mapping is performed on Zx, so that the numbers belong to the same section node, after the execution structure mapping, the task control attribution relationship within each combination is identified, the parallel control link between the task numbers is established through the connection logic atlas between the task numbers and the space topology under the same section, in the actual construction process, the control relationship between the task pairs adopts the task order index matching mode, the combination A contains tasks T21, T22 and T23, the task order is T21→T22→T23, and the corresponding section is Z301, so the combination is merged into the parallel control task chain under Z301, the task parallel control rule set can be established through the chain structure, each rule set is expressed as the mapping structure {Z301:[T21,T22,T23]}, and the construction site task parallel control result is generated.

[0167] The construction site task scheduling system based on multiple agents is used for executing the construction site task scheduling method based on multiple agents, and the system comprises:

[0168] The task triggering module obtains the starting time range of the tasks to be executed by the agents in the construction area, the resource arrival time and the unit work efficiency record, judges the overlapping interval of the task starting time and the resource arrival time, defines the tasks matched with the unit work efficiency record as triggerable tasks, sets the task execution priority, and generates a multi-agent task execution priority group.

[0169] The conflict detection module calls the task starting time, the work space number and the expected execution duration in the multi-agent task execution priority group, compares the start and end times of the tasks in the same region, marks the task combination and the corresponding space section number that cause the intersection conflict, and establishes a construction site scheduling behavior section structure table.

[0170] The state judgment module calls the work completion flag, the operation content quantity and the positioning record time of the construction site scheduling behavior section structure table, marks the tasks that meet the conditions as active states, and generates a scheduling task triggering state chain.

[0171] The path construction module extracts the construction area label, the work type label and the rhythm control identifier according to the startable task number in the scheduling task triggering state chain, compares the number of continuous items of the path segments, and obtains a task content logical connection atlas.

[0172] The parallel control module calls the task pair section number in the task content logical connection atlas, performs the adjacency judgment of the space section and the comparison between the rhythm control identifier and the real-time work rhythm, merges the tasks with the same rhythm to the same section, and generates a construction site task parallel control result.

[0173] The above merely describes the preferred embodiments of the present application, and is not intended to limit the present application in other forms. Any skilled person in the art can modify or change the disclosed technical content into equivalent embodiments with equivalent changes, and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present application, without departing from the technical solution content of the present application, still falls within the protection scope of the present application.

Claims

1. A construction site task scheduling method based on multi-agent systems, characterized in that, Includes the following steps: S1: Obtain the start time range, resource arrival time and unit operation efficiency record of the tasks to be executed by the intelligent agents in the construction area, make time period coverage judgment on the task start time and resource arrival time, match the unit operation efficiency record and task type requirements, and form a multi-agent task execution priority group. S2: Based on the multi-agent task execution priority group, call the estimated execution time of the job and the job space number, and compare the overlapping intervals of the start time and end time of the tasks in the same area to determine whether there are concurrent or cross conflicts in the same scheduling cycle, and generate a site scheduling behavior segment structure table. S3: Call the completed task number in the construction site scheduling behavior segment structure table, extract the corresponding work completion flag, operation content quantity and location record time, set the approved task as the active task, and generate a scheduling task trigger state chain. S4: Based on the task number that can be started in the scheduling task triggering state chain, extract the construction area label, operation type label and rhythm control identifier, merge the three labels to construct a path segment, and then compare the number of consecutive items in the path segment to generate a task content logical continuity graph. S5: Call the task number in the task content logical succession diagram, perform adjacency judgment on the segment number of the tasks that form a succession relationship, and enter the scheduling reconstruction of the task group that is determined to be successful. Compare the operation rhythm identifier with the real-time segment operation rhythm, verify the synchronization group and merge it into the same segment to generate the parallel control result of the construction site task. The parallel control results of the construction site tasks include synchronous operation groups within the section, consistency verification results of scheduling rhythm, and the set of scheduling reconstructed sections.

2. The construction site task scheduling method based on multi-agent technology according to claim 1, characterized in that, The multi-agent task execution priority group includes task start time interval, resource availability judgment result, and priority corresponding to task type. The construction site scheduling behavior segment structure table includes mutual exclusion identifier, task number mapping relationship, and work space segment number. The scheduling task trigger state chain includes activated task number, next-order task start mark, and task state mapping chain. The task content logical continuity graph includes path segment sequence, content continuity relationship between tasks, and work rhythm control identifier.

3. The construction site task scheduling method based on multi-agent technology according to claim 1, characterized in that, The specific steps for obtaining the multi-agent task execution priority group are as follows: S111: Obtain the start time range, resource arrival time and unit work efficiency record of the tasks to be executed by the intelligent agent in the construction area, and make interval coverage judgment between the resource arrival time and the task start time range of each task. Filter out the tasks whose task start time range is completely covered by the resource arrival time and generate the resource coverage task interval. S112: Call the resource coverage task range, extract the corresponding task type requirements and construction site task records, compare the consistency between the task identifier and the task type field, filter the tasks whose task types match the construction site task records, and generate a set of matching task types. S113: Based on the matching task type task set, extract the job types corresponding to multiple tasks, divide the tasks into groups according to job types, extract the resource input, unit job efficiency, required task duration and job obstacle factor of the tasks in the group, calculate the job execution priority of the tasks in each job type group, sort the tasks in the job type in descending order of value, and obtain the multi-agent task execution priority group.

