Construction process control method for water overflowing pavement and water overflowing culvert in alpine region

Abstract

The present application relates to the field of automatic control and process regulation of road base construction, in particular to a construction process control method for water-crossing pavement and water-filled culvert in high-cold regions. The discharge state, transportation heat and humidity state and equipment space-time data of cement stabilized gravel batch are obtained, the process coordinates associated with batch, paving time and working surface grid are established, the material compaction allowance field is generated, and the structure sensitivity field is generated according to the change of culvert boundary, design thickness and supporting stiffness; the two types of state fields are coupled into paving and compaction tasks with time sequence constraints, the paving rhythm, compaction equipment sequence, access path and working mode are coordinated through task constraint graph, and the subsequent tasks are updated according to the track vibration response and compaction growth deviation. The present application forms a continuous closed loop for material state attenuation, structure position risk and equipment action, reduces the missed compaction period of low-temperature water loss batch and excessive work of stiffness mutation position, and keeps the construction state of full-width large-thickness base traceable.

Description

Technical Field

[0001] This invention relates to the field of automatic control and process regulation in road base construction, specifically a method for controlling the construction process of flooded culverts in high-altitude and cold regions. Background Technology

[0002] In high-altitude and cold regions, the cement-stabilized crushed stone base course for flood-crossing culverts typically involves determining construction parameters through test sections, followed by a sequence of mixing, covered transportation, full-width paving, initial compaction, secondary compaction, final compaction, and curing. The mixing plant controls the output based on the mix proportions, and transport vehicles are covered. On-site quality indicators include discharge temperature, moisture content, loose paving thickness, paving speed, and compaction degree. The paver operates continuously at a predetermined speed and screed parameters, while the roller covers the work surface according to a pre-arranged combination of machine types, travel speed, and number of compaction passes. When deviations are detected at the culvert top, wing wall adjacent areas, and transition zones at entrances and exits, management personnel adjust water supply, paving speed, or additional compaction based on individual test results. More digital construction methods also record the location of transport vehicles, paving trajectories, roller tracks, and vibration responses, and use grid colors to display locations of missed compaction, under-compaction, or excessive compaction passes. The aforementioned controls typically revolve around material batches, equipment operations, or local inspection points. The temperature drop and water loss process of a batch from discharge to entry into the working surface, the specific location where the material falls into the full-width base layer, and the corresponding culvert structure conditions at that location do not form a continuous correspondence.

[0003] Under the combined effects of low temperature, wind cooling, and transportation delays, different batches with the same mix proportion will have different equivalent ages and remaining compaction time. Furthermore, large-thickness full-width paving causes the same batch to be dispersed to multiple locations during lateral material placement and vehicle overlap. The susceptibility of the culvert top, water-facing side, wing wall adjacency zone, and stiffness transition zone to under-compaction, over-vibration, and process delays varies. Existing control methods, when adjusting fixed parameters or single-index limits, cannot incorporate the degree of material degradation over time with the structural sensitivity of the final material placement location into the same control criteria. Nor can they determine the order of paving pace, compaction equipment entry timing, access paths, and vibration modes. The main technical problem arising from this is that the construction system cannot identify priority work surface grids when the material still has adjustable margins and cannot translate the identification results into mutually non-conflicting equipment actions. Adding compaction after abnormal compaction or surface conditions have occurred may have exceeded the material's workable timeframe or introduced excessive work at locations of abrupt stiffness changes, making it difficult to simultaneously control the internal density of the base layer and the continuity of the structural transition. Summary of the Invention

[0004] The purpose of this invention is to provide a method for controlling the construction process of flooded culverts in high-altitude and cold regions, which can solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for controlling the construction process of flood culverts for water-crossing pavements in high-altitude and cold regions includes: obtaining the discharge status of cement-stabilized crushed stone batches and the sequence of environmental and material states during transportation, and establishing process coordinates according to batch identification, paving time and work surface grid; The material compaction margin field of each working face grid is generated based on the state sequence, and the structural sensitivity field is generated based on the culvert boundary and base course design information. The material compaction margin field is coupled with the structure sensitivity field to form a paving and compaction control task with operation timing constraints. The vibration response of the compaction equipment when performing control tasks is collected, and the control tasks corresponding to the grid of the subsequent working face are updated using the vibration response.

[0006] Preferably, the generation of the material compactibility margin field includes: taking the mixing and discharge time as the starting point of the batch state, and binding the discharge temperature, initial moisture content and mix proportion records to the batch identifier; Based on the transportation positioning trajectory, the temperature of the carriage, the ambient temperature, the wind speed, the coverage status, and the dwell events are time-aligned to form a thermal and humid boundary sequence; The equivalent age and heat and moisture loss of the batch are recursively derived based on the heat and moisture boundary sequence, and the recursive results are corrected with the measured material condition when the batch enters the paver. The corrected batch status is mapped to the corresponding work surface grid according to the paving time, so as to obtain the compaction margin of each grid as the construction time changes.

[0007] Preferably, the generation of the structural sensitivity field includes: extracting the culvert top boundary, wing wall adjacent boundary, inlet and outlet gradient boundary, water-facing side boundary, and design thickness variation line from the culvert and base design data; Each boundary is converted into distance and direction features in the working surface mesh, and stiffness transition features are constructed according to the change in support stiffness between the base layer and the culvert. A grid sensitivity is generated based on the distance feature, the direction feature, and the stiffness transition feature, and a continuity constraint is applied to adjacent grids along the full width direction to form a structural sensitivity field consistent with the paving mileage coordinates.

[0008] Preferably, the formation of the paving and compaction control task includes: converting the compaction margin of each working face grid into the latest effective working time, converting the structural sensitivity into the process completion priority, and writing the latest effective working time and the process completion priority into the grid task description; Based on the continuous material supply status of the paver, the current position of the roller, and the grid adjacency relationship, a task constraint diagram is constructed to determine the paving rhythm, the entry order of the compaction equipment, and the grid access path without interrupting the continuous paving of the entire width. The task constraint diagram is updated on a rolling basis according to the construction progress, and the current control task is issued to the corresponding equipment.

[0009] Preferably, the recursion between the equivalent age and the heat and moisture loss includes: establishing a heat capacity state based on the batch loading and the exposed boundary of the carriage, and forming a heat transfer boundary that changes with transportation events using the ambient temperature, wind speed and coverage status. The thermal migration state is constructed based on the temperature difference between the material surface and the interior, and the change in water content and the thermal migration state are jointly projected onto the state time axis of cement-stabilized crushed stone. When a vehicle stops, coverage is enabled, or coverage is restored, the state before the event is saved and the corresponding boundary conditions are switched. After obtaining the corrected observations at the paver entrance, the state residuals of the event intervals are allocated to obtain a continuous batch equivalent age sequence.

[0010] Preferably, the mapping of batch status to the working face grid includes: reading the vehicle unloading mark and unloading time window in the paver hopper, and calculating the material flow rate within the time window based on the auger distributor speed, paving width, travel speed and target loose paving thickness; The material flow rate during the overlapping unloading of adjacent vehicles is distributed to the full-width grid according to the mass conservation relationship, and the batch composition ratio is stored for each grid. The calibration states of each batch are fused according to the batch composition ratio, and the time when the screed passes through the grid is used as a unified state time reference to generate a compactable margin field with dual indexes of batch source and spatial location.

