A multi-machine cooperative intelligent scheduling management model, method and application

Abstract

This invention discloses a multi-machine collaborative intelligent scheduling management model, method, and application, relating to the field of construction equipment automation control and collaborative scheduling technology. It includes a scheduling main architecture comprising a field perception access layer, a twin state organization layer, a collaborative path decision layer, and an execution feedback correction layer, handling the entire process from equipment data to scheduling implementation. The collaborative scheduling link, by embedding a task sorting and allocation unit, a global path timing generation unit, and a rolling conflict resolution unit in the collaborative path decision layer, achieves task allocation, path orchestration, and conflict resolution. The model described in this invention achieves significantly better results in equipment data organization, target association expression, task unit generation, path timing orchestration, spatiotemporal occupancy judgment, and feedback correction archiving.

Description

Technical Field

[0001] This invention relates to the field of automated control and collaborative scheduling technology for construction equipment, specifically a multi-machine collaborative intelligent scheduling management model, method, and application. Background Technology

[0002] With the continuous expansion of the construction scale of prefabricated buildings, high-rise buildings and large-scale complex projects, construction sites are gradually shifting from single-equipment operations to multi-equipment collaborative operations.

[0003] Construction machines, scaffolding machines, concrete placing booms, construction hoists, and other equipment operate concurrently within the same construction area, and there is a strong correlation between their operating status, spatial location, work progress, and construction tasks.

[0004] To improve the continuity of construction organization and the degree of automation of on-site control, existing technologies are gradually introducing methods such as equipment status acquisition, positioning monitoring, digital twin modeling, path planning, and scheduling control to centrally manage construction equipment and display equipment operation status and task execution status through a visual interface.

[0005] Field equipment data, location data, status data, and progress data often come from different sources, and there is a lack of unified organizational relationships between the data, making it difficult to directly form targets, target statuses, and target relationships that can be used for multi-device collaborative scheduling.

[0006] Existing scheduling methods mostly focus on single device operation control or task dispatch, lacking a continuous processing link from task unit organization and global path timing generation to spatiotemporal occupancy judgment.

[0007] When multiple devices are running in the same construction area, adjacent work surfaces, or intersecting paths, problems such as path conflicts, time overlaps, and unclear area occupancy are likely to occur. Adjustments still need to be made based on manual experience, and the scheduling results are difficult to be fed back to the subsequent parameter correction and data archiving process in a timely manner. Summary of the Invention

[0008] In view of the above-mentioned problems, the present invention is proposed.

[0009] Therefore, the technical problem solved by this invention is that existing collaborative scheduling and management models for construction equipment suffer from problems such as scattered equipment data organization, insufficient correlation of target status, low connectivity between task allocation and path arrangement, and insufficient continuity of conflict judgment and feedback correction. The problem is how to achieve continuous collaborative scheduling processing from equipment data access, target status organization, task unit generation, global path timing arrangement, spatiotemporal occupancy judgment to feedback correction and archiving.

[0010] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a multi-machine collaborative intelligent scheduling and management model, including a scheduling main architecture.

[0011] It includes a collaborative path decision-making layer, which handles the entire process of intelligent decision-making for multi-machine collaborative paths.

[0012] The collaborative scheduling link achieves task allocation, path orchestration, and conflict resolution by embedding a task sorting and allocation unit, a global path timing generation unit, and a rolling conflict resolution unit in the collaborative path decision layer.

[0013] As a preferred embodiment of the multi-machine collaborative intelligent scheduling and management model described in this invention, it further includes a task unit organization unit, which is located between the twin state organization layer and the task sorting and allocation unit.

[0014] Read the targets formed by the twin state organization layer, generate a set of task units based on the relationship between the targets, and write them into the task definition field set.

[0015] As a preferred embodiment of the multi-machine collaborative intelligent scheduling and management model described in this invention, it further includes a spatiotemporal occupancy organization unit, which is located between the global path timing generation unit and the rolling conflict resolution unit.

[0016] The spatiotemporal occupancy organization unit reads the global path timing output by the global path timing generation unit and converts the elements and planned execution times in the global path timing into spatiotemporal occupancy windows; The spatiotemporal occupancy window establishes a correspondence with the target in the twin state organization layer.

[0017] As a preferred embodiment of the multi-machine collaborative intelligent scheduling and management model described in this invention, the scheduling main architecture further includes a field perception access layer.

[0018] The field sensing access layer categorizes and receives device data.

[0019] As a preferred embodiment of the multi-machine collaborative intelligent scheduling and management model described in this invention, the scheduling main architecture further includes a twin state organization layer.

[0020] The received device data is categorized and written into the object index field set, and targets, target statuses, and target relationships are formed based on a unified spatial benchmark.

[0021] As a preferred embodiment of the multi-machine collaborative intelligent scheduling and management model described in this invention, the collaborative path decision layer receives the target, target state, and target association relationship formed by the twin state organization layer and connects to the collaborative scheduling link, thereby outputting task sorting results, allocation results, global path timing, conflict identifiers, and scheduling execution information.

[0022] As a preferred embodiment of the multi-machine collaborative intelligent scheduling and management model described in this invention, the scheduling main architecture further includes an execution feedback correction layer.

[0023] Receive feedback data after scheduling is implemented, and associate the feedback data with the corresponding task unit set, global path timing, and spatiotemporal occupancy window.

[0024] Based on the feedback data, parameters are generated to correct the input and archived data.

[0025] Another objective of this invention is to provide a multi-machine collaborative intelligent scheduling and management method, which can organize equipment data into targets, target states, and target relationships, and generate task unit sets, global path timing, and spatiotemporal occupancy windows sequentially based on the target relationships. This solves the problems in current construction equipment collaborative scheduling technology, such as the lack of continuous mapping between equipment data and scheduling tasks, and the difficulty in directly participating in conflict prediction and feedback correction of path arrangement results.

[0026] As a preferred embodiment of the multi-machine collaborative intelligent scheduling and management method described in this invention, it includes receiving and classifying device data through a field sensing access layer.

[0027] The twin state organization layer forms the target, target state, and target association relationship, and generates a set of task units.

[0028] Through the collaborative scheduling link in the collaborative path decision layer, the task sorting and allocation unit, the global path timing generation unit, and the rolling conflict resolution unit are invoked to generate and adjust the global path timing.

[0029] The system outputs scheduling execution information and receives feedback data by performing a feedback correction layer.

[0030] It also includes an execution feedback correction layer that associates feedback data with task unit sets, global path timing, and conflict identifiers to form parameter correction inputs.

[0031] The collaborative path decision layer modifies the parameters of the task sorting and allocation unit, the global path timing generation unit, and the rolling conflict resolution unit based on the parameters, and archives them after the scheduling cycle ends.

[0032] Another objective of this invention is to provide a multi-machine collaborative intelligent scheduling and management application, which includes at least two of the following: building construction machine, scaffolding climbing machine, concrete placing machine, and construction hoist.

[0033] The beneficial effects of the present invention are as follows: The multi-machine collaborative intelligent scheduling and management model provided by the present invention achieves better results in terms of equipment data organization, target association expression, task unit generation, path timing arrangement, spatiotemporal occupancy judgment, and feedback correction archiving. Attached Figure Description

[0034] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0035] Figure 1 This is a schematic diagram of the collaborative scheduling link embedding structure of a multi-machine collaborative intelligent scheduling management model provided in Embodiment 1 of the present invention.

[0036] Figure 2 This is a diagram of the scheduling simulation interface of a multi-machine collaborative intelligent scheduling management model provided in Embodiment 1 of the present invention.

[0037] Figure 3 This is a schematic diagram of the task unit organization unit connection structure of a multi-machine collaborative intelligent scheduling and management model provided in Embodiment 2 of the present invention.

[0038] Figure 4 This is a matrix diagram of a multi-machine collaborative intelligent scheduling and management model provided in Embodiment 2 of the present invention.

[0039] Figure 5 This is a task unit condition matrix diagram of a multi-machine collaborative intelligent scheduling and management model provided in Embodiment 2 of the present invention.

[0040] Figure 6 This is an experimental verification diagram of the task unit organization of a multi-machine collaborative intelligent scheduling and management model provided in Embodiment 2 of the present invention.

[0041] Figure 7 This is a schematic diagram of the spatiotemporal occupancy organization unit connection structure of a multi-machine collaborative intelligent scheduling and management model provided in Embodiment 3 of the present invention.

[0042] Figure 8 This is a schematic diagram of the global path timing to spatiotemporal occupancy window conversion for a multi-machine collaborative intelligent scheduling and management model provided in Embodiment 3 of the present invention.

[0043] Figure 9 This is a schematic diagram of the spatiotemporal occupancy window cube conflict retrieval for a multi-machine collaborative intelligent scheduling and management model provided in Embodiment 3 of the present invention.

[0044] Figure 10 This is a schematic diagram of the field perception access layer structure of a multi-machine collaborative intelligent scheduling and management model provided in Embodiment 4 of the present invention.

[0045] Figure 11 This is a schematic diagram of the twin state organization layer structure of a multi-machine collaborative intelligent scheduling and management model provided in Embodiment 5 of the present invention.

[0046] Figure 12 This is a schematic diagram of the collaborative path decision layer structure of a multi-machine collaborative intelligent scheduling and management model provided in Embodiment 6 of the present invention.

[0047] Figure 13 This is a schematic diagram of the execution feedback correction layer structure of a multi-machine collaborative intelligent scheduling and management model provided in Embodiment 7 of the present invention. Detailed Implementation

[0048] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0049] Example 1

[0050] Reference Figures 1-2 As an embodiment of the present invention, a multi-machine collaborative intelligent scheduling management model is provided, including a scheduling main architecture 100 and a collaborative scheduling link 200, wherein the collaborative scheduling link 200 is embedded in the collaborative path decision layer 101 of the scheduling main architecture 100.

[0051] In this embodiment, the scheduling main architecture 100 includes at least a collaborative path decision layer 101, which is used to handle the entire process of intelligent decision-making for multi-machine collaborative paths; the field perception access layer 103, the twin state organization layer 102, and the execution feedback correction layer 104 can be further described as supporting layers of the scheduling main architecture 100 in subsequent embodiments.

[0052] In practice, the field perception access layer 103 is deployed between the equipment side and the edge gateway side at the construction site to receive equipment data from multiple devices at the construction site.

[0053] The equipment data includes at least one of the following: equipment operation data, location data, status data, work progress data, and environmental data.

[0054] Equipment operation data can be output by a PLC controller, frequency converter, or equipment body control unit. Position data can be output by a UWB positioning base station, laser rangefinder, encoder, or visual positioning device. Status data can include equipment start / stop status, lifting status, load status, limit status, and fault status. Work progress data can come from construction task sheets, floor construction plans, or on-site work confirmation information. Environmental data can include wind speed, temperature and humidity, obstacle status, and construction area occupancy status.

[0055] After receiving the data from the aforementioned devices, the on-site sensing access layer 103 writes a unified collection time, device identifier, data source identifier, and construction area identifier into the data from different sources.

[0056] For data from different sampling frequencies, the field sensing access layer 103 uses a unified scheduling period as a time reference for data alignment.

[0057] Specifically, when the sampling period for PLC operation data is 100ms, the sampling period for UWB position data is 200ms, and the update period for work progress data is 5s, the field perception access layer 103 uses a scheduling period of 1s as a unified processing period, merges the equipment operation data, position data, and work progress data within the same scheduling period into the same equipment data packet, and outputs the equipment data packet to the twin state organization layer 102.

