A Concrete Pouring Path Optimization Method Based on BIM Model
By using a BIM model-based concrete pouring path optimization method, the problem of difficulty in judging the changes in concrete flow direction and velocity in complex nodes was solved, enabling more accurate pouring path planning and improving the filling quality and structural reliability inside the nodes.
Patent Information
- Application Number
- CN202611610854.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-19
- Publication Date
- 2026-11-17
AI Technical Summary
In large-scale construction projects, during the concrete pouring process at complex nodes, existing technologies struggle to accurately determine the preferred flow direction and velocity changes of concrete at the intersection of branch lines. This results in some branch lines being filled preferentially while others lag behind, leading to problems such as localized accumulation, underfilling, voids, or inadequate sealing.
Based on the BIM model, by acquiring the spatial orientation characteristics and gravity diversion tendency of the node branches, and combining computational fluid dynamics algorithms to simulate the force distribution of the slurry, internal flow direction data is generated to determine the filling sequence and the movement trajectory of the external placing boom, thus optimizing the concrete pouring path.
It improves the accuracy of concrete pouring path planning for complex nodes, reduces problems such as local accumulation, underfilling of branch paths and voids, and improves the compactness and reliability of one-time pouring.
Smart Images

Figure CN123116876A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of building construction and intelligent construction technology, specifically to a method for optimizing concrete pouring paths based on BIM models. Background Technology
[0002] In large-scale construction projects, complex structural parts such as beam-column connection areas, irregular transition nodes, and multi-branch intersection nodes usually bear a large load transfer. Their internal spatial forms are irregular, and different branches have significant differences in extension direction, slope, cross-sectional dimensions, and intersection relationships. In order to ensure the integrity and load-bearing capacity of the node, such nodes usually require concrete pouring to be completed in one go as much as possible. Therefore, whether the concrete can be reasonably distributed among the branches inside the node and fully and continuously filled directly affects the forming quality and structural reliability of the node.
[0003] In current construction processes, the pouring location, placement sequence, and placement boom movement of complex nodes are typically determined primarily based on the node's shape, construction drawings, and the on-site experience of construction personnel. For complex nodes with multiple branch intersections, due to the different spatial orientations of each branch, the concrete entering the node is easily affected by both the direction of gravity and the spatial structure of the branches, resulting in varying degrees of flow distribution among different branches. Current methods struggle to accurately determine the preferred flow direction of concrete at the branch intersection before pouring, leading to some branches being filled preferentially while others with larger deflection angles, higher positions, or greater flow resistance experience delayed filling. Furthermore, after entering the node via pumping pipes, the flow state of the concrete is not solely determined by the node's geometry but is also influenced by the initial pumping pressure, gravity, and the inherent flow characteristics of the slurry. As the concrete passes through branch bends, bifurcations, and areas of cross-sectional change, its flow velocity and stress state continuously change, potentially resulting in significant flow direction deflection. Because the interior of a node is typically closed or semi-closed, construction workers cannot directly observe the flow direction and velocity changes of the grout within each branch, making it difficult to determine in a timely manner whether the concrete is spreading in the expected direction. Furthermore, the time required for different node branches to reach full filling is usually inconsistent, with sequential filling relationships and time differences between branches. Existing pouring methods often lack effective prediction of the filling sequence within the node, making it difficult to accurately grasp the continuous spread of concrete from one branch to adjacent branches. When the external placing boom continues to move according to the preset position and fixed rhythm, its trajectory is prone to discrepancies with the actual filling progress within the node, resulting in continued feeding into areas that are already nearly full, while branches that are not yet fully filled fail to receive sufficient grout in a timely manner.
[0004] In addition, the cavity volume varies in different areas of complex nodes, and the actual filling amount of concrete per unit time will also change with the internal spread state. If the movement process of the placing boom and the discharge speed cannot be coordinated with the actual filling situation, problems such as local concrete accumulation, underfilling of branch roads, voids, or incomplete wrapping may easily occur. Summary of the Invention
[0005] The purpose of this invention is to provide a method for optimizing concrete pouring paths based on BIM models, thereby solving the problems existing in the prior art.
[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for optimizing concrete pouring paths based on a BIM model, comprising: S1, acquiring BIM model data of nodes, extracting spatial orientation features of each node branch based on the BIM model data, calculating gravity diversion tendency corresponding to different spatial orientations based on the branch intersection location, and constructing a node spatial map based on the gravity diversion tendency; S2, acquiring initial pumping pressure data and gravity data of concrete, and using computational fluid dynamics algorithms to simulate the flow field of the initial pumping pressure data and gravity data within the node spatial map to obtain a slurry stress distribution map; S3, judging the velocity change within each node branch based on the slurry stress distribution map and the node spatial map, and generating an internal... S4. Determine the time nodes when each node branch reaches full load status through internal flow data, and generate a filling sequence based on the time nodes; S5. Obtain the spatial coordinates corresponding to the filling sequence, adjust the memory weights of the long short-term memory network according to the full load time interval of adjacent node branches, perform temporal prediction on the spatial coordinates, and obtain the propagation process trajectory; S6. Calculate the movement vector of the external placing rod based on the propagation process trajectory, and generate the rod trajectory through the movement vector; S7. Obtain the node cavity volume corresponding to the rod trajectory, calculate the volume increment per unit time based on the node cavity volume and the propagation process trajectory, and if the volume increment is less than the preset volume threshold, determine the discharge speed, optimize the rod trajectory based on the discharge speed, and obtain the concrete pouring path.
[0007] Preferably, step S1 includes acquiring BIM model data, parsing the BIM model data to obtain the three-dimensional coordinates of each node branch; extracting the spatial orientation feature vector of each node branch at the intersection position based on the three-dimensional coordinates; determining whether the angle between the spatial orientation feature vector and the vertical vector is less than a preset angle threshold; if the angle between the spatial orientation feature vector and the vertical vector is less than the preset angle threshold, it is determined that gravity dominates, and the gravity diversion tendency value is calculated based on the gravity dominance; and constructing a node spatial map based on the gravity diversion tendency value.
[0008] Preferably, step S2 includes acquiring initial pumping pressure data and gravity data of the concrete; constructing a node space diagram based on the initial pumping pressure data and gravity data to obtain a set of spatial grid nodes; solving the spatial grid node set using a computational fluid dynamics algorithm to obtain a set of node velocity vectors; calculating the node pressure gradient matrix based on the set of node velocity vectors; calculating local shear stress data if the node pressure gradient matrix exceeds a preset pressure gradient threshold; and performing flow field simulation within the node space diagram based on the local shear stress data and the node pressure gradient matrix to obtain a slurry stress distribution diagram.
[0009] Preferably, step S3 includes extracting each node branch based on the slurry stress distribution diagram and node space diagram, solving each node branch to obtain the cross-sectional flux; determining the velocity change by comparing the initial flow velocity and the real-time flow velocity based on the cross-sectional flux; determining whether the velocity change is greater than a preset velocity threshold; if the velocity change is greater than the preset velocity threshold, obtaining the deflection angle of each node branch; and generating internal flow direction data by vector synthesis based on the deflection angle and each node branch.
[0010] Preferably, step S4 includes acquiring internal flow direction data, extracting the current flow rate and maximum carrying capacity of each node branch from the internal flow direction data; calculating the available remaining space based on the current flow rate and maximum carrying capacity; obtaining the predicted arrival time through the available remaining space; marking the predicted arrival time as the target time node; determining the time node when each node branch reaches the full load state based on the target time node; and generating a filling sequence based on the time node.
[0011] Preferably, step S5 includes obtaining the spatial coordinates corresponding to the filling sequence to construct an initial spatial coordinate set; calculating the full-load time interval of adjacent node branches based on the initial spatial coordinate set; if the full-load time interval is greater than a preset time threshold, adjusting the memory weights of the long short-term memory network using the full-load time interval, and performing time-series prediction on the initial spatial coordinate set through the adjusted memory weights of the long short-term memory network to obtain a predicted coordinate sequence; and splicing the historical sequence based on the predicted coordinate sequence to obtain the propagation process trajectory.
[0012] Preferably, step S6 includes acquiring the trajectory of the propagation process, calculating an initial displacement sequence based on the trajectory of the propagation process, and determining a movement vector based on the initial displacement sequence.
[0013] Preferably, step S6 further includes obtaining the member posture and joint angle by solving the movement vector; determining whether the joint angle exceeds the angle threshold; if the joint angle exceeds the angle threshold, correcting the member posture to obtain the target posture; generating control commands based on the target posture; calculating the movement vector of the external fabric rod based on the spread process trajectory using the control commands; and generating the member trajectory using the movement vector.
[0014] Preferably, step S7 includes acquiring the rod trajectory, calculating the node cavity volume based on the rod trajectory, acquiring the propagation process trajectory, extracting fluid filling data based on the node cavity volume and the propagation process trajectory, and calculating the volume increment.
[0015] Preferably, step S7 further includes determining the discharge speed based on fluid filling data if the volume increment is less than a preset volume threshold; correcting the rod trajectory based on the discharge speed to obtain the target rod trajectory; and generating a concrete pouring path based on the target rod trajectory.
[0016] As can be seen from the above technical solution, the present invention has the following beneficial effects:
[0017] This BIM-based concrete pouring path optimization method extracts the spatial orientation characteristics of each branch of a node and constructs a node spatial map by combining it with gravity diversion tendency. This allows the spatial distribution relationship of different branches within a node to be used for subsequent slurry flow analysis. Furthermore, it combines initial pumping pressure and gravity data to perform flow field simulation, obtaining the slurry stress distribution and velocity changes within each branch. Based on this, it determines the internal flow direction, the time node for each branch to reach full load, and the filling sequence, thus more accurately reflecting the diversion and filling process of concrete within complex nodes. On this basis, it optimizes the pouring path based on the full load time interval between adjacent branches. The memory weights of the short-term memory network are adjusted to predict the temporal spread of the slurry and generate the trajectory of the external placing boom accordingly. This ensures that the movement of the placing boom is coordinated with the actual spread sequence of the slurry inside the node. Simultaneously, the volume increment per unit time is determined based on the node cavity volume and the slurry spread process, and the discharge rate is determined accordingly to further optimize the boom trajectory. This reduces issues such as local accumulation, underfilling of branches, voids, or incomplete wrapping caused by mismatch between the placing trajectory and the internal filling progress, thereby improving the accuracy of concrete pouring path planning for complex nodes and the compactness and reliability of one-time casting. Attached Figure Description
[0018] Figure 1 This is a flowchart of the method of the present invention;
[0019] Figure 2 This is a diagram illustrating the process of determining the internal flow direction of a node in this invention. Detailed Implementation
[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0021] Example 1:
[0022] like Figure 1 and Figure 2 As shown, the present invention provides a technical solution: a method for optimizing concrete pouring paths based on a BIM model, comprising:
[0023] S1. Obtain the BIM model data of the nodes, extract the spatial orientation characteristics of each node branch based on the BIM model data, calculate the gravity diversion tendency corresponding to different spatial orientations based on the branch intersection location, and construct the node spatial map based on the gravity diversion tendency.