[0011] Preferably, the generation of the grid sensitivity includes: establishing a direction-dependent propagation kernel with each structural boundary as the source, and taking different spatial scales for the propagation direction along the culvert boundary and in accordance with the water flow direction; The rate of change of support stiffness and the rate of change of design thickness are used as amplitude constraints of the propagation kernel, and the scouring effect on the water-facing side is written as a directional weight into the propagation kernel. Nonlinear synthesis is performed on the propagation results generated by multiple structural boundaries on the same grid, and the abrupt changes in the synthesis results are limited by the maximum allowable difference between adjacent grids, so as to obtain the structural sensitivity field in which the culvert top region, the gradient region and the general base region are continuously connected.

[0012] Preferably, the task constraint diagram includes task nodes with grids to be paved, grids to be initially compacted, grids to be re-compacted, and grids to be finally compacted, and directed constraint edges with process sequence, equipment safety distance, overlapping relationship of adjacent grids, and the latest effective operation time. When a new batch status, equipment location, or compaction response is written to the process coordinates, only the affected task nodes and their reachable nodes are recalculated. Based on the process completion priority and time margin of each task node, candidate actions for equipment are generated, and candidate actions that cause equipment conflicts, undervoltage, or cross incomplete grids are excluded by mutual exclusion edges, forming a rolling equipment control sequence.

[0013] Preferably, the allocation of the state residual includes: comparing the material temperature and moisture content obtained at the paver inlet with the batch-progressive state to form thermal state residual and wet state residual; The residual allocation weights are established based on the duration, boundary switching type, and state sensitivity of each event interval in the transportation trajectory, and the correction amount of each event interval is obtained with the continuity of state before and after the event as a constraint. After writing the correction amount back to the corresponding event interval, the equivalent age sequence is recalculated, while retaining the boundary parameter versions before and after correction. Extract stable residuals from the same batch that has completed compaction, and progressively update the boundary parameter versions used in subsequent batches.

[0014] Preferably, the update of the equipment control sequence includes: matching the vibration response of the road roller to the nodes of the executed tasks according to the wheel track range, and forming a compaction growth sequence based on the response increment of the continuous compaction process within the same grid; The compaction growth sequence, together with the compactibility margin, structural sensitivity, compaction method, and equipment speed carried by the task node, forms a response record; For records that deviate from the response trend of adjacent grids, candidate causes for batch status, candidate causes for paving thickness, and candidate causes for compaction execution are generated respectively, and the update object is determined based on the consistency between each candidate cause and the observation in the process coordinates; The updated results are written to the affected task nodes, and the candidate actions and delivery order for the corresponding reachable nodes are recalculated.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. By incorporating material discharge status, transport thermal and moisture boundaries, paving time, and work surface grids into a unified process coordinate system, the compaction margin field can continuously characterize the remaining work capacity of each material batch at different locations. The structural sensitivity field converts the culvert top boundary, wing wall adjacent boundary, inlet and outlet transition boundary, water-facing direction, and support stiffness changes into spatial constraints. The coupling of these two fields forms the latest work time and process completion priority. Based on this, the task constraint diagram coordinates the paver's continuous material supply and the roller's entry sequence, access path, and operation mode, prioritizing the allocation of remaining workable time to grids with stronger structural changes and faster material margin shrinkage. Construction control shifts from local remediation after a single indicator exceeds its limit to a timing arrangement when the material state is still adjustable. This reduces the probability of batches with low-temperature water loss missing their compaction time slots and avoids excessive local work caused by indiscriminately increasing compaction at locations of abrupt changes in culvert stiffness, ensuring the continuous compaction process of the thick, full-width base course along the travel direction and transverse direction.

[0016] 2. After the batch identifier is associated with the paver unloading time window, auger feed flow rate, and full-width grid position, the material composition formed by the overlap of adjacent vehicles can still be traced according to quality relationships, and the grid status is no longer replaced by a single surface inspection. The thermal and wet state residuals of transportation events are reconfigured according to dwell time, coverage switching, and boundary sensitivity, and the state recursion basis used for subsequent batches can be corrected step by step; the roller vibration response is written back to the executed task nodes according to the wheel track range, and the compaction growth deviation can be respectively corresponding to the batch status, paving thickness, or compaction execution record. The feedback chain formed by this retains the parameter versions, material sources, and equipment actions before and after correction, so that the cause of the anomaly can be located along the same process coordinates, and only the affected task and its subsequent nodes are recalculated. This processing reduces equipment waiting and process disturbance caused by global rearrangement, and at the same time, it enables the culvert top area, transition area and general base area to adopt continuous control tasks, and establishes a verifiable consistency relationship between construction records and actual material placement and compaction response. Attached Figure Description

[0017] Figure 1 Diagram showing the source of material compactibility margin; Figure 2 Flowchart for batch status and grid mapping; Figure 3 Diagram showing the sources of structural sensitivity; Figure 4 This is a closed-loop diagram of task constraints and compaction feedback. Detailed Implementation

[0018] In one embodiment, the construction object is the top of a flood culvert in a high-altitude and cold region and the cement-stabilized crushed stone base course connecting it before and after. The base course adopts a thick, full-width continuous paving. The control process extends from the discharge of the mixing plant to the feedback of the curing state. The on-site control platform acquires the discharge status of the cement-stabilized crushed stone batches and the environmental and material state sequence during transportation. It establishes process coordinates according to batch identification, paving time, and work surface grid. Based on the state sequence, it generates the material compaction margin field of each work surface grid and generates the structural sensitivity field based on the culvert boundary and base course design information. It couples the material compaction margin field and the structural sensitivity field to form a paving and compaction control task with work sequence constraints. When the compaction equipment executes the control task, it continuously collects vibration response. The control platform uses the vibration response to update the control task corresponding to the subsequent work surface grid. In this way, the state change of the material from discharge to final compaction, the position of the material in the full-width base course, the culvert structural conditions, and the equipment actions share the same set of spatiotemporal indexes.

[0019] Specifically, the control platform uses the road centerline mileage as the longitudinal coordinate in the construction coordinate system, the lateral position from the left boundary to the right boundary of the full-width base layer as the lateral coordinate, and the design base layer thickness direction as the vertical reference. Based on the actual trajectory of the paver, the continuous working surface is discretized into a grid. The grid boundary maintains a traceable geometric relationship with the culvert top edge line, wing wall projection line, inlet and outlet gradient line, and design thickness change line. The process time adopts a unified time benchmark after time calibration of the mixing plant, transportation terminal, paver, and roller. Each data item simultaneously carries the data source, collection time, location, batch identifier, and quality mark. Data that cannot be directly mapped to the grid is first retained in the batch time axis. After the vehicle unloading event and the paver establish a correlation through the event, it is written into the grid status. The original observations that have already been written are not overwritten by subsequent calculations, and the correction values ​​are attached to the same observation record with the new version.

[0020] In this embodiment, the mixing plant generates a non-repeatable batch identifier when each transport batch is loaded. The discharge time, discharge temperature, initial moisture content, aggregate gradation record, cement dosage record, loading amount, and mixing uniformity record are bound to this batch identifier. The discharge status is stored in parallel using original measured values ​​and normalized status values. Normalization is only used for cross-dimensional calculations, while the original values ​​are still used for construction verification. After the batch leaves the mixing plant, the vehicle positioning trajectory is written into the batch timeline according to a unified time reference. The temperature of the material in the truck bed, ambient temperature, wind speed, cover opening or closing status, and parking waiting events are registered as timestamped observations or events. If multiple temperature observations exist at the same time, a dimensional and reasonable range check is performed first, and then the observations are classified into the material surface state or the material internal state based on the observation location, instead of directly calculating the average.

