Undercompaction area automatic identification and recompaction path optimization method and system

Abstract

The application discloses a method and system for automatic identification of under-compacted areas and optimization of compaction paths, and relates to the technical field of intelligent construction of water conservancy projects such as roads and embankments. The method comprises the following steps: obtaining compaction degree detection data and road roller kinematics parameters, identifying under-compacted blocks, and determining target geometric boundaries; modeling the transfer path of the road roller between nodes as a work corridor with a physical width, calculating the intersection of the work corridor and the under-compacted blocks, and calculating the incidental coverage gain of the transfer path based on the intersection; on this basis, determining the net increase in effective work amount of each block when it is visited, and solving the optimal compaction visit sequence with the goal of minimizing the total work cost including the transfer journey and the net increase in effective work amount. The application weakens the binary opposition between transfer and work by calculating the incidental compaction value (area or defect elimination amount) of the transfer path, and realizes the overall optimization of the total work cost.

Description

Technical Field

[0001] This invention relates to the field of intelligent construction technology for water conservancy projects, and in particular to a method and system for automatic identification of under-compacted areas and optimization of recompaction paths. Background Technology

[0002] In the quality monitoring of highway subgrade, embankment, dam filling and other engineering projects, compaction degree is one of the important indicators for evaluating the quality of the project. Re-compacting the under-compacted areas identified after the initial compaction can reduce structural hazards and achieve uniform compaction.

[0003] With the development of modern intelligent construction technology, the use of unmanned road rollers for automatic compaction has become a trend. Their efficient compaction path planning not only reduces ineffective empty travel but also lowers energy consumption. Current intelligent compaction systems typically employ location-based path planning strategies, simplifying identified undercompacted areas into isolated geometric centroids and using a standard Traveling Salesman Problem (TSP) model to plan the roller's access sequence. In this type of technology, the system divides the roller's operation into two independent states: transfer and compaction. The transfer path serves as a transport route connecting various work points. The planning objective is usually limited to minimizing the Euclidean distance or travel time of the transfer path. Effective compaction is only considered to be performed when the roller reaches the target area.

[0004] However, existing technologies tend to overlook the physical characteristics of road rollers at the modeling level. For example, during the transfer process, the roller is always in contact with the ground, and its travel trajectory itself has physical width and compaction function. The model ignores the incidental coverage that the transfer path may cause to other under-compacted areas along the way, resulting in the system planning redundant operation paths. In other words, the path planning result is not an overall optimization. Summary of the Invention

[0005] Purpose of the invention: To provide a method and system for automatic identification of undercompacted areas and optimization of recompaction paths, in order to solve the above-mentioned problems existing in the prior art.

[0006] Technical solution: Firstly, a method for automatic identification of undercompacted areas and optimization of recompaction paths, comprising:

[0007] Obtain compaction test data and kinematic parameters of the roller in the compaction operation area;

[0008] Several undercompacted blocks were identified based on compaction test data, and the target geometric boundaries of each undercompacted block were determined.

[0009] Using the path planning model, the centroid of each undercompacted block is determined as the node to be visited, the transfer path of the roller between nodes is modeled as a working corridor with physical width, and the path length of each transfer path is calculated.

[0010] The spatial intersection of the area enclosed by the operation corridor and the target geometric boundary is calculated, and the incidental coverage gain of the transfer path on the undercompacted block is calculated based on the intersection.

[0011] Based on incidental coverage gain, determine the net increase in effective workload when each undercompacted block is accessed;

[0012] The optimal multi-pressure access sequence is generated by minimizing the total operating cost, including the transfer travel and the net increase in effective workload.

[0013] Control commands are generated based on the optimal re-compaction access sequence to guide the road roller to perform re-compaction operations.

[0014] According to one aspect of this application, the transfer path of the road roller between nodes is modeled as an operational corridor with physical width, specifically including:

[0015] Based on the minimum turning radius in the kinematic parameters of the road roller, an inter-node transfer trajectory that satisfies the kinematic constraints is generated;

[0016] Using the working width in the kinematic parameters of the road roller as the buffer distance, a buffer analysis is performed on the transfer trajectory to generate a strip polygon, which is then used as the working corridor.

[0017] According to one aspect of this application, the incidental coverage gain of the transition path to the undercompacted block is calculated based on the intersection, specifically including:

[0018] Calculate the geometric area of ​​the intersection;

[0019] Calculate the original total area of ​​the region enclosed by the target's geometric boundaries;

[0020] Calculate the ratio of the geometric area to the original total area, and use this ratio as the incidental coverage gain.

[0021] According to one aspect of this application, the incidental coverage gain of the transition path to the undercompacted block is calculated based on the intersection, specifically including:

[0022] Based on compaction test data, a compaction defect density field covering each undercompacted block is constructed. The compaction defect density in the compaction defect density field represents the degree to which the compaction at the corresponding location deviates from the qualified standard.

[0023] Based on the compaction defect density field, the cumulative compaction defect density within the intersection range is calculated as the defect elimination amount covered by the transfer path;

[0024] Based on the compaction defect density field, the cumulative compaction defect density within the target geometric boundary is calculated as the total defect amount of the undercompacted block;

[0025] Calculate the ratio of defect elimination amount to total defect amount, and use the ratio as incidental coverage gain.

[0026] According to one aspect of this application, a compaction defect density field covering each undercompacted block is constructed, specifically including:

[0027] Obtain the preset compaction degree qualification threshold;

[0028] Calculate the difference between the acceptable compaction threshold and the measured values ​​at each test point in the compaction test data;

[0029] When the difference is greater than zero, the difference is retained; when the difference is less than or equal to zero, the difference is set to zero, thus obtaining the compaction defect density corresponding to each detection point, and constructing a compaction defect density field accordingly.