[0058] The twin state organization layer 102 receives the device data packets output by the field perception access layer 103 and establishes the target, target status and target association relationship based on the spatial reference of the construction site.

[0059] The spatial reference can be determined using floor coordinates, construction area coordinates, and equipment operating boundary coordinates in the BIM model, or it can be determined using the unified world coordinate system of the construction site. The twin state organization layer 102 writes the equipment identifier, spatial location, status field, and construction area identifier from the equipment data package into the object index field set, and forms the target based on the object index field set.

[0060] In this embodiment, the target is an objectified representation used to participate in the scheduling process. The target can correspond to at least one of equipment, construction area, work task, personnel location, or material location.

[0061] The target status includes at least one of the following: target current location, target running status, target occupied status, target schedulable status, and target abnormal status.

[0062] The target relationships include the regional affiliation between equipment and construction area, the execution relationship between equipment and work task, the spatial correspondence between construction area and work task, the occupation relationship between personnel location and construction area, and the avoidance relationship between material location and equipment running path.

[0063] For example, when the first building machine is located in the construction area on the east side of the fifth floor and is performing the formwork transportation task, and the second building machine is located in the construction area on the south side of the fifth floor and is performing the material lifting task, the twin state organization layer 102 forms the first target and the second target respectively, and records the target states of the two.

[0064] If the first target and the second target need to pass through the same floor transition area in subsequent scheduling cycles, the twin state organization layer 102 records the common association between the two targets and the floor transition area in the target association relationship.

[0065] The collaborative path decision layer 101 receives the target, target status and target association for multi-machine collaborative path intelligent decision-making, and embeds the collaborative scheduling link 200 in the collaborative path decision layer 101.

[0066] The collaborative scheduling link 200 achieves task allocation, path orchestration, and conflict resolution by embedding a task sorting and allocation unit 201, a global path timing generation unit 202, and a rolling conflict resolution unit 203 in the collaborative path decision layer 101.

[0067] Specifically, the task sorting and allocation unit 201 reads the construction tasks within the current scheduling cycle, as well as the target status and target relationships used for decision-making. Construction tasks may include floor construction tasks, area operation tasks, material delivery tasks, equipment transfer tasks, and work surface switching tasks. The task sorting and allocation unit 201 sorts and allocates the construction tasks according to the target status and target relationships, and generates task sorting results and allocation results.

[0068] In one specific implementation, the task sorting and allocation unit 201 determines the task sorting result based on the urgency of the task, the availability of the target, the distance to the target, the completion status of the preceding task, and the occupancy status of the construction area.

[0069] If the construction area corresponding to a certain task unit is in an accessible state and the target required to perform the task is in an idle state, then the task sorting and allocation unit 201 includes the task unit in the set of allocable tasks.

[0070] If the prerequisite task of a certain task unit is not completed, or the required target is in a fault, occupied or waiting state, the task sorting and allocation unit 201 retains the task unit in the set of tasks to be triggered.

[0071] After the task sorting and allocation unit 201 completes the sorting and allocation, it outputs the task sorting results and allocation results to the global path timing generation unit 202.

[0072] The global path timing generation unit 202 generates the global path timing based on the task sorting results, allocation results, target current location, target running boundary, and construction area spatial boundary.

[0073] The global path timing includes elements and planned execution times. Elements may include at least one of the following: floor stop point, horizontal movement segment, elevation segment, work surface switching point, waiting point, and control trigger point.

[0074] The planned execution time includes at least one of the following: entry time, departure time, dwell time, start time of ascent / departure, end time of ascent / departure, and operation switching time.

[0075] For example, for the task of moving the first target from the construction area on the east side of the fifth floor to the central work area on the fifth floor, the global path timing generation unit 202 breaks down its path into the starting stop point, horizontal movement segment, floor transfer point, work surface switching point and target stop point, and writes the planned execution time for each element.

[0076] For the task of moving the second objective from the construction area on the south side of the fifth floor to the vicinity of the same central work area, the global path timing generation unit 202 also generates the corresponding elements and planned execution times.

[0077] If two targets pass through the same construction area in adjacent time periods, the rolling conflict resolution unit 203 will then make the judgment and handle the situation.

[0078] The rolling conflict resolution unit 203 receives the global path timing and combines it with the target state used for decision-making to make rolling predictions of future time windows.

[0079] Specifically, the rolling conflict resolution unit 203 reads the global path timing within several future scheduling cycles according to a preset rolling cycle, and determines whether there are any issues such as the same construction area being occupied, paths intersecting, planned execution times overlapping, or insufficient safety distance between different targets based on the elements, planned execution times, and target status in the global path timing.

[0080] In this embodiment, the rolling period can be set to 1s, 2s or 5s, and the future prediction time window can be set to 30s, 60s or 120s.

[0081] The rolling conflict resolution unit 203 updates the target's current position, target status, and global path timing in each rolling cycle, and reassesses whether there is a risk of conflict within the future prediction time window.

[0082] When the rolling conflict resolution unit 203 determines that the first target and the second target enter the same construction area at the same time of plan execution, or that the space occupied by the two targets overlaps, or that the distance between the two targets is less than the safe distance threshold, the rolling conflict resolution unit 203 generates a conflict identifier.

[0083] Conflict identifiers may include conflict target identifiers, conflict area identifiers, conflict time, conflict type, and associated task units.

[0084] Conflict types can include area occupancy conflicts, path intersection conflicts, time sequence overlap conflicts, and safety distance conflicts.

[0085] After generating conflict identifiers, the rolling conflict resolution unit 203 adjusts the global path timing based on the conflict identifiers. The adjustment methods include path fine-tuning, timing adjustment, and job surface switching.

[0086] Path fine-tuning is used to change the local path elements traversed by the target; timing adjustment is used to postpone or advance the planned execution time of a target entering the conflict zone.

[0087] Work area switching is used to switch the target to another work area when an alternative work area exists. If path fine-tuning can eliminate the conflict, path fine-tuning is performed first.

[0088] If conflicts still exist after path fine-tuning, timing adjustments will be performed; if timing adjustments still cannot meet scheduling requirements, job face switching or waiting control will be performed.

[0089] After the adjustment is completed, the rolling conflict resolution unit 203 outputs the adjusted global path timing and scheduling execution information.

[0090] The scheduling execution information may include target movement instructions, waiting instructions, deceleration instructions, pause instructions, work surface switching instructions, and resumption of execution instructions.

[0091] The scheduling execution information is converted into control data that can be recognized by the field control terminal via the edge gateway and sent to the control interface of the corresponding device.

[0092] The execution feedback correction layer 104 receives feedback data after the scheduling implementation.

[0093] Feedback data includes at least one of the following: actual start time, actual end time, actual location, execution status, conflict resolution result, device status change, and anomaly record. The execution feedback correction layer 104 associates the feedback data with the corresponding task sorting result, allocation result, global path timing, and conflict identifier.

[0094] When feedback data shows that a target has not arrived at the designated position according to the adjusted global path timing, or when the execution status is delayed, paused, abnormal or deviated, the execution feedback correction layer 104 generates parameter correction input and returns the parameter correction input to the collaborative path decision layer 101 so that subsequent scheduling cycles can correct the task sorting, path orchestration and conflict resolution.

[0095] To verify the feasibility of embedding the collaborative scheduling link 200 into the collaborative path decision layer 101 in this embodiment, this embodiment further sets up an experiment on multi-platform cross-operation conflict prediction and automatic resolution.

[0096] like Figure 2As shown, the experimental subjects selected were three construction equipment in the standard construction floor of a high-rise building as test targets, which were denoted as target A, target B and target C respectively. The conflict area was marked as D and the range was displayed in red.

[0097] The work areas of targets A, B, and C are enclosed in blue, while the transit areas of targets A, B, and C are enclosed in yellow.

[0098] Target A performs a horizontal material transfer task, Target B performs a vertical transfer task, and Target C performs a work surface switching task. The experimental area includes the eastern work area, the central transition area, and the southern work area, with the central transition area being an intersection area that multiple targets may pass through together.

[0099] The experimental data includes: equipment data, location data, status data, and operation progress data for targets A, B, and C; spatial boundaries of the construction area; safety distance thresholds between targets; task unit sets corresponding to each target; and the initial global path timing.

[0100] In the experiment, the safe distance threshold was set to 2.0m, the rolling cycle to 1s, and the future prediction time window to 60s. The position data of each target were written into the object index field set according to a unified spatial reference, and the twin state organization layer 102 formed the target, target state, and target association relationship.

[0101] During the experiment, the three treatment methods were compared and tested: The first method is manual scheduling, in which on-site management personnel manually adjust the order in which equipment enters the central conversion area based on equipment location and work tasks. The second method is static path scheduling, which generates the global path timing only based on the initial task order and does not perform rolling conflict resolution. The third type is the collaborative scheduling link 200 described in this embodiment, which is continuously processed by the task sorting and allocation unit 201, the global path timing generation unit 202 and the rolling conflict resolution unit 203, and generates the adjusted global path timing and scheduling execution information before the conflict occurs.

[0102] The experimental evaluation metrics include conflict prediction accuracy, conflict resolution success rate, average scheduling adjustment time, control information generation delay, and task completion time deviation.

[0103] The conflict prediction accuracy rate is the ratio of the number of conflicts that are correctly identified to the number of conflicts that are manually verified and confirmed.

[0104] The conflict resolution success rate is the ratio of the number of times no space occupancy conflicts or safe distance conflicts occurred after adjustment to the total number of conflict resolutions.

[0105] The average scheduling adjustment time is the time required from identifying the conflict flag to outputting the adjusted global path timing.

[0106] The control information generation delay is the time required from the generation of the adjusted global path timing to the output of the scheduling execution information.

[0107] Task completion time deviation is the difference between the actual completion time and the planned completion time.

[0108] The test results are shown in Table 1 below: Table 1 Scheduling Test Data Table

[0109] The above experiments show that although manual scheduling can handle some conflicts based on on-site experience, it is difficult to cover path intersections and area occupancy issues in future time windows in a timely manner.

[0110] Static path scheduling can generate initial paths and timing plans, but it lacks rolling updates for changes in the target state. When the target execution progress deviates, the timing of subsequent paths is prone to inconsistencies with the actual state.

[0111] This embodiment embeds a task sorting and allocation unit 201, a global path timing generation unit 202, and a rolling conflict resolution unit 203 into the collaborative path decision layer 101, enabling task allocation, path orchestration, and conflict resolution to be executed continuously in the same link. It can generate a conflict identifier before the target enters the conflict area and output the adjusted global path timing and scheduling execution information.

[0112] Specifically, during a test, both target A and target B were scheduled to enter the central transition zone at 38 seconds, and the horizontal movement segment of target A and the lifting transfer exit area of ​​target B had a spatial envelope overlap.

[0113] The rolling conflict resolution unit 203 identifies the conflict risk during the 10s rolling prediction and generates a conflict identifier.

[0114] Subsequently, the rolling conflict resolution unit 203 reads the task sorting results of the corresponding task units of target A and target B, determines that the priority of the lifting and transferring task of target B is higher than that of the horizontal material transfer task of target A, and thus postpones the planned execution time of target A entering the central conversion area by 12 seconds, while keeping the original path of target B unchanged.

[0115] After the adjustment, target A enters the central transition zone at 50 seconds, and target B leaves the central transition zone at 44 seconds. The two no longer conflict in the same construction area and overlapping time period.

[0116] The corresponding scheduling execution information includes the instruction for target A to wait, the instruction for target A to resume movement, and the instruction for target B to continue execution.