[0021] Preferably, the transportation state sequence uses event intervals as the basic recursive units. The control platform reads the vehicle movement state and coverage state between adjacent events, and marks continuous driving, short-term waiting, coverage opening, coverage closing, and arrival queuing as different boundary conditions. The starting point of the interval inherits the material internal temperature, surface temperature, moisture content, and equivalent age of the ending point of the previous interval. Within the interval, thermal migration is estimated based on the ambient temperature difference, wind action, the exposed boundary of the carriage, and the material heat capacity. The change in moisture content is estimated based on the relationship between surface temperature and moisture migration. The ending point of the interval simultaneously saves the recursive value and the boundary condition version. When an event occurs, the previous interval is not resampled; only the boundary conditions used in the next event interval are switched. Therefore, the state time axis can be kept continuous when the coverage state changes or the vehicle stops during transportation.

[0022] Material compaction margin is represented by a dimensionless value. A value close to 1 indicates that the batch status still has a large margin from the current project's allowable work boundary. A value close to 0 indicates that the relevant process needs to be completed immediately. A negative value indicates that the current grid will no longer enter the normal rolling schedule and will switch to abnormal handling. Each time the control platform obtains a new transportation observation, paver entry correction observation, or grid time advancement event, it calculates the margin after the batch is mapped to the grid according to the following formula. Each input is normalized within the applicable range of the same batch of material, and the weight is jointly determined by the response records of the test section and the completed batches, while maintaining the same meaning within the same project version.

[0023] ; in, This represents the material compaction margin of mesh g at time t. This represents the normalized equivalent age of batch b that makes up the grid at time t. This represents the normalized heat loss of the batch relative to its initial state. This represents the normalized moisture loss of the batch relative to its initial state. , and These represent the contribution weights of equivalent age, heat loss, and moisture loss to the workability state, respectively, and the sum of the three is 1. For example, when a batch is mapped to the mesh... Take 0.30, Take 0.40, Take 0.20, Take 0.40, Take 0.35, If we take 0.25, then The value is 0.69. The control platform retains the normal schedule but determines the equipment entry order based on the subsequent structural sensitivity. If transportation delays occur... and If it continues to increase, then It decreases over time and reduces the time margin for the corresponding task.

[0024] Please refer to Figure 1 , Figure 1 The horizontal axis is represented by the source sample number, and the sample number corresponds to the order of the batch status mapped to the work surface grid after paving time. To avoid mutual occlusion between variables of different magnitudes, Figure 1 A three-coordinate display is used: the vertical axis of the upper coordinate system represents the material's compaction margin. The vertical axis of the central coordinate system represents the normalized equivalent age. and normalized heat loss The vertical axis of the lower coordinate system represents the normalized water loss. . , and In this calculation, the contribution weights of the three types of state variables are used to combine material age, heat loss, and moisture changes into the margin for continued compaction operations within the current grid. Figure 1 The upper curve is visible. The fact that the curve remained in the higher range in most source samples indicates that the batch still retained a compactable margin after entering the grid; when the curve dipped locally, it indicated that the corresponding batch had experienced a longer transportation wait, low temperature and air cooling, or margin shrinkage after inlet condition correction.

[0025] The two curves in the middle respectively reflect and right The path of influence. It increases with batch waiting time and paving time, representing the cumulative equivalent age of the material from discharge to placement; Influenced by factors such as carriage temperature, ambient temperature, wind speed, coverage status, and waiting time, local peak values ​​correspond to increased thermal state loss caused by transportation events or entry corrections. When both factors rise simultaneously in several sample segments, it indicates that the material state has shifted from a single-time decay to a combined effect of time and low-temperature environment. Subsequent compaction tasks should prioritize ensuring the initial compaction window.

[0026] lower part The curve is listed separately because the magnitude of water loss is significantly smaller than that of water loss. , and If placed on the same vertical axis, they will be close to the bottom and difficult to identify. Although the value is small, , , Still participating under the combined effect Deduct; when and Simultaneous rise indicates that the low-temperature air cooling and water loss processes may jointly compress the material's workability window.

[0027] Furthermore, the equivalent age is not simply the time elapsed after material discharge. The control platform projects the material state changes within each event interval onto a unified state time axis. Intervals where the temperature is within the reactivity range of the mix proportion and the moisture content is stable are accumulated at the normal rate. Intervals where low temperatures slow down the reaction but are accompanied by surface water loss are included in the age component and the heat and moisture loss component, respectively. When the cover is opened, causing a sudden change in the surface boundary, the pre-event state is preserved and the heat transfer and moisture migration conditions are switched from the event moment. Intervals waiting for unloading after arrival are continued to be counted, and the arrival of the vehicle is not considered the end of transportation. Only after the batch enters the paver is the transportation phase closed by the entrance observation. Therefore, batches with the same transportation time but experiencing different air cooling, cover, and waiting events will achieve different results. , and combination.

[0028] In a preferred embodiment, the paver inlet is positioned to obtain observation locations for material temperature and moisture content. When the vehicle begins unloading, the batch identifier is read and the unloading time window is opened. The inlet observation is compared with the batch recursive state in the time window to form thermal state residuals and wet state residuals. The control platform establishes residual allocation weights according to the duration of each transport event interval, boundary switching type, and state sensitivity to the boundary. The interval correction amount is obtained with the constraints of continuous internal state before and after the event and the sum of correction amounts of each interval equal to the inlet residual. The correction amount is written back to the corresponding event interval and the equivalent age is recursively calculated. At the same time, the pre-correction version, the post-correction version, the inlet observations used, and the allocation weights are saved. Batches of the same type that have been compacted and have stable responses can contribute residual samples. Subsequent batches only update the same type of boundary condition version without modifying the parameters of different material sources or different coverage events.

[0029] Please refer to Figure 2 When batch status is mapped to the working face grid, the control platform reads the vehicle unloading identifier and unloading start and end times in the paver hopper, forms a batch input stream with the vehicle load and unloading time window, and calculates the material flow rate delivered to the front of the screed at each time based on the auger distributor speed, paving width, paver travel speed and target loose paving thickness. The paver position trajectory gives the longitudinal grid when the material leaves the screed, and the left and right conveying status of the auger distributor gives the lateral distribution relationship. The residence of material from the hopper to the screed is described by the internal flow queue of the equipment. The batch identifier moves forward with the flow queue instead of being directly bound to the vehicle unloading point to avoid incorrectly mapping the material in the hopper to the road position when the vehicle just arrived.

[0030] When adjacent vehicles overlap in the hopper, the control platform saves the remaining mass and input flow rate of each batch within the overlap time window. Based on the mass conservation principle, it calculates the batch composition ratio entering each transverse grid after passing through the auger. The sum of each ratio within the same grid is 1. Any mass difference that cannot be explained by the vehicle record enters a pending verification state and is not assigned to any batch. The corrected equivalent age, heat loss, and moisture loss of each batch are merged according to the composition ratio, and a unified state time reference is established using the moment the screed passes through the grid. The mixed grid simultaneously retains the composition ratio of each batch for residual backtracking. Based on the fusion state calculation, if it is subsequently discovered that the ingress observation of one of the batches is distorted, only the grid containing the components of that batch and its reachable tasks are recalculated.