[0024] S2. Obtain the initial pumping pressure data and gravity data of concrete, and use computational fluid dynamics algorithm to simulate the flow field of the initial pumping pressure data and gravity data in the nodal space diagram to obtain the force distribution diagram of the slurry.
[0025] S3. Based on the slurry stress distribution diagram and node space diagram, determine the velocity change in each node branch. If the velocity change is greater than the preset velocity threshold, generate internal flow direction data.
[0026] S4. Determine the time point when each node branch reaches full load state through internal flow data, and generate a filling sequence based on the time point.
[0027] S5. Obtain the spatial coordinates corresponding to the filling sequence, adjust the memory weights of the long short-term memory network according to the full load time interval of adjacent node branches, perform temporal prediction on the spatial coordinates, and obtain the propagation process trajectory.
[0028] S6. Calculate the movement vector of the external material bar based on the propagation process trajectory, and generate the bar trajectory through the movement vector;
[0029] S7. Obtain the volume of the node cavity corresponding to the rod trajectory. Calculate the volume increment per unit time based on the node cavity volume and the propagation process trajectory. If the volume increment is less than the preset volume threshold, determine the discharge speed. Optimize the rod trajectory based on the discharge speed to obtain the concrete pouring path.
[0030] In this embodiment, the BIM model data corresponding to the node to be poured is first read, and the node cavity, each node branch, the branch intersection position, and the three-dimensional spatial coordinates corresponding to each node branch are extracted from the BIM model data.
[0031] Using the intersection of branches as the spatial analysis benchmark, the spatial orientation of each node branch is determined from the intersection to the direction of its extension, and this spatial orientation is used as the spatial orientation feature of the node branch. For node branches extending in a straight line, the spatial orientation is determined according to the direction of the extension of the branch centerline. For node branches extending in a curved manner, continuous segments are divided according to the actual curved path of the branch, and the spatial orientation of each segment is determined, so that the extraction results are consistent with the actual geometric shape of the node branch.
[0032] After obtaining the spatial orientation characteristics of each node branch, the spatial orientation of each node branch is compared with the gravity direction of the construction area, and the gravity diversion tendency is determined based on the directional relationship between the two.
[0033] Specifically, the extension direction of the node branch is decomposed into the spatial direction of gravity, and the direction of gravity's effect on the flow of concrete along the node branch is determined based on the decomposition results. When the extension direction of the node branch contains a directional component consistent with the gravity direction, it is determined that gravity has a unidirectional driving effect on the flow of concrete along the node branch. When the extension direction of the node branch is perpendicular to the gravity direction, it is determined that gravity does not form a diversion driving force in the axial direction of the node branch. When the extension direction of the node branch contains a directional component opposite to the gravity direction, it is determined that gravity has a reverse effect on the flow of concrete along the node branch. Based on the above directional relationships of each node branch, a corresponding gravity diversion tendency is formed.
[0034] The branch intersection locations, each node branch, and their spatial connection relationships are converted into graph structure data; the branch intersection locations and feature locations in the node branches are set as graph nodes, the channels between interconnected feature locations are set as graph edges, and the spatial orientation features and gravity diversion tendency of the corresponding node branches are written into the corresponding graph edge attributes, thereby forming a node spatial graph.
[0035] This node space diagram enables a unified data correspondence between the geometric positional relationships of node structures, the connectivity between node branches, and the direction of gravity.
[0036] After the node space map is constructed, the initial pumping pressure data when concrete enters the node area and the corresponding gravity data of the construction area are obtained.
[0037] Based on the BIM model, the space range available for concrete flow inside the node cavity and each node branch is determined, and this space range is set as the computational fluid dynamics analysis region. The position where the concrete enters the node cavity is set as the fluid inlet, the initial pumping pressure data is set as the inlet pressure condition, the gravity data is set as the volume force condition in the entire fluid analysis region, and the inner walls of the node cavity and each node branch are set as the flow boundary.
[0038] The flow parameters of the concrete are established based on the density, viscosity and rheological parameters of the concrete to be poured, and the force and flow processes of the concrete after entering the node are calculated using computational fluid dynamics algorithms.
[0039] During the calculation process, the pressure state, flow velocity state, flow direction and gravity state corresponding to different spatial positions inside the node cavity and each node branch are obtained respectively, and the above calculation results are matched according to the three-dimensional spatial position in the BIM model. After the spatial matching is completed, the force information, flow velocity information and flow direction information corresponding to each calculation position are summarized to form the slurry force distribution map. The slurry force distribution map is used to represent the force state and motion state of concrete at different spatial positions inside the node under the combined action of initial pumping pressure and gravity.
[0040] The stress distribution diagram of the slurry is spatially correlated with the nodal space diagram. The flow field region corresponding to each nodal branch is determined based on the branch boundary in the nodal space diagram, and the flow velocity data of the concrete is extracted within the flow field region.
[0041] Continuous flow velocity detection positions are set along the extension direction of the node branch at a predetermined spatial interval, and the flow velocity corresponding to each flow velocity detection position is read. The flow velocities corresponding to two adjacent flow velocity detection positions are subtracted, and the absolute value of the difference is taken as the flow velocity change between adjacent positions. The flow velocity change between all adjacent flow velocity detection positions is calculated sequentially according to the extension direction of the branch, so as to obtain the flow velocity change inside the corresponding node branch.
[0042] The velocity change of each node branch is compared with the preset velocity threshold one by one; when any velocity change is greater than the preset velocity threshold, the corresponding node branch is determined as the node branch with effective flow change, and the slurry flow direction at the location of the velocity change is read.
[0043] Subsequently, by combining the branch connectivity relationships in the node spatial diagram, the direction of concrete flowing from the current spatial location to the next spatial location is determined, and the node branch identification, inflow location, outflow location, flow direction, flow velocity data and corresponding time data are combined to form internal flow direction data. Through the internal flow direction data, the spatial areas that concrete passes through sequentially inside the node after entering each node branch from the branch intersection location and the corresponding time relationship are determined.
[0044] The time points at which each node branch reaches full load are determined based on internal flow data.
[0045] First, read the internal boundaries of each node branch in the BIM model, and calculate the internal volume of concrete that each node branch can accommodate based on the internal boundaries. Use this internal volume as the target filling volume of the corresponding node branch.
[0046] Subsequently, the volume of concrete entering the node branch per unit time is determined based on the flow velocity recorded in the internal flow direction data and the flow cross-sectional area corresponding to the node branch. Then, the volume of concrete entering the node branch is continuously accumulated according to the time sequence to obtain the cumulative filling volume of the corresponding node branch.
[0047] When the cumulative filling volume of a certain node branch reaches the target filling volume of that node branch, the moment when the target filling volume is reached is determined as the full-load time node of that node branch.
[0048] Perform the above calculations on all node branches to obtain the full load time node corresponding to each node branch; then arrange the corresponding node branches in order of full load time nodes from first to last, and generate a filling sequence sequence from the arrangement results; the filling sequence sequence consists of the sequence information of the node branches and their corresponding full load time information, and is used to describe the actual filling order relationship of concrete between different node branches.
[0049] Based on the filling sequence, the spatial coordinates of each node branch in the BIM model are read sequentially, and the spatial coordinates of each node branch are arranged in chronological order of the full load time nodes to form spatial coordinate time series data. For any two adjacent node branches in the filling sequence, the full load time nodes of the preceding node branch and the following node branch are read respectively. The full load time interval between the two adjacent node branches is obtained by subtracting the full load time of the preceding node branch from the full load time of the following node branch. The full load time interval between all adjacent node branches in the filling sequence is calculated in the same way.
[0050] The obtained full-load time intervals are numerically converted according to a uniform time scale so that the time intervals between different adjacent node branches can be input into the long short-term memory network within a uniform range; the converted full-load time intervals are used as the basis for adjusting memory weights.
[0051] When the full-load time interval between adjacent node branches corresponds to the first time interval value, the participation weight of the historical state of the previous node branch in the spatial prediction process of the next node branch is set according to the first time interval value; when the full-load time interval between adjacent node branches corresponds to the second time interval value, the participation weight corresponding to the historical state is reset according to the second time interval value.
[0052] This ensures that the actual full-load time interval corresponding to each pair of adjacent node branches directly participates in the memory state update process of the Long Short-Term Memory network.
[0053] The spatial coordinate time-series data corresponding to the filling sequence are input into the Long Short-Term Memory (LSTM) network, which has completed the memory weight adjustment. The LSTM network sequentially reads the spatial position of the node branch that has been filled, the corresponding full-load time node, and the full-load time interval between adjacent node branches, and predicts the spatial coordinates corresponding to the next moment based on the previous spatial state and the current time state. After obtaining multiple predicted spatial coordinates in chronological order, the adjacent predicted spatial coordinates are connected sequentially to form the spread process trajectory. This spread process trajectory is used to represent the spatial position change process corresponding to the concrete expanding from the currently filled area to the subsequent unfilled area.
[0054] Calculate the movement vector of the external fabric rod based on the trajectory of the propagation process.
[0055] Specifically, for two adjacent target positions in the spread process trajectory, the three-dimensional spatial coordinates of the previous target position and the next target position are read respectively. The coordinate values of the next target position in the three coordinate directions are subtracted from the coordinate values of the previous target position in the corresponding coordinate directions. The spatial direction and moving distance of the cloth rod from the previous target position to the next target position are determined by the difference in the three coordinate directions.
[0056] This generates a corresponding movement vector; following the temporal order of the spread process trajectory, the above processing is performed sequentially on all adjacent target positions to obtain multiple consecutive movement vectors.
[0057] All movement vectors are arranged in the order of generation, and the starting and ending positions of each movement vector are connected in chronological order to generate the trajectory of the external placing boom. Each target position in the trajectory corresponds to a predicted filling position in the spread process trajectory. The movement order in the trajectory is consistent with the filling order sequence, thus establishing a correspondence between the movement process of the external placing boom and the predicted spread process of the concrete inside the node.
[0058] After obtaining the member trajectory, the corresponding node cavity is determined based on the spatial position traversed by the member trajectory, and the three-dimensional geometric boundary of the node cavity is read from the BIM model.
