An automatic calculation method for the amount of dredging work in highway construction

Abstract

The present invention discloses a method for automatically calculating the amount of dredging work in highway construction, comprising the following steps: S1. Using an unmanned aerial vehicle (UAV) to conduct aerial surveys of the highway construction area before and after dredging, collect raw data, and compile a range file defining the boundaries of the dredging area; S2. Preprocessing the raw data and performing range file verification; S3. Constructing a triangulated ground model based on the range file and ground point cloud data, and identifying the dredging area using a residual neural network model; S4. Using a spider colony algorithm, each spider individual corresponds to a set of residual network structure parameters; S5. Each spider individual updates its position based on its own optimal parameters and those of its neighboring spider individuals; S6. Pixel-level segmentation of the dredging area based on the residual neural network structure parameters; and S7. Generating a preview file of the results based on the calculation results. The present invention combines the residual neural network model, the spider colony algorithm, and three-dimensional volume estimation technology to achieve automatic calculation of the amount of dredging work in highway construction.

Description

Technical Field

[0001] The present invention relates to the field of highway construction, and in particular to an automatic calculation method for dredging engineering quantities in highway construction. Background Art

[0002] In highway construction projects, the calculation of construction quantities is not only the foundation for cost estimation but also a crucial factor in project management, schedule control, and quality assessment. Dredging, as a crucial component of the foundation construction phase, requires accurate calculation of its quantities, which is directly related to the rationality of the overall project budget and the effectiveness of construction scheduling. However, existing technologies still rely primarily on traditional manual measurement methods for obtaining and calculating dredging quantities. This method faces challenges in efficiency, accuracy, safety, and data integrity, severely hindering the development of information technology and automation in modern highway construction.

[0003] The traditional method of calculating the quantity of dredging projects is mainly based on manual on-site measurement. GPS-RTK equipment is usually used for field sampling. The terrain information of the area to be dredged is obtained by point measurement. Then, manual calculation or CAD modeling software is used to estimate the cross section and finally complete the volume calculation. This method has significant bottlenecks in the operational process. First, the data collection efficiency is extremely low. In large-scale linear engineering scenarios, such as highways, mountain tunnel entrances and exits, or dredging areas around culverts, manual methods are used to complete the full-area point distribution. It is not only time-consuming and sparsely covered, but the sampling density cannot meet the requirements of refined measurement. In order to improve the accuracy of engineering quantity accounting, a large amount of point information is required. However, this high-density sampling is almost impossible to complete under traditional methods, and the progress of the project is therefore seriously affected.

[0004] Secondly, traditional methods present significant safety risks. Dredging construction sites are often plagued by soft soil, stagnant water, and accumulated silt. This complex environment, particularly in scenarios like river channel reconstruction and ditch excavation, makes manual measurement prone to becoming bogged down in the mud. This poses a threat to the safety of workers and increases the risk of equipment failure. Furthermore, due to the dramatic topography of the dredging area, instrument placement and securing are difficult, making them susceptible to mud interference. This leads to significant deviations in measurement data, frequent repetitive work, and difficulty ensuring measurement accuracy.

[0005] Thirdly, data quality under traditional measurement models is difficult to systematically guarantee, leaving ample room for human intervention and untraceable errors. The recording, transmission, and processing of measurement results often rely on manual transcription and subsequent manual modeling, resulting in poor data integrity and weak logical connections. This makes it easy for human negligence to cause information loss or error accumulation, thus compromising the accuracy and reliability of the entire desilting volume calculation. Against the backdrop of increasing demands for digital engineering construction and full-process supervision, relying on traditional manual methods for desilting volume calculations is clearly unable to meet the precision, efficiency, and intelligence requirements of modern engineering.