[0117] During another test, when Target C was performing a work area switching task, due to the temporary occupation of the south work area, the original planned path of Target C and the material temporary storage area had a space occupation conflict.

[0118] After generating the conflict identifier, the rolling conflict resolution unit 203 prioritizes path fine-tuning, switching the horizontal movement segment of target C from the south channel to the east channel, and regenerating the corresponding elements and planned execution time.

[0119] If the path elements and planned execution times after the path fine-tuning no longer conflict with other objectives, the adjusted global path timing is output; if conflicts still exist, the timing adjustment or job switching continues.

[0120] After the scheduling execution is completed, the execution feedback correction layer 104 receives feedback data from target A, target B and target C, and establishes association between the feedback data and the corresponding task sorting results, allocation results, global path timing and conflict identifiers.

[0121] For example, when the actual waiting time for target A is 4 seconds longer than the adjusted planned waiting time, the execution feedback correction layer 104 writes the deviation into the feedback data and forms the parameter correction input.

[0122] In the next scheduling cycle, the collaborative path decision layer 101 reads the parameter correction input and corrects the waiting time reservation for similar path elements.

[0123] Thus, the scheduling main architecture 100 can form a continuous processing process from device data access, twin state organization, collaborative path decision-making to execution feedback correction.

[0124] Through the above implementation method, this embodiment does not only perform independent path planning for multiple devices, but forms a hierarchical processing structure in the scheduling main architecture 100, consisting of a field perception access layer 103, a twin state organization layer 102, a collaborative path decision layer 101, and an execution feedback correction layer 104. Furthermore, through the collaborative scheduling link 200, a task sorting and allocation unit 201, a global path timing generation unit 202, and a rolling conflict resolution unit 203 are embedded in the collaborative path decision layer 101.

[0125] Thus, equipment data can be organized into targets, target states, and target relationships. Task allocation results can continue to participate in global path timing generation, global path timing can continue to participate in conflict rolling prediction, and feedback data after scheduling can be returned to subsequent scheduling cycles, thereby ensuring that the entire process from equipment data to scheduling implementation has a feasible data flow and control loop.

[0126] Example 2

[0127] like Figures 3-6 In one embodiment of the present invention, a connection method for the task unit organization unit 300 is provided, specifically as follows: A task unit organization unit 300 is set between the twin state organization layer 102 and the task sorting and allocation unit 201.

[0128] In this embodiment, the data output by the twin state organization layer 102 is not directly input into the task sorting and allocation unit 201, but is first input into the task unit organization unit 300.

[0129] The task unit organization unit 300 adopts a processing flow of target association graph construction, job task splitting, candidate target matching, entry condition determination, task sequence writing, task field verification, and task unit set output. It reads the targets formed by the twin state organization layer 102, generates task unit sets based on the association relationships between targets, and writes them into the task definition field set.

[0130] In practice, the twin state organization layer 102 outputs the target table, target state table and target association table to the task unit organization unit 300.

[0131] The target table should include at least the fields of target identifier, target type, floor, region, and spatial coordinates.

[0132] The target status table should include at least the following fields: target identifier, running status, occupied status, schedulable status, abnormal status, and update time.

[0133] The target association table should include at least the following fields: first target identifier, second target identifier, association type, association source, association effective time, association expiration time, and association status.

[0134] The association types include region affiliation, candidate execution, spatial adjacency, avoidance constraint, and task dependency.

[0135] After receiving the above three types of data tables, the task unit organization unit 300 first uses the primary key comparison method to complete the target validity check.

[0136] Specifically, the task unit organization unit 300 uses the target identifier in the target table as the primary key and queries the target status table one by one to see if there is a target identifier with the same identifier.

[0137] If the same target identifier does not exist in the target status table, the target will be added to the target set to be reviewed.

[0138] When there is a target with the same identifier in the target status table but the abnormal status field is fault, location missing or load over limit, the target is written to the unschedulable target set.

[0139] When the same target identifier exists in the target status table and the abnormal status field is empty and the schedulable status field is schedulable, the target is written into the callable target set.

[0140] Only targets from the callable target set participate in the generation of subsequent task units.

[0141] Task unit organization unit 300 then uses the adjacency list method to construct the target association graph.

[0142] Specifically, each target in the target table is treated as a graph node, each relationship in the target relationship table is treated as a graph edge, and edge attributes are written for the graph edges according to the relationship type.

[0143] For region affiliation relationships, the edge attribute is written with the region and spatial boundary number.

[0144] For candidate execution relationships, the edge attribute is written with the executable task type and the allowed action type.

[0145] For spatial adjacency relationships, the edge attribute is written with the adjacent region number and the passage boundary number.

[0146] For avoidance constraints, the edge attributes are written with the safety distance threshold and avoidance level.

[0147] For task dependencies, the edge attribute is written with the identifiers of the preceding and succeeding tasks.

[0148] Using the adjacency list mentioned above, the task unit organization unit 300 can quickly query the region, task, access area and dependent tasks associated with any target based on any target identifier.

[0149] In one specific implementation, the task unit organization unit 300 synchronously displays the target table, target status table, target association table, regional target table, and construction task plan table through the task unit organization matrix operation interface. It also displays the association matrix between targets, regions, and tasks, the task unit condition matrix, field validation results, and the valid task unit set output to the task sorting and allocation unit 201 on the same interface. Figures 4-5 As shown.

[0150] Figures 4-5 The associated matrix E1 is used to display the relationship between the target and the construction area and the work task, while the task unit condition matrix E2 is used to display the entry conditions corresponding to the area occupancy, target status, preceding tasks, safety distance and time window.

[0151] After the target association graph is constructed, the task unit organization unit 300 reads the construction task plan table.

[0152] The construction task plan should include at least the following fields: task identifier, floor number, construction area identifier, work surface identifier, task type, required target type, planned start time, planned end time, prerequisite task identifier, and work action field.

[0153] Task unit organization unit 300 adopts a four-level splitting rule of floor, construction area, work surface, and execution action to divide a work task into multiple task units.

[0154] Specifically, when a certain task includes four actions at the same time: entering the construction area, moving horizontally, performing the task, and exiting the construction area, the task unit organization unit 300 generates an entry task unit, a movement task unit, a work task unit, and an exit task unit, and generates a unique task unit identifier for each task unit.

[0155] The task unit identifier is generated using a fixed field concatenation method, with the format: Project Number + Floor Number + Construction Area Number + Action Type Number + Serial Number.

[0156] For example, if the project number is P01, the floor number is F05, the construction area number is R02, the action type is "entry action," and the sequence number is 001, the generated task unit identifier will be P01-F05-R02-IN-001. The task unit organization unit 300 will write the generated task unit identifier into the task unit identifier field of the task definition field set.

[0157] For each task unit, the task unit organization unit 300 uses a target candidate matching method to determine the target candidate identifier.

[0158] Specifically, the required target types are first read from the construction task plan table, and then targets with the same target types are selected from the set of callable targets.

[0159] Then, the target association graph is read to determine whether the target has a regional affiliation relationship, spatial adjacency relationship or candidate execution relationship with the construction area corresponding to the task unit.

[0160] When the target type is consistent, the target status is schedulable, and there is a valid association between the target and the task area, the target is written into the target candidate identifier field.

[0161] When there are multiple candidate targets, they are sorted in ascending order according to the path distance from the current location of the target to the task area entrance, and the sorted target identifiers are written into the target candidate identifier field in sequence.

[0162] When no candidate target exists, the generation status of the task unit is written as pending completion.

[0163] Task unit organization unit 300 uses a rule table matching method to write entry conditions.

[0164] The rule table includes area occupancy rules, target status rules, prerequisite task rules, safety distance rules, and time window rules.

[0165] The area occupancy rule is used to determine whether the construction area corresponding to the task unit is in a free, shared, occupied or closed state.

[0166] The target state rule is used to determine whether a candidate target is in a schedulable state.

[0167] Prerequisite task rules are used to determine whether a prerequisite task has been completed.

[0168] The safe distance rule is used to determine whether the distance between a candidate target and the location of personnel, materials, or other targets is greater than the safe distance threshold.

[0169] Time window rules are used to determine whether the current time or the planned start time falls within the allowed job time window.

[0170] When the area occupancy rule, target status rule, prerequisite task rule, safety distance rule, and time window rule are all satisfied, the task unit organization unit 300 will write "entry allowed" into the entry condition field.

[0171] When the area occupancy rule is not met but other rules are met, the condition field will be written to wait for the occupancy to be released.

[0172] If the rules of the preceding task are not met, the entry condition field will be written to wait for the preceding task to complete.

[0173] When the target state rules are not met, the entry condition field will be written to wait for the target to become schedulable again.

[0174] When the safe distance rule is not met, the entry condition field will be written to wait for the avoidance condition to be met.

[0175] When the time window rules are not met, the entry condition field will be written to the waiting job time window.

[0176] The task unit organization unit 300 uses a directed acyclic graph sorting method to write the preceding task identifier and the succeeding task identifier.

[0177] Specifically, the work tasks and the subdivided task units in the construction task schedule are used as nodes in a directed graph, and the relationships between preceding tasks are used as directed edges.

[0178] The task unit organization unit 300 first establishes sequential edges for the entry task unit, movement task unit, operation task unit and exit task unit within the same operation task, and then establishes cross-task edges for the prerequisite relationships between different operation tasks.

[0179] Then, a topological sorting method is used to check for the existence of closed-loop dependencies.

[0180] If topological sorting can be completed, then write the predecessor node of each task unit into the predecessor task identifier field and the successor node into the successor task identifier field.

[0181] If topological sorting fails, the relevant task units will be written into the abnormal task set, and a closed-loop dependency warning will be recorded in the task organization log.

[0182] The task unit organization unit 300 also writes an execution constraint field for each task unit. The execution constraint field includes allowed target types, allowed access areas, safe distance threshold, maximum waiting time, whether path fine-tuning is allowed, whether timing adjustment is allowed, and whether task waiting is allowed.

[0183] The specific writing method is as follows: the target type is allowed to be taken from the required target type in the construction task plan table; the passable area is allowed to be taken from the spatial adjacency relationship in the target association graph.

[0184] The safe distance threshold is taken from the project safety parameter table; the maximum waiting time is taken from the construction organization parameter table. Whether path fine-tuning, timing adjustment, and task waiting are allowed are jointly determined by the task type and area status.

[0185] For example, material transport tasks allow for path fine-tuning and timing adjustments, lifting and transfer tasks allow for timing adjustments but do not allow for arbitrary switching of passage areas, and safety avoidance tasks allow tasks to wait and prioritize writing safety locking constraints.

[0186] After the task definition field set is written, the task unit organization unit 300 verifies the task unit set using field integrity verification, association validity verification, and sequence relationship verification.

[0187] The field integrity check uses a non-empty field check method to determine whether the task unit identifier, target candidate identifier, construction area identifier, action type, entry condition, planned start time, and planned end time are empty.

[0188] The association validity check uses the primary key lookup method to determine whether the target candidate identifier exists in the target table, whether the construction area identifier exists in the area target table, and whether the predecessor task identifier and successor task identifier exist in the task unit set.

[0189] The sequence relationship verification uses a topological sorting result judgment method to determine whether there are closed-loop dependencies or broken dependencies between task units.

[0190] The verified task units are written into the valid task unit set.

[0191] Tasks that fail validation are written to the abnormal task set, along with the exception reason field.

[0192] The task unit organization unit 300 outputs the set of valid task units to the task sorting and allocation unit 201. After receiving the set of valid task units, the task sorting and allocation unit 201 directly reads the target candidate identifier, entry condition, predecessor task identifier, successor task identifier and execution constraint fields from the task definition field set, without repeatedly parsing the target, target status and target association.