[0059] The volume of the node cavity is calculated based on the internal space enclosed by the three-dimensional geometric boundary of the node cavity; then, according to the time sequence corresponding to the propagation process trajectory, the spatial coverage of concrete in the node cavity at two consecutive sampling times is determined, and the corresponding concrete filling volume is calculated based on the spatial coverage at each sampling time.
[0060] Subtract the concrete filling volume corresponding to the previous sampling time from the concrete filling volume corresponding to the subsequent sampling time to obtain the change in filling volume between the two sampling times; then divide the change in filling volume by the time length between the two sampling times to obtain the volume increment per unit time; calculate all adjacent sampling times in the propagation process trajectory in the same way to obtain the continuous volume increment data corresponding to the rod trajectory; the volume increment per unit time is used to represent the rate of change of the actual volume of concrete occupying the node cavity space within the corresponding time period.
[0061] The calculated volume increment is compared with a preset volume threshold.
[0062] When the volume increment is greater than or equal to the preset volume threshold, the discharge control state corresponding to the current member trajectory is maintained, and subsequent pouring continues according to the member trajectory; when the volume increment is less than the preset volume threshold, the total volume of the current node cavity and the current filled volume are read, and the total volume of the node cavity is subtracted from the current filled volume to obtain the remaining volume to be filled in the node cavity; then, based on the remaining volume to be filled, the position corresponding to the current propagation process trajectory, and the set subsequent filling time, the concrete volume required to be output per unit time is determined, and the concrete volume required to be output per unit time is determined as the discharge rate.
[0063] The trajectory of the rod is optimized based on the determined discharge speed.
[0064] First, based on the spatial range of each node cavity and node branch, the rod trajectory is divided into multiple continuous trajectory segments, and a correspondence is established between each trajectory segment and the discharge speed of the corresponding node cavity. For trajectory segments with a volume increment less than a preset volume threshold, the discharge duration required to complete the corresponding remaining volume to be filled is calculated based on the determined discharge speed, and the dwell time and movement sequence of the placing rod in the corresponding trajectory segment are determined based on the discharge duration. After the placing rod completes the output of the corresponding volume of concrete in the current node area, it moves along the next movement vector to the next target position in the filling sequence.
[0065] The discharge speed configuration, discharge duration configuration, and movement sequence configuration are executed sequentially for each trajectory segment in the rod trajectory, so that each trajectory segment corresponds to a determined movement position, movement direction, discharge speed, and discharge duration.
[0066] After processing all trajectory segments, the trajectory segments are connected in the order of filling to obtain the final concrete pouring path. This concrete pouring path also includes the spatial movement sequence of the placing boom between different node areas and the concrete output control relationship corresponding to each node area.
[0067] The BIM model provides data on node cavities, node branches, branch intersections, node spatial coordinates, and node cavity volume. The relationship between the spatial orientation characteristics of node branches and the gravity direction is used to determine the gravity diversion tendency. Initial pumping pressure data and gravity data are used to form the flow boundary and stress conditions for computational fluid dynamics algorithms. The slurry stress distribution diagram is used to determine the velocity changes and flow direction in each node branch. The comparison results of velocity changes with preset velocity thresholds are used to generate internal flow direction data. The internal flow direction data is used to calculate the time nodes when each node branch reaches full load. Each full load time node is used to generate the filling sequence. The full load time interval between adjacent node branches is used to adjust the memory weights of the long short-term memory network. The continuous spatial coordinates output by the long short-term memory network are used to generate the propagation process trajectory. The propagation process trajectory is used to calculate the movement vector of the external placing rod and generate the rod trajectory. The node cavity volume and the propagation process trajectory are used to calculate the volume increment per unit time.
[0068] When the volume increment is less than the preset volume threshold, the discharge speed is determined based on the remaining volume to be filled, and the discharge duration and movement sequence in the rod trajectory are adjusted using the discharge speed, thereby generating a concrete pouring path corresponding to the concrete filling process inside the node.
[0069] To verify the improvement effect of this application on the concrete filling process and material placement path of complex nodes, a comparative test was conducted between a traditional fixed path pouring group and a path optimization group of this application.
[0070] The experiment used a node model with three branches. The node cavity dimensions were 300 mm × 300 mm × 300 mm, and the internal cross-sectional dimensions of each branch were 100 mm × 100 mm. The effective length of each branch was 300 mm. The first branch extended horizontally, the second branch inclined downward at 45 degrees from the node cavity, and the third branch inclined upward at 45 degrees from the node cavity to create different spatial orientation conditions. The node model was built as a BIM model with the same geometric dimensions, and a transparent observation plate was set on the side of the model to record the time when concrete entered each branch and when each branch reached full load.
[0071] The experimental concrete used was the same batch of fine aggregate concrete, and the three branch models used the same mix proportion. A small concrete pump was used to deliver the concrete, and the pump pipe end was fixed to a triaxial positioning bracket so that the discharge end could move according to the set spatial coordinates. The traditional fixed path pouring group distributed the concrete to the corresponding areas of the three branches in a predetermined fixed order, while maintaining a constant discharge speed. The path optimization group of this application first imported the node BIM model, constructed the node spatial diagram according to the spatial orientation of each node branch, and then obtained the slurry stress distribution diagram based on the pumping pressure and gravity conditions. According to the aforementioned implementation method, the internal flow direction, branch full load time, filling sequence, spread process trajectory and rod trajectory were determined, and the discharge speed was adjusted according to the volume increment per unit time.
[0072] Both sets of tests started with the node cavity in an empty state. Timing and video recording were started at the same time as pumping began, recording the time when concrete began to enter each node branch and the time when concrete reached the end mark of each branch. The time when concrete reached the end of the branch and filled the branch cross section was taken as the full load time of that branch. Timing was stopped after all three node branches reached the full load state, and the time for a single pour was recorded.
[0073] The consistency of branch filling is evaluated by the time difference between the full load times of the three branches. Specifically, the time when the branch first reaches full load and the time when it last reaches full load are selected from the full load times of the three node branches. The time obtained by subtracting the first full load time from the last full load time is taken as the maximum time difference of branch full load. This time can directly reflect the difference in filling progress between branches with different spatial orientations.
[0074] During the pouring process, a graduated collection container is set outside the node model to collect all the concrete overflowing from the node entrance or the end of the branch. After the pouring is completed, the volume of concrete in the collection container is read directly as the overflow amount of a single test.
[0075] After the model was poured, it was left to stand for 24 hours before the side formwork was removed. The exposed surfaces of the three branches and node cavities were inspected. Unfilled holes with a diameter of 5 mm or more were recorded as visible voids, and the total number of visible voids in a single specimen was counted. At the same time, it was checked whether there were areas at the ends of each branch that did not reach the specified filling position. If such a situation existed, it was recorded as one instance of underfilling. Each pouring method was tested three times consecutively, and the average value of the three tests was used for comparison. The number of times underfilling occurred was recorded as a cumulative occurrence across the three tests.
[0076] The experimental data are shown in the table below:
[0077] Table 1
[0078] Comparison Projects Fixed path casting This application optimizes the path Average pouring completion time 172 seconds 154 seconds Maximum time difference when branch lines are fully loaded 52 seconds 25 seconds Average overflow 1.8 liters 0.7 liters Number of underfilling tests in 3 trials 3 times 0 times Average number of visible voids 11 4
[0079] The results above show that, under the condition that the node model, concrete material and pumping equipment are kept consistent, the average pouring completion time corresponding to the optimized path of this application is reduced from 172 seconds to 154 seconds, and the maximum time difference of full load between the three node branches is reduced from 52 seconds to 25 seconds. This indicates that after determining the material placement sequence based on the internal flow direction and full load time, the filling progress of each spatial orientation branch can be adjusted according to the actual slurry spread state.
[0080] The average overflow during the fixed path pouring process was 1.8 liters, while the average overflow for the optimized path in this application was 0.7 liters. This result shows that after the volume of the node cavity and the volume increment per unit time are involved in the discharge speed adjustment, the continuous feeding of the material by the placing boom in the area that is already close to full load is reduced.
[0081] Three fixed-path pouring tests recorded three instances of underfilling of branch sections, while the three tests corresponding to the optimized path in this application did not record any underfilling of branch sections. After demolding, the average number of visible voids decreased from 11 to 4. This comparative result corresponds to the changes in the time difference and overflow of the branch section under full load, indicating that adjusting the rod trajectory and discharge speed according to the slurry propagation process can reduce the situation where concrete is concentrated in a single branch section while other branches are under-supplied.
[0082] Example 2:
[0083] S1 includes acquiring BIM model data, parsing the BIM model data to obtain the three-dimensional coordinates of each node branch; extracting the spatial orientation feature vector of each node branch at the intersection position based on the three-dimensional coordinates; determining whether the angle between the spatial orientation feature vector and the vertical vector is less than a preset angle threshold; if the angle between the spatial orientation feature vector and the vertical vector is less than the preset angle threshold, it is determined that there is a gravity dominance effect, and the gravity diversion tendency value is calculated based on the gravity dominance effect; and constructing a node spatial map based on the gravity diversion tendency value.
[0084] In this embodiment, after obtaining the BIM model data of the node to be poured, the geometric entity data and spatial connection data of the node component are first parsed, and the local coordinates in the BIM model are uniformly transformed to the same three-dimensional coordinate system.
[0085] Identify the branch intersection area based on the connection relationship between node components, and extract the center path of each node branch; if the BIM model already contains the branch centerline, directly read the coordinates of the control points on the centerline; if the centerline is not directly recorded, continuously cut cross sections along the extension direction of the node branch, calculate the geometric center of each cross section, and connect the geometric centers according to the extension order of the branch to form the center path; record the coordinate points on the center path as horizontal coordinates, vertical coordinates and vertical coordinates in sequence, thereby obtaining the three-dimensional coordinates of each node branch.
[0086] The location of the intersection of branch roads is determined according to the spatial connection relationship of the central paths of each branch road.
[0087] When the central paths of two or more branches enter the same connected cavity, the area enclosed by the intersections of each central path with the boundary of the connected cavity is defined as the convergence area, and the geometric center of this convergence area is taken as the convergence position. For any node branch, taking the convergence position as the starting point, the first central path point that leaves the convergence area is determined along the outward direction of the central path of the node branch, and this central path point is taken as the direction reference point. The horizontal, vertical, and longitudinal coordinates of the convergence position are subtracted from the horizontal, vertical, and longitudinal coordinates of the direction reference point to obtain the position difference in the three coordinate directions. The three position differences are then squared and summed, and the square root of the result is taken to obtain the spatial distance from the convergence position to the direction reference point. Finally, the position differences in the three coordinate directions are divided by this spatial distance, and the three values together constitute the spatial orientation feature vector of the node branch at the convergence position.