[0030] According to one aspect of this application, the net increase in effective workload when each undercompacted block is accessed is determined based on incidental coverage gain, specifically including:

[0031] Based on the current candidate recompression access order, determine all preceding transition paths before reaching the current undercompressed block;

[0032] Based on the incidental coverage gain of the preceding transition path to the undercompacted block, the cumulative coverage rate of the undercompacted block is calculated using a probabilistic complementary model.

[0033] The remaining uncovered proportion of under-compacted blocks is calculated based on the cumulative coverage rate. Combined with the original total amount of work done in the under-compacted blocks, the net increase in effective work is calculated.

[0034] According to one aspect of this application, several undercompacted blocks are identified based on compaction test data, and the target geometric boundary of each undercompacted block is determined, specifically including:

[0035] Obtain the preset compaction degree qualification threshold;

[0036] Compaction degree test data is compared with the compaction degree qualified threshold point by point to generate a binary segmentation map;

[0037] Connectivity analysis was performed on the binarized segmented graph, and the extracted connected regions were identified as undercompacted blocks.

[0038] Extract the edge contours of the connected regions and define the edge contours as the target geometric boundaries.

[0039] According to one aspect of this application, several undercompacted blocks are identified based on compaction test data, and the target geometric boundary of each undercompacted block is determined, specifically including:

[0040] The compaction gradient field is calculated based on compaction test data. The compaction gradient field characterizes the degree of drastic change in compaction in space.

[0041] Based on compaction test data, the under-compacted core area was initially identified;

[0042] The search proceeds point by point outward along the edge normal of the undercompacted core region. When the gradient magnitude at the search location is less than the preset gradient convergence threshold and the compaction value is greater than the preset safety margin threshold, the search stops and the extended edge is obtained.

[0043] The extended edge is defined as the target geometric boundary.

[0044] According to one aspect of this application, the optimal complex pressure access order is generated by employing a marginal benefit greedy strategy, which includes:

[0045] Initialize the set of unvisited blocks and the current location of the road roller;

[0046] Traverse each candidate block in the set of unvisited blocks and calculate the marginal benefit of using a candidate block as the next access target;

[0047] Marginal benefits are directly proportional to the numerator and inversely proportional to the denominator; the numerator includes the net increase in effective operations after reaching the candidate block, and the sum of the incidental coverage gain of the transfer path from the current location to the candidate block on other unvisited blocks; the denominator includes the length of the transfer path from the current location to the candidate block.

[0048] Select the candidate block with the greatest marginal benefit as the next access target, and remove it from the set of unvisited blocks until the set of unvisited blocks is empty.

[0049] In a second aspect, there is an automatic identification and recompaction path optimization system for undercompacted areas. The system includes a processor coupled to a memory. The memory stores instructions, and when the instructions are executed by the processor, they implement any of the methods described in the first aspect.

[0050] The beneficial effect is that by calculating the incidental compaction value (area or defect elimination) of the transfer path, the binary opposition between transfer and operation is weakened, and the overall operation cost is optimized. Attached Figure Description

[0051] Figure 1 This is a flowchart of the method for automatic identification of undercompacted areas and optimization of recompaction paths in this application.

[0052] Figure 2 This is a flowchart illustrating the calculation of the incidental coverage gain of the transition path for the undercompacted block in this application.

[0053] Figure 3 This is a flowchart illustrating the construction of a compaction defect density field covering each undercompacted block in this application.

[0054] Figure 4This is a flowchart for determining the net increase in effective workload when each undercompacted block is accessed, as required by this application.

[0055] Figure 5 This is a flowchart illustrating the determination of the target geometric boundaries of each undercompacted block in this invention. Detailed Implementation

[0056] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.

[0057] In addition to the problems mentioned in the background technology, the applicant also found that existing solutions usually treat undercompacted areas as homogeneous geometric planes and use the coverage area as the evaluation standard for the completion of the operation. This ignores the non-uniformity of the spatial distribution of compaction defects, such as severe defects in the core area and slight defects at the edges. This makes it impossible for the planning algorithm to distinguish the value difference between eliminating severe defects and covering the edge areas, resulting in high-risk areas not being processed quickly and increasing the number of iterations for subsequent closed-loop detection.

[0058] To solve these problems, combined with Figures 1 to 5 The present invention will be specifically described through the following embodiments.

[0059] Example 1

[0060] According to one aspect of this application, this embodiment provides a method for automatic identification of undercompacted areas and optimization of recompaction paths, used to describe the system and its main workflow in this application.

[0061] In this embodiment, the main hardware architecture of the automatic identification and recompaction path optimization system for undercompacted areas includes an onboard compaction degree detection device, a high-precision positioning system, a vehicle-mounted controller, and a road roller actuator.

[0062] Specifically, the onboard compaction testing device is installed at the vibratory wheel of the road roller. Employing a continuous compaction control sensor, it can collect the acceleration signal of the interaction between the vibratory wheel and the soil in real time and convert it into a compaction test value. The high-precision positioning system uses a global navigation satellite system with carrier phase differential technology to obtain the three-dimensional coordinates (x, y, z) of the road roller's contact point, with a positioning accuracy at the centimeter level.

[0063] The onboard controller includes a processor and a memory. The memory stores a computer program that executes the compaction path planning algorithm, and the processor runs the program to implement the method described in this embodiment. The onboard controller is connected to the detection device, positioning device, and actuator via a bus interface to receive sensing data and issue control commands. The roller actuator includes a steering hydraulic system, a travel drive system, and a vibration control system, which respond to control commands and drive the roller to travel along the planned path for vibration compaction.