[0006] Based on this, existing technologies have gradually explored the use of drone aerial surveys, lidar scanning, point cloud modeling, and other means to acquire three-dimensional data. However, most methods remain at the data acquisition stage and have yet to form an automated engineering quantity accounting system that integrates "collection-processing-identification-calculation-upload." In particular, the identification of dredging engineering quantities still has the following shortcomings: First, the identification of dredging area boundaries relies on manual annotation, which cannot adapt to complex environments such as different landforms, materials, and occlusion conditions. Second, the modeling process after the fusion of point cloud and image data lacks an intelligent structural parameter tuning mechanism, resulting in weak generalization of the recognition model and prone to misjudging or missing the dredging scope in complex terrain. Third, the calculation of dredging volume fails to effectively combine recognition accuracy, boundary fitting, and point density factors. Volume calculation still relies on static rules or empirical models and lacks intelligent feedback capabilities.

[0007] Therefore, how to provide an automatic accounting method for the desilting engineering quantity in highway construction is a problem that those skilled in the art urgently need to solve. Summary of the Invention

[0008] One purpose of the present invention is to propose a method for automatic accounting of dredging engineering quantities in highway construction. The present invention combines a residual neural network model, a spider colony algorithm and a three-dimensional volume estimation technology. By constructing a multi-layer convolution structure to extract the spatial boundary features of the dredging area, and using jump connections to enhance the edge information retention capability, high-precision identification of dredging areas under complex terrain is achieved. At the same time, a swarm intelligence optimization algorithm that simulates the vibration behavior of social spiders is introduced to dynamically tune the residual network structure parameters so that the model can adapt to the recognition requirements in different dredging scenarios. The system establishes a spatial voxel structure based on three-dimensional point cloud data, performs volume calculation based on the recognition results, and realizes automatic accounting and result preview. This method has the characteristics of high recognition accuracy, strong parameter self-optimization capability, rigorous volume calculation, and full process automation, which effectively improves the efficiency and reliability of engineering quantity accounting.

[0009] According to an embodiment of the present invention, a method for automatically calculating the amount of dredging work in highway construction includes the following steps:

[0010] S1. Use drones to conduct aerial surveys of the highway construction area before and after desilting, collect raw data, and collect scope files defining the desilting area boundaries;

[0011] S2. Preprocess the original data and perform range file verification to generate ground point cloud data of the desilting area;

[0012] S3. Construct a triangulated ground model based on the range file and ground point cloud data, input the triangulated ground model into the residual neural network model, and output the initial mask map of the dredging area. Each residual module of the residual neural network model extracts the features of the dredging area through the convolution layer and retains the spatial boundary information through the skip connection technology;

[0013] S4. Using the spider colony algorithm, the search direction of the optimal parameters is guided according to the dredging area recognition accuracy and the initial mask map. In the spider colony algorithm, each spider individual corresponds to a set of residual network structure parameters.

[0014] S5, the spider individual updates its position according to the optimal parameters of itself and neighboring spider individuals until the population converges, and outputs the optimized residual neural network model structure parameters;

[0015] S6. Perform pixel-level segmentation of the desilting area based on the optimized residual neural network model, calculate the three-dimensional total volume of the desilting area based on the point cloud spatial coordinate information, and output the calculation results of the desilting engineering quantity;

[0016] S7. Generate a result preview file based on the calculation results.

[0017] Optionally, the raw data specifically includes RGB images, lidar point cloud data, ground check point data, spatial coordinates, inertial measurement unit, flight path planning and timestamp.

[0018] Optionally, the desilting area characteristics specifically include height changes, area undulations and desilting traces.

[0019] Optionally, the S2 specifically includes:

[0020] S21. Convert the format of the original data into a universal .las ground point cloud data format;

[0021] S22, performing a quality check on the ground point cloud data, using the collected ground check point data to perform point cloud precision check and calculation, and removing noise points and abnormal data through a filtering algorithm;

[0022] S23. Based on the scope document, the construction unit shall verify whether the scope is qualified according to the set measurement criteria including area size, construction depth and volume calculation standards.