[0031] In another embodiment, the structural sensitivity field is generated from the culvert and base course design data imported before construction. The control platform extracts the outer contour projection of the culvert top, the adjacent boundaries of the wing walls, the gradual boundary of the inlet and outlet, the water-facing boundary, the backwater boundary, the design thickness change line, and the base course support condition zoning. All boundaries are uniformly converted to the construction coordinate system. For each working face grid, the shortest distance to each boundary, the relationship between the boundary normal direction and the grid travel direction, and the relationship between the water flow direction and the boundary direction are calculated. Then, stiffness transition characteristics are generated according to the change of base course support conditions along the mileage direction, and thickness change characteristics are generated according to the difference between adjacent grids of design thickness. When the design data lacks local stiffness values, the confirmed structural zoning category is used instead of fabricating continuous values.

[0032] To enable structural conditions to participate in the same scheduling calculations as material margins, structural sensitivity is also normalized to a dimensionless value. The control platform converts boundary proximity, support stiffness variation, design thickness variation, and water-facing exposure into inputs within the range of 0 to 1, and synthesizes mesh sensitivity according to the following formula. The weights are determined by the degree of concern in engineering design and the response of the test section. Any adjustment of the weights will form a new version and will only affect the task generated after the adjustment.

[0033] ; in, This indicates the structure sensitivity of the mesh g. This indicates the normalized proximity of the grid to the culvert top, wing walls, and gradient boundaries. This indicates the degree of normalized variation in support stiffness. This indicates the degree of normalization variation in the design thickness. Indicates the degree of exposure on the water-facing side. , , and For the corresponding weights, such as the gradient mesh at the top of a culvert. Take 0.80, Take 0.70, Take 0.50, If we take 0.60, the corresponding weights are 0.35, 0.30, 0.20, and 0.15 respectively. The value is 0.665, which makes the mesh uniform. Under certain conditions, it enters the compaction candidate set earlier than the general base grid, while still being constrained by equipment safety distance and process sequence.

[0034] Please refer to Figure 3 , Figure 3 The horizontal axis represents the grid sample number, indicating the structural feature samples extracted along the construction mileage and horizontal grid sequence; the vertical axis represents the structural sensitivity and its input contribution, with values ​​interpreted as dimensionless from 0 to 1. The solid black line represents the grid structural sensitivity. The gray dashed line, dotted line, and marker line represent the degree of proximity of the boundary, respectively. Degree of variation in support stiffness Degree of variation in design thickness and water-facing side exposure degree . , , and As structural attention weights, the above four types of inputs are combined into a structurally sensitive result that can participate in task ranking.

[0035] Figure 3 middle The curve shows a significant rise in some grid segments, indicating that these grids are closer to the culvert top, adjacent boundaries of the wing walls, or gradual transition boundaries at the inlet and outlet, or are simultaneously affected by changes in support stiffness and design thickness. These high-value segments do not simply represent increased compaction energy, but rather suggest that the control platform needs to more strictly handle the sequence of operations, safe equipment spacing, overlap of adjacent grids, and the latest effective working time when generating task nodes, to avoid localized excessive work or prolonged loose paving at locations of abrupt changes in culvert stiffness.

[0036] The relative relationships between the input curves can be used to distinguish sensitive sources: At higher levels, proximity to structural boundaries is the primary factor; The lifting indicates that the transition between the base course and the culvert support conditions has a stronger impact on the grid. The variation reflects the difference in design thickness along the mileage or laterally, and is often related to paving thickness control and continuous adjustment of screed posture. The rise indicates that the exposed water-facing side necessitates greater consideration of the flow direction and the directionality of the boundary propagation core in this area. Separating and displaying each curve facilitates judgment. Is it dominated by a single structural factor, or formed by the superposition of multiple structural factors?

[0037] Figure 3 The correspondence with the structural sensitivity formula is that, The risk labels are not manually assigned on-site, but rather... , , , The spatial state field is generated jointly by the curve and its weights. The undulations of the curve reflect the continuous transition between the culvert crest region, the transition zone, the upstream side, and the general base layer region. The subsequent task constraint diagram will... Converted into process completion priority, and compared with These factors collectively influence the candidate ranking of nodes awaiting paving, initial compaction, secondary compaction, and final compaction. Through... Figure 3 It can be used to check whether sensitive areas are concentrated in the affected grid after the design boundary version change, thus avoiding irrelevant areas being updated by mistake.

[0038] Preferably, the boundary proximity is not achieved using a uniform circular attenuation. Instead, a direction-dependent propagation kernel is established with the structural boundary as the source. The propagation kernel reflects changes in support conditions along the direction perpendicular to the culvert boundary and reflects the differences between upstream and downstream conditions along the direction parallel to the water flow. The rate of change of support stiffness and the rate of change of design thickness control the amplitude of the propagation kernel, and the degree of exposure on the upstream side controls the directional weight. When the same grid is affected by multiple boundaries, the propagation results are synthesized using bounded nonlinear synthesis to prevent highly sensitive boundaries from being diluted by simple averaging of multiple low values. After synthesis, the differences between adjacent grids in the full-width direction and the direction of travel are checked. Local abrupt changes exceeding the allowable range of continuous variation are jointly constrained by adjacent propagation sources. The actual boundary position of the structure still maintains a high gradient, while isolated peaks are not formed in general areas.

[0039] In this embodiment, the control platform generates the already generated data each time the paving front advances. and The task description is written into the grid. The material margin is calculated based on the latest effective operation time according to the current time. The structural sensitivity is mapped to the process completion priority. The task description also includes batch composition, target loose paving thickness, available compaction methods, current process status and adjacent grid status. Batches that have not yet entered the paver only form a predicted task. After the screed passes through the grid, the task changes to the pending initial compaction state. After the compaction equipment completes one wheel track coverage, it changes to the pending secondary compaction or pending final compaction state according to the equipment status. The task status transition is triggered by verifiable equipment trajectory and response records. Manual confirmation by personnel can only handle abnormal states and cannot replace normal trajectory evidence.

[0040] Please refer to Figure 4The task constraint graph uses the grids to be paved, to be initially compacted, to be recompacted, and to be finally compacted as task nodes. Process edges are formed by the sequence of operations within the same grid, mutually exclusive edges are formed by the safety distance between equipment, adjacent edges are formed by the requirement that adjacent grids must maintain wheel track overlap, time edges are formed by the latest effective working time, and material supply edges are formed by the continuous material supply of the paver and the arrival prediction of transport vehicles. Each directed constraint edge stores its source and effective interval. When a new batch status, equipment location, or compaction response enters the process coordinates, the control platform starts from the directly associated task nodes and searches for affected nodes along the reachability relationship. Only the affected subgraph is recalculated, while the other equipment tasks that have been issued and satisfy the constraints remain unchanged, avoiding frequent reordering of the entire work surface due to local state changes.

[0041] The control platform calculates a dynamic priority value for each executable task, placing grids with lower material margins, higher structural sensitivity, larger response deviations, or those close to the latest operation time in a higher position. The calculation results are only used for ranking candidate tasks. Whether the equipment can execute is still determined by the process edges, mutually exclusive edges, adjacent edges, and feeding edges in the task constraint diagram. The dynamic priority value is obtained according to the following formula, where the stable small quantity is used to keep the calculation bounded when the material margin is close to 0. All variables continue to use the aforementioned physical meaning without being redefined.