4. The construction site task scheduling method based on multi-agent technology according to claim 3, characterized in that, The specific steps for obtaining the site scheduling behavior segment structure table are as follows: S211: Based on the multi-agent task execution priority group, call the estimated execution time of the job and the job space number, filter the task combinations in the same area, obtain the start time and end time of the task, compare the time overlap interval, determine whether cross-conflict occurs in the same scheduling cycle, and obtain the cross-conflict task combinations. S212: Based on the cross-conflicting task combination, identify the task combination that generates cross-conflict within the same scheduling cycle, extract the corresponding job space number, classify and register it according to the segment range, and obtain a list of conflicting job space segment numbers. S213: Call the list of conflicting work space segment numbers, compare the work space range and execution time between tasks in the conflict combination, extract the spatial overlap length and task priority value, and calculate the mutual exclusion conflict identifier value between task pairs in combination with the spatial intersection density, set the mutual exclusion identifier, and establish a site scheduling behavior segment structure table.

5. The construction site task scheduling method based on multi-agent technology according to claim 4, characterized in that, The specific steps for obtaining the scheduling task trigger state chain are as follows: S311: Call the completed task number in the construction site scheduling behavior segment structure table, extract the corresponding work completion flag, operation content quantity and location record time, match the three types of data with the numbering and reconstruct them by unifying the numbering dimension to form a work status data combination item; S312: Based on the multiple number contents in the job status data combination item, determine the job completion flag as the identifier value, perform structural offset processing on the number of operation contents and the task number group value, calculate the single task combination response quantity, record the task number with a single task combination response quantity greater than zero, and obtain the activated task number item. S313: Call the activated task number item, retrieve the corresponding next task number, assign a start identifier field to the next task, establish a mapping path relationship between the activated task number and the next task, and generate a scheduling task trigger state chain.

6. The construction site task scheduling method based on multi-agent technology according to claim 5, characterized in that, The specific steps for obtaining the task content logical continuity graph are as follows: S411: Based on the task number that can be started in the scheduling task triggering state chain, extract the corresponding construction area label, operation type label and rhythm control identifier, establish a corresponding label item set using the task number and the three label contents, and then splice the three label contents in the order of task number to generate a path segment sequence set. S412: Call the path segment sequence set, extract the sequential position of the tag items in the sequence for continuous task path segments, perform consistency judgment between adjacent segments, identify path segment combinations with related features of tag content, obtain the tag comparison relationship in the path segment group, and generate tag consistency comparison results; S413: Based on the label consistency comparison results and the task number sequence information, determine whether the label combination features between adjacent task path segments have a continuous relationship. Construct a unidirectional connection path for task numbers with a continuous relationship, organize the task nodes according to the path structure, and generate a task content logical continuity graph.

7. The construction site task scheduling method based on multi-agent technology according to claim 6, characterized in that, The specific steps for obtaining the parallel control results of the construction site tasks are as follows: S511: Based on the task number in the task content logical succession diagram, call each pair of task numbers that form a succession relationship and the segment number in which they are located. According to the order between the numbers, by comparing whether the segment numbers of adjacent tasks form a continuous relationship, perform a segment number adjacency judgment on the task pairs that form a succession relationship and generate adjacent task pair number groups. S512: Based on the adjacent task pair numbering group, extract the operation rhythm identifier corresponding to each group of tasks and the operation rhythm value of the real-time segment. By comparing the consistency between the task operation rhythm identifier and the real-time segment operation rhythm value, filter the task combinations with synchronized operation rhythms to obtain a list of synchronized task combinations. S513: Call the set of segment numbers and task numbers in the synchronous task combination list, merge the tasks in the same combination into the corresponding segment, identify the task number control relationship under the corresponding segment, and generate the parallel control result of the construction site tasks.

8. A construction site task scheduling system based on multi-agent systems, characterized in that, The system is used to implement the multi-agent-based construction site task scheduling method according to any one of claims 1-7, and the system includes: The task triggering module obtains the start time range, resource arrival time and unit operation efficiency record of the tasks to be executed by the intelligent agents in the construction area, judges the coverage interval of the task start time and resource arrival time, defines the tasks that match the unit operation efficiency record as triggerable tasks, sets the task execution priority, and generates a multi-agent task execution priority group. The conflict detection module calls the task start time, job space number and expected execution time in the multi-agent task execution priority group, compares the start and end times of tasks in the same area across intervals, marks the task combinations that cause cross-conflicts and their corresponding space segment numbers, and establishes a site scheduling behavior segment structure table. The status judgment module calls the work completion flag, operation content quantity and location record time of the construction site scheduling behavior segment structure table, marks the tasks that meet the conditions as active, and generates a scheduling task trigger status chain. The path construction module extracts the construction area label, operation type label and rhythm control identifier based on the startable task number in the scheduling task trigger state chain, compares the number of consecutive items in the path segment, and obtains the task content logical continuity graph. The parallel control module calls the task pair segment number in the task content logical sequence diagram, performs spatial segment adjacency judgment and compares the rhythm control mark with the real-time operation rhythm, merges rhythm-synchronized tasks into the same segment, and generates the construction site task parallel control result.

Images