[0064] In this embodiment, the main technical terms are defined as follows to eliminate ambiguity and unify subsequent descriptions:

[0065] A working corridor refers to the strip-shaped area swept across the ground by the physical compaction wheel of a road roller during its relocation. Unlike traditional path planning, which treats paths as geometric lines without width, the working corridor in this embodiment is modeled as a polygon with physical width. For example, for a relocation path from point A to point B, its working corridor is an area extending W / 2 width to both sides of the path's centerline, where W is the working width of the road roller.

[0066] Incidental coverage gain refers to the effective operational contribution of a road roller's working corridor to an undercompacted block when it passes through the space. It is expressed as the geometric coverage area and can also be represented as the amount of defect elimination in engineering quality.

[0067] Net increase in effective work volume refers to the amount of compaction work that is required for a certain under-compacted block after deducting the aforementioned incidental cover gain in order to meet the compaction qualification standard. This amount of work volume can be calculated as the remaining area or the remaining defect volume.

[0068] This embodiment provides an automatic identification and recompaction path optimization system for undercompacted areas. The system includes a processor coupled to a memory. The memory stores instructions, and when the instructions are executed by the processor, an automatic identification and recompaction path optimization method for undercompacted areas is implemented.

[0069] Example 2

[0070] According to another aspect of this application, this embodiment proposes a method for automatic identification of undercompacted areas and optimization of recompaction paths. In one possible implementation, the specific steps include:

[0071] Step 101: Obtain compaction test data and kinematic parameters of the roller in the compaction operation area.

[0072] Specifically, the vehicle-mounted controller receives data streams from the onboard compaction testing device and positioning system in real time via a data interface. Since the raw data may contain noise interference and coordinate deviations, preprocessing is required. This preprocessing includes coordinate registration, mapping compaction values ​​to spatial coordinates (x, y); and noise reduction, such as using median filtering or Gaussian filtering algorithms to smooth data fluctuations and remove outliers. This results in compaction testing data, represented as a two-dimensional grid matrix K, where each element K(i, j) represents the percentage compaction value at the corresponding spatial location.

[0073] The kinematic parameters of the road roller include the minimum turning radius R. _min , represents the minimum radius of the arc that the roller can travel when the steering angle reaches its limit; working width W, represents the effective width of the compaction wheel contacting the ground; current position coordinates of the roller P. _cur Entry point P _in and the point of entry P _out .

[0074] Step 102: Based on the compaction test data, identify several undercompacted blocks and determine the target geometric boundary of each undercompacted block.

[0075] In this embodiment, image segmentation techniques are typically used to discretize the continuously distributed compaction data into independent, operable units. Specifically, a qualified threshold is set, and regions in the raster matrix K below this threshold are marked as undercompacted regions. A connected component analysis algorithm, such as eight-neighbor search, is used to aggregate adjacent undercompacted raster cells into independent connected components, each of which represents an undercompacted block. The target geometric boundary refers to the sequence of closed polygon vertices surrounding the edge of the block; the set of identified blocks is denoted as {R}. _1 ,R _2 ,...,R _n}

[0076] Step 103: Using the path planning model, the transfer path of the road roller between nodes is modeled as a working corridor with physical width. In other words, using the path planning model, the centroid of each undercompacted block is determined as the node to be visited, the transfer path of the road roller between nodes is modeled as a working corridor with physical width, and the path length of each transfer path is calculated.

[0077] In this embodiment, the path planning problem is modeled as a generalized traveling salesman problem. The set of nodes to be visited includes the centroids and entry / exit points of all undercompacted blocks; that is, the centroids of each undercompacted block are determined as the nodes to be visited. Nodes are the core spatial reference points of the path planning model. For any two nodes, the system generates a path that satisfies the minimum turning radius R. _minA constrained feasible travel trajectory. Unlike traditional methods for calculating the trajectory length in the Traveling Salesman Problem, this embodiment constructs the corresponding work corridor polygon T for this trajectory. _ij Its geometric feature is a band-shaped region with a width of W centered on the trajectory line.

[0078] Step 104: Calculate the spatial intersection of the area enclosed by the working corridor and the target geometric boundary, and calculate the incidental coverage gain of the transfer path on the undercompacted block based on the intersection.

[0079] Specifically, the system calculates the polygon T of the work corridor. _ij With any undercompacted block R _k The geometric intersection region of the polygon enclosed by the target geometric boundaries. If there is no intersection, the incidental coverage gain is zero; if there is an intersection, the contribution of the intersection region to R-reduction is calculated according to the selected metric, such as area or defect value. _k The contribution value of the undercompacted state, i.e., the incidental coverage gain.

[0080] Step 105: Based on incidental coverage gain, determine the net increase in effective workload when each undercompacted block is accessed.

[0081] After determining the access order and transition paths, for each block to be accessed, the system accumulates the incidental coverage gain generated by all transition paths for that block. The net increase in effective job volume is the original total job volume for that block minus the accumulated incidental coverage gain. If the incidental coverage gain has completely covered the block, the net increase in effective job volume is 0, and the block can be skipped or simply passed.

[0082] Step 106: With the objective of minimizing the total operating cost, including the transfer travel and the net increase in effective work volume, solve to generate the optimal re-pressure access sequence.

[0083] In this embodiment, the merits of any access order θ are evaluated using a general objective function J(θ), and the corresponding calculation formula is as follows:

[0084] J(θ) = ∑(L _trans )+∑(L _cover );

[0085] Where, ∑(L _trans ) represents the sum of the lengths of all transfer routes, representing transportation costs; ∑(L _cover J(θ) represents the sum of the net increase in effective activity across all blocks, after discounting, and represents the activity cost. The algorithm finds the sequence θ that minimizes J(θ). * This is the optimal recompression access order.

[0086] It should be noted that the transfer trip refers to the distance the roller travels to the work area. There is no actual compaction work; it is only for preparation for the operation and is a transportation cost incurred during the re-compaction operation.