[0088] Process each node branch in the manner described above, so that the spatial orientation feature vectors all point outward from the intersection position.
[0089] After obtaining the spatial orientation feature vector, the vertical direction is determined based on the vertical coordinate axis of the BIM model and the gravity direction of the construction site. When the vertical coordinate axis of the BIM model is consistent with the gravity direction of the site, the vertically downward direction is directly used as the vertical direction; when there is a coordinate transformation relationship between the two, the gravity direction of the site is transformed to the BIM model coordinate system according to the BIM model coordinate transformation matrix, and then the transformed vertically downward direction is used as the vertical direction.
[0090] Then, the angle between the spatial orientation feature vector and the vertical direction is calculated.
[0091] Specifically, the three directional components of the spatial orientation feature vector in the horizontal, vertical, and longitudinal directions are multiplied by the three directional components in the vertical direction, respectively. The three products are added together, and the resulting value is divided by the product of the length of the spatial orientation feature vector and the length of the vertical vector. Then, the inverse cosine operation is performed on the resulting ratio to obtain the vector angle corresponding to the branch of the node. Since the aforementioned spatial orientation feature vector has been normalized, the length of the spatial orientation feature vector is taken as 1 in the actual calculation.
[0092] The preset angle threshold is determined by calibration before the node space diagram is constructed. During calibration, a single branch flow model is established using the same concrete density, rheological parameters and design pumping pressure as the project to be poured. The angle between the branch center axis and the vertical direction is increased from 0 degrees to 90 degrees in increments of 1 degree. For each angle, the pressure effect of gravity along the branch axis and the pressure effect of pumping pressure along the branch axis are calculated.
[0093] The pressure exerted by gravity along the branch axis is determined by the concrete density, gravitational acceleration, and the vertical height difference between the branch inlet and outlet, i.e., by multiplying the absolute values of the concrete density, gravitational acceleration, and vertical height difference in sequence. The pressure exerted by pumping pressure along the branch axis is determined by the difference between the branch inlet pressure and the branch outlet pressure. The two pressure effects are compared one by one. When the axial pressure exerted by gravity is not less than the axial pressure exerted by pumping pressure, the corresponding angle is recorded as the gravity-dominated state. When the axial pressure exerted by gravity is less than the axial pressure exerted by pumping pressure, the corresponding angle is recorded as the non-gravity-dominated state. The calibration results are sequentially retrieved from 0 degrees to 90 degrees, and the last angle that continuously satisfies the gravity-dominated state is determined as the preset threshold. This threshold remains unchanged under the same concrete parameters and design pumping pressure. When the concrete parameters or design pumping pressure change, the threshold is re-determined according to the above calibration process.
[0094] After the threshold calibration is completed, the vector angle of each node branch is compared with the preset threshold.
[0095] Specifically, when the vector angle is less than a preset angle threshold, the corresponding node branch is marked as having a gravity-dominant effect; when the vector angle is greater than or equal to the preset angle threshold, the corresponding node branch is marked as not having a gravity-dominant effect. Through the above determination, branches with only spatial tilt characteristics are distinguished from branches whose axial flow is dominated by gravity, and the subsequent gravity diversion tendency value is calculated according to the determination result.
[0096] For node branches where gravity is determined to be dominant, the amount of gravity acting in the branch extension direction is calculated based on the angle between the spatial orientation feature vector and the vertical direction.
[0097] In the calculation, the cosine value of the angle between the vectors is obtained, and the absolute value of the cosine value is multiplied by the gravitational acceleration to obtain the amount of gravity along the axial direction of the node branch. Then, the directional relationship between the outward extension direction of the node branch and the vertical downward direction is determined. When the node branch extends downward from the confluence position, the corresponding axial amount is recorded as a positive value; when the node branch extends upward from the confluence position, the corresponding axial amount is recorded as a negative value. In this way, the obtained value reflects both the direction and amount of gravity along the axial direction of the branch.
[0098] For each node branch connected at the same junction, the absolute values of the axial forces corresponding to the branches determined to be dominated by gravity are added together to obtain the total gravity force at that junction. Then, the absolute value of the axial force corresponding to any node branch is divided by the total gravity force to obtain the diversion weight of that node branch at the same junction. When a node branch extends downward, the diversion weight is retained as a positive value, and when a node branch extends upward, the diversion weight is recorded as a negative value. The final value is used as the gravity diversion tendency value of that node branch. For node branches that do not meet the determination conditions of gravity dominance, their gravity diversion tendency value is recorded as 0.
[0099] After adopting this calculation method, the gravity diversion tendency values of each node branch at the same intersection location have a unified calculation benchmark. The numerical sign indicates the direction of gravity on the concrete flow in the branch, and the absolute value of the numerical value indicates the gravity diversion weight of the branch at the intersection location.
[0100] After obtaining the gravity diversion tendency values of each node branch, a node space diagram is constructed.
[0101] The intersection of branch roads is set as a graph node, and the spatial passages extending from the central path of each node branch road to the next intersection or the branch road termination point are set as graph edges. When there are more than two intersections within the same node, corresponding graph nodes and graph edges are established sequentially according to the actual branch road connection relationship recorded in the BIM model. Each graph edge stores the three-dimensional coordinates, spatial orientation feature vector, vector angle, gravity dominance determination result, and gravity diversion tendency value of its corresponding node branch road. The connection direction of the graph edge is set according to the direction of concrete entering the node branch road from the intersection, so that the node spatial graph simultaneously records the spatial position, topological connection relationship, and gravity diversion state of the node branch road.
[0102] Through the above processing, the node geometric data in the BIM model is first converted into the three-dimensional coordinates of each node branch. The spatial orientation feature vector is determined by the intersection position and the center path of the branch. Then, the gravity dominance is determined according to the angle between the spatial orientation feature vector and the vertical direction. Based on the axial gravity action and the total gravity action of each branch at the same intersection position, the gravity diversion tendency value is obtained. Finally, the obtained value is written into the graph edge attribute of the corresponding node branch to form a node spatial map for subsequent flow field simulation.
[0103] Example 3:
[0104] S2 includes acquiring initial pumping pressure and gravity data of concrete; constructing a node space diagram based on the initial pumping pressure and gravity data to obtain a set of spatial grid nodes; solving the spatial grid node set using computational fluid dynamics algorithms to obtain a set of node velocity vectors; calculating the node pressure gradient matrix based on the node velocity vector set; calculating local shear stress data if the node pressure gradient matrix exceeds a preset pressure gradient threshold; and performing flow field simulation within the node space diagram based on the local shear stress data and the node pressure gradient matrix to obtain a slurry stress distribution diagram.
[0105] In this embodiment, the initial pumping pressure data and gravity data of the concrete are obtained.
[0106] Specifically, the initial pumping pressure data is taken from the pumping pipeline pressure data before the concrete enters the node to be poured, and it is used as the inlet pressure condition when the concrete enters the node spatial diagram; the gravity data includes the gravitational acceleration and gravity direction at the construction location, where the gravity direction is converted to a vertically downward spatial direction according to the coordinate system used in the BIM model; the initial pumping pressure data is loaded to the corresponding feeding position in the node spatial diagram, and the gravity data is loaded to the flow area corresponding to the node cavity and each node branch, so that the spatial position in the node spatial diagram simultaneously has geometric connection relationship, inlet pressure condition and gravity action condition.
[0107] Mesh division is performed based on the internal boundaries of the node cavities and node branches corresponding to the node space diagram.
[0108] Specifically, the inner surfaces of the node cavities and the inner surfaces of the node branches recorded in the BIM model are used as the fluid calculation boundaries. Continuous grid cells are divided within the space enclosed by these boundaries, and the vertices or center of each grid cell are determined as spatial grid nodes. Each spatial grid node records its three-dimensional coordinates, its associated node branch, adjacent grid nodes, and boundary type. Grid nodes located at the feed position record the initial pumping pressure, grid nodes located on the inner walls of the cavities and branches record the wall boundary conditions, and the remaining grid nodes are used as internal fluid nodes. The set of spatial grid nodes is composed of all spatial grid nodes and their adjacent connections.
[0109] The grid size is determined based on the minimum internal cross-sectional size of the node branch.
[0110] Read the internal cross-sectional dimensions of each node branch in the BIM model, and use the smallest internal cross-sectional dimension as the basis for determining the grid scale. Divide the smallest internal cross-section into at least 10 continuous grid units in any cross-sectional direction. At branch intersections, branch bends, and locations where the cross-sectional dimensions change, further subdivide the resulting grid by one level. After completing the grid division, compare the flow velocity calculation results at the same spatial location before and after subdivision. When the flow velocity difference obtained from two consecutive grid subdivisions reaches the preset numerical convergence condition, the current grid is determined as the computational grid used by the spatial grid node set.
[0111] After obtaining the set of spatial grid nodes, a computational fluid dynamics algorithm is used for solving the problem. The density, yield shear stress, and plastic viscosity of concrete are input into the fluid calculation model, the initial pumping pressure is used as the inlet boundary, and gravity is used as the external force acting on the concrete volume. For any internal grid node, mass conservation calculation is performed based on the mass inflow and mass outflow between the node and its adjacent grid nodes, and momentum conservation calculation is performed based on the pumping pressure, gravity, viscous resistance, and momentum transfer relationship between nodes. Thus, the velocity components of the grid node along the three coordinate directions are obtained.
[0112] The same process is applied to all internal grid nodes. When the velocity changes of each grid node obtained from two consecutive iterations meet the pre-set numerical convergence conditions, the current velocity field iteration is stopped. The three directional velocity components of each spatial grid node are stored in correspondence with their three-dimensional coordinates to form a set of node velocity vectors.
[0113] The nodal pressure gradient matrix is calculated by inversely based on the set of nodal velocity vectors and the momentum conservation relationship.
[0114] For any spatial grid node, its adjacent grid nodes in the horizontal, vertical, and longitudinal directions are obtained, and the velocity components corresponding to the current grid node and its adjacent grid nodes are read. First, the velocity difference between adjacent grid nodes in the corresponding directions is divided by the actual distance between the two nodes to obtain the rate of change of velocity in that spatial direction. Then, the inertial and viscous effects of the grid node are calculated based on the rate of change of velocity in different directions, and the gravity component acting on the node is read. According to the momentum conservation relationship, the total amount of force required to drive the flow of the grid node is subtracted from the inertial, viscous, and gravity effects to obtain the pressure changes in the horizontal, vertical, and longitudinal directions. These changes are then divided by the spatial distance between adjacent grid nodes in the corresponding directions to obtain the pressure gradient components in the three directions.
[0115] Perform the above calculations on all spatial grid nodes, and arrange the pressure gradient components in three directions according to the grid node number to obtain the node pressure gradient matrix.