[0193] Therefore, the task sorting and allocation unit 201 can use the effective task unit set as standardized input to perform subsequent task sorting and allocation.

[0194] This embodiment also includes a dynamic update method. Specifically, when the twin state organization layer 102 updates the target state or target association in a new scheduling cycle, the task unit organization unit 300 queries the target association graph through the target identifier, locates the task units that have an association edge with the changed target, and only re-executes target candidate matching, entry condition writing, and field validation for these task units.

[0195] If the target changes from schedulable to faulty, the task unit organization unit 300 will move the task unit referencing the target from the set of valid task units to the set of tasks to be reviewed.

[0196] If a region changes from occupied to free, the task unit organization unit 300 will re-evaluate the task units whose entry conditions are waiting for the occupancy to be released, and rewrite them as enterable after the conditions are met.

[0197] This dynamic update method avoids regenerating all task units in each scheduling cycle, thus improving the efficiency of task unit organization.

[0198] To verify the feasibility of this embodiment, an experiment was set up to verify the accuracy of task unit organization and the effectiveness of scheduling input.

[0199] Based on the technical link selection in the patent analysis report regarding multi-source heterogeneous data access and digital twin mapping for multi-machine collaboration on construction platforms, this experiment focuses on verifying whether the task unit organization unit 300 can convert the target and target association relationship formed by the twin state organization layer 102 into a task unit set that can be directly read by the task sorting and allocation unit 201.

[0200] likeFigure 6 As shown, this embodiment uses the task unit organization function curve verification interface to compare and verify the manual task organization scheme, the no-task-unit organization scheme, and the scheme of this embodiment.

[0201] The task unit organization function curve verification interface includes: a curve of effective task unit quantity versus dynamically updated task quantity (F1), a bar chart comparing key indicators (F2), a curve of task sorting input time versus convergence (F3), and a chart comparing the order relationship error rate (F4). Through this interface, the changes in task unit generation completeness rate, target matching accuracy rate, entry condition judgment accuracy rate, order relationship error rate, and task sorting input time can be observed simultaneously.

[0202] The experiment was completed using a BIM construction space model, target state playback data, construction task plan, target association diagram generation program, and task unit organization verification program.

[0203] The BIM construction space model is used to export floor boundaries, construction area boundaries, and work surface boundaries.

[0204] The target status replay data is used to provide the location status, running status, occupied status, schedulable status, and abnormal status of target A, target B, and target C in different scheduling cycles.

[0205] The construction task schedule is used to provide work tasks, planned time, and prerequisite relationships.

[0206] The target association graph generation program is used to generate target association graphs using the adjacency list method.

[0207] The task unit organization verification procedure is used to generate a valid task unit set using four-level splitting rules, rule table matching method, topology sorting method, and field validation method.

[0208] The experimental subjects included three types of equipment targets and three regional targets.

[0209] The three types of equipment targets are Target A, Target B, and Target C, where Target A corresponds to a building construction machine, Target B corresponds to a concrete placing boom, and Target C corresponds to a construction hoist.

[0210] The three target areas are the eastern work area, the central transfer area, and the southern work area.

[0211] The experiment included twelve tasks, including horizontal transport, lifting and transferring, fabric placement, work surface switching, and area exit.

[0212] Each task includes the planned start time, planned end time, required target type, construction area, and prerequisite task information.

[0213] The experimental steps are as follows.

[0214] First, export the area target table from the BIM construction space model. The area target table includes area target identifiers, floor numbers, area boundaries, area occupancy status, and area open status.

[0215] Import the status replay data of target A, target B and target C into the target status table. The target status table includes target identifier, spatial coordinates, running status, occupied status, schedulable status, abnormal status and update time.

[0216] Secondly, the target association graph generation program reads the target table, the regional target table, and the target status table. It uses a spatial inclusion judgment method to establish the regional affiliation relationship between the target and the region, a distance threshold judgment method to establish the spatial adjacency relationship between the target and the adjacent region, and a task requirement field matching method to establish the candidate execution relationship between the target and the task. Finally, it generates the target association graph.

[0217] Next, the task unit organization verification procedure reads the construction task plan and, according to the four-level splitting rule of floor, construction area, work surface, and execution action, breaks down the twelve work tasks into task units.

[0218] For each task unit after splitting, the task unit organization verification program reads the target association graph, writes the target candidate identifier using the target candidate matching method, writes the entry condition using the rule table matching method, writes the predecessor task identifier and successor task identifier using the topology sorting method, and writes the execution constraint field using the project safety parameter table and construction organization parameter table.

[0219] Then, the task unit organization verification procedure uses field integrity verification, association validity verification, and sequence relationship verification to filter the task definition field set and obtain a valid task unit set.

[0220] For task units with missing fields, invalid targets, closed regions, or closed-loop dependencies, the task unit organization verification procedure writes them into the abnormal task set and records the reason for the abnormality.

[0221] Finally, the set of valid task units is input into the task sorting and allocation unit 201. The task sorting and allocation unit 201 checks whether it can directly read the target candidate identifier, entry condition, predecessor task identifier, successor task identifier and execution constraint field, and outputs the task sorting result and allocation result.

[0222] Three comparison schemes were set up in the experiment.

[0223] The first group is a manual task organization scheme, in which task units are manually generated by humans based on the construction task plan and target status.

[0224] The second group is a task-less organization unit scheme, in which the target, target status and target association formed by the twin state organization layer 102 are directly input into the task sorting and allocation unit 201, and the task sorting and allocation unit 201 temporarily parses the task units.

[0225] The third group is the solution of this embodiment, that is, a task unit organization unit 300 is set between the twin state organization layer 102 and the task sorting and allocation unit 201. The task unit organization unit 300 generates a valid task unit set by using the target association graph, four-level splitting rules, rule table matching and field verification methods.

[0226] The experimental evaluation metrics include the completeness rate of task unit generation, the accuracy rate of target matching, the accuracy rate of entry condition judgment, the error rate of sequence relationship, and the time consumed by task sorting input.

[0227] The task unit generation completeness rate is the ratio of the number of task units that are successfully generated with complete fields to the number of task units that should be generated.

[0228] The target matching accuracy is the ratio of the number of task units with correctly identified target candidate identifiers to the number of task units that have been generated.

[0229] The accuracy rate of entry condition judgment is the ratio of the number of task units whose entry conditions match the results of manual review to the number of task units that have been generated.

[0230] The sequence relationship error rate is the ratio of the number of task units with errors in the preceding and succeeding task identifiers to the number of generated task units.

[0231] The task sorting input time is the time required for the task sorting allocation unit 201 to start sorting calculation from receiving input data.

[0232] The test results are shown in Table 2 below: Table 2 Test Data Table of Organization and Reorganization Plan

[0233] In one specific test, the construction task schedule included a material transfer task for the central transfer zone.

[0234] The task requires that after target B completes the fabric preparation, target A enters the central transfer area to perform horizontal transport operations.

[0235] The target association relationships output by the twin state organization layer 102 show that target A has a regional affiliation relationship with the eastern work area, target A has a spatial adjacency relationship with the central conversion area, and target B has a current occupancy relationship with the central conversion area.

[0236] After reading the above target association, the task unit organization unit 300 divides the task into three task units according to the four-level splitting rule: target A enters the central conversion area, target A performs horizontal transportation, and target A exits the central conversion area. The task unit organization unit 300 writes each task unit into the task definition field set.

[0237] Since target B is currently occupying the central conversion area, task unit organization unit 300 writes the entry condition for target A to enter the central conversion area as waiting for the occupancy to be released, and writes the task unit corresponding to target B into the pre-task identifier field.

[0238] When target B leaves the central conversion zone in a subsequent scheduling cycle, task unit organization unit 300 re-evaluates the entry condition through a dynamic update method and rewrites it as enterable.

[0239] In another specific test, when target C performs the work surface switching task, the south work area is temporarily written to an occupied state.

[0240] After reading the regional affiliation relationship between target C and the south work area, the spatial adjacency relationship between target C and the east work area, and the occupancy status of the south work area, the task unit organization unit 300 marks the original planned task unit as pending review and generates candidate task units switched through the east work area.

[0241] The execution constraint field of this candidate task unit contains allowed path fine-tuning, allowed timing adjustment, and maximum waiting time.

[0242] Subsequently, the task unit is input into the task sorting and allocation unit 201 for subsequent sorting, allocation, and path generation.

[0243] Through the above implementation, the task unit organization unit 300 is not a simple data forwarding node, but uses specific methods such as primary key comparison, adjacency list graph construction, four-level task splitting, target candidate matching, rule table matching, topology sorting, field integrity verification, association validity verification, and dynamic local update to convert the target and target association relationship formed by the twin state organization layer 102 into a task unit set that the task sorting and allocation unit 201 can directly read.

[0244] Example 3

[0245] like Figures 7-9 As an embodiment of the present invention, an optimized technical solution for embodiment 2 is also provided, specifically: a time-space occupancy organization unit 400 is set between the global path timing generation unit 202 and the rolling conflict resolution unit 203.

[0246] The spatiotemporal occupancy organization unit 400 reads the global path timing output by the global path timing generation unit 202 and converts the elements and planned execution times in the global path timing into spatiotemporal occupancy windows.

[0247] The spatiotemporal occupancy window establishes a correspondence with the target in the twin state organization layer 102, enabling the rolling conflict resolution unit 203 to perform conflict retrieval and resolution processing based on the spatiotemporal occupancy window corresponding to the target.

[0248] In this embodiment, the data output by the global path timing generation unit 202 is not directly input to the rolling conflict resolution unit 203, but is first input to the space-time occupancy organization unit 400.

[0249] The spatiotemporal occupancy organization unit 400 converts the "element + planned execution time" in the global path sequence into a spatiotemporal occupancy window that includes spatial occupancy range, time occupancy interval, and target status reference. This enables the rolling conflict resolution unit 203 to directly determine whether there are regional occupancy conflicts, path intersection conflicts, or safety distance conflicts between different targets based on the spatiotemporal occupancy window.

[0250] In practice, the global path timing generation unit 202 outputs three types of data tables to the spatiotemporal occupancy organization unit 400: the global path timing table, the path element table, and the target path association table.

[0251] The global path sequence table should include at least the target identifier, task unit identifier, element identifier, element type, planned entry time, planned exit time, planned stay duration, planned action type, and sequence version number.

[0252] The path element table should include at least the element identifier, element type, element centerline, element boundary, floor, construction area, direction of travel, permitted target type, and spatial height range.

[0253] The target path association table should include at least the target identifier, task unit identifier, path sequence number, element identifier, path generation time, path source, and path status.

[0254] After receiving the above three types of data tables, the spacetime occupancy organization unit 400 first uses the composite primary key check method to verify the data validity.

[0255] Specifically, using the target identifier, task unit identifier, element identifier, and time sequence version number as a combined primary key, we check each entry in the global path time sequence table to see if the planned entry time and planned exit time are empty, and whether the planned entry time is earlier than the planned exit time.

[0256] If the planned entry time is empty, the record is written to the missing start time set; if the planned exit time is empty, the next element of the planned entry time of the same target and the same task unit is read as the temporary exit time, and a temporary completion mark is written to the record.

[0257] If the planned entry time is later than the planned exit time, the record will be written to the abnormal time series set, and the spatiotemporal occupancy window will not be generated for the time being.

[0258] The spatiotemporal occupancy organization unit 400 also uses the element identifier as the primary key to query the path element table.