[0023] Optionally, the S3 specifically includes:

[0024] S31, read the specified area boundary data from the range file to determine the ground area range to be processed;

[0025] S32, triangulating the processed ground point cloud data using a triangulation algorithm, connecting every three adjacent ground points into a triangle, and generating a triangulated ground model covering the ground;

[0026] S33, clipping the triangulated ground model according to the boundary provided by the scope file, removing the portion outside the boundary, and retaining only the ground data within the desilting area;

[0027] S34. For the cropped triangulated ground model, adjust the size and angle of the triangles to eliminate triangles with an area greater than 5 square meters and acute triangles with a minimum internal angle less than 20 degrees.

[0028] S35. Input the adjusted triangulated ground model into the residual neural network model, and output an initial mask map of the dredging area.

[0029] Optionally, the S35 specifically includes:

[0030] S351, the triangulated ground model is passed into the residual neural network model as input data, and the convolution layer uses nodes to extract the desilting area features from the input triangulated ground model;

[0031] ;

[0032] in, is the desilting area feature of node i in the l-th layer residual neural network model, is the neighbor node set of node i, i is a node in the triangulated ground model, For the The convolution kernel weight matrix of the layer that weights nodes i and j, is the degree of node i, is the degree of node j, For the The desilting area characteristics of node j in the layer, For the The bias term of the layer;

[0033] S352. Skip connection technology is used in the convolution operation to preserve the spatial boundary information of the dredging area. Skip connection allows input data to be directly transferred to the output data stream;

[0034] S353, generating multiple feature maps through convolution operations, wherein the feature maps contain different spatial features of the desilting area, including the shape of the desilting area, the fineness of the boundary, and the depth change;

[0035] S354 flattens the two-dimensional or three-dimensional feature map generated by the convolution operation, that is, converts the pixel values of each feature map into a one-dimensional vector in sequence, and then passes the one-dimensional vector as input to the fully connected layer;

[0036] S355, the fully connected layer combines each input node received with the set weight matrix Perform matrix multiplication and generate an initial mask map of the desilting area.

[0037] Optionally, the S4 specifically includes:

[0038] S41. Each spider individual corresponds to a set of residual network structure parameters, including convolution kernel size, number of residual modules, number of feature channels and learning rate;

[0039] S42. The fitness value of each spider individual is determined by comparing the ratio of the intersection and union between the generated initial mask map and the manually marked dredging area. A ratio greater than 0.5 indicates that the recognition result of the residual neural network model is close to the actual dredging area and the fitness value is high;

[0040] S43, the spider group simulates the vibration signal propagation behavior of social spiders based on the fitness value of each individual spider, so that the spider individuals with high fitness values guide the spiders to move to areas with high fitness values;

[0041] S44. The spider individual guides the search direction of the optimal parameters by guiding the spider to move towards the area with high fitness value:

[0042] ;

[0043] in, is the vibration signal propagation state within the spider colony, In the propagation state The search guidance direction vector of the spider individual p is: is the set of all nodes in the graph structure of the residual neural network model, i is a node in the triangulated ground model, N(i) is the set of adjacent nodes of node i, For the The desilting area characteristics of node i in the layer residual neural network model, For the The desilting area feature of node j in the layer residual neural network model, j is a neighboring node adjacent to node i, For the propagation state The guidance weight of node j to node i.

[0044] Optionally, the S5 specifically includes:

[0045] S51. In each iteration, the individual spider compares its current fitness value with the fitness value of the best individual spider in the neighborhood, and performs a round of position update operation in the structural parameter space based on the current individual position and the guidance direction of the neighboring individuals;

[0046] S52. The position of each spider individual represents a set of structural parameters of the residual neural network model. The spider individual updates its position based on the optimal fitness value of itself and its neighboring individuals:

[0047] ;

[0048] in, is the position of spider individual p in generation t, is the best position of spider individual q in history, is the global best position obtained by the entire spider population, 、 and is the acceleration constant, is the position of spider individual p in the t+1 generation, is the neighborhood individual set of spider individual p, p is the individual number in the spider group, g is the current global optimal spider individual number, is the coupling influence weight of spider individual q on p in the current iteration, To estimate the long-term average drift trend of historical position changes, is the position change of spider individual p in the i-th iteration, t is the number of the current iteration round, is the number of historical iterations;

[0049] S53, the individual spider updates its position through multiple iterations and adjusts its search direction according to the recognition accuracy after each iteration until the population reaches the maximum number of iterations, 100;

[0050] S54. During each iteration, the individual spider updates its position according to the dredging area recognition accuracy and the initial mask map, while monitoring the improvement of the dredging area recognition effect;

[0051] S55. After the iteration is completed, the optimized residual neural network model structure parameters are output.