[0087] In another possible implementation, with the objective of minimizing total operating costs, including transfer travel and net increase in effective workload, the following are specifically included:

[0088] Establish the objective function, which consists of two parts;

[0089] The first part is the sum of the lengths of all transition paths in the recompression access sequence;

[0090] The second part is the sum of the work trips after converting the net increase in effective work volume for each under-compacted block;

[0091] Find the sequence of access nodes that minimizes the value of the objective function, which is the optimal complex pressure access order.

[0092] Step 107: Generate control instructions based on the optimal re-compaction access sequence to guide the road roller to perform re-compaction operations.

[0093] Specifically, the onboard controller translates the optimal sequence into a specific sequence of low-level control commands, including driving speed, steering angle, and vibration activation or deactivation timing. The roller actuator responds to these commands and proceeds sequentially to each undercompacted block. For each block, the system also plans its internal coverage path, such as using a bow-shaped path, to complete the compaction task that increases the net effective workload. The bow-shaped path is a well-known full-coverage path planning method in the art. It uses the working width W as the spacing between adjacent compaction rows and performs reciprocating compaction along the long axis of the block until all uncovered areas within the block are covered. Those skilled in the art can select a suitable coverage path pattern based on the shape of the block.

[0094] Furthermore, in practical engineering, especially in linear filling projects such as embankments, multiple road rollers are often used to operate synchronously or collaboratively to improve construction efficiency. This embodiment is applicable to multi-road roller collaborative scenarios. That is, through automatic identification of under-compacted areas and optimization of re-compaction paths, re-compaction tasks can be reasonably allocated to each road roller, reducing repeated compaction and missed areas, and achieving a balanced improvement in overall compaction quality.

[0095] Example 3

[0096] Based on the above embodiments, this embodiment further explains the process of identifying several undercompacted blocks based on compaction degree detection data and determining the target geometric boundary of each undercompacted block. In one possible implementation, the process specifically includes the following steps:

[0097] Step 201: Obtain the preset compaction degree qualification threshold.

[0098] Obtain soil type information (clay, sand, or gravel) and fill layer information for the current work area. The onboard controller has a pre-stored compaction standard reference table. Based on the soil type and layer information, this table is consulted to obtain the design compaction standard value K corresponding to the current working condition. _std and the acceptable compaction threshold K _th This threshold is used to determine whether a grid point is qualified.

[0099] Step 202: Compactness test data is compared point by point with compaction qualification threshold to generate a binary segmentation map.

[0100] Specifically, the system iterates through the preprocessed compaction degree grid matrix K, and compares the value of each grid cell K(x,y) with K... _th Compare the values ​​and construct a binary matrix B with the same dimension as K. If K(x,y) is less than K... _th If the position is deemed unqualified, assign B(x,y) the value 1; if K(x,y) is greater than or equal to K... _th If the position is deemed acceptable, B(x,y) is assigned a value of 0. The work area is then divided into acceptable background and unacceptable foreground using the binarized segmentation image B.

[0101] Step 203: Perform connected component analysis on the binarized segmentation graph and identify the extracted connected regions as undercompacted blocks.

[0102] In the binarized segmentation map B, defective raster points typically appear in clusters. In this embodiment, the system uses an eight-neighbor connectivity analysis algorithm to scan matrix B. This algorithm merges all spatially adjacent raster points with a value of 1 into the same object, including diagonal directions, and assigns them the same tag ID. Each independent connected object constitutes an undercompacted block R. _k A minimum area threshold is preset to eliminate tiny connected regions with an area smaller than this threshold, thus filtering out isolated noise points caused by sensor noise. This minimum area threshold is preset by those skilled in the art, and its specific value can be adjusted according to process standards, physical dimensions, etc.

[0103] Step 204: Extract the edge contours of the connected regions and determine the edge contours as the target geometric boundaries.

[0104] For each marked connected region, the system extracts its outermost grid point sequence and transforms it into a vector polygon vertex sequence. This polygon is the basic target geometric boundary of the block. The boundary based on a fixed threshold is usually jagged and a hard boundary, meaning that the inside of the boundary is completely unqualified, while the outside of the boundary is completely qualified.

[0105] In another possible embodiment, the process of identifying several undercompacted blocks based on compaction test data and determining the target geometric boundary of each undercompacted block can also be as follows:

[0106] Step 201a: Calculate the compaction gradient field based on the compaction test data. The compaction gradient field characterizes the degree of drastic change in compaction in space.

[0107] In this embodiment, the Sobel operator or the Prewitt operator is used to perform a convolution operation with the compaction grid matrix K to calculate the partial derivatives of compaction in the x-direction ΦK / Φx and y-direction ΦK / Φy, respectively. Correspondingly, the formula for calculating the compaction gradient magnitude G(x,y) at each grid point is:

[0108] G(x,y)=sqrt((ΦK / Φx) 2 +(ΦK / Φy) 2 );

[0109] Where sqrt represents the square root operation, Φ represents the partial derivative, and the gradient magnitude G(x,y) represents the change in compaction degree per unit distance. The larger the value, the more drastic the change in compaction degree at that location, which usually corresponds to the boundary between the qualified and unqualified areas.

[0110] Step 202a: Based on the compaction test data, the under-compacted core area is preliminarily determined.

[0111] Furthermore, the determined basic target geometric boundary is used as the initial seed, or the connected regions extracted from the binarized segmentation map are used as the undercompacted core region. The compaction degree within this core region is below the acceptable standard, and therefore requires recompaction.

[0112] Step 203a: Perform a point-by-point search outward along the edge normal of the undercompacted core region. When the gradient magnitude at the search location is less than the preset gradient convergence threshold and the compaction value is greater than the preset safety margin threshold, stop the search and obtain the extended edge.