[0259] If the corresponding element identifier does not exist in the path element table, then the record is written to the element missing set.

[0260] If a corresponding element identifier exists but the element boundary is empty, a temporary boundary is generated based on the element's centerline and the target's external dimensions, and a temporary boundary marker is written to it.

[0261] If both the element's center line and boundary are empty, the record will not be included in the spatiotemporal occupancy window generation process.

[0262] After completing the data validity verification, the spatiotemporal occupancy organization unit 400 uses the element type branch parsing method to determine the space occupancy range.

[0263] The element types include horizontal movement segments, waiting points, lifting segments, work surface switching points, path nodes, and control trigger points.

[0264] For the horizontal movement segment, the spatiotemporal occupancy organization unit 400 reads the center line of the element in the path element table and reads the target length, target width and target running direction recorded in the target table in the twin state organization layer 102.

[0265] Then, a linear sweep envelope is generated by buffering along both sides of the element's centerline using half the target width plus a safety distance threshold plus a positioning error margin.

[0266] If the target is a building machine, the safety distance threshold is set to 2m, and the positioning error margin is determined based on the average UWB positioning error over the last 30 seconds; if the average positioning error is greater than 0.3m, the positioning error margin is increased to 0.5m.

[0267] For the waiting point, the spatiotemporal occupancy organization unit 400 reads the coordinates of the waiting point and the boundary of the waiting area. If the waiting area boundary exists, it is used as the basic occupancy range and expanded outward according to the safety distance threshold.

[0268] If the waiting area boundary does not exist, a rectangular or circular waiting area is generated centered on the waiting point coordinates, using the target's dimensions and a safe distance threshold.

[0269] For the lifting segment, the space-time occupancy organization unit 400 reads the coordinates of the lifting start point, the lifting end point, the lifting channel boundary, and the height range.

[0270] If the target is a construction hoist, a three-dimensional envelope is generated with the boundary of the hoisting channel as the bottom surface and the planned hoisting height range as the height.

[0271] If the lifting section corresponds to a temporary lifting channel, then add a construction safety margin outside the bottom envelope and write the temporary channel mark.

[0272] For the work surface switching point, the spatiotemporal occupancy organization unit 400 reads the work surface boundary before switching, the work surface boundary after switching, and the intermediate transition area boundary, and then sets the union of the three to form the work surface switching occupancy range.

[0273] If the intermediate transition region is occupied in the twin state organization layer 102, the spatiotemporal occupancy window is written into the pending review state and passed to the rolling conflict resolution unit 203 for priority judgment.

[0274] For a control trigger point, the spatiotemporal occupancy organization unit 400 determines whether the control trigger point corresponds to an actual spatial dwelling action.

[0275] If the control trigger point is only used to send PLC start instructions or speed setting instructions, then a spatial envelope is not generated separately, and only the control trigger time is written.

[0276] If the control trigger point corresponds to a wait instruction, pause instruction, or security lock instruction, then a time-space occupancy window is generated according to the wait point method.

[0277] After determining the space occupancy range, the spatiotemporal occupancy organization unit 400 uses the time interval expansion method to generate the occupancy time interval.

[0278] Specifically, the plan's entry and exit times are read, and the time extension is determined based on the scheduling cycle, positioning sampling cycle, instruction issuance delay, and historical execution deviation.

[0279] In this embodiment, the scheduling period is 1s, the UWB positioning sampling period is 200ms, the edge gateway command issuance delay fluctuates between 80ms and 250ms, and the historical execution deviation is the average difference between the actual departure time and the planned departure time of the most recent 20 similar elements.

[0280] If the historical execution deviation is less than 1s, the time extension is 1s; if the historical execution deviation is between 1s and 3s, the time extension is 2s; if the historical execution deviation is greater than 3s, the time extension is 3s, and this element is written into the high deviation element set.

[0281] The spatiotemporal occupancy organization unit 400 extends the planned entry time forward by the time extension amount to obtain the occupancy start time.

[0282] The planned departure time is extended backward by the aforementioned time extension to obtain the occupancy end time. For the lifting section of the construction hoist, if the load value exceeds 80% of the rated load, an additional 1 second of end time extension is added.

[0283] For the switching point of the concrete placing boom, if the boom rotation angle exceeds 60°, an additional 2 seconds of end time extension is added. In this way, the time-space occupancy window can cover common delays in on-site execution, rather than just using the ideal planned time.

[0284] Subsequently, the spatiotemporal occupancy organization unit 400 combines the spatial occupancy range and the occupancy time interval into a spatiotemporal occupancy window. Each spatiotemporal occupancy window is written into the spatiotemporal occupancy window field set.

[0285] The spatiotemporal occupancy window field set includes occupancy window identifier, target identifier, task unit identifier, element identifier, element type, spatial envelope, floor identifier, construction area identifier, occupancy start time, occupancy end time, occupancy status, time sequence version number, safety distance threshold, positioning error margin, target status reference, and data trust mark.

[0286] In one specific implementation, the spatiotemporal occupancy organization unit 400 expands the target identifier, element identifier, element type, construction area, planned entry time, and planned departure time in the global path time sequence table in a time axis manner, and forms an extended time window according to the time interval expansion method.

[0287] The generated results are categorized into valid windows, temporary completions, pending verification, and invalid windows, and the overlapping intervals between different targets are displayed on the timeline. Figure 8 As shown.

[0288] Figure 8 In the process, the element E-MZ-03 corresponding to target A and the element E-MZ-02 corresponding to target B form an overlapping interval between 37s and 45s. The old version of target C's ascending and descending segment is marked as the failure window, and the new version's horizontal moving segment is marked as the effective window.

[0289] The occupancy window identifier is generated using a fixed field concatenation method, with the format being target identifier + task unit identifier + element identifier + time sequence version number.

[0290] For example, when target A executes task unit TU-A-001 and passes through element E-MZ-03, and the current global path timing version is V02, the occupied window identifier is generated as A-TU-A-001-E-MZ-03-V02.

[0291] If a new time-series version of the same target, task unit, and element appears in subsequent path replanning, the spatiotemporal occupancy organization unit 400 retains the historical record of the old version's occupancy window, writes the latest version into the valid state, and writes the old version into the invalid state.

[0292] The spatiotemporal occupancy organization unit 400 further establishes a correspondence with the target in the twin state organization layer 102.

[0293] Specifically, the spatiotemporal occupancy organization unit 400 queries the target table and target state table in the twin state organization layer 102 using the target identifier as the primary key.

[0294] If a corresponding target exists in the target table and the abnormal status in the target status table is empty, then the target status reference field is written to the address of the corresponding target status record.

[0295] If the target exists but the target status table update time is more than 3 seconds later than the current scheduling cycle, the data trust mark is written to low trust, and the spatiotemporal occupancy window is output to the rolling conflict resolution unit 203 with a "state lag" mark.

[0296] If the target does not exist, the spatiotemporal occupancy window will be written into the target unmatched set, will not participate in automatic conflict resolution, and will only output a manual review prompt.

[0297] To ensure that the rolling conflict resolution unit 203 can quickly query relevant occupancy relationships within the actual scheduling cycle, the spatiotemporal occupancy organization unit 400 adopts a combined indexing method of time bucket index + construction area index + spatial envelope index to manage the spatiotemporal occupancy window.

[0298] Specifically, based on a 1-second scheduling cycle as the basic time granularity, the occupancy time interval of each spatiotemporal occupancy window is divided into multiple time buckets.

[0299] The construction area identifier is used as the area index.

[0300] The spatial index is the bounding rectangle or the three-dimensional bounding box of the spatial envelope.

[0301] In each rolling cycle, the rolling conflict resolution unit 203 uses the future prediction time window, construction area identifier, and target identifier as query conditions, and the spatiotemporal occupancy organization unit 400 returns a set of candidate spatiotemporal occupancy windows in the same area, adjacent areas, and within the safe distance range.

[0302] In one specific implementation, the spatiotemporal occupancy organization unit 400 uses a PostgreSQL / PostGIS database to store the spatiotemporal occupancy window.

[0303] The spatial envelope of the planar movement segment is stored in a Polygon field, the path centerline is stored in a LineString field, and the rise and fall segments are stored in a bottom Polygon plus a height range field.

[0304] The time interval is stored using two fields: the start time of occupation and the end time of occupation. The database establishes indexes for the target identifier, construction area identifier, time bucket, and spatial envelope. In cases where unstable network conditions at the construction site prevent timely writing to the database from the edge side, the spatiotemporal occupancy organization unit 400 first caches the spatiotemporal occupancy window data for the most recent 120 seconds locally on the edge gateway.

[0305] Once the network is restored, the cached data will be synchronized to the database according to the time sequence version number.

[0306] This embodiment also includes a window version update method.

[0307] Specifically, when the global path timing generation unit 202 outputs a new global path timing due to path fine-tuning, timing adjustment or job surface switching, the time-space occupancy organization unit 400 reads the new timing version number and queries the old version time-space occupancy window under the same target and the same task unit.

[0308] If the new version no longer contains the element identifiers of the old version, then the spacetime occupancy window of the old version will be updated to an invalid state.

[0309] If the element identifier still exists but the planned execution time changes, the start and end times of occupation are recalculated; if the element's spatial range changes, the spatial envelope is regenerated.

[0310] If the target status changes, for example, from running to waiting, the occupancy status will be updated from planned occupancy to waiting occupancy.

[0311] Through this version update method, the rolling conflict resolution unit 203 always reads the effective spatiotemporal occupancy window that is consistent with the current global path timing.

[0312] To verify the feasibility of this embodiment, an experiment on spatiotemporal occupancy window conversion and retrieval stability was conducted.

[0313] This experiment does not use the integrated operation platform experiment in Example 1, nor the task unit organization accuracy experiment in Example 2. Instead, it verifies the core processing process of converting elements in the global path time sequence and planned execution times into spatiotemporal occupancy windows and establishing a correspondence with the target.

[0314] This experiment selects the technical links corresponding to conflict rolling prediction and dynamic resolution in the patent analysis report, as well as conflict rolling prediction and active resolution based on the spatiotemporal cube model, as the experimental basis.

[0315] The report also points out that this case needs to strengthen the coupling relationship between future time window conflict prediction and PLC command issuance, BIM coordinate mapping and construction platform constraints. Therefore, this experiment chooses BIM boundary, path timing, PostGIS index and on-site disturbance data for joint verification, rather than simply using ideal simulation data.

[0316] The experiment was completed using BIM spatial boundary data, global path time series samples, UWB positioning playback data, PostGIS spatiotemporal index library, Python geometry processing program and QGIS time series visualization interface.

[0317] BIM spatial boundary data is used to provide construction area boundaries, passageway boundaries, floor heights, and element spatial geometry.

[0318] Global path time series samples are used to provide the target, elements, and planned execution time; UWB positioning playback data are used to provide the target's actual position drift and sampling interval fluctuations.

[0319] Python geometry processing programs are used to generate spatial envelopes based on element types and perform time interval expansion; the PostGIS spatiotemporal index library is used to store and retrieve spatiotemporal occupancy windows.

[0320] The QGIS time-series visualization interface is used to replay the correspondence between spatiotemporal occupancy windows and targets, and to manually review the window conversion results.

[0321] The experimental subjects included three types of targets and five types of path elements.

[0322] The three types of targets are Target A, Target B, and Target C. Target A corresponds to a building construction machine, Target B corresponds to a concrete placing machine, and Target C corresponds to a construction hoist.

[0323] The five types of path elements include horizontal movement segments, waiting points, elevation segments, work surface switching points, and control trigger points.