[0042] ; in, This represents the dynamic priority value of grid g at time t. This represents the normalized deviation of the existing compaction response of the grid relative to the trend of adjacent grids. This indicates the degree of normalization of the current time as it approaches the latest valid job time. Indicates the response bias correction weight. Indicates that the weight is adjusted as the time approaches. This represents a stable small quantity that prevents the denominator from being zero, such as a certain grid. 0.70 0.35 For 0.20, It is 0.60. Take 0.50, Take 0.40, If we take 0.05, then The value is 2.165. If another mesh structure has lower sensitivity and larger material margin, its dynamic priority value is smaller and it is shifted backward while satisfying the time edge.

[0043] The paving control task is generated from the paving nodes in the task constraint diagram. The control platform determines the current cycle time allowable range based on the continuous material supply status, the material flow rate in front of the screed, and the predicted arrival time of the next batch. Within this range, it selects a paving speed that will not cause the grid to be compacted behind to exceed the latest effective working time. When the grid with high structural sensitivity reaches the front of the screed, the task description simultaneously carries the full-width elevation, cross slope, and loose paving thickness constraints that need to be maintained. The screed attitude adjustment uses a control quantity that changes continuously along the mileage without generating a step at the grid boundary. The auger distributor status is used to maintain the consistency between the lateral material flow rate and the batch composition mapping. When the vehicle material supply is interrupted on site, the control platform compares the amount of available material in front with the time slack of the already formed compaction task, and then decides to reduce the cycle time, control the shutdown, or isolate the unfinished joint position.

[0044] The order in which compaction equipment enters is determined by the set of executable tasks and the current position of the equipment. The control platform establishes the current position, direction of travel, current compaction mode, continuous working time and accessible grid set for each roller. Candidate actions include moving to an adjacent grid, maintaining the current wheel track, switching to an allowed compaction mode or exiting a conflict area. Mutually exclusive edges prevent two machines from entering the same channel from the opposite direction within a safe distance. Adjacent edges require wheel tracks to overlap with the previous valid wheel track. Process edges prevent re-compaction from crossing grids that have not yet completed initial compaction. Time edges reject long-distance scheduling that would cause high-priority tasks to expire. The filtered control sequence is issued in a rolling window. After each action is completed, the equipment returns the trajectory and working status to release the corresponding constraint edges.

[0045] In the culvert crest and stiffness-gradient mesh, the compaction method selection is read simultaneously. , When the adjacent grids have formed a dense growth sequence, and the material margin is sufficient but the structural sensitivity is high, the control task adopts a continuous and constrained compaction trajectory and reduces sudden energy switching. When the material margin shrinks, the initial compaction that can form a stable contact is completed first, and then subsequent operations are arranged. The existing response indicates that when the grid compaction growth slows down, the vibration compaction is no longer repeated based solely on the preset number of passes. The task sequence on the upstream and downstream sides is jointly defined by the directional propagation core and the direction of water flow. Before the equipment crosses the culvert boundary, the task status of the adjacent grids is checked to avoid the final compaction of one side being completed while the adjacent side remains in a loose state for a long time.

[0046] After the compaction response is collected, it is matched to the executed task node according to the geometric range of the roller wheel track and a unified time reference. When the same sampling point falls into multiple grids, the response contribution is allocated according to the wheel track coverage area. Observations during equipment turning, vibration stopping, or acceleration and deceleration are marked with unstable state and the original records are retained. They are not directly incorporated into the stable response sequence. The vibration response in the normal driving range is arranged according to the passing order in the grid. The control platform reads the compaction method, equipment speed, material compaction margin, structural sensitivity, base layer thickness status, and batch composition corresponding to the passing, forming a single response record that can be traced back to the material, location, and equipment action.

[0047] The response increments of continuous compaction processes within the same grid are used to construct a density growth sequence. The control platform describes the state changes brought about by this round of operation using the response difference between two adjacent stable passes, and judges whether the current increment deviates by the trend of adjacent grids in the same structural region. When the absolute value of the response is affected by the base support conditions, it is not directly compared across regions, but rather based on... The source of propagation is selected from comparable neighborhoods. The grid is saved as a sequence to be accumulated only after a valid pass through each grid cell. The sequence is only generated after a series of comparable records are reached. When the final compression is completed and subsequent verification results are obtained, the growth sequence before the final compression is bound to the verification results, so that the response trend can be corrected for the same batch and the same structure grid without replacing the field observation with the verification value.

[0048] when When a grid deviates from the comparable neighborhood trend, the control platform establishes candidate causes for batch status, paving thickness, and compaction execution. The candidate causes for batch status check the batch composition of the grid, the entrance observation residual, and the response consistency of other grids in the same batch. The candidate causes for paving thickness check the screed passing through records, design thickness changes, and the status of adjacent full-width grids. The candidate causes for compaction execution check wheel track coverage, compaction method, speed, and unstable state markers. Each type of candidate cause only reads existing observations in the process coordinates and forms a consistency score. If the score cannot distinguish the cause, the grid is transferred to the field for verification without automatically changing the material model. Only when a cause is supported by multiple source records is the corresponding batch boundary version, paving task parameters, or compaction candidate action updated.

[0049] Preferably, the update adopts the principle of local reachability; when the batch boundary version changes, the uncompressed mesh containing that batch is searched and recalculated. When structural design data changes during verification, the incomplete meshes covered by the corresponding propagation kernel are located and recalculated. When confirming the cause of the compaction, only the candidate actions reachable by the current device are updated. , , and Recalculate Then, the task constraint diagram is used to check the process edges and mutually exclusive edges. Completed tasks with complete evidence are not reopened. Grids that need to be reworked on site are entered into the diagram as independent abnormal tasks and isolated from normal continuous paving tasks, so that closed-loop correction will not destroy the confirmed construction status.

[0050] In a complete operational scenario, a transport batch, after being discharged from the mixing plant, undergoes low-temperature driving, a short stop, and waiting upon arrival. The coverage status switches once during a stop inspection. The control platform recursively calculates the batch status based on event intervals and allocates residuals at the paver inlet using measured temperature and humidity conditions. Vehicle unloading overlaps with the previous vehicle, and the material flow queue maps the two batches onto a continuous grid of the general base course, transition zone, and culvert top zone. The culvert top zone... Higher and waiting Lowering and achieving greater The task constraint diagram prioritizes the initial compaction nodes in the culvert top area while maintaining continuous material supply from the paver. After the roller completes its work along the allowed path, it sends back a response. If the growth sequence is consistent with the comparable neighborhood of the culvert top, the subsequent re-compaction nodes are released. If they are inconsistent, the review object is determined based on the batch composition and wheel track records.

[0051] In the construction scenario on the water-facing side, the structural sensitivity propagation kernel directionally expands the boundary influence along the water flow direction, and the adjacent boundary grid on the water-facing side... Above the grid at the same distance on the backwater side, the control platform still retains the effect of stiffness and thickness variations. The common constraints prevent the information of other structures from being overridden by the water-facing position alone. When the paver passes the full width of the culvert top, the material composition ratio and state time reference of the transverse grid are updated simultaneously. The compaction path selection requires that the wheel tracks maintain adjacent overlap when crossing the full width direction. The grids near the wing wall projection boundary cannot be directly determined by the low response trend of general areas. The control platform constructs comparable neighborhoods from grids with the same propagation source and similar support conditions to keep the response deviation judgment consistent with the structural position.