[0113] Specifically, for each node on the boundary of the core area, the system extends the search outwards towards the qualified area along its normal direction. In actual engineering, there is usually a gradual transition zone from unqualified to qualified compaction, although this transition zone may numerically be slightly higher than K. _th However, it is structurally unstable and has a large gradient value. In order to achieve thorough recompaction, this embodiment shifts the boundary to a region where the compaction field tends to be stable.

[0114] Accordingly, the search must stop simultaneously to satisfy both the gradient convergence condition and the safety margin condition. The gradient convergence condition is satisfied, specifically G(x,y). <G _th G _thThe gradient convergence threshold indicates that the change in compaction degree has become gradual; the safety margin condition is met, i.e., K(x,y)≥K. _th +Δ, where Δ is the safety margin coefficient, indicating that the compaction degree has exceeded the acceptable level.

[0115] Step 204a: Determine the extended edge as the target geometric boundary.

[0116] Specifically, the system connects all search stopping points into closed loops, forming a new, larger polygon, which serves as the final optimized target geometric boundary. This boundary is more robust than the fixed threshold boundary and can reduce the edge defect rate in subsequent loop closure detection.

[0117] Based on this, to assist in subsequent path planning and sequencing, this embodiment further calculates the severity index of each undercompacted block, and the corresponding calculation formula is as follows:

[0118] SI _k =α×(Area(R _k' ) / Area _max )+β×((K _th -K avg k ) / K _th );

[0119] Among them, SI _k R is the severity index of the k-th block; _k 'This is the expanded block; Area(R) _k' Area represents the area of ​​the block. _max K represents the maximum area among all blocks, used for normalization. _avg k α represents the average compaction degree within the block; α and β are weighting coefficients. (Simplified Chinese) _k Arranging blocks in descending order can identify blocks that are large in area and have severe undervoltage.

[0120] In one alternative implementation, SI is... _k Initial sorting for the marginal benefit greedy strategy. Specifically, in the initialization phase of step 501, according to SI... _k The candidate priority of the first access block is determined in descending order so that the system prioritizes evaluating the block with the highest severity; when the marginal benefits are similar, SI is selected first. _k Larger blocks. Example 4

[0121] Based on the above embodiments, this embodiment further describes a path planning implementation method based on geometric area coverage. Specifically, this means that the road roller generates operational revenue as long as it covers the undercompacted area in geometric space.

[0122] In one possible implementation, the transfer path of the road roller between nodes is modeled as a work corridor with physical width, specifically including the following steps:

[0123] Step 301: Based on the minimum turning radius in the kinematic parameters of the road roller, generate a node transfer trajectory that satisfies the kinematic constraints; use the working width in the kinematic parameters of the road roller as a buffer distance to perform buffer analysis on the transfer trajectory, generate a strip polygon, and use the strip polygon as the working corridor.

[0124] Specifically, for any two nodes P _i and P _j For example, from the centroid of block A to the centroid of block B, the system determines the direction based on the minimum turning radius R of the roller. _min The Dubins curve algorithm is used to generate the shortest feasible path connecting two points. The Dubins curve is composed of a circular arc segment C and a straight line segment S. After determining the centerline trajectory, two parallel curves are generated by offsetting the center of this trajectory to the left and right by a distance of W / 2, where W is the working width. The closed area enclosed by the two parallel curves and the lines connecting the start and end points is the strip polygon. This polygon is used to simulate the actual contact area of ​​the roller's compaction wheel during the transfer process.

[0125] Step 302: Calculate the geometric area of ​​the intersection; calculate the original total area of ​​the region enclosed by the target geometric boundary; calculate the ratio of the geometric area to the original total area, and use the ratio as the incidental coverage gain.

[0126] In this embodiment, incidental coverage gain is defined as dimensionless area coverage. Specifically, let the transfer path be from node i to node j, and the resulting work corridor be T. _ij The k-th undercompacted block is R _k The system uses a polygon Boolean operation algorithm to calculate T. _ij With R _k The geometric intersection region. Let the area of ​​this intersection region be denoted as Area(Intersection). Calculate block R. _k Total area Area (R) _k Accordingly, the formula for calculating the incidental coverage gain is as follows:

[0127] C _ij k =Area(Intersection) / Area(R _k );

[0128] Among them, C _ij k For incidental coverage gain, its value ranges from 0 to 1, representing the proportion of the area that the transition path incidentally covers in the block.

[0129] Step 303: In the iterative process of solving the optimal recompression access order, all preceding transfer paths before reaching the current undercompacted block are determined according to the current candidate recompression access order; based on the incidental coverage gain of the preceding transfer paths to the undercompacted block, the cumulative coverage rate of the undercompacted block is calculated using a probabilistic complementary model; based on the cumulative coverage rate, the remaining uncovered proportion of the undercompacted block is calculated, and combined with the original total number of operations of the undercompacted block, the net increase in effective operations is calculated.

[0130] When determining the repacking order, a block may be swept by multiple transition paths sequentially. To estimate the cumulative effect, this embodiment uses a probabilistic complementary model. For example, assuming each coverage is an independent random event, for the k-th block, the cumulative coverage Cov after traversing t transition paths is... _t The corresponding expression is:

[0131] Cov _t =1-(1-C _1 k )×(1-C _2 k )×...×(1-C _t k );

[0132] Among them, (1-C _s k ) represents the probability that the s-th path does not cover the block, and the product represents the probability that the block is never covered by any path.

[0133] Based on this, the system calculates the net increase in effective workload for this block, using the following formula:

[0134] A _net k =Area(R _k )×(1-Cov _t );

[0135] Among them, A _net k This represents the additional area that the roller needs to compact within the block when specifically visiting it. Dividing this area by the working width W estimates the stroke length L required for zigzag compaction within the block. _cover Substitute it into the total cost function for optimization.