[0324] The experiment selected 168 global path time series records, including 132 normal records, 11 records with missing planned times, 9 records of temporary changes to path elements, 10 records of location data drift, and 6 records of delayed update of regional status.

[0325] The above-mentioned anomaly records are from common UWB short-term point loss, edge gateway latency, temporary occupation of construction areas, and actual equipment execution lag in the simulated field, making the experimental process closer to the real construction scenario.

[0326] The experimental steps are as follows.

[0327] First, import the BIM spatial boundary data into the PostGIS spatiotemporal index library to form a construction area boundary table and a path element table.

[0328] The construction area boundary table includes fields for construction area identifier, floor identifier, area boundary, area open status, and area capacity.

[0329] The path element table includes fields for element identifier, element type, element centerline or element boundary, construction area, permitted target type, travel direction, and height range.

[0330] Secondly, the 168 global path time series records are imported into the Python geometry processing program. The Python geometry processing program first performs union primary key checks, planned time checks, and element identifier checks.

[0331] For records where the planned departure time is missing but the planned entry time of the next element exists, the program uses the planned entry time of the next element as the temporary departure time and writes a temporary completion flag.

[0332] For records where element boundaries are missing but element centerlines exist, the program uses a centerline buffering method to generate temporary spatial boundaries.

[0333] For records where the target identifier cannot be matched, the program outputs them as items for manual review and does not participate in automatic conflict resolution.

[0334] Next, the Python geometry processing program generates the spatial envelope according to the element type.

[0335] The horizontal moving segment uses a centerline buffering method to generate a linear sweep envelope.

[0336] The waiting point uses a point buffering method to generate the waiting occupancy envelope.

[0337] The vertical occupancy envelope of the rising and falling sections is generated by adding the bottom projection and the height range.

[0338] The transition occupancy envelope is generated by using the union of the boundaries of the regions before and after the switch at the work surface switching point.

[0339] The control trigger point determines whether to generate a spatial envelope based on whether it corresponds to a pause action.

[0340] Subsequently, the program generates the occupied time interval based on the scheduling cycle, positioning sampling cycle, instruction issuance delay, and historical execution deviation.

[0341] Then, the spatiotemporal occupancy organization unit 400 writes the generated spatiotemporal occupancy window into the PostGIS spatiotemporal index library and establishes time bucket index, construction area index and spatial envelope index.

[0342] The experiment simultaneously recorded the transformation results of each path's time sequence record, including five states: successful generation, temporary completion, pending verification, failure update, and abnormal discard.

[0343] For temporarily supplemented spatiotemporal occupancy windows, the system displays them as dashed boundaries in the QGIS time-series visualization interface.

[0344] For spatiotemporal occupancy windows awaiting review, a yellow indicator is displayed. For valid spatiotemporal occupancy windows, a solid line boundary is displayed.

[0345] Finally, the rolling conflict resolution unit 203 uses the future 60s time window as the query condition and sends a retrieval request to the spatiotemporal occupancy organization unit 400 every 1s to retrieve the candidate spatiotemporal occupancy window set in the same region and adjacent regions.

[0346] The QGIS time-series visualization interface synchronously replays the occupancy range of targets A, B, and C in different time windows, and allows manual verification of whether the candidate spatiotemporal occupancy windows are consistent with the target, path elements, and planned execution time.

[0347] In this embodiment, the spatiotemporal occupancy organization unit 400 also constructs the spatiotemporal occupancy window into a spatiotemporal occupancy voxel cube according to the construction area coordinates and time coordinates. In the spatiotemporal occupancy voxel cube, the X / Y direction represents the spatial location of the construction area, and the T direction represents the planned execution time or extended occupancy time.

[0348] Different targets have spatiotemporal occupancy windows displayed in different colors, while failed occupancy windows and pending review occupancy windows are displayed with distinguishing status markers.

[0349] When the spatiotemporal occupancy windows corresponding to two targets overlap simultaneously in both spatial range and temporal interval, the spatiotemporal occupancy organization unit 400 outputs the overlapping interval as a set of candidate conflict occupancy windows and passes it to the rolling conflict resolution unit 203, such as... Figure 9 As shown.

[0350] Three comparison schemes were set up in the experiment.

[0351] The first group is a direct path segment judgment scheme, which judges conflict candidates only based on the path centerline in the global path time series and the planned execution time, without generating the target space envelope.

[0352] The second group is a two-dimensional area occupancy scheme, which only determines whether a target has entered the same area based on the boundary of the construction area, without distinguishing the target size, positioning error, or element type.

[0353] The third group is the scheme of this embodiment, which is to convert elements and planned execution times into spatiotemporal occupancy windows with spatial envelope, time extension, target state reference and data trust mark through the spatiotemporal occupancy organization unit 400, and establish a combined spatiotemporal index.

[0354] The experimental evaluation metrics include the completeness rate of occupancy window conversion, the accuracy of target correspondence, the consistency rate of spatial envelope verification, the processability rate of abnormal records, the recall rate of conflicting candidates, the false alarm rate, and the time spent searching the next 60 seconds of the window.

[0355] The occupancy window conversion integrity rate is the ratio of the number of records that successfully generate valid or temporary completed spatiotemporal occupancy windows to the total number of global path time-series records.

[0356] The target correspondence accuracy is the ratio of the number of spatiotemporal occupancy windows that are correctly associated with targets in twin state organization layer 102 to the total number of spatiotemporal occupancy windows.

[0357] The spatial envelope verification consistency rate is the ratio of the number of spatiotemporal occupancy windows that, after manual verification, are consistent with the BIM boundary, target size, and safety distance requirements to the number of sampled verification windows.

[0358] The abnormal record processing rate is the ratio of the number of records that are correctly completed, marked, or removed from records such as missing planned time, location drift, temporary element changes, and delayed update of regional status to the total number of abnormal records.

[0359] The conflict candidate recall rate is the ratio of the number of real conflict candidates retrieved to the number of real conflicts verified by manual review.

[0360] The false alarm rate is the ratio of the number of conflict candidates output by the system that, after manual review, do not constitute a real conflict to the total number of conflict candidates output by the system.

[0361] The time taken for window retrieval in the next 60 seconds is the time required for the scrolling conflict resolution unit 203 to query the relevant spatiotemporal occupied window set in the next 60 seconds.

[0362] The test results are shown in Table 3 below: Table 3. Efficiency and Time Consumption Data for Each Scheme

[0363] The test results above are not perfect scores under ideal conditions, but rather obtained under conditions of missing planned moments, location drift, temporary changes to elements, and delayed updates to regional states.

[0364] This demonstrates that the spatiotemporal occupancy organization unit 400 can maintain a high conversion integrity rate and retrieval efficiency even under engineering disturbance conditions. At the same time, it avoids directly inputting abnormal data into the automatic resolution process through data trust marking and pending verification status.

[0365] In a specific test, the global path timing generation unit 202 output that target A passes through the middle transition zone element E-MZ-03 from 38s to 50s, and target B passes through the middle transition zone element E-MZ-02 from 32s to 44s.

[0366] After reading the two records, the spatiotemporal occupancy organization unit 400 generates spatial envelopes based on the external dimensions, safety distance threshold, and UWB positioning error margin of target A and target B, respectively, and extends the planned execution time of the two targets to the 37th to 51st and the 31st to 45th, respectively.

[0367] The query revealed that the two spatiotemporal occupancy windows overlapped between the 37th and 45th seconds, and their spatial envelopes both fell within the central transition zone and overlapped.

[0368] The spatiotemporal occupancy organization unit 400 establishes a correspondence between these two sets of spatiotemporal occupancy windows and target A and target B, and outputs them as a set of candidate occupancy windows in the same region and the same future time window to the rolling conflict resolution unit 203.

[0369] In another specific test, the original global path timing of target C included the southern work area element ES-04, with a planned execution time from the 35th to the 58th second.

[0370] Due to temporary material stockpiling on site, the status of the south work area in twin state organization layer 102 was updated to occupied at 34s, but there was a delay of about 2s in transmitting this status to the database.

[0371] When reading the spatiotemporal occupancy window of target C, the spatiotemporal occupancy organization unit 400 compares the regional state update time with the current scheduling cycle and finds that there is a risk of delayed update of the regional state. Therefore, it writes the spatiotemporal occupancy window corresponding to target C into the pending review state.

[0372] Subsequently, the global path timing generation unit 202 generates a new path version that includes the east side channel element EE-02. After reading the new version, the spatiotemporal occupancy organization unit 400 marks the old version's south side work area occupancy window as invalid and regenerates a valid spatiotemporal occupancy window based on the east side channel element.

[0373] The QGIS time-series visualization interface playback results show that the occupied area of ​​target C has changed from the south work area to the east passage, and the corresponding target status reference has been updated synchronously.

[0374] Through the above implementation method, the spatiotemporal occupancy organization unit 400 does not simply forward the global path time sequence to the rolling conflict resolution unit 203. Instead, it uses specific methods such as joint primary key verification, element type branch parsing, centerline buffering, waiting point buffering, three-dimensional envelope of rising and falling segments, union of work surface switching areas, time interval expansion, target status primary key mapping, spatiotemporal index writing, data trust marking, and window version updating to convert the path time sequence data of element + planned execution time into a spatiotemporal occupancy window that can be directly retrieved and judged by the rolling conflict resolution unit 203.

[0375] Therefore, the path orchestration results output by the global path timing generation unit 202 can enter the rolling conflict resolution process in the form of standardized spatiotemporal occupancy data.

[0376] Example 4

[0377] like Figure 10 As shown in the figure, an embodiment of the present invention provides an internal hierarchical structure of a field sensing access layer 103. The field sensing access layer 103 includes a data access unit, a location access unit, a status access unit, and a progress access unit, which are used to classify and receive device data and form device periodic data packets.

[0378] The data access unit uses a Modbus TCP protocol adapter and an OPC UA protocol adapter to receive equipment operation data. For data output from the PLC controller, it reads the speed, running direction, start / stop status, load value, operating mode, fault code, and control receipt fields according to the register mapping table; for OPC UA node data, it subscribes to node value changes according to the node address table and writes the protocol type, field name, field value, unit, and data quality flag.

[0379] The location access unit uses UWB positioning reception and coordinate preprocessing methods to receive location data. The location access unit converts the positioning tag into a device identifier based on the tag number and the unified access identifier mapping table, and performs median filtering on the five most recent positioning points. When the instantaneous speed calculated from the current positioning point and the previous valid positioning point exceeds 1.5 times the maximum allowable speed of the device, the positioning point is written into the abnormal location queue.

[0380] The status access unit uses status field normalization and finite state machine methods to receive device status data. It receives limit switches, emergency stop signals, safety circuits, door locks, load sensors, fault registers, and communication heartbeat fields, and categorizes the target status as idle, running, waiting, paused, faulty, locked, or offline. Continuous confirmation rules are set for emergency stop, limit switch, and door lock signals to avoid misjudgments caused by single jitter.

[0381] The progress access department uses a task event reception and progress status mapping method to receive work progress data. It receives construction task plans, on-site mobile terminal confirmation data, work completion events, and manual review records, and generates progress statuses such as pending start, in progress, completed, delayed, or pending exception after verification by task number.

[0382] All four access units write categorized data to the access buffer. The access buffer uses a unified access identifier and scheduling cycle number as the cache key and sets up slots for running data, location data, status data, and progress data. At the end of each 1-second scheduling cycle, the field sensing access layer 103 takes the last valid value of the 100ms running data, takes the median coordinate of the 200ms positioning data, and uses the previous valid value for the unupdated status and progress, writing a reuse mark.