[0116] To determine whether the pressure change has reached a level that causes a change in the local shear state of the concrete, a pressure gradient threshold is determined before performing flow field simulation.
[0117] The pressure gradient threshold is calibrated using a sample with the same mix proportion as the actual poured concrete. First, the yield shear stress and plastic viscosity of the concrete are obtained through rheological testing. Then, a calibration channel is established with cross-sectional dimensions corresponding to the branch node to be analyzed, using the same meshing method and fluid parameters as the node space diagram. Keeping the outlet pressure of the calibration channel constant, the inlet pressure is gradually increased from the set inlet pressure, calculating the flow velocity and shear stress within the channel after each increase. When the calculated shear stress of a continuous grid node in the channel first reaches the concrete yield shear stress, forming a continuous flow path from the inlet to the outlet, the pressure difference between the inlet and outlet is read. This pressure difference is then divided by the length of the flow path from the inlet to the outlet, and the resulting pressure change rate is used as the pressure gradient threshold. The calibration process is repeated whenever the concrete mix proportion, yield shear stress, or branch node cross-sectional conditions change.
[0118] After the nodal pressure gradient matrix is formed, for each spatial grid node, the pressure gradient components in the three directions are squared respectively. The three squared results are added together and the square root is taken as the pressure gradient value of the spatial grid node. This pressure gradient value is then compared with a preset pressure gradient threshold.
[0119] When the pressure gradient value does not exceed the preset threshold, the pressure gradient and velocity data of the grid node are retained; when the pressure gradient value exceeds the preset threshold, the grid node is designated as a shear stress calculation node, and local shear stress data is calculated at that location.
[0120] Local shear stress is determined based on the velocity variation of the shear stress calculation node and its adjacent grid nodes.
[0121] Specifically, the velocity components of adjacent grid nodes in the three coordinate directions are read respectively. The velocity difference between adjacent nodes is divided by the actual spatial distance between the two nodes to obtain the rate of change of velocity along each spatial direction. The velocity change rates of the intersection directions that produce shear deformation are combined according to the shear direction to obtain the local shear rate at that location. Then, the local shear rate is multiplied by the plastic viscosity of the concrete and added to the yield shear stress of the concrete. The result is used as the local shear stress of the corresponding spatial grid node. The local shear stress is then correlated with the three-dimensional coordinates, pressure gradient, and node velocity vector of the grid node to form local shear stress data.
[0122] After the local shear stress data is calculated, it is mapped together with the nodal pressure gradient matrix to the nodal space diagram.
[0123] For each spatial grid node, the corresponding pressure gradient, flow velocity direction, flow velocity value, and local shear stress are recorded. For spatial grid nodes whose pressure gradient does not exceed the preset threshold, the pressure gradient and flow velocity calculation results are recorded, and their corresponding stress state is maintained according to the current rheological parameters. Then, using the initial pumping pressure as the inlet driving force and gravity as the volume force, the pressure transmission, velocity transmission, and shear state of concrete in the node cavity and each node branch are continuously iterated according to the connection relationship between each spatial grid node in the node spatial diagram.
[0124] During the flow field simulation, once the pressure and velocity of a spatial grid node are calculated, the motion state of the adjacent nodes is updated based on the pressure gradient and local shear stress between that node and its neighboring nodes, and then transmitted to subsequent grid nodes according to the connectivity in the node spatial diagram.
[0125] Each round of calculation redetermines the flow velocity, pressure gradient, and shear stress of each grid node until the results of two consecutive rounds of calculation satisfy the numerical convergence condition of the flow field. Finally, the pressure gradient, gravity, local shear stress, flow velocity, and flow direction corresponding to each location are summarized according to the three-dimensional coordinates of the spatial grid nodes to generate a slurry stress distribution map.
[0126] Therefore, the initial pumping pressure and gravity data are first loaded into the spatial grid node set corresponding to the node space diagram. The node velocity vector set is obtained through computational fluid dynamics algorithm. Then, the pressure gradient in each direction is calculated by using the velocity change between nodes, the momentum conservation relationship, and the effect of gravity. When the pressure gradient exceeds the threshold obtained by calibration based on the actual yield state of concrete, the shear rate is calculated based on the local velocity change. The local shear stress is obtained by combining the yield shear stress and plastic viscosity. Finally, the pressure gradient and local shear stress are returned to the node space diagram for flow field iteration to obtain the slurry stress distribution diagram used for subsequent node branch velocity change judgment.
[0127] Example 4:
[0128] S3 includes extracting each node branch based on the slurry stress distribution diagram and node space diagram, solving each node branch to obtain the cross-sectional flux; determining the velocity change by comparing the initial flow velocity and the real-time flow velocity based on the cross-sectional flux; determining whether the velocity change is greater than a preset velocity threshold; if the velocity change is greater than the preset velocity threshold, obtaining the deflection angle of each node branch; and generating internal flow direction data by vector synthesis based on the deflection angle and the node branches.
[0129] In this embodiment, the slurry stress distribution diagram and the node space diagram are first mapped to each other in three-dimensional coordinates. Based on the branch confluence positions, node branch boundaries and branch connection relationships recorded in the node space diagram, the flow field region corresponding to each node branch is determined in the slurry stress distribution diagram.
[0130] For each node branch, starting from the branch intersection, select a position one grid cell away from the intersection along the extension direction of the node branch centerline, and establish a section perpendicular to the node branch centerline at this position. The area enclosed by this section and the inner wall of the node branch is determined as the flux calculation section.
[0131] When solving for the flux calculation section, read the flow velocity and grid area of each grid cell contained in the section.
[0132] For any grid cell, first determine the component of the flow velocity along the normal direction of the cross section, then multiply the flow velocity component by the area of the grid cell to obtain the volume of concrete passing through the grid cell per unit time. Calculate the flux of all grid cells within the cross section in the same way, and add up the concrete volumes of each grid cell per unit time. The result is taken as the cross section flux of the node branch at the current moment. Therefore, the lateral movement component of the slurry along the cross section direction is not included in the cross section flux; only the concrete flow rate actually passing through the cross section of the node branch is counted.
[0133] Both the initial and real-time flow velocities are calculated from the cross-sectional flux. After the concrete enters the node, the computational fluid dynamics algorithm described in Example 3 is used for the first flow field solution. When the node flow velocities obtained from two consecutive iterations meet the predetermined numerical convergence conditions, the cross-sectional flux at the corresponding time is read, and the cross-sectional flux is divided by the area of the flux calculation cross section. The resulting flow velocity is used as the initial flow velocity of the node branch. The cross-sectional flux of the node branch is recalculated at each subsequent sampling time, and the same cross-sectional area is used for conversion to obtain the real-time flow velocity at the corresponding time. The initial flow velocity is subtracted from the real-time flow velocity, and the absolute value of the difference is taken as the flow velocity change of the current node branch.
[0134] The preset flow rate threshold is obtained through stable pumping calibration before the pouring path is calculated.
[0135] Specifically, during calibration, the same concrete mix proportion, initial pumping pressure, node spatial structure, and mesh generation parameters as in actual construction are used. Under the condition that the pumping pressure remains at the set value and the flow direction of each node branch does not change, the real-time flow velocity at 20 sampling times is continuously acquired. The absolute difference between the real-time flow velocity and the initial flow velocity at each sampling time is calculated, and the maximum value is selected from the 20 absolute differences. Then, the maximum flow velocity difference generated by two consecutive mesh subdivisions in the mesh convergence calculation of Example 3 is obtained. The two maximum differences are added together, and the result is determined as the flow velocity threshold. When the concrete mix proportion, initial pumping pressure, or node branch cross-sectional dimensions change, the above calibration process is repeated to redetermine the corresponding flow velocity threshold.
[0136] Subsequently, the velocity change of each node branch is compared with the velocity threshold; when the velocity change is less than or equal to the velocity threshold, the original flow direction data of the node branch is maintained; when the velocity change is greater than the velocity threshold, it is determined that the node branch meets the conditions for recalculating the flow direction, and the deflection angle of the node branch is further obtained.
[0137] The deflection angle is calculated based on the geometric orientation of the node branch and the actual flow direction of the concrete in that branch.
[0138] First, the centerline extension direction of the node branch at the branch confluence position is read from the node space diagram, and this direction is used as the reference direction of the branch. Then, the velocity direction and velocity value of each grid cell in the flux calculation section are read from the slurry stress distribution diagram. For each grid cell, the velocity of the grid cell is decomposed into three coordinate directions in the three-dimensional coordinate system, and the velocity component in each coordinate direction is multiplied by the cross-sectional flux of the grid cell. The products of all grid cells in the same coordinate direction are accumulated and then divided by the total cross-sectional flux of the flux calculation section to obtain the average flow components in the three coordinate directions. The actual flow direction of the node branch at the current moment is determined by the three average flow components.
[0139] After obtaining the actual flow direction, calculate the spatial angle between the actual flow direction and the reference direction of the node branch.
[0140] Specifically, the corresponding components of the two directions in the three coordinate directions are multiplied respectively, and the three products are added together. The result is divided by the product of the lengths of the two directions, and the resulting ratio is processed by inverse cosine to obtain the angle between the two directions. This angle is used as the deflection angle of the branch at the node. At the same time, the deflection direction is determined according to the rotation direction of the actual flow direction relative to the branch reference direction in the three-dimensional coordinate system, so that the deflection angle has both a clear angle value and a spatial direction.
[0141] After obtaining the deflection angle of each node branch, vector synthesis is performed on the node branches connected at the same branch intersection position. For any node branch, the flow direction of the node branch is first corrected according to its reference direction, deflection angle, and deflection direction, and then the cross-sectional flux of the node branch is used as the numerical value of the flow vector. For node branches flowing towards the branch intersection position, the flow vector direction is set from the node branch to the intersection position. For node branches flowing out of the branch intersection position, the flow vector direction is set from the intersection position to the node branch.
[0142] After unifying the directions, the flow vector of each node branch is decomposed into three coordinate directions of the BIM model. The vector components in the same coordinate direction are added together, and the composite flow direction at the intersection of the branches is determined by the three accumulated results.
[0143] The composite flow direction characterizes the internal flow direction of the slurry after passing the junction of the branches. The cross-sectional flux, velocity change, deflection angle, deflection direction of each branch node, and composite flow direction together form the internal flow direction data.
[0144] When the node space diagram contains more than two branch intersections, each branch intersection is processed sequentially according to the connection order of concrete from the feed position to the subsequent node branches. The synthetic flow direction obtained from the previous branch intersection is used as the upstream flow direction information of the next branch intersection. The cross-sectional flux calculation, velocity change judgment, deflection angle calculation and vector synthesis are then completed to obtain the internal flow direction data of concrete continuously transmitted between branches within the node.