[0136] Example 5

[0137] This embodiment describes a value-based coverage path planning method based on compaction defect density. It elevates the optimization objective of path planning from geometric coverage area to engineering quality value. Specifically, by introducing a compaction defect density field and a grid state matrix, it identifies and processes areas with severe undercompaction, reducing coverage calculation errors when multiple paths overlap.

[0138] In one possible implementation, the incidental coverage gain of the transition path to the undercompacted block is calculated based on the intersection, specifically including:

[0139] Step 401: Based on the compaction test data, construct a compaction defect density field covering each undercompacted block. The compaction defect density in the compaction defect density field represents the degree to which the compaction at the corresponding location deviates from the qualified standard.

[0140] Specifically, in order to describe the urgency of recompaction in different regions, this embodiment no longer treats the undercompacted blocks as homogeneous geometric planes, but rather represents them as a mass defect field with thickness.

[0141] Furthermore, a compaction defect density field covering each undercompacted block is constructed, specifically including:

[0142] Step 4011: Obtain the preset compaction qualification threshold and calculate the difference between the compaction qualification threshold and the measured value at each test point in the compaction test data. Retain the difference when it is greater than zero; set the difference to zero when it is less than or equal to zero, thus obtaining the compaction defect density corresponding to each test point. Based on this, construct the compaction defect density field D. _k The formula for calculating the defect density value at any point (x, y) in the matrix is:

[0143] D _k (x,y)=max(0,K) _th -K(x,y));

[0144] Among them, D _k (x,y) represents the compaction defect density at coordinate (x,y); K _th is the acceptable compaction threshold; K(x,y) is the measured compaction value; the max function makes the defect density in the acceptable area zero, that is, it does not produce a negative contribution.

[0145] Step 402: Based on the compaction defect density field, calculate the cumulative compaction defect density within the intersection range as the defect elimination amount covered by the transfer path; based on the compaction defect density field, calculate the cumulative compaction defect density within the target geometric boundary range as the total defect amount of the undercompacted block; calculate the ratio of defect elimination amount to total defect amount, and use the ratio as incidental coverage gain.

[0146] In this embodiment, incidental coverage gain is redefined as value coverage. For the work corridor T formed by path (i→j) _ij and block R _k The system no longer calculates the intersection area, but instead performs spatial integration on the defect density within the intersection region, which is represented as a summation in the discrete grid to obtain the defect elimination amount. Correspondingly, the block R is calculated. _k Total defects Q _total The corresponding calculation formula is:

[0147] Gain=Q _covered / Q _total ;

[0148] Wherein, Gain represents the corridor in block R. _k The percentage of defects eliminated in Q _covered Q represents the defect elimination amount. _total For block R _k Total number of defects

[0149] Step 403: Establish a coverage status grid matrix for each undercompacted block, and initialize each grid cell in the matrix to an uncovered state; according to the recompaction access order, when a transfer path is determined, update the grid cells in the coverage status grid matrix located within the intersection of the work corridor and the target geometric boundary to a covered state; count the grid cells in the coverage status grid matrix that are still uncovered, and calculate the net increase in effective work volume based on the uncovered grid cells.

[0150] This embodiment abandons the use of a probabilistic model and instead employs a deterministic grid state matrix M for cumulative calculation. For each block R _k The system establishes a binary matrix M with the same resolution as the compaction data. _k Initially, all values ​​are 0. When a work corridor along a transfer path covers a grid (x, y), regardless of whether the grid was previously covered, the system marks its state as M. _k (x,y)=1.

[0151] When calculating the net increase in effective workload, the system only counts M. _k A grid with (x,y)=0. Net increase in defects Q. _net The calculation formula is as follows:

[0152] Q _net =∑(D _k (x,y))×ΔA;

[0153] Where ΔA is the area of ​​the grid cell, ∑(D _k (x,y) represents all conditions M satisfying M _kThe summation of (x,y) with (x,y)=0 is used to reduce the error of repeated calculations when multiple paths intersect and overlap.

[0154] Furthermore, when the net increase in effective work volume equals the net increase in compaction defects, the sum of the work strokes after conversion of the net increase in effective work volume for each under-compacted block is calculated as follows:

[0155] Step 404: Calculate the average compaction defect density of the uncovered area within each undercompacted block; divide the net increase in compaction defect by the working width in the roller's kinematic parameters, and then divide by the average compaction defect density to obtain the equivalent working stroke for each undercompacted block; sum the equivalent working strokes for all undercompacted blocks.

[0156] Since the transfer travel in the objective function is in meters, while the calculated net increase in effective work volume is measured by the integral value of compaction defects, the two cannot be directly added together. To eliminate this difference in dimensions, this embodiment introduces a dimension conversion.

[0157] Specifically, calculate the average defect density D of the remaining uncovered area within block k. _avg The corresponding calculation formula is:

[0158] D _avg =Q _net / Area _net ;

[0159] Among them, Area _net This is the total area of ​​the uncovered grid. Accordingly, the system uses this average density to convert the net increase in defects into an equivalent physical travel L. _cover :

[0160] L _cover =Q _net / (W×D _avg );

[0161] This formula is expressed as follows when the average defect density is D. _avg Eliminate the Q region _net The amount of defects, and the distance the road roller needs to travel.

[0162] Furthermore, during the execution phase, for this L _cover For the corresponding blocks, the system plans to use a bow-shaped path for full-coverage compaction to eliminate any remaining defects.

[0163] Example 6

[0164] This embodiment integrates the calculated gain and cost into a unified algorithm framework to generate the final executable instructions, and achieves a closed-loop process through a quality feedback mechanism.