[0383] The device's periodic data packets include a unified access identifier, scheduling cycle number, running data field group, location data field group, status data field group, progress data field group, data quality flag, missing data flag, and generation time. Data with a reception delay of less than 500ms is written to the current cycle; data between 500ms and 1500ms is appended with a delay flag; and data exceeding 1500ms is written to the late data set.

[0384] The experiment employed a PLC register data playback program, a UWB positioning playback program, an edge gateway access program, a timing database, and a data quality monitoring interface. The experiment lasted 30 minutes and included settings for PLC communication jitter, short-term UWB packet loss, positioning jumps, progress event delays, and status field jitter. Results showed that the scheme in this embodiment achieved a classification reception integrity rate of approximately 97.60%, a periodic packet generation success rate of approximately 96.90%, and an average time alignment error of approximately 146ms, which is superior to the single-channel hybrid access scheme and the access scheme based solely on device number.

[0385] Example 5

[0386] like Figure 11 As shown, this is an embodiment of the present invention, providing an internal hierarchical structure of a twin state organization layer 102. The twin state organization layer 102 includes an index generation unit, a spatial mapping unit, and an association organization unit, used to write the classified and received device data into an object index field set, and to form targets, target states, and target association relationships based on a unified spatial reference.

[0387] The index generation unit receives periodic data packets from devices and queries the access identifier mapping table using the unified access identifier as the primary key. If a match is successful, it reads the device number, device type, default floor, default construction area, and target coding rule to generate a target candidate identifier; if a match fails, the data packet is written to the unregistered access set and a record to be reviewed index is generated.

[0388] The object index field set includes object index identifier, unified access identifier, target candidate identifier, target type, collection time, scheduling cycle number, floor candidate identifier, construction area candidate identifier, spatial coordinates, operation status field, location status field, progress status field, abnormal status field, and data quality flag. When short-term location gaps exist, the index generation department temporarily fills in the gaps using the previous valid coordinates; when consecutive gaps exceed three scheduling cycles, the data is added to the target set awaiting review.

[0389] The Spatial Mapping Department establishes a unified spatial reference based on the BIM model coordinate system, the construction site measurement coordinate system, and the UWB local coordinate system. Specifically, it reads the UWB local coordinates and BIM world coordinates of no less than four spatial calibration points, uses least-squares rigid transformation to solve for the rotation matrix and translation vector, and removes abnormal calibration points with registration residuals exceeding 0.3m.

[0390] The spatial mapping department converts the UWB local coordinates in the object index record into BIM world coordinates, and determines the construction area identifier through point-surface inclusion judgment and the floor identifier through floor elevation matching. For targets near the area boundary, a regional hysteresis judgment method is used. When the boundary crossing distance is less than 0.5m, a boundary preservation mark is written; the construction area identifier is only updated when the boundary crossing exceeds 0.5m for two consecutive scheduling cycles or the movement direction continuously points to the adjacent area.

[0391] The associated organization department generates or updates target records using the target candidate identifier as the primary key, and forms location status, running status, occupied status, schedulable status, and abnormal status based on spatial mapping status, running status field, progress status field, and abnormal status field. The schedulable status is determined by rules such as communication validity, fault null, location validity, area validity, and task permission.

[0392] The associated organization further generates regional affiliation relationships, spatial adjacency relationships, candidate task execution relationships, occupation relationships, and avoidance relationships. Regional affiliation relationships are generated based on construction area identifiers; spatial adjacency relationships are generated based on regional adjacency lists; candidate execution relationships are generated based on construction task plans and target schedulable status; occupation relationships are generated based on the intersection of the target spatial envelope and the regional boundary; and avoidance relationships are generated based on a comparison of minimum distance and safe distance thresholds.

[0393] The experiment used BIM spatial reference data, UWB positioning playback data, equipment cycle data package samples, spatial calibration point data, PostGIS spatial database, and a target association visualization interface to verify the stability of BIM spatial reference registration and target association organization. The experiment included UWB positioning drift, construction area boundary errors, cross-regional movement, local positioning loss, and short-term status anomalies. The results show that the object index writing completeness rate of this embodiment is approximately 97.80%, the BIM spatial mapping accuracy is approximately 96.20%, the target association accuracy is approximately 95.70%, and the single-cycle organization time is approximately 0.63 seconds.

[0394] Example 6

[0395] like Figure 12 As shown in the figure, an embodiment of the present invention provides an internal hierarchical structure for a collaborative path decision layer 101. The collaborative path decision layer 101 receives the target, target state, and target association relationship formed by the twin state organization layer 102, and connects to the collaborative scheduling link 200, outputting task sorting results, allocation results, global path timing, conflict identifiers, and scheduling execution information.

[0396] The decision data carrier receives the target table, target status table, and target relationship table, and generates a decision snapshot using the scheduling cycle number as the primary index. If the target status update time is more than 3 seconds later than the current cycle, it is written to the status lag set; if the target relationship version is lower than the target version, it is written to the relationship pending update set. After the field validation passes, the data is divided into schedulable target partitions, task candidate partitions, spatial constraint partitions, and security constraint partitions.

[0397] The link embedding interface department sets up an interface description table, with fields including interface number, access unit, input fields, output fields, calling method, timeout time, and failure handling strategy. Task sorting and allocation unit 201 uses synchronous calls via local API with a timeout of 500ms; global path timing generation unit 202 uses asynchronous calls via message queue with a timeout of 1500ms; rolling conflict resolution unit 203 is triggered on a 1-second rolling cycle, with a future prediction time window of 60s.

[0398] When the task sorting interface times out, the link embedding interface retains the set of unexecuted tasks from the previous cycle and resubmits the call request. When the path generation interface returns a partially feasible path, it adds the partially feasible path to the set of paths to be reviewed. When no feasible path is found, the corresponding target is added to the waiting state. Each call is written to the link running status record table.

[0399] The decision output unit receives task sorting results, allocation results, global path timing, conflict identifiers, and scheduling execution information, and performs consistency checks. It checks whether the task unit identifier exists in the current decision snapshot, whether the allocation target is still schedulable, whether the path version number is the current valid version, and whether the conflict status has been resolved or the target has been written to the waiting set.

[0400] The scheduling execution information includes target identifier, task unit identifier, action type, target path element, action start time, action end condition, control fields, execution priority, and instruction version number. The control fields are written with options such as start, stop, wait, resume execution, speed setting, direction setting, or safety lock, depending on the action type. After successful verification, the information is output to the execution feedback correction layer 104 or the edge gateway.

[0401] The experiment uses target state replay data, target association samples, a collaborative scheduling link interface simulation program, a path planning service simulator, a conflict resolution service simulator, and a scheduling result verification program to verify the consistency of the collaborative path decision link interface and the stability of the result output. Perturbations include target state lag, delayed updates of candidate execution relationships, path planning service timeouts, partial failures in path generation, failure of conflict resolution strategies, and edge gateway acknowledgment delays.

[0402] Test results show that the consistency pass rate of the direct call to the collaborative scheduling link scheme is about 81.30%, and the completeness rate of scheduling execution information generation is about 82.70%; the scheme with only a unified interface is about 88.90% and 90.10% respectively; the scheme in this embodiment is about 95.80% and 97.20% respectively, with an average decision link time of about 0.72s, which is still less than the 1s scheduling cycle.

[0403] Example 7

[0404] like Figure 13 As shown in the figure, an embodiment of the present invention provides an internal hierarchical structure for an execution feedback correction layer 104. The execution feedback correction layer 104 includes a feedback receiving unit, a parameter correction unit, and a data archiving unit, used to receive feedback data after scheduling implementation, associate the feedback data with the corresponding task unit set, global path timing, and spatiotemporal occupancy window, and form parameter correction input and archived data.

[0405] The feedback receiving unit is equipped with PLC receipt receiving channels, position feedback receiving channels, task status receiving channels, and conflict handling receiving channels. PLC receipts use the target identifier and instruction version number to query the scheduling execution information table; position feedback uses the target identifier and positioning time to query the path element in the global path sequence; task status uses the task unit identifier to query the task unit set; and conflict handling feedback uses the conflict identifier to query the conflict record table.

[0406] The feedback receiving department uses a joint field association method to establish the correspondence between feedback data and task unit sets, global path timing, and spatiotemporal occupancy windows. Task unit sets are associated with task unit identifiers; global path timing is associated with target identifiers, path version numbers, and planned execution times; and spatiotemporal occupancy windows are associated with target identifiers, element identifiers, and occupancy window identifiers. Feedback data with these three types of associations is written into a complete associated feedback set.

[0407] The parameter correction unit extracts task time deviation, path execution deviation, conflict resolution deviation, and equipment response deviation from the associated feedback data. It uses the moving average of the actual execution time of the 20 most recent execution records for the same task type as the task time correction input; it writes path elements that deviate from the path element centerline by more than 0.8m for three consecutive scheduling cycles into the high-deviation path element set; it adjusts strategy priorities based on conflict strategy failures or recurrences; and it generates equipment response time correction input based on PLC start-up or stop response delays.

[0408] Parameter correction inputs are written to the parameter correction table, with fields including correction object, correction source, correction type, original parameter value, suggested parameter value, associated feedback record, generation time, and correction status. Safety distance threshold, speed limit, and device locking conditions only generate correction inputs pending review; estimated task time, path traversal cost, and conflict strategy priority are written to the parameter library when the sample size meets the requirements.

[0409] The data archiving department establishes an archiving index based on project number, scheduling cycle number, target identifier, task unit identifier, and conflict identifier, and generates task execution archive packages, path execution archive packages, time and space occupancy archive packages, conflict resolution archive packages, and parameter correction archive packages. Current construction cycle data is written to the hot data table, and data is written to the cold archive repository after the construction cycle ends; data related to conflict recurrence, manual intervention, safety lockouts, or abnormal equipment responses are marked with key review tags.

[0410] The experiment employed a PLC acknowledgment log replay program, a UWB trajectory feedback replay program, task execution logs, conflict resolution logs, parameter correction services, and an archived database to verify the correlation between execution feedback and the stability of parameter write-back. Experimental disturbances included PLC acknowledgment delay, UWB position offset, task confirmation lag, conflict resolution recurrence, and missing manual intervention records.

[0411] Test results show that the feedback association success rate of the feedback-only scheme is about 82.40%, and the archive integrity rate is about 78.60%; the feedback association but parameter correction scheme is about 91.70% and 88.90% respectively; the scheme in this embodiment is about 96.30% and 95.40% respectively, the parameter correction input effectiveness rate is about 89.60%, the recurrence conflict identification rate is about 86.70%, and the single-cycle processing time is about 0.49s.

[0412] Example 8

[0413] As one embodiment of the present invention, a multi-machine collaborative intelligent scheduling and management method is provided. This method employs a multi-machine collaborative intelligent scheduling and management model, first receiving and classifying device data, and then forming targets, target states, and target relationships based on the classified device data.

[0414] Specifically, equipment operation data is read from the PLC controller via Modbus TCP or OPC UA; location data is obtained through UWB positioning tags and base stations; status data is obtained through limit switches, emergency stop signals, safety circuits, and fault registers; and progress data is obtained through construction task schedules, on-site confirmation terminals, and work completion events. All of the above data is used to form equipment periodic data packets according to a unified access identifier and scheduling cycle number.

[0415] After the equipment cycle data packet is written into the object index field set, a target spatial record is formed through a rigid transformation from UWB local coordinates to BIM world coordinates. The target status is then formed based on communication status, fault codes, location status, construction area status, and task status. Subsequently, target association relationships are formed based on area affiliation, spatial adjacency, task candidate execution, occupancy, and avoidance relationships.