[0052] Optionally, the S6 specifically includes:

[0053] S61. Based on the optimized residual neural network model, perform layer-by-layer convolution operations on the triangulated ground model of the desilting area to obtain the classification results of each pixel point and output a binary mask image. Each pixel position only has two labels: desilting area or non-desilting area.

[0054] S62: extract all pixel points marked as desilting areas in the binary mask image, match them with corresponding coordinate points in the ground point cloud data, and construct a desilting area spatial point set. The desilting area spatial point set consists of the three-dimensional coordinates corresponding to all points determined to be desilting areas in the binary mask image;

[0055] S63, dividing the three-dimensional space with fixed side lengths, constructing a regular voxel grid, projecting and mapping the spatial point set of the desilting area onto the voxel grid, determining whether each voxel contains at least one valid spatial point, and counting the number of all occupied voxel units;

[0056] S64. Calculate the volume of each voxel based on the side length of the voxel unit, and sum the volumes of all occupied voxel units to obtain the three-dimensional total volume of the desilting area:

[0057] ;

[0058] in, is the total three-dimensional volume of the desilting area, K is the total number of all divided spatial voxels, is the spatial partition accuracy threshold, is the index of a valid spatial point in the spatial voxel unit k, k is the kth spatial voxel unit obtained by dividing the space, is the kth spatial voxel The volume weighting factor of each point, is the spatial volume of the k-th spatial voxel unit, is the number of valid points contained in the k-th spatial voxel, is the position of spider individual p in generation t+1;

[0059] S65, extracting the boundary contour of the desilting area based on the edge pixel points of the desilting area in the binary mask image, and corresponding the contour to the coordinates of the three-dimensional space point set to form a boundary map of the desilting area;

[0060] S66: Arrange the identified binary mask image, the boundary image of the desilting area, and the three-dimensional total volume of the desilting area into a unified data structure, and output it as the calculation result of the desilting engineering quantity.

[0061] Optionally, the S7 specifically includes:

[0062] S71. Outputting the binary mask map in the calculation result of the desilting engineering quantity into two-dimensional image data. Specifically, encoding the binary mask map into a layer image format file, and performing spatial registration and color overlay with the image of the drone aerial survey of the highway construction area before desilting, to generate visual two-dimensional image data.

[0063] S72, encapsulating the boundary map of the desilting area, the three-dimensional total volume of the desilting area, and the visualized two-dimensional image data into a structured data result file;

[0064] S73. Generate a unique task identification number for each set of data in the structured data result file, bind the structured data result file to the task identification number, and upload it to a remote database;

[0065] S74. Indexing the task identification number in the remote database, loading the structured data result file on the user interface, displaying the visualized two-dimensional image data as the background image, overlaying the boundary map of the desilting area in wireframe form, and displaying the three-dimensional total volume statistics of the desilting area on the numerical panel. The three together constitute a visualization display structure for the results;

[0066] S75: Generate a result preview number for each set of data in the structured data result file based on the desilting identification time, encapsulate the visualized two-dimensional image data, the boundary map of the desilting area, and the three-dimensional total volume of the desilting area together with the corresponding task identification number and result preview number to generate a result preview file;

[0067] S76. The results preview file is displayed in a unified format in the accounting interface, and supports online viewing, downloading and archiving of the results preview file and task archiving operations through the results preview number.