[0165] In one possible implementation, the optimal complex pressure access order is generated by employing a marginal benefit greedy strategy, which includes:

[0166] Step 501: Initialize the set of unvisited blocks and the current location of the road roller; traverse each candidate block in the set of unvisited blocks and calculate the marginal benefit of using the candidate block as the next access target.

[0167] In this embodiment, the system maintains an unvisited set U = {1, 2, ..., n}. At each decision step, the system needs to evaluate the impact of the current position P... _cur The cost-effectiveness of traveling to each candidate block k, i.e., the marginal benefit η. _k .

[0168] The marginal benefit is directly proportional to the numerator and inversely proportional to the denominator. The numerator includes the net increase in effective operations after reaching the candidate block, and the sum of the incidental coverage gain of the transfer path from the current location to the candidate block on other unvisited blocks. The denominator includes the length of the transfer path from the current location to the candidate block.

[0169] When the value-coverage method is used, the marginal benefit η _k The calculation formula is as follows:

[0170] η _k =(Q _k,_net +∑(Q _j,_incidental )) / L _trans (cur,k);

[0171] Among them, Q _k,_net This is the net increase in defects eliminated through pre-defined operations after reaching block k; ∑(Q _j,_incidental ) is the sum of defects eliminated incidentally by the transition path (cur→k) on all other blocks j in set U; L _trans (cur,k) is the transition path length.

[0172] Step 502: Select the candidate block with the greatest marginal benefit as the next access target and remove it from the set of unvisited blocks until the set of unvisited blocks is empty.

[0173] In each iteration, argmax(η) is selected. _k The corresponding block is used as the next node, and the current position P is updated. _cur Based on the selected transition path, update the raster state matrix or cumulative coverage of all remaining blocks, and repeat the process until a complete initial access sequence is generated.

[0174] Furthermore, solving for the optimal complex pressure access order also includes local optimization of the initial sequence generated using a greedy strategy with marginal benefits, specifically including:

[0175] Step 503: Traverse any two non-adjacent transition paths in the initial sequence and perform path swapping to generate neighborhood solutions; calculate the job cost corresponding to the neighborhood solution based on the objective function of the total job cost; if the job cost corresponding to the neighborhood solution is lower than the job cost of the current sequence, update the current sequence using the neighborhood solution until the job cost cannot be further reduced through path swapping, and obtain the optimal complex pressure access order.

[0176] The initial sequence represents the preliminary access sequence of the compressed blocks generated solely through a greedy strategy based on marginal benefits. It is used for subsequent local optimization as an approximate optimal solution under the greedy strategy, not the final globally optimized compressed access sequence, as the greedy algorithm may get trapped in local optima. Therefore, in this embodiment, a 2-opt local search algorithm is introduced to correct the initial solution. Specifically, the system attempts to disconnect two edges in the path sequence, such as u→v and x→y, and reconnect u→x and v→y, checking whether this exchange can reduce the total operation cost J(θ). If the new cost is lower, the exchange is accepted. This process is repeated until no exchange can further reduce the cost.

[0177] In some embodiments, after instructing the roller to perform a re-compaction operation, the method further includes:

[0178] Step 504: Trigger the intelligent compaction system to perform secondary detection and obtain updated compaction test data; Based on the updated compaction test data, calculate the compaction failure rate of the current area; Determine whether the iteration termination condition is met. The iteration termination condition includes the compaction failure rate being lower than the preset standard or the iteration round reaching the preset upper limit; If the iteration termination condition is not met, use the updated compaction test data as input and return to the step of identifying several under-compacted blocks based on the compaction test data to generate a new round of re-compaction path.

[0179] After the road roller completes one round of compaction, the system automatically collects new data. The iteration termination condition includes a hard indicator, namely the non-compliance rate R. _fail Is the value 0 or lower than the preset allowed value, or has the maximum number of iterations been reached?

[0180] Furthermore, determining whether the iteration termination condition is met also includes:

[0181] Step 505, in other words, calculate the compaction improvement rate for this round based on the compaction failure rate of the previous round and the compaction failure rate of the current area; determine whether the compaction improvement rate for this round is lower than the preset improvement threshold; if it is lower than the improvement threshold, it is determined that the iteration termination condition is met, stop generating a new round of re-compaction path, and output abnormal area alarm information.

[0182] In this embodiment, an improvement rate judgment mechanism is introduced to prevent the system from performing ineffective cycles in hard, non-compactable areas or springy soil areas. The formula for calculating the improvement rate is as follows:

[0183] ΔR=(R _prev -R _cur ) / R _prev ;

[0184] Where ΔR is the improvement rate, R _prev R represents the non-compliance rate from the previous round. _cur This represents the non-compliance rate for the current round. If ΔR is lower than the preset improvement threshold, it indicates that the previous round of re-compaction had almost no effect, the soil compaction degree has reached its limit, or there are soil abnormalities. Accordingly, the system terminates the iteration, outputs the alarm coordinates of the abnormal area, or suggests that construction personnel take physical treatment measures such as replacement, drying, or mixing with lime.

[0185] This embodiment introduces a compaction defect density field and a value-weighted evaluation model. By describing compaction deviation as defect density, it upgrades area coverage to defect value elimination. Combined with a greedy marginal benefit solution strategy, the system can plan a path through severely undercompacted core areas. This addresses the lack of differentiated evaluation of defect severity, allowing high-risk areas to be addressed, reducing the rework rate caused by substandard recompaction in the core area, and decreasing the number of iterations required for closed-loop detection.