[0416] Based on the relationships between objectives, a task unit set is generated and written into the task definition field set. The task definition field set includes task unit identifier, objective candidate identifier, construction area identifier, action type, entry condition, termination condition, predecessor task identifier, successor task identifier, planned start time, planned end time, and execution constraint fields.

[0417] Tasks are sorted and allocated based on the task unit set and target status. First, task units that meet the entry criteria are selected. Then, task priority is calculated based on task urgency, target distance, target idle time, construction area occupancy status, completion time of preceding tasks, and equipment load status. When multiple targets can execute the same task unit, a cost matrix is ​​established, and the Hungarian matching method is used to determine the allocation result.

[0418] A global path time series is generated based on the task sorting and allocation results. Specifically, the construction area is divided into path nodes, path edges, waiting points, elevation segments, horizontal movement segments, work surface switching points, and control trigger points. A time extension graph is constructed based on the target's current position, task area, construction area boundary, equipment operating boundary, and safety distance constraints. Candidate paths are generated using A* search or Dijkstra search, and the path with the minimum total cost is selected.

[0419] Elements and planned execution times in the global path timeline are converted into spatiotemporal occupancy windows. Conflict assessments are then performed using a 1-second rolling cycle and a 60-second predicted time window. When the spatiotemporal occupancy windows of different targets overlap in time and their spatial envelopes intersect, or when the minimum distance is less than the safe distance threshold, or when they enter a single target's occupancy area at the same time, a conflict flag is generated. Based on the conflict flag, path fine-tuning, timeline adjustments, or task waiting are performed on the global path timeline.

[0420] After conflict resolution, the adjusted global path timing and scheduling execution information are output. The scheduling execution information includes target identifier, task unit identifier, action type, path element, action start time, action end condition, control fields, and instruction version number, which are then converted by the edge gateway into PLC control fields such as start, stop, wait, resume execution, speed setting, direction setting, or safety lock.

[0421] Example 9

[0422] As one embodiment of the present invention, a feedback correction and archiving process is provided for a multi-machine collaborative intelligent scheduling and management method. This process receives feedback data after scheduling is implemented, and associates the feedback data with task unit sets, global path timing, and spatiotemporal occupancy windows to form feedback association data.

[0423] Feedback data includes PLC control acknowledgments, UWB actual location data, task execution status, conflict resolution records, and manual confirmation records. When establishing associations, task unit sets are associated using task unit identifiers, global path timing is associated using target identifiers, path version numbers, and planned execution times, and spatiotemporal occupancy windows are associated using target identifiers, element identifiers, and occupancy window identifiers; records that cannot be associated are written to an isolated feedback set.

[0424] Based on feedback and related data, parameters are generated to correct the input. For task time deviations, the planned start time, planned end time, actual start time, and actual end time are compared to calculate the moving average of the execution time of the most recent 20 executions of the same task type and the estimated execution time is corrected accordingly. For path execution deviations, it is determined whether the distance between the actual location and the centerline of the path element continuously exceeds 0.8m, and the passage cost of the corresponding path element is increased accordingly.

[0425] For conflict resolution deviations, if path fine-tuning fails but timing adjustment succeeds, the priority of the path fine-tuning strategy under the corresponding conflict type is reduced and the priority of the timing adjustment strategy is increased. For device response deviations, if the target does not enter the corresponding state for more than 2 seconds after receiving the start or stop command, a device response correction input is generated and the time margin in the subsequent path timing is increased.

[0426] While the scheduling cycle is in progress, the parameters used in task sorting and allocation, global path timing generation, and rolling conflict resolution are updated based on the parameter correction input, and the parameter version number is written. When the scheduling cycle ends, the device data, targets, target status, target relationships, task unit sets, global path timing, spatiotemporal occupancy windows, conflict identifiers, and feedback data are archived.

[0427] The archived data includes equipment data archives, target status archives, task execution archives, path timing archives, time and space occupancy archives, conflict resolution archives, and feedback correction archives, and is indexed according to project number, floor number, scheduling cycle number, target identifier, task unit identifier, and conflict identifier. Data for the current construction cycle is written to the hot data table, and data after the construction cycle ends is written to the cold archive database.

[0428] In one test, target A was scheduled to enter the central transition zone at 318 seconds, but the actual PLC start receipt returned at 319.2 seconds, and the UWB trajectory showed that it did not enter the corresponding path element until 321 seconds. The system associates this feedback with the corresponding task unit, path timing, and time-space occupancy window to form equipment response correction input and task time correction input, and writes the task into the high-latency task set.

[0429] Example 10

[0430] As one embodiment of the present invention, a multi-machine collaborative intelligent scheduling and management application is provided, which coordinates the scheduling of at least two types of equipment, including building construction machines, scaffolding climbing machines, concrete placing booms, and construction hoists. Each type of equipment is connected to the scheduling target model and is programmed with target identifier, target type, operating area, traversable path, operation action, equipment status, and safety constraints.

[0431] When multiple devices are involved, including a building construction machine and a concrete placing machine, the building construction machine performs tasks such as moving floor construction platforms, hoisting components, or adjusting platforms, while the concrete placing machine performs concrete placement. The model writes the central transition zone, the work surface entrance, and the concrete placing machine's turning range into spatial constraint partitions, and determines whether there are conflicts in area occupancy or work surface entry order based on the spatiotemporal occupancy windows of the two. If the concrete placing machine's task cannot be interrupted, the building construction machine's entry time is delayed, and a waiting instruction is generated.

[0432] When multiple devices include a building construction machine and a construction hoist, the lifting segment of the construction hoist is represented as a vertical spatiotemporal occupancy window, and the horizontal movement segment of the building construction machine is represented as a planar spatiotemporal occupancy window. The model determines within a 60-second prediction window whether the two devices will spatially overlap in the same or adjacent areas; if the construction hoist is performing an uninterrupted personnel or material transfer task, the building construction machine will be put into a waiting state.

[0433] When multiple devices include climbing scaffolding machines and construction hoists, the model writes the climbing scaffolding machine attachment area, lifting area, and construction hoist docking area into the spatial constraint partition. If the climbing scaffolding machine lifting area and the construction hoist docking area are on the same floor or adjacent floors and the safe distance is less than the threshold, a conflict flag is generated, and the construction hoist is ordered to wait, the climbing scaffolding machine is ordered to pause, or the docking floor is adjusted according to the task priority.

[0434] In one specific application, target A is a building construction machine, target B is a concrete placing boom, and target C is a construction hoist. After receiving PLC operation data, UWB location data, and task data from the three devices, the model forms target A, target B, and target C and their respective target states. After task sorting and allocation, a global path timing sequence is generated and converted into a spatiotemporal occupancy window. When both target A and target B occupy the central transition zone between 37s and 45s, the model generates a region occupancy conflict flag, keeps target B's current task unchanged, delays target A's entry time by 12s, and outputs wait and resume execution instructions.

[0435] In another application, target A is a climbing scaffold machine, and target B is a construction hoist. Target A is scheduled to adjust its lifting and lowering operations between the 5th and 6th floors, while target B is scheduled to dock on the 5th floor for material transfer. The model maps the climbing scaffold machine's lifting area and the construction hoist's docking area to the BIM world coordinate system. If the spatiotemporal occupancy windows of both are less than the safety distance threshold within the same time window, a safety distance conflict indicator is generated, and the docking time of target B is adjusted according to the task stage, or a pause instruction for target A is generated.

[0436] Through the above application methods, this embodiment can be applied to at least two combinations of equipment such as building construction machines, climbing scaffolding machines, concrete placing machines, and construction hoists, and can carry out multi-machine collaborative intelligent scheduling and management around area occupation, path elements, time sequence elements, work surface switching, and multi-objective conflict resolution.

[0437] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A multi-machine collaborative intelligent scheduling and management model, characterized in that, include: The scheduling main architecture (100) includes a collaborative path decision layer (101) that handles the entire process of intelligent decision-making for multi-machine collaborative paths; The collaborative scheduling link (200) realizes task allocation, path orchestration and conflict resolution by embedding a task sorting and allocation unit (201), a global path timing generation unit (202) and a rolling conflict resolution unit (203) in the collaborative path decision layer (101).

2. The multi-machine collaborative intelligent scheduling and management model as described in claim 1, characterized in that: It also includes, A task unit organization unit (300) is located between the twin state organization layer (102) and the task sorting and allocation unit (201); Read the targets formed by the twin state organization layer (102), generate a set of task units based on the relationship between the targets, and write them into the task definition field set.

3. The multi-machine collaborative intelligent scheduling and management model as described in claim 2, characterized in that: It also includes, The spatiotemporal occupancy organization unit (400) is located between the global path timing generation unit (202) and the rolling conflict resolution unit (203); The spatiotemporal occupancy organization unit (400) reads the global path timing output by the global path timing generation unit (202) and converts the elements and planned execution times in the global path timing into spatiotemporal occupancy windows; The spatiotemporal occupancy window establishes a correspondence with the target in the twin state organization layer (102).

4. The multi-machine collaborative intelligent scheduling and management model as described in any one of claims 1 to 3, characterized in that: The scheduling main architecture (100) also includes a field perception access layer (103), through which device data is classified and received.

5. The multi-machine collaborative intelligent scheduling and management model as described in any one of claims 1 to 3, characterized in that: The scheduling main architecture (100) also includes a twin state organization layer (102), which writes the classified and received device data into the object index field set and forms the target, target state and target association relationship based on a unified spatial reference.

6. The multi-machine collaborative intelligent scheduling and management model as described in claim 5, characterized in that: The collaborative path decision layer (101) receives the target, target status and target association relationship formed by the twin state organization layer (102) and connects to the collaborative scheduling link (200), thereby outputting task sorting results, allocation results, global path timing, conflict identifier and scheduling execution information.

7. The multi-machine collaborative intelligent scheduling and management model as described in any one of claims 1 to 3, characterized in that: The scheduling main architecture (100) also includes an execution feedback correction layer (104), which receives feedback data after scheduling is implemented, establishes association between the feedback data and the corresponding task unit set, global path timing and spatiotemporal occupancy window, and forms parameter correction input and archive data based on the feedback data.

8. A multi-machine collaborative intelligent scheduling and management method, characterized in that, The multi-machine collaborative intelligent scheduling and management model as described in any one of claims 1 to 7 includes: The device data is received and classified through the field sensing access layer (103); The twin state organization layer (102) forms the target, target state and target association relationship, and generates a set of task units; Through the collaborative scheduling link (200) in the collaborative path decision layer (101), the task sorting and allocation unit (201), the global path timing generation unit (202) and the rolling conflict resolution unit (203) are invoked to generate and adjust the global path timing; The system outputs scheduling execution information and receives feedback data by executing the feedback correction layer (104).

9. The multi-machine collaborative intelligent scheduling and management method as described in claim 8, characterized in that: It also includes, The execution feedback correction layer (104) associates the feedback data with the task unit set, global path timing and conflict identifier to form parameter correction input; The collaborative path decision layer (101) modifies the parameters of the task sorting and allocation unit (201), the global path timing generation unit (202), and the rolling conflict resolution unit (203) based on the parameter correction input, and archives them after the scheduling cycle ends.

10. An application of a multi-machine collaborative intelligent scheduling and management model in the construction field for the collaborative scheduling of multiple devices, wherein the multiple devices are at least two of building construction machines, climbing scaffolding machines, concrete placing booms, and construction hoists.

Images