[0068] The beneficial effects of the present invention are:

[0069] This invention proposes a method and tool for automatically calculating dredging quantities in highway construction. This method systematically addresses prominent issues such as low efficiency, poor accuracy, significant safety hazards, and data non-traceability in traditional dredging quantity calculation processes, significantly improving the scientific and intelligent level of quantity calculation in dredging construction. By integrating key technologies such as drone aerial surveying, high-precision point cloud acquisition, triangulated ground modeling, residual neural network recognition, and spider colony algorithm optimization, this invention establishes a fully automated quantity calculation system, from raw data collection, recognition modeling, quantity accounting, to output results. This effectively reduces the reliance on manual operations in traditional quantity calculation methods, improves data processing efficiency, and improves the objectivity of calculation results.

[0070] In terms of efficiency, this invention uses an unmanned aerial vehicle system equipped with lidar or image sensors to conduct rapid, comprehensive aerial surveys of the desilting area, enabling the rapid acquisition of high-density point cloud data over a large area. Compared to traditional GPS-RTK-based point acquisition methods, aerial surveys significantly reduce manpower input and significantly shorten field measurement time. Furthermore, this invention combines deep learning recognition with 3D modeling analysis during the internal processing phase to achieve efficient, batch-based calculation of desilting project quantities, improving overall project scheduling efficiency.

[0071] In terms of precision control, the present invention uses a residual neural network to perform multi-level recognition of spatial features in triangulated ground models, and uses a spider colony algorithm to adaptively optimize network structural parameters. This allows the model to maintain strong recognition and generalization capabilities across diverse terrain structures and desilting scenarios. The optimized neural network structure allows for more accurate boundary extraction of desilting areas, and the three-dimensional volume calculations performed using a point cloud coordinate system are more consistent and stable, effectively avoiding the underestimation or overestimation of volume that can occur in traditional manual calculations due to empirical bias or data acquisition errors.

[0072] In terms of safety, this invention fully considers the high-risk environments often found in muddy and waterlogged dredging operations. By replacing traditional on-site surveying with remote aerial surveying, it significantly reduces the need for personnel to enter hazardous areas and enhances safety during operations. Furthermore, the collected raw data and calculation results are fully uploaded to a cloud-based calculation platform. This data is automatically recorded and transmitted back, ensuring excellent traceability and integrity, avoiding the risk of data falsification, omissions, or tampering due to human intervention. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0074] Figure 1 This is a flow chart of a method for automatically calculating the amount of dredging work in highway construction proposed by the present invention;

[0075] Figure 2 This is a schematic diagram of an automatic calculation method for dredging engineering quantities in highway construction proposed by the present invention;

[0076] Figure 3 This is a data flow diagram of an automatic calculation method for dredging engineering quantities in highway construction proposed by the present invention. DETAILED DESCRIPTION

[0077] The present invention will now be described in further detail with reference to the accompanying drawings, which are simplified schematic diagrams that illustrate the basic structure of the present invention in a schematic manner.

[0078] refer to Figure 1-3 A method for automatically calculating the amount of dredging work in highway construction comprises the following steps:

[0079] S1. Use drones to conduct aerial surveys of the highway construction area before and after desilting, collect raw data, and collect scope files defining the desilting area boundaries;

[0080] S2. Preprocess the original data and perform range file verification to generate ground point cloud data of the desilting area;

[0081] S3. Construct a triangulated ground model based on the range file and ground point cloud data, input the triangulated ground model into the residual neural network model, and output the initial mask map of the dredging area. Each residual module of the residual neural network model extracts the features of the dredging area through the convolution layer and retains the spatial boundary information through the skip connection technology;

[0082] S4. Using the spider colony algorithm, the search direction of the optimal parameters is guided according to the dredging area recognition accuracy and the initial mask map. In the spider colony algorithm, each spider individual corresponds to a set of residual network structure parameters.