[0145] Example 5:
[0146] S4 includes acquiring internal flow direction data, extracting the current flow rate and maximum carrying capacity of each node branch from the internal flow direction data; calculating the available remaining space based on the current flow rate and maximum carrying capacity; obtaining the predicted arrival time through the available remaining space; marking the predicted arrival time as the target time node; determining the time node when each node branch reaches the full load state based on the target time node; and generating a filling sequence based on the time node.
[0147] In this embodiment, the generated internal flow direction data is first obtained, and the flow direction, cross-sectional flux and sampling time corresponding to each node branch are extracted according to the node branch identifier.
[0148] For node branches whose flow direction is from the branch confluence point to the inside of the node branch, the cross-sectional flux corresponding to the current sampling time is taken as the current flow rate of the node branch; for the case where the flow direction is from the node branch to the branch confluence point, the cross-sectional flux corresponding to that direction is taken as the outflow and is not included in the current filling amount of the node branch.
[0149] Therefore, the current flow rate represents the actual volume of concrete entering the node branch per unit time at the current sampling moment.
[0150] The maximum carrying capacity of each node branch is determined by the internal geometry of the branch in the BIM model.
[0151] Specifically, the closed area defined by the inner wall of the node branch is read, and its internal volume is calculated according to the actual three-dimensional geometric boundary of the node branch. For node branches with constant cross-sectional dimensions, the effective length of the node branch is multiplied by the corresponding internal cross-sectional area to obtain the internal volume. For node branches with cross-sectional dimensions that change along the extension direction of the branch, continuous segments are divided along the centerline of the branch, and the volume of the space enclosed by the cross-sections at both ends of each segment is calculated. The volumes of each segment are then summed. The summed result is used as the maximum carrying capacity of the node branch, and a data association is established based on the node branch number and the corresponding current flow rate.
[0152] When calculating the available remaining space based on the current flow rate and maximum carrying capacity, it is also necessary to determine the volume of concrete that has already entered the branch of that node at the current moment.
[0153] The sampling time when concrete first flows into the node branch in the internal flow direction data is taken as the filling start time. The flow rate between two adjacent sampling times is read in the sampling order. For each sampling time period, the average flow rate at the start and end of the time period is multiplied by the time interval between two sampling times to obtain the volume of concrete entering the node branch in that time period. The concrete volumes corresponding to each time period from the filling start time to the current sampling time are accumulated sequentially to obtain the current filled volume of the node branch. The current filled volume is then subtracted from the maximum bearing capacity to determine the available remaining space of the node branch. When the current filled volume reaches the maximum bearing capacity, the available remaining space is determined to be 0.
[0154] After obtaining the available remaining space, calculate the time required for the corresponding node branch to reach full load based on the current flow rate.
[0155] When the current flow rate is greater than 0, the available remaining space is divided by the current flow rate, and the resulting time is taken as the remaining filling time required for the node branch to reach its maximum carrying capacity from the current filling state. The remaining filling time is then added to the time corresponding to the current sampling time, and the resulting time is taken as the predicted arrival time. When the current flow rate is equal to 0, the remaining filling time division calculation is not performed, and the predicted arrival time corresponding to the previous sampling time of the node branch is retained until the internal flow direction data again indicates that concrete has entered the node branch and then the calculation is recalculated.
[0156] During continuous sampling, for each new set of internal flow direction data obtained, the current flow rate and the current filled volume are recalculated, and the predicted arrival time is re-determined based on the updated available remaining space.
[0157] The concrete volume that has been accumulated into the node branch at the previous sampling time is directly retained and does not participate in the volume integration of subsequent time periods, so that the prediction results always take the current actual filling state as the starting point for calculation.
[0158] The predicted arrival time calculated for each node branch is marked as the target time node for that node branch. For node branches whose current filled volume has not yet reached the maximum carrying capacity, the target time node is the predicted arrival time calculated in the current sampling period. For node branches whose current filled volume has reached the maximum carrying capacity, the sampling time when the cumulative filled volume first reaches the maximum carrying capacity is taken as the target time node, and the prediction arrival time of that node branch is stopped from being updated.
[0159] The actual time point when each node branch reaches full load is determined based on the target time point; the filled volume of each node branch is continuously updated; when the filled volume of a certain node branch at the previous sampling time is less than the maximum carrying capacity, while the filled volume at the current sampling time reaches the maximum carrying capacity, time interpolation is performed between the two sampling times.
[0160] Specifically, the maximum bearing capacity is first subtracted from the volume already filled at the previous sampling time to obtain the volume of concrete that still needs to be filled from the previous sampling time. Then, this volume is divided by the average flow rate corresponding to the two sampling times to obtain the time required from the previous sampling time to reach the full load state. This time is added to the previous sampling time to obtain the actual time node when the branch reaches the full load state. This process avoids directly limiting the full load time to the discrete sampling time.
[0161] Once the time nodes of all node branches are determined, the corresponding node branches are arranged in order of their time nodes from first to last, forming a filling sequence. When the time nodes of two node branches are different, the node branch with the earlier time node is used as the preceding branch. When the time nodes of two node branches are the same, the first inflow time of both is read from the internal flow data, and the node branch with the earlier first inflow time is used as the preceding branch. When both the time node and the first inflow time are the same, the node branches are arranged in the order of their pre-determined node branch numbers in the node space graph.
[0162] By using the above sorting method, any node branch can obtain a definite filling order.
[0163] Therefore, the cross-sectional flux in the internal flow data is first converted into the current flow rate of the node branch, and the internal volume of the branch in the BIM model is used as the maximum carrying capacity. The current filled volume is obtained by accumulating the inflow volume of each sampling time period, and the available remaining space is obtained by subtracting the current filled volume from the maximum carrying capacity. The predicted arrival time is calculated based on the available remaining space and the current flow rate to form the target time node. After the node branch actually reaches the maximum carrying capacity, the corresponding full load time node is determined. Finally, the filling sequence is generated according to the order of the time nodes.
[0164] Example 6:
[0165] S5 includes obtaining the spatial coordinates corresponding to the filling sequence to construct an initial spatial coordinate set; calculating the full-load time interval of adjacent node branches based on the initial spatial coordinate set; if the full-load time interval is greater than a preset time threshold, adjusting the memory weight of the long short-term memory network using the full-load time interval, and performing time-series prediction on the initial spatial coordinate set through the long short-term memory network after adjusting the memory weight to obtain the predicted coordinate sequence; and splicing the historical sequence based on the predicted coordinate sequence to obtain the propagation process trajectory.
[0166] In this embodiment, the obtained filling order sequence is first read, and the BIM model is accessed according to the arrangement order of the node branches in the sequence.
[0167] For each node branch, the full-load judgment section corresponding to the node branch when it reaches full load is used as the coordinate extraction position. The horizontal, vertical and longitudinal coordinates of the geometric center of the section in the unified coordinate system of the BIM model are read, and the three coordinate values are associated with the node branch number and the full-load time node. The spatial coordinates of each node branch are arranged from front to back according to the filling sequence. The resulting ordered coordinate data constitutes the initial spatial coordinate set, so that the coordinate order in the initial spatial coordinate set is consistent with the actual concrete filling sequence.
[0168] The full-load time interval of adjacent node branches is calculated item by item according to the filling order sequence.
[0169] Specifically, the full-load time nodes of two adjacent node branches in the sequence are read, and the full-load time node of the previous node branch is subtracted from the full-load time node of the next node branch. The resulting time length is used as the full-load time interval between the two node branches. All adjacent node branches in the filling sequence are processed in the same way to ensure that there is a corresponding full-load time interval between each pair of adjacent spatial coordinates.
[0170] The preset time threshold is determined before the Long Short-Term Memory network performs the current node prediction.
[0171] Select historical normal pouring records that use the same node structure, concrete mix ratio, and initial pumping pressure setting as the node to be poured. Calculate the full-load time interval of adjacent node branches in each historical pouring according to the method described in Example 5. Simultaneously, call the internal flow direction data from Example 4, retaining only historical records where the flow velocity change does not exceed the corresponding flow velocity threshold and the filling order is consistent with the connection direction of the node space diagram. Extract all full-load time intervals of adjacent node branches from the retained historical records, and determine the largest time interval as the time threshold. If the node structure, concrete mix ratio, or initial pumping pressure setting changes, re-determine the time threshold according to the above process.
[0172] After obtaining the time threshold, the full-load time interval of each adjacent node branch in the current filling sequence is compared with the time threshold.
[0173] When the full load time interval is less than or equal to the time threshold, the Long Short-Term Memory network uses the original memory weights saved after training for temporal operations; when the full load time interval is greater than the time threshold, the memory weights are adjusted using the current full load time interval.
[0174] The process of adjusting the memory weight is as follows: divide the time threshold by the current full-load time interval, and use the resulting value as the memory weight correction coefficient.
[0175] Since the current full load time interval is greater than the time threshold when the adjustment is triggered, the memory weight correction coefficient is between 0 and 1. The original memory weights used by the Long Short-Term Memory Network to retain the historical memory state of the previous moment at the current prediction position are read. The original memory weights are multiplied by the memory weight correction coefficients, and the result is used as the adjusted memory weights at the current prediction position. The adjusted memory weights are then used to weight the historical memory state of the previous moment.
[0176] Therefore, the full-load time interval directly participates in the calculation of the retention of historical filling state; when the full-load time interval increases, the retention weight of the historical state of the previous node branch in the prediction of the next node branch changes accordingly according to the above calculation results.
[0177] The Long Short-Term Memory (LSTM) network is pre-trained using data from nodes that have already been poured.
[0178] Training samples are generated using the same method as the initial spatial coordinate set, with the spatial coordinates of the preceding node branches arranged in the actual filling order and their full-load time intervals as input, and the spatial coordinates of the subsequent node branches as the output target. After training, the network parameters and original memory weights are saved. When performing current node prediction, the initial spatial coordinate set is input into the network item by item in the filling order, along with the full-load time intervals corresponding to adjacent node branches. For positions exceeding the time threshold, the adjusted memory weights are used; for positions not exceeding the time threshold, the original memory weights are used.
[0179] In each temporal prediction, the Long Short-Term Memory Network outputs the horizontal, vertical, and longitudinal coordinates of the next prediction position based on the current input spatial coordinates, the previous memory state, and the corresponding memory weights.
[0180] The three coordinate values are combined to form a predicted spatial coordinate, which is then used as the input data for the next prediction time. This process is repeated to complete the subsequent spatial location predictions. The predicted spatial coordinates are arranged in chronological order to form a predicted coordinate sequence.