[0052] In the alternative scenario of rapidly decreasing ambient temperature, new observations during the transport and paving phases revealed multiple incomplete grids. Synchronously increase, and the control platform updates the corresponding data according to the batch composition relationship. However, instead of directly elevating all tasks to the same priority, it combines the priorities of each grid. Current process status and The candidate order is recalculated. High-sensitivity grids that have completed initial compaction and are close to the final compaction deadline are given priority. General grids that have not yet been paved are delayed in entering the working face by adjusting the material feeding cycle. Batches that are predicted to have no normal margin before entering the paver are marked as abnormal batches and prevented from participating in the normal material flow queue. The on-site handling results are transmitted back as status records instead of being fabricated by the control platform for compensation.

[0053] When some observations are missing, the control platform selects the processing method according to the position of the missing data in the state chain. When there is a short-term lack of environmental observations during transportation but the vehicle trajectory and coverage status are complete, the most recent boundary version with the same type corrected by the entrance residual can be used for recursion and the state quality flag can be reduced. When there is a lack of batch identifier or unloading time window, the batch to grid determination mapping is not performed, but the relevant grid is marked as a component to be checked. When there is a lack of stable response of the road roller but the trajectory is complete, wheel track coverage can be registered but no compaction growth increment is generated. When the structural design boundary cannot be confirmed, the corresponding propagation area is not automatically assigned general area sensitivity. The state quality flag is entered into the task description and the automatic distribution range is limited. The affected sub-map is recalculated after manual review or supplementary observations.

[0054] To ensure that records from multiple devices can form the same process coordinates, the control platform periodically checks the clock deviation of each data source. It verifies the time relationship using common events such as vehicles arriving at the paver, screeds passing through fixed mileage markers, and rollers passing through known control points. When the deviation is within the allowable range, the correction mapping is saved and the original timestamp is retained. Data with deviations exceeding the interpretable range is transferred to an isolation queue. Position data is checked against the coordinate system and direction of travel before being placed into the grid. Duplicate messages are deduplicated based on the data source identifier, acquisition time, and load summary. Out-of-order messages are inserted into the state timeline according to the event time and trigger local recalculation. If real-time messages that have been used for equipment safety constraints are later corrected, the original issuing decision and correction reason are retained for traceability.

[0055] During the curing stage, the same grid index is used as the end of the closed loop. When the coverage is completed, the start and end positions of the coverage, the completion time, and the time relationship with the final compaction task are recorded. The surface temperature and humidity status, integrity check, and subsequent quality review results obtained during the curing period are bound to the corresponding grid. The control platform compares the end state of similar batches, similar structural sensitivity, and similar compaction growth sequences to determine whether there is a continuous deviation in the transport heat and humidity boundary version, inlet residual allocation, or response trend. The confirmed deviation is only updated in the model version of subsequent batches that have not yet started. The historical calculations and equipment actions of the completed grids remain unchanged. Therefore, the state chain from material discharge to curing at any base position can be restored along the batch identifier, grid position, task node, and version number.

[0056] Furthermore, the construction management terminal provides batch view, grid view, and equipment view based on process coordinates. The batch view displays material records, transportation events, inlet calibration, and the mapped grid. The grid view displays batch composition, ... , The system displays task status, wheel track coverage, and response sequence. The equipment view shows issued actions, constraint sources, and execution feedback. When operators confirm abnormal batches, structural boundary changes, or on-site verification results, the confirmation record must include personnel, time, and basis, and form a new version. The system does not allow manual values ​​without a source to overwrite the original observations. The exported construction records are generated from the above versioned data, so that the material source, construction time, equipment actions, and quality verification in the records correspond to the same work surface location.

[0057] In a preferred deployment, the mixing plant, transport vehicles, pavers, and rollers each retain local data caches. The on-site control platform is responsible for unified time, process coordinates, state field calculations, and task constraint diagrams. When communication is temporarily interrupted, the equipment continues to execute the short-term task that has recently passed the safety constraint check and records local actions. Any new tasks that require cross-device coordination are suspended from generation. After communication is restored, cached records are merged according to the data source sequence number and event time. During replay, the batch state and grid state are restored first, then the equipment execution evidence is restored, and then the task subgraph is recalculated starting from the affected node. If the on-site actions during offline period conflict with the original constraints, the relevant grid enters the verification state instead of automatically declaring the process completed.

[0058] During the test section preparation phase, the control platform does not directly use unverified weight and boundary versions. Instead, it divides the test section into calibration areas encompassing general base layers, design thickness variations, and simulated stiffness transitions. Each transport batch is still registered with its discharge status and transport events according to the formal construction method. Each grid is still organized into batches based on material flow queues. Compaction equipment forms different state combinations according to the approved test section process. Subsequent sampling and verification results are compared with the corresponding grids. , And dense growth sequence binding, the control platform ensures , and The correction margin contribution weight is adjusted under the condition that the sum is 1 and all terms are non-negative, while ensuring that the order of structural design concerns is not reversed. to After calibration, the initial version of the project is frozen. Online updates during the main construction are only made within the same boundary and material range to avoid individual abnormal batches changing the entire control basis.

[0059] Normalization processing employs mapping rules bound to physical boundaries. Equivalent age, heat loss, and moisture loss are determined at zero and boundary positions based on mix proportion tests and permissible construction conditions, respectively. Boundary positions represent the acceptable state limits for normal rolling scheduling, rather than the sensor range. When observations are below zero, they are treated as zero and the original negative deviation is retained. When observations exceed boundary positions, the normalized value is allowed to exceed 1 to trigger anomaly judgment. Boundary proximity in structural inputs is mapped by the directional distance from the grid to the boundary. Stiffness and thickness variations are mapped by the differences between adjacent design units. Water-facing exposure is jointly mapped by the design water flow direction, slope orientation, and protection zone. Each mapping rule records the source file, applicable range, and version. When the model is called, it first checks whether the rule matches the current material and structural region.

[0060] The material flow queue is implemented using a conservation account to record the hopper inlet, hopper stock, left and right outputs of the auger spreader, and the forming volume behind the screed. When the vehicle starts unloading, the inlet account for the corresponding batch is added. When unloading ends, the inlet is closed but the batch balance in the hopper is not immediately cleared. The output of the auger spreader is deducted from the batch balance according to the left and right conveying status. The forming volume behind the screed is converted into consumed mass based on the paving speed, full width, and loose paving thickness and checked against the output account. The account difference is temporarily stored during the allowable data delay period. If the account is not closed after the explainable delay, the mapping quality is marked as abnormal. When the lateral material distribution is uneven, the grid composition is adjusted according to the left and right output ratio without changing the total mass of the vehicle. The flow chain of that batch in the paving equipment is closed only after the batch balance is zeroed and all outputs have been located.

[0061] The rolling solution of the task subgraph starts with the currently executable node as the entry point. The control platform first deletes candidate actions that violate hard constraints, including equipment mutual exclusion, process sequence, isolation of unconfirmed structural boundaries, and isolation of abnormal batches. Then, based on... The remaining nodes are sorted. If the sorting is the same, the node with the shorter equipment movement distance and the ability to maintain the continuity of adjacent wheel tracks is selected. After the candidate action is added to the rolling window, the corresponding mutual exclusion edge and equipment time interval are occupied immediately. Subsequent candidates can only be solved on the remaining resources. The last operation margin of the nodes that have not yet been executed is reserved at the end of the window. When any node is expected to be unable to obtain equipment resources within the margin, the control platform applies a cycle time limit to the upstream paving task and solves again. If the limit still cannot restore feasibility, the relevant batch or grid is transferred to manual handling state without generating actions that violate hard constraints.