[0083] S5, the spider individual updates its position according to the optimal parameters of itself and neighboring spider individuals until the population converges, and outputs the optimized residual neural network model structure parameters;

[0084] S6. Perform pixel-level segmentation of the desilting area based on the optimized residual neural network model, calculate the three-dimensional total volume of the desilting area based on the point cloud spatial coordinate information, and output the calculation results of the desilting engineering quantity;

[0085] S7. Generate a result preview file based on the calculation results.

[0086] This invention uses a spider colony algorithm to optimize the structural parameters of the residual neural network. By simulating the coordinated vibration behavior of individual spiders, it guides the structural search direction and achieves adaptive structural optimization of the network in the task of identifying dredging areas. Based on point cloud data obtained by drone aerial surveys, the system constructs a triangulated ground model. The residual neural network extracts spatial features layer by layer, retains boundary information through jump connections, and outputs an initial mask map. Combining pixel-level recognition and three-dimensional spatial reconstruction technology, it accurately calculates the volume of the dredging area. It automatically generates a preview file of the results, comprehensively improving the intelligence, automation, and precision of engineering calculations.

[0087] In this embodiment, the raw data specifically includes RGB images, lidar point cloud data, ground check point data, spatial coordinates, inertial measurement unit, flight path planning and timestamp.

[0088] In this invention, the raw data includes RGB images, LiDAR point cloud data, ground checkpoint data, spatial coordinates, inertial measurement unit information, flight path planning, and timestamps, forming a data foundation for multi-source fusion. By introducing a spatiotemporal collaborative calibration mechanism, precise alignment of images and point clouds within the same coordinate system is achieved, improving the spatial consistency of subsequent modeling and recognition. Utilizing timestamps and path information, the acquisition process can be effectively tracked, supporting data backtracking and verification, improving the accuracy and controllability of overall data processing, and providing reliable guarantees for the high-precision automatic calculation of dredging project quantities.

[0089] In this embodiment, the desilting area characteristics specifically include height changes, regional undulations and desilting traces.

[0090] In the present invention, the characteristics of the dredging area include height changes, regional undulations, and dredging traces, and a multi-dimensional spatial feature expression system is constructed. By capturing terrain height differences and local slope information, combined with the surface disturbance characteristics after dredging operations, the boundaries of the dredging area can be accurately extracted. This method avoids relying solely on a single height value for judgment, improving the robustness and integrity of regional identification. After multi-feature fusion is input into the residual neural network, the model's ability to perceive complex landform changes is enhanced, effectively improving the accuracy of dredging identification and providing an accurate basis for volume calculation and construction assessment.

[0091] In this embodiment, S2 specifically includes:

[0092] S21. Convert the format of the original data into a universal .las ground point cloud data format;

[0093] S22, performing a quality check on the ground point cloud data, using the collected ground check point data to perform point cloud precision check and calculation, and removing noise points and abnormal data through a filtering algorithm;

[0094] S23. Based on the scope document, the construction unit shall verify whether the scope is qualified according to the set measurement criteria including area size, construction depth and volume calculation standards.

[0095] The present invention achieves data quality assurance before calculating the dredging engineering quantity by constructing a point cloud data standardization and automatic verification process. The original data is uniformly converted into a universal .las format to ensure multi-platform compatibility, and is combined with ground verification points for precision verification to effectively eliminate abnormalities and non-ground points. The system uses filtering and classification algorithms to remove noise and thinning, extract stable ground point clouds, and improve the integrity and accuracy of ground modeling. At the same time, the scope files submitted by the construction unit are automatically compared with the measurement criteria in the quantity calculation platform, covering the area range, depth setting and volume calculation standards, to achieve accurate verification of the calculation boundaries. This method improves the data consistency, accuracy and intelligence level in the point cloud preprocessing stage, provides high-quality basic data support for subsequent identification and volume calculation, and ensures that the engineering quantity calculation is more standardized and reliable.