[0181] To ensure that the predicted coordinates remain within the actual fillable space of the node, each predicted spatial coordinate is matched with the node spatial map. If the predicted spatial coordinate falls within the effective flow area of the node cavity or node branch, the coordinate is retained. If the predicted spatial coordinate is outside the effective flow area, the effective grid node with the smallest distance from the predicted spatial coordinate is found in the node spatial map, and the three-dimensional coordinates of the effective grid node are used to replace the original predicted coordinates.
[0182] After the above processing, all coordinates in the predicted coordinate sequence are located within the concrete infillable space defined by the BIM model.
[0183] The historical sequence consists of the spatial coordinates of the nodes that have reached full load before the current prediction time, and is arranged in chronological order of the actual full load time nodes.
[0184] When performing sequence splicing, the last spatial coordinate of the historical sequence is read first, followed by the first spatial coordinate of the predicted coordinate sequence. When the two correspond to the same node in the spatial graph, the first spatial coordinate in the predicted coordinate sequence is deleted to avoid duplicate connection positions. When the two correspond to different grid nodes, the predicted coordinate sequence is directly connected to the end of the historical sequence. Subsequently, the remaining predicted coordinates are appended sequentially according to the prediction time order.
[0185] After the splicing is completed, the actual full-load time nodes in the historical sequence and the predicted time in the predicted coordinate sequence are used together as time indexes to arrange all spatial coordinates in sequence and connect them sequentially according to the connectivity of adjacent coordinates in the node space graph.
[0186] The resulting continuous spatial trajectory is the propagation process trajectory. The first part of the trajectory corresponds to the actual filling process that has already occurred, and the second part of the trajectory corresponds to the subsequent propagation position predicted by the long short-term memory network based on the full load time interval and spatial coordinates. This provides a definite spatial position and temporal sequence for the calculation of the external fabric rod movement vector.
[0187] Example 7:
[0188] S6 includes acquiring the propagation process trajectory, calculating the initial displacement sequence based on the propagation process trajectory, determining the movement vector based on the initial displacement sequence, solving for the member attitude and joint angle based on the movement vector, determining whether the joint angle exceeds the angle threshold, and if the joint angle exceeds the angle threshold, correcting the member attitude to obtain the target attitude, generating control commands based on the target attitude, calculating the movement vector of the external material rod based on the propagation process trajectory using the control commands, and generating the member trajectory using the movement vector.
[0189] In this embodiment, the propagation process trajectory obtained in Embodiment 6 is acquired, and the three-dimensional spatial coordinates are arranged in chronological order according to the time sequence of each trajectory point.
[0190] For two adjacent trajectory points, subtract the coordinates of the preceding trajectory point from the horizontal, vertical, and axial coordinates of the following trajectory point to obtain displacement components in three coordinate directions. Square each of the three displacement components, sum them, and then take the square root of the sum to obtain the spatial displacement distance between adjacent trajectory points. Perform the calculation on all adjacent trajectory points in the order of their arrangement, and store the displacement components, spatial displacement distances, and corresponding start and end coordinates of each segment in sequence to form an initial displacement sequence.
[0191] When determining the movement vector based on the initial displacement sequence, the starting point corresponding to each displacement segment is taken as the vector starting point, and the corresponding ending point is taken as the vector ending point. The displacement components in the three coordinate directions are divided by the spatial displacement distance of that segment to obtain the direction components of the movement direction in the three coordinate directions. A correspondence is established between these direction components and the spatial displacement distance to obtain the movement vector for that trajectory segment. When the spatial displacement distance between adjacent trajectory points is 0, the subsequent duplicate trajectory point is deleted, and no movement vector is generated for that position.
[0192] After obtaining the movement vector, read the structural parameters of the external placing rod, including the installation position of the placing rod base in the BIM coordinate system, the length of each rod section, the rotation direction of each joint, the connection order of the rods, and the current joint angle.
[0193] The three-dimensional coordinates corresponding to the endpoint of the movement vector are transformed to the equipment coordinate system with the base of the material placing rod as the reference, and the transformed coordinates are determined as the target position of the material placing rod's discharge end. Based on the length of each rod section, the direction of the joint rotation axis, and the current joint angle, the spatial position of each rod endpoint and the rod axis are calculated section by section from the base to the discharge end, thereby obtaining the current rod posture and the current discharge end coordinates.
[0194] The joint angles are solved using inverse kinematics. Using the current joint angle as the initial calculation value, the coordinate differences between the target position and the current discharge end in three coordinate directions are calculated, and the spatial distance between the discharge end and the target position is determined based on the three coordinate differences.
[0195] Then, the angles of each joint are adjusted in sequence from the base side joint to the discharge end side joint. Each time, the angle of each joint is changed by one step according to the minimum angle control step specified by the material placing rod controller. After the change, the position of each rod section and the coordinates of the discharge end are recalculated. When the spatial distance between the adjusted discharge end and the target position is less than the spatial distance before the adjustment, the current adjustment direction is retained; otherwise, the joint is adjusted in the opposite direction. The calculation of each joint is completed in sequence and updated cyclically until the spatial distance between the discharge end and the target position reaches the end positioning error range set by the equipment. The current angle of each joint is used as the solution angle corresponding to the movement vector, and the current position and direction of each rod section constitute the rod posture.
[0196] The joint angle threshold is determined based on the allowable range of the placing boom machinery and the current construction space.
[0197] First, read the minimum and maximum mechanical rotation angles of each joint recorded in the equipment's technical documents to form the allowable mechanical angle range. Then, place the 3D model of the placing boom into the current BIM construction scene. Starting from the minimum mechanical rotation angle, gradually change the corresponding joint angle according to the minimum angle control step size specified by the controller. At each angle position, check whether each section of the boom intersects with the beams, columns, walls, formwork, and other fixed components in the BIM model. The continuous angle range without spatial intersection is determined as the allowable construction angle range. The overlap range between the allowable mechanical angle range and the allowable construction angle range is taken as the final angle threshold range for that joint. After the position of the placing boom base or the position of surrounding components changes, re-determine the joint angle threshold according to the above process.
[0198] Each joint angle obtained from inverse kinematics is compared with its corresponding angle threshold range.
[0199] When the joint angle is within the threshold range, the current solved angle is retained; when the joint angle exceeds the upper limit of the threshold, the joint angle is corrected to the upper limit of the threshold; when the joint angle is below the lower limit of the threshold, the joint angle is corrected to the lower limit of the threshold; after the correction of the over-limit joint is completed, the corrected joint is fixed at the corresponding threshold boundary, and the inverse kinematics calculation is re-executed for the remaining joints so that the material output end of the material rod approaches the end point of the original movement vector again; when all the re-calculated joint angles are within their respective threshold ranges, and the position of the material output end meets the positioning error requirements of the equipment end, the position, direction and joint angle of each section of the rod at this time are determined as the target posture.
[0200] When generating control commands based on the target posture, the current angle of each joint is compared with the target angle to determine the required rotation direction and rotation angle of each joint. The control data is then written according to the joint connection sequence of the fabric rod, and the corresponding propagation process trajectory point number is associated with it.
[0201] The controller drives each joint to rotate to the target angle based on the control data; after the target posture is executed, the forward kinematics is recalculated based on the length of each link and the actual joint angle to obtain the actual three-dimensional spatial coordinates of the material output end of the material bar.
[0202] The actual coordinates of the material delivery end of the material bar are compared with the coordinates of the next target in the propagation process trajectory. The coordinate differences in the three coordinate directions are calculated respectively. Then, the spatial direction and distance of the material delivery end to continue moving are determined based on the coordinate differences, thereby forming the external material bar movement vector after joint angle constraint correction.
[0203] The target attitude solution, joint angle verification, attitude correction and movement vector update are executed sequentially according to the time sequence of the propagation process trajectory, so that each movement vector corresponds to a set of link attitudes that meet the joint angle threshold.
[0204] Finally, the moving vectors are arranged in the order of their propagation trajectory, with the endpoint of the previous moving vector serving as the starting point of the next moving vector, and the coordinates of each actual discharge end are connected in sequence to form the rod trajectory of the external material distribution rod.
[0205] Therefore, the propagation process trajectory is first converted into an initial displacement sequence and a movement vector, and then the member posture and joint angle are obtained based on the structural parameters of the placing boom. For joints that exceed the angle threshold, the member posture is corrected by angle limiting and resolving, and the actual movement vector is re-determined based on the corrected target posture, finally obtaining the member trajectory that satisfies the movement range and construction space constraints of the placing boom.
[0206] Example 8:
[0207] S7 includes acquiring the rod trajectory, calculating the node cavity volume based on the rod trajectory; acquiring the propagation process trajectory, extracting fluid filling data based on the node cavity volume and the propagation process trajectory, and calculating the volume increment; if the volume increment is less than a preset volume threshold, determining the discharge speed based on the fluid filling data; correcting the rod trajectory based on the discharge speed to obtain the target rod trajectory, and generating a concrete pouring path based on the target rod trajectory.
[0208] In this embodiment, after obtaining the generated rod trajectory, the coordinates of each material discharge position in the rod trajectory are matched with the BIM model to determine the node cavity corresponding to each trajectory segment.
[0209] In specific processing, the starting and ending coordinates of the material delivery end of the fabric rod within a certain trajectory segment are used as search conditions. The BIM model is then used to locate node cavities that are simultaneously connected to this trajectory segment, and the three-dimensional enclosed space defined by the inner surface of the node cavity is read. For node cavities with unchanged cross-sectional dimensions, the internal cross-sectional area and effective length of the cavity are read, and the cross-sectional area is multiplied by the effective length to obtain the node cavity volume. For node cavities with cross-sectional changes, bends, or branch intersections, continuous spatial segments are divided along the cavity's central path. The internal volume enclosed between the two ends of each segment is calculated separately, and the volumes of each segment are summed to obtain the total volume of the node cavity. After calculation, the node cavity number, node cavity volume, and corresponding rod trajectory segment are associated.
[0210] Subsequently, the trajectory of the spread process is obtained, and the spatial coordinates in the trajectory are mapped to the corresponding node cavity in chronological order.
[0211] The node cavity is discretized according to the spatial grid used in Example 3. Each grid cell has a defined spatial coordinate and cell volume. For any sampling time, the positions of that time and all previous trajectories are read from the propagation process trajectory, and combined with the branch connection direction in the node spatial diagram, the grid cells that the concrete has reached are determined. The volumes of the reached grid cells are accumulated, and the result is used as the filled volume at that sampling time. The remaining unfilled volume at that sampling time is obtained by subtracting the filled volume from the total volume of the node cavity. The sampling time, the position of the rod trajectory, the filled volume, the remaining unfilled volume, and the corresponding node cavity number are combined to form fluid filling data.