[0062] After receiving the control task, the paver first checks the task version, target grid range, continuous material supply status, and current equipment working mode. If the version is outdated or the grid range is inconsistent with the current position, execution is rejected and the reason is returned. After verification, the target cycle time and continuous attitude constraints are converted into control commands acceptable to the equipment. During execution, feedback is generated based on the actual speed, screed position, attitude changes, and spiral material distribution status. The control platform compares the target grid with the actual passed grid. If a short-term deviation occurs, the actual trajectory is used as the basis for material placement and subsequent prediction tasks are updated. The target trajectory is not used to overwrite the factual record. If a continuous deviation occurs, new automatic cycle time changes are paused and the current safe state is retained. After the position and material supply relationship are restored and confirmed, the task boundary is re-established from the actual construction front.

[0063] When the roller receives the control sequence, it checks the equipment identification, task node status, allowed entry direction, and evidence of previous task completion. The control sequence only includes actions that can be executed within the current scrolling window. The equipment sends back a start event every time it enters a grid and a finish event and compaction method during the process every time it leaves a grid. If another equipment occupies a mutually exclusive area in advance, the control platform retains the path of the first-occupied equipment based on the already occurred entry event and withdraws the conflict candidate. It does not repeatedly switch priorities between the two equipment. If the equipment cannot arrive in sequence, it recalculates the remaining reachable nodes. When all reachable nodes are blocked by hard constraints, it issues a waiting status and transmits a resource shortage event to the paving cycle control. After the waiting is lifted, the path is rebuilt with the new equipment position instead of continuing to use the expired trajectory.

[0064] Before the vibration response enters the compaction growth sequence, it undergoes checks for time continuity, wheel track stability, and amplitude rationality. The continuity check confirms that the observations belong to the same grid passage and there are no long-term discontinuities. The wheel track stability check excludes transitional data near turning, start-stop, and compaction mode switching. The amplitude rationality check identifies sudden distortions based on the recent stable interval of the same equipment. The retained observations are aggregated at equal intervals according to wheel track length, and the dispersion within the aggregation range is saved. If the dispersion is too large, the passage is only recorded as coverage evidence and does not participate in the incremental comparison. Only when the dispersion meets the requirements will it form a response increment with the previous stable passage. The responses of different compaction equipment are not directly mixed into the same sequence. Instead, the correspondence between equipment is established through a common verification grid before being used for trend judgment.

[0065] The cause consistency scoring adopts a combination of evidence coverage and conflict deduction. For batch status causes, the abnormal mesh must correspond to other meshes in the same batch in terms of thermal and humid conditions and response direction. For thickness causes, the screed record, forming volume account, and adjacent transverse mesh must have at least verifiable consistent changes. For compaction execution causes, the wheel track gap, mode switch, or unstable state must coincide with the time period of the abnormality. If a certain cause lacks necessary evidence, it will not be selected because other causes have lower scores. When multiple causes have evidence at the same time, the mesh will be transferred to composite review and the actions of related models and equipment will be restricted respectively. After the on-site review results are written, the score will be recalculated and the selection process will be saved. Subsequent batch version updates will only adopt cause records with closed evidence chains.

[0066] Data verification after construction is carried out using the grid as the smallest closed-loop object. Each completed grid must have an interpretable batch composition, screed passing event, task status from initial compaction to final compaction, wheel track coverage, compaction response or reason for missing response, and curing coverage record. If any necessary chain is missing, the grid remains in a state of pending verification. Batch-level verification checks whether the loading quality and the quality consumed by all mapped grids are closed. Equipment-level verification checks whether the issued actions, start and end events, and trajectories are closed. Structure-level verification checks whether there are process records that cross incomplete grids in highly sensitive propagation areas. The verification results are associated with the sampling location and handover inspection records. When exporting, the original observation summary, calculation version, and anomaly handling record are given at the same time to avoid only retaining the final color or a single pass mark.

[0067] In the optional method that does not directly connect to the equipment execution interface, the control platform still generates control tasks according to the same process coordinates, state field, and task constraint diagram. However, it displays the paving cycle time, compaction entry sequence, access path, and compaction method to the on-site operators. After the operators confirm, the terminal generates an execution acceptance event with a timestamp. When rejecting or changing a task, a reason corresponding to material supply, equipment, materials, or on-site safety must be selected. The actual equipment trajectory and status still serve as the criteria for determining whether the task is completed. Manual confirmation does not replace trajectory evidence. The control platform updates resource availability based on the difference between the accepted event and the actual execution, without changing the... and The physical source allows semi-automatic organization methods and direct equipment linkage methods to share the same closed-loop logic and traceability structure.

[0068] For the compaction continuity of thick base courses, the control platform records the loose thickness as part of the grid state and saves it together with the source of structural sensitivity propagation. If the actual thickness record shows discontinuous changes between adjacent grids in the lateral direction after the screed passes, the relevant grids establish additional adjacency relationships in the task constraint diagram. The roller access path needs to first cover the adjacent areas on both sides of the change line and then extend outward. When comparing the compaction growth sequence, grids with similar thickness states and consistent propagation sources are selected as references. After the cause of the thickness is confirmed, the compaction candidate actions of grids that have not yet been completed are regenerated according to the actual thickness state. The completed grids enter the verification chain without passing through to increase the number of subsequent passes to cover up the existing differences. When the base course thickness record cannot be verified, the control platform reduces the automatic judgment range of the area and retains the equipment trajectory and response for joint verification on site.

[0069] Recovery after a construction interruption begins with the actual construction front and material status. The control platform freezes control sequences not yet executed at the time of the interruption and continues to recursively calculate the thermal and moisture status of already discharged batches and paved grids. Once recovery conditions are met, the hopper inventory, batch locations, paver locations, roller locations, and evidence of completed tasks are re-read. Abnormal states are established for batches and grids that exceeded the normal margin boundary during the interruption. Grids still within the normal range are processed according to the new... , Reconstruct the task subgraph with equipment resources. For tasks that were occupied but not completed before the interruption, first check the actual trajectory. If it is confirmed that the task has not been executed, release the resources. If it is confirmed that the task has been partially executed, split the wheel track coverage to the corresponding grid. Restore the schedule from the checked state instead of returning to the planned state before the interruption.

[0070] When design changes or on-site remeasurements correct the culvert boundary, the control platform saves the original structural data version and converts the new boundary to the same construction coordinate system. It then uses geometric differences to locate the affected meshes in the propagation kernel, retaining the original meshes that have already been completed. Along with the original mission evidence and associated changes, meshes that have not yet been laid out or finalized will be recalculated using the new version. , , , and The task subgraph only replaces the structural sensitivity sources of these grids and their reachable nodes. The evidence of equipment mutual exclusion and process sequence remains unchanged. If the new boundary conflicts with the area currently occupied by the equipment, the control tasks that have not yet started are withdrawn and the safety boundary is confirmed on site. After the confirmation is completed, a new executable sequence is generated based on the actual equipment location.

[0071] Through the above implementation method, batch discharge status, transportation heat and humidity boundary, equivalent age, material flow queue, working surface grid, structural boundary propagation, task constraint diagram and compaction response are continuously connected according to the same data chain. The material compaction margin answers how much operational margin the current material still has in a specific grid. The structural sensitivity limits the grid's tolerance to process delays and changes in operation mode. Dynamic priority values ​​organize candidate tasks without replacing safety and process constraints. Response residuals and cause attribution return the execution results to the batch status, paving status or compaction action. The control platform updates adjacent grids that have not yet been completed accordingly, so that material state decay, full-width thick fabric and culvert stiffness changes in high-altitude and cold environments can be controlled in the same closed loop, and all original observation, calculation versions and equipment evidence remain verifiable.