[0096] In this embodiment, S3 specifically includes:

[0097] S31, read the specified area boundary data from the range file to determine the ground area range to be processed;

[0098] S32, triangulating the processed ground point cloud data using a triangulation algorithm, connecting every three adjacent ground points into a triangle, and generating a triangulated ground model covering the ground;

[0099] S33, clipping the triangulated ground model according to the boundary provided by the scope file, removing the portion outside the boundary, and retaining only the ground data within the desilting area;

[0100] S34. For the cropped triangulated ground model, adjust the size and angle of the triangles to eliminate triangles with an area greater than 5 square meters and acute triangles with a minimum internal angle less than 20 degrees.

[0101] S35. Input the adjusted triangulated ground model into the residual neural network model, and output an initial mask map of the dredging area.

[0102] The present invention realizes the spatial structured expression of ground point cloud data by constructing a triangulated ground model, providing geometric support for the identification of dredging areas. The system reads the dredging boundary information from the range file, accurately locates the area to be processed, and uses the triangulation algorithm to convert the ground point cloud into a grid structure, thereby improving data processing efficiency and spatial expression integrity. By automatically clipping the model boundary, only valid data within the dredging area is retained, avoiding irrelevant terrain interference and enhancing recognition focus. At the same time, the system optimizes the morphology of the triangular units, eliminates sharp angles and lengthy structures, and improves the stability of the subsequent recognition stage. The residual neural network uses the triangulated network model as input, deeply extracts spatial features, and realizes accurate identification of dredging areas in complex landforms. This method improves the accuracy of spatial modeling, data processing standardization and recognition effect, and lays a high-quality foundation for dredging volume accounting.

[0103] In this embodiment, the S35 specifically includes:

[0104] S351, the triangulated ground model is passed into the residual neural network model as input data, and the convolution layer uses nodes to extract the desilting area features from the input triangulated ground model;

[0105] ;

[0106] in, is the desilting area feature of node i in the l-th layer residual neural network model, is the neighbor node set of node i, i is a node in the triangulated ground model, For the The convolution kernel weight matrix of the layer that weights nodes i and j, is the degree of node i, is the degree of node j, For the The desilting area characteristics of node j in the layer, For the The bias term of the layer;

[0107] S352. Skip connection technology is used in the convolution operation to preserve the spatial boundary information of the dredging area. Skip connection allows input data to be directly transferred to the output data stream;

[0108] S353, generating multiple feature maps through convolution operations, wherein the feature maps contain different spatial features of the desilting area, including the shape of the desilting area, the fineness of the boundary, and the depth change;

[0109] S354, flattening the two-dimensional or three-dimensional feature map generated by the convolution operation, that is, converting the pixel values of each feature map into a one-dimensional vector in sequence, and then passing the one-dimensional vector as input to the fully connected layer;

[0110] S355, the fully connected layer combines each input node received with the set weight matrix Perform matrix multiplication and generate an initial mask map of the desilting area.

[0111] The present invention uses the triangulated ground model as input and combines it with the convolution structure of the residual neural network model to extract the features of the dredging area, thereby realizing in-depth recognition of the dredging boundaries in complex landforms. The convolution layer aggregates the features of spatially adjacent nodes at multiple levels in the network, and uses the jump connection technology to enhance the ability to retain boundary information and prevent the loss of features during the downsampling process. The multidimensional feature map output by the network comprehensively expresses the morphological structure and edge changes of the dredging area. The multidimensional features are flattened and matrix operated through the fully connected layer to extract significant features and generate an initial mask map. When constructing the dredging area recognition model, this method not only considers the spatial structure, but also combines the expression capabilities of features at different levels to enhance the model's recognition robustness for irregular terrain, local disturbances and regional boundaries, significantly improves recognition accuracy and spatial consistency, and provides stable input for subsequent volume calculations.