[0212] The volume increment is calculated based on the already filled volume at adjacent sampling times.
[0213] Read the filled volume at the previous sampling time and the filled volume at the next sampling time. Subtract the filled volume at the previous sampling time from the filled volume at the next sampling time to obtain the actual increase in filled volume between the two sampling times. Then divide the increased filled volume by the time length between the two sampling times, and the result is used as the volume increment per unit time. Calculate all adjacent sampling times in the same way, so that each volume increment is associated with the corresponding rod trajectory segment, node cavity, and sampling time period.
[0214] The preset volume threshold is determined before the current node cavity begins to be poured.
[0215] First, read the total volume of the node cavity calculated from the BIM model. Then, obtain the planned start time and planned completion time of the pouring assigned to the node cavity in the construction task. Subtract the planned start time of the pouring from the planned completion time to obtain the planned pouring time of the node cavity. Divide the total volume of the node cavity by the planned pouring time, and the resulting unit time filling volume is used as the volume threshold of the node cavity. When the total volume of the node cavity or the planned pouring time changes, the volume threshold is re-determined according to the same calculation process. After adopting this determination method, the volume threshold directly corresponds to the unit time filling amount required for the current node cavity to complete the pouring within the planned time.
[0216] Compare the volume increment of the current time period with the volume threshold corresponding to the cavity of that node.
[0217] When the volume increment is greater than or equal to the volume threshold, the discharge parameters corresponding to the current rod trajectory segment are retained; when the volume increment is less than the volume threshold, the current filled volume, the remaining unfilled volume and the current sampling time are read from the fluid filling data, and the remaining pouring time is calculated based on the planned pouring completion time.
[0218] Specifically, the remaining available pouring time for the current node cavity is obtained by subtracting the planned completion time from the current sampling time. Then, the remaining volume to be filled is divided by the remaining available pouring time, and the resulting concrete output volume per unit time is determined as the current required discharge rate.
[0219] After obtaining the discharge rate, the equipment's operating range is checked. The minimum stable discharge rate and the maximum allowable discharge rate recorded in the technical parameters of the concrete pumping equipment are read.
[0220] When the calculated discharge rate is between the minimum stable discharge rate and the maximum allowable discharge rate, the calculated discharge rate is directly determined as the target discharge rate. When the calculated discharge rate exceeds the maximum allowable discharge rate, the maximum allowable discharge rate is determined as the target discharge rate, and the remaining discharge duration is recalculated by dividing the remaining volume to be filled by the maximum allowable discharge rate. When the calculated discharge rate is less than the minimum stable discharge rate, the minimum stable discharge rate is determined as the target discharge rate, and the remaining volume to be filled is divided by the minimum stable discharge rate to obtain the discharge duration required to complete the remaining filling of the current node cavity.
[0221] Therefore, the target discharge rate is always within the continuous operating range allowed by the pumping equipment.
[0222] When correcting the rod trajectory according to the determined target discharge speed, the target posture and joint angle constraints determined in Example 7 remain unchanged, and only the movement time, dwell time and discharge duration of each trajectory segment are adjusted.
[0223] For the rod trajectory segment corresponding to the current node cavity, divide the remaining volume to be filled in the node cavity by the target discharge speed to obtain the total discharge time required to complete the remaining filling; then allocate the dwell time of each discharge position according to the volume to be filled corresponding to each trajectory position in the propagation process trajectory, that is, first calculate the volume of the grid to be filled corresponding to a certain discharge position, and then divide the volume of the grid to be filled by the target discharge speed to obtain the discharge duration of the cloth rod at that position.
[0224] The material discharge positions in the rod trajectory are corrected sequentially according to the time sequence of the propagation process trajectory.
[0225] Once the grid corresponding to the current discharge position is filled, the discharge at that position ends, and the material distribution rod is driven to move to the next discharge position according to the next movement vector determined in Example 7. When the cumulative filling volume of all effective grid units in the node cavity reaches the total volume of the node cavity, the discharge process of the trajectory segment corresponding to that node cavity ends, and the next node cavity in the filling sequence is entered. All trajectory segments after correction of discharge speed, dwell time and movement timing are rearranged according to the original spatial connection order to obtain the trajectory of the target rod.
[0226] After the target rod trajectory is formed, the three-dimensional coordinates of each discharge position, the corresponding target posture, the sequence of movement, the target discharge speed, the discharge duration, and the node cavity number are associated.
[0227] The concrete pouring path is formed by sequentially connecting the corresponding target rod trajectory segments according to the order of each node cavity in the filling sequence. This pouring path also limits the spatial movement position of the placing rod and the concrete output parameters corresponding to each position, so that the actual volume increment of the node cavity reaches the filling requirement corresponding to the current planned pouring time.
[0228] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. A method for optimizing concrete pouring paths based on a BIM model, characterized in that, include: S1. Obtain the BIM model data of the nodes, extract the spatial orientation characteristics of each node branch based on the BIM model data, calculate the gravity diversion tendency corresponding to different spatial orientations based on the branch intersection location, and construct the node spatial map based on the gravity diversion tendency. S2. Obtain the initial pumping pressure data and gravity data of concrete, and use computational fluid dynamics algorithm to simulate the flow field of the initial pumping pressure data and gravity data in the nodal space diagram to obtain the force distribution diagram of the slurry. S3. Based on the slurry stress distribution diagram and node space diagram, determine the velocity change in each node branch. If the velocity change is greater than the preset velocity threshold, generate internal flow direction data. S4. Determine the time point when each node branch reaches full load state through internal flow data, and generate a filling sequence based on the time point. S5. Obtain the spatial coordinates corresponding to the filling sequence, adjust the memory weights of the long short-term memory network according to the full load time interval of adjacent node branches, perform temporal prediction on the spatial coordinates, and obtain the propagation process trajectory. S6. Calculate the movement vector of the external material bar based on the propagation process trajectory, and generate the bar trajectory through the movement vector; S7. Obtain the volume of the node cavity corresponding to the rod trajectory. Calculate the volume increment per unit time based on the node cavity volume and the propagation process trajectory. If the volume increment is less than the preset volume threshold, determine the discharge speed. Optimize the rod trajectory based on the discharge speed to obtain the concrete pouring path.
2. The method for optimizing concrete pouring paths based on a BIM model according to claim 1, characterized in that: S1 includes: Obtain BIM model data and parse the BIM model data to obtain the three-dimensional coordinates of each node branch; Extract the spatial orientation feature vector of each node branch at the intersection position based on the three-dimensional coordinates; Determine whether the angle between the spatial orientation feature vector and the vertical vector is less than a preset angle threshold; If the angle between the spatial orientation feature vector and the vertical vector is less than a preset angle threshold, it is determined that gravity has a dominant effect, and the gravity diversion tendency value is calculated based on the dominant effect of gravity. A node space graph is constructed using numerical values of gravity-induced flow tendency.
3. The method for optimizing concrete pouring paths based on a BIM model according to claim 1, characterized in that: S2 includes: Obtain the initial pumping pressure and gravity data of concrete, construct a node space diagram based on the initial pumping pressure and gravity data, and obtain a set of spatial grid nodes; Computational fluid dynamics algorithms are used to solve the spatial grid node set to obtain the set of node velocity vectors; The node pressure gradient matrix is obtained by calculating based on the set of node velocity vectors; If the nodal pressure gradient matrix exceeds the preset pressure gradient threshold, local shear stress data are calculated. Based on local shear stress data and nodal pressure gradient matrix, flow field simulation is performed in the nodal space diagram to obtain the force distribution diagram of the slurry.
4. The method for optimizing concrete pouring paths based on a BIM model according to claim 1, characterized in that: S3 includes: Based on the slurry stress distribution diagram and nodal space diagram, the branches of each node are extracted, and the cross-sectional flux is obtained by solving the branches of each node. The velocity change is determined by comparing the initial velocity with the real-time velocity based on the cross-sectional flux. Determine whether the change in flow rate exceeds a preset flow rate threshold; If the flow rate change is greater than the preset flow rate threshold, the deflection angle of each node branch is obtained. Internal flow direction data is generated by vector synthesis based on the deflection angle and the branches of each node.
5. The method for optimizing concrete pouring paths based on a BIM model according to claim 1, characterized in that: S4 includes: Obtain internal flow data, and extract the current flow rate and maximum carrying capacity of each node branch from the internal flow data; Calculate the available remaining space based on the current turnover rate and maximum carrying capacity; The predicted arrival time can be obtained using the available remaining space; Mark the predicted arrival time as the target time node; Determine the time when each node branch reaches full load based on the target time node, and generate a filling sequence based on the time node.
6. The method for optimizing concrete pouring paths based on a BIM model according to claim 1, characterized in that: S5 includes: Obtain the spatial coordinates corresponding to the filling sequence and construct an initial set of spatial coordinates; Calculate the full-load time interval of adjacent node branches based on the initial spatial coordinate set; If the full load time interval is greater than the preset time threshold, the memory weight of the long short-term memory network is adjusted by the full load time interval, and the initial spatial coordinate set is temporally predicted by the long short-term memory network after adjusting the memory weight to obtain the predicted coordinate sequence. The spread process trajectory is obtained by splicing historical sequences with predicted coordinate sequences.
7. The method for optimizing concrete pouring paths based on a BIM model according to claim 1, characterized in that: S6 includes: Obtain the trajectory of the propagation process, and calculate the initial displacement sequence based on the trajectory. The movement vector is determined based on the initial displacement sequence.
8. The method for optimizing concrete pouring paths based on a BIM model according to claim 7, characterized in that: S6 further includes: The attitude of the link and the joint angles are obtained by solving the translation vector. Determine if the joint angle exceeds the angle threshold. If the joint angle exceeds the angle threshold, correct the member attitude to obtain the target attitude. Generate control commands based on the target attitude. Calculate the movement vector of the external material rod based on the spread process trajectory using the control commands. Generate the member trajectory using the movement vector.
9. The method for optimizing concrete pouring paths based on a BIM model according to claim 1, characterized in that: S7 includes: Obtain the rod trajectory, and calculate the nodal cavity volume based on the rod trajectory; The propagation process trajectory is obtained, and fluid filling data is extracted based on the node cavity volume and the propagation process trajectory. The volume increment is then calculated.
10. The method for optimizing concrete pouring paths based on a BIM model according to claim 9, characterized in that: The S7 also includes: If the volume increment is less than the preset volume threshold, the discharge rate is determined based on the fluid filling data. The trajectory of the rod is corrected according to the discharge speed to obtain the target rod trajectory, and the concrete pouring path is generated based on the target rod trajectory.