[0186] This invention models the transfer path of a road roller between nodes as a working corridor with physical width and calculates the spatial intersection of the working corridor and the undercompacted blocks, describing the incidental compaction value of the transfer path. This weakens the binary opposition between transfer and operation in traditional methods, allowing the transfer path itself to be included in the calculation of effective operation contribution, reducing redundancy in the total operation journey. By dynamically calculating the net increase in effective operation volume of each block based on incidental coverage gain and incorporating it and the transfer journey into the objective function for global optimization, this invention avoids the local optimum problem of traditional methods that only minimize the transfer distance while ignoring changes in operation volume, and achieves overall optimization of the total operation cost including transfer and operation.

[0187] The embodiments of the present invention have been described in detail above. However, the present invention is not limited to the specific details of the above embodiments. Within the scope of the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all fall within the protection scope of the present invention.

Claims

1. A method for automatic identification of undercompacted areas and optimization of recompaction paths, characterized in that, include: Obtain compaction test data and kinematic parameters of the roller in the compaction operation area; Several undercompacted blocks were identified based on compaction test data, and the target geometric boundaries of each undercompacted block were determined. Using the path planning model, the centroid of each undercompacted block is determined as the node to be visited, the transfer path of the roller between nodes is modeled as a working corridor with physical width, and the path length of each transfer path is calculated. The spatial intersection of the area enclosed by the operation corridor and the target geometric boundary is calculated, and the incidental coverage gain of the transfer path on the undercompacted block is calculated accordingly. Based on incidental coverage gain, determine the net increase in effective workload when each undercompacted block is accessed; The optimal multi-pressure access sequence is generated by minimizing the total operating cost, including the transfer travel and the net increase in effective workload. Based on this, control commands are generated to guide the road roller to perform secondary compaction operations; The transfer path of the road roller between nodes is modeled as a work corridor with physical width, specifically including: Based on the minimum turning radius in the kinematic parameters of the road roller, an inter-node transfer trajectory that satisfies the kinematic constraints is generated; Using the working width in the kinematic parameters of the road roller as the buffer distance, a buffer analysis is performed on the transfer trajectory to generate a strip polygon, which is then used as the working corridor. The calculation of the incidental coverage gain of the transition path on the undercompacted block includes: Based on compaction test data, a compaction defect density field covering each undercompacted block is constructed; Based on the compaction defect density field, the cumulative compaction defect density within the intersection range is calculated as the defect elimination amount covered by the transfer path; Based on the compaction defect density field, the cumulative compaction defect density within the target geometric boundary is calculated as the total defect amount of the undercompacted block; Calculate the ratio of defect elimination amount to total defect amount, and use the ratio as incidental coverage gain; Determine the net increase in effective workload for each undercompacted block when it is accessed, specifically including: Based on the current candidate recompression access order, determine all preceding transition paths before reaching the current undercompressed block; Based on the incidental coverage gain of the preceding transition path to the undercompacted block, the cumulative coverage rate of the undercompacted block is calculated using a probabilistic complementary model. The remaining uncovered proportion of under-compacted blocks is calculated based on the cumulative coverage rate. Combined with the original total amount of work done in the under-compacted blocks, the net increase in effective work is calculated.

2. The method according to claim 1, characterized in that, The incidental coverage gain of the transition path to the undercompacted block is calculated based on the intersection, specifically including: Calculate the geometric area of ​​the intersection; Calculate the original total area of ​​the region enclosed by the target's geometric boundaries; Calculate the ratio of the geometric area to the original total area, and use this ratio as the incidental coverage gain.

3. The method according to claim 1, characterized in that, Construct a compaction defect density field covering each undercompacted block, specifically including: Obtain the preset compaction degree qualification threshold; Calculate the difference between the acceptable compaction threshold and the measured values ​​at each test point in the compaction test data; When the difference is greater than zero, the difference is retained; when the difference is less than or equal to zero, the difference is set to zero, thus obtaining the compaction defect density corresponding to each detection point, and constructing a compaction defect density field accordingly.

4. The method according to claim 1, characterized in that, Determine the target geometric boundaries for each undercompacted block, specifically including: Obtain the preset compaction degree qualification threshold; Compaction degree test data is compared with the compaction degree qualified threshold point by point to generate a binary segmentation map; Connectivity analysis was performed on the binarized segmented graph, and the extracted connected regions were identified as undercompacted blocks. Extract the edge contours of the connected regions and define the edge contours as the target geometric boundaries.

5. The method according to claim 1, characterized in that, Determine the target geometric boundaries for each undercompacted block, specifically including: The compaction gradient field is calculated based on compaction test data. The compaction gradient field characterizes the degree of drastic change in compaction in space. Based on compaction test data, the under-compacted core area was initially identified; The search proceeds point by point outward along the edge normal of the undercompacted core region. When the gradient magnitude at the search location is less than the preset gradient convergence threshold and the compaction value is greater than the preset safety margin threshold, the search stops and the extended edge is obtained. The extended edge is defined as the target geometric boundary.

6. The method according to claim 1, characterized in that, Solving for the optimal complex pressure access order involves employing a marginal benefit greedy strategy, including: Initialize the set of unvisited blocks and the current location of the road roller; Traverse each candidate block in the set of unvisited blocks and calculate the marginal benefit of making it the next target for access; Marginal benefits are directly proportional to the numerator and inversely proportional to the denominator; the numerator includes the net increase in effective operations after reaching the candidate block, and the sum of the incidental coverage gain of the transfer path from the current location to the candidate block on other unvisited blocks; the denominator includes the length of the transfer path from the current location to the candidate block. Select the candidate block with the greatest marginal benefit as the next access target, and remove it from the set of unvisited blocks until the set of unvisited blocks is empty.

7. A system for automatic identification and recompaction path optimization of undercompacted areas, characterized in that, The system includes a processor coupled to a memory storing instructions which, when executed by the processor, implement the method as described in any one of claims 1 to 6.

Images