[0112] In this embodiment, the S4 specifically includes:

[0113] S41. Each spider individual corresponds to a set of residual network structure parameters, including convolution kernel size, number of residual modules, number of feature channels and learning rate;

[0114] S42. The fitness value of each spider individual is determined by comparing the ratio of the intersection and union between the generated initial mask map and the manually marked desilting area. A ratio greater than 0.5 indicates that the recognition result of the residual neural network model is close to the actual desilting area and the fitness value is high:

[0115] ;

[0116] Among them, IoU is the fitness value, Z is the initial mask map, and G is the manually marked dredging area;

[0117] S43, the spider group simulates the vibration signal propagation behavior of social spiders based on the fitness value of each individual spider, so that the spider individuals with high fitness values guide the spiders to move to areas with high fitness values;

[0118] S44. The spider individual guides the search direction of the optimal parameters by guiding the spider to move towards the area with high fitness value:

[0119] ;

[0120] in, is the vibration signal propagation state within the spider colony, In the propagation state The search guidance direction vector of the spider individual p is: is the set of all nodes in the graph structure of the residual neural network model, i is a node in the triangulated ground model, N(i) is the set of adjacent nodes of node i, For the The desilting area characteristics of node i in the layer residual neural network model, For the The desilting area feature of node j in the layer residual neural network model, j is a neighboring node adjacent to node i, For the propagation state The guidance weight of node j to node i.

[0121] The present invention introduces a spider colony algorithm to intelligently optimize the structural parameters of the residual neural network model. Each spider individual corresponds to a set of structural parameter combinations, covering core elements such as convolution kernel size, number of residual modules, number of feature channels and learning rate. By calculating the intersection-over-union ratio between the initial mask map and the manually marked dredging area, a fitness evaluation mechanism is established. Individuals with high fitness will generate stronger vibration signals, simulate the information dissemination behavior of social spiders, and guide the remaining individuals to move to the optimal solution area. The spider individual performs search updates based on the feature expression of the residual neural network model in the graph structure, dynamically adjusts the direction of the structural combination, and realizes global convergence and adaptive optimization of the structural parameters. This method avoids the uncertainty and inefficiency of manual parameter adjustment, improves the robustness and generalization ability of the dredging area recognition model in different scenarios, and provides more accurate recognition results for subsequent volume calculations.

[0122] In this embodiment, the S5 specifically includes:

[0123] S51. In each iteration, the individual spider compares its current fitness value with the fitness value of the best individual spider in the neighborhood, and performs a round of position update operation in the structural parameter space based on the current individual position and the guidance direction of the neighboring individuals;

[0124] S52. The position of each spider individual represents a set of structural parameters of the residual neural network model. The spider individual updates its position based on the optimal fitness value of itself and its neighboring individuals:

[0125] ;

[0126] in, is the position of spider individual p in generation t, is the best position of spider individual q in history, is the global best position obtained by the entire spider population, 、 and is the acceleration constant, is the position of spider individual p in the t+1 generation, is the neighborhood individual set of spider individual p, p is the individual number in the spider group, g is the current global optimal spider individual number, is the coupling influence weight of spider individual q on p in the current iteration, To estimate the long-term average drift trend of historical position changes, is the position change of spider individual p in the i-th iteration, t is the number of the current iteration round, is the number of historical iterations;

[0127] S53, the individual spider updates its position through multiple iterations and adjusts its search direction according to the recognition accuracy after each iteration until the population reaches the maximum number of iterations, 100;

[0128] S54. During each iteration, the individual spider updates its position according to the dredging area recognition accuracy and the initial mask map, while monitoring the improvement of the dredging area recognition effect;

[0129] S55. After the iteration is completed, the optimized residual neural network model structure parameters are output.

[0130] The present invention uses a spider colony algorithm to iteratively optimize the structural parameters of the residual neural network and construct a multi-dimensional parameter search mechanism driven by fitness. In each round of iteration, the spider individual performs a position update operation along the optimal direction in the structural parameter space according to the fitness difference between itself and the individuals in the neighborhood, forming a synergy of local guidance and global convergence. The individual position represents a set of network structure parameter combinations, including key factors such as convolution kernel and number of channels, and dynamically adjusts the search path in combination with historical drift trends. By setting the maximum number of iterations and the recognition accuracy threshold, the system continuously monitors the intersection-over-union ratio of the mask map and the manually marked area