Claims

1. A method for controlling the construction process of flood-crossing culverts in high-altitude and cold regions, characterized in that, include: Obtain the discharge status of cement-stabilized crushed stone batches and the sequence of environmental and material status during transportation, and establish process coordinates according to batch identification, paving time and work surface grid; The material compaction margin field of each working face grid is generated based on the state sequence, and the structural sensitivity field is generated based on the culvert boundary and base course design information. The material compaction margin field is coupled with the structure sensitivity field to form a paving and compaction control task with operation timing constraints. The vibration response of the compaction equipment when performing control tasks is collected, and the control tasks corresponding to the grid of the subsequent working face are updated using the vibration response.

2. The method for controlling the construction process of flood-crossing culverts in high-altitude and cold regions according to claim 1, characterized in that, The generation of the material compactibility margin field includes: taking the mixing and discharge time as the starting point of the batch state, and binding the discharge temperature, initial moisture content and mix proportion records to the batch identifier; Based on the transportation positioning trajectory, the temperature of the carriage, the ambient temperature, the wind speed, the coverage status, and the dwell events are time-aligned to form a thermal and humid boundary sequence; The equivalent age and heat and moisture loss of the batch are recursively derived based on the heat and moisture boundary sequence, and the recursive results are corrected with the measured material condition when the batch enters the paver. The corrected batch status is mapped to the corresponding work surface grid according to the paving time, so as to obtain the compaction margin of each grid as the construction time changes.

3. The method for controlling the construction process of flood-crossing culverts in high-altitude and cold regions according to claim 1, characterized in that, The generation of the structural sensitivity field includes: extracting the culvert top boundary, wing wall adjacent boundary, inlet and outlet gradient boundary, water-facing side boundary, and design thickness variation line from the culvert and base design data; Each boundary is converted into distance and direction features in the working surface mesh, and stiffness transition features are constructed according to the change in support stiffness between the base layer and the culvert. A grid sensitivity is generated based on the distance feature, the direction feature, and the stiffness transition feature, and a continuity constraint is applied to adjacent grids along the full width direction to form a structural sensitivity field consistent with the paving mileage coordinates.

4. The method for controlling the construction process of flood-crossing culverts in high-altitude and cold regions according to claim 1, characterized in that, The formation of the paving and compaction control task includes: converting the compaction margin of each working face grid into the latest effective working time, converting the structural sensitivity into the process completion priority, and writing the latest effective working time and the process completion priority into the grid task description. Based on the continuous material supply status of the paver, the current position of the roller, and the grid adjacency relationship, a task constraint diagram is constructed to determine the paving rhythm, the entry order of the compaction equipment, and the grid access path without interrupting the continuous paving of the entire width. The task constraint diagram is updated on a rolling basis according to the construction progress, and the current control task is issued to the corresponding equipment.

5. The method for controlling the construction process of flood-crossing culverts in high-altitude and cold regions according to claim 2, characterized in that, The recursion between the equivalent age and the heat and moisture loss includes: establishing a heat capacity state based on the batch loading and the exposed boundary of the carriage, and forming a heat transfer boundary that changes with transportation events using the ambient temperature, wind speed and coverage status. The thermal migration state is constructed based on the temperature difference between the material surface and the interior, and the change in water content and the thermal migration state are jointly projected onto the state time axis of cement-stabilized crushed stone. When a vehicle stops, coverage is enabled, or coverage is restored, the state before the event is saved and the corresponding boundary conditions are switched. After obtaining the corrected observations at the paver entrance, the state residuals of the event intervals are allocated to obtain a continuous batch equivalent age sequence.

6. The method for controlling the construction process of flood-crossing culverts in high-altitude and cold regions according to claim 3, characterized in that, The mapping of batch status to the working face grid includes: reading the vehicle unloading mark and unloading time window in the paver hopper, and calculating the material flow rate within the time window based on the auger distributor speed, paving width, travel speed and target loose paving thickness; The material flow rate during the overlapping unloading of adjacent vehicles is distributed to the full-width grid according to the mass conservation relationship, and the batch composition ratio is stored for each grid. The calibration states of each batch are fused according to the batch composition ratio, and the time when the screed passes through the grid is used as a unified state time reference to generate a compactable margin field with dual indexes of batch source and spatial location.

7. The method for controlling the construction process of flood-crossing culverts in high-altitude and cold regions according to claim 3, characterized in that, The generation of the grid sensitivity includes: establishing a direction-dependent propagation kernel with each structural boundary as the source, and taking different spatial scales for the propagation direction along the culvert boundary and in accordance with the water flow direction; The rate of change of support stiffness and the rate of change of design thickness are used as amplitude constraints of the propagation kernel, and the scouring effect on the water-facing side is written as a directional weight into the propagation kernel. Nonlinear synthesis is performed on the propagation results generated by multiple structural boundaries on the same grid, and the abrupt changes in the synthesis results are limited by the maximum allowable difference between adjacent grids, so as to obtain the structural sensitivity field in which the culvert top region, the gradient region and the general base region are continuously connected.

8. The method for controlling the construction process of flood-crossing culverts in high-altitude and cold regions according to claim 4, characterized in that, The task constraint diagram includes task nodes with grids to be paved, grids to be initially compacted, grids to be re-compacted, and grids to be finally compacted, and directed constraint edges with the sequence of operations, equipment safety distance, overlapping relationship of adjacent grids, and the latest effective operation time. When a new batch status, equipment location, or compaction response is written to the process coordinates, only the affected task nodes and their reachable nodes are recalculated. Based on the process completion priority and time margin of each task node, candidate actions for equipment are generated, and candidate actions that cause equipment conflicts, undervoltage, or cross incomplete grids are excluded by mutual exclusion edges, forming a rolling equipment control sequence.

9. A method for controlling the construction process of flood-crossing culverts in high-altitude and cold regions according to claim 5, characterized in that, The allocation of the state residuals includes: comparing the material temperature and moisture content obtained at the paver inlet with the batch-based recursive state to form thermal state residuals and wet state residuals; The residual allocation weights are established based on the duration, boundary switching type, and state sensitivity of each event interval in the transportation trajectory, and the correction amount of each event interval is obtained with the continuity of state before and after the event as a constraint. After writing the correction amount back to the corresponding event interval, the equivalent age sequence is recalculated, while retaining the boundary parameter versions before and after correction. Extract stable residuals from the same batch that has completed compaction, and progressively update the boundary parameter versions used in subsequent batches.

10. A method for controlling the construction process of flood-crossing culverts in high-altitude and cold regions according to claim 8, characterized in that, The update of the equipment control sequence includes: matching the vibration response of the road roller to the nodes of the executed tasks according to the wheel track range, and forming a compaction growth sequence based on the response increment of the continuous rolling process within the same grid; The compaction growth sequence, together with the compactibility margin, structural sensitivity, compaction method, and equipment speed carried by the task node, forms a response record; For records that deviate from the response trend of adjacent grids, candidate causes for batch status, candidate causes for paving thickness, and candidate causes for compaction execution are generated respectively, and the update object is determined based on the consistency between each candidate cause and the observation in the process coordinates; The updated results are written to the affected task nodes, and the candidate actions and delivery order for the corresponding reachable nodes are recalculated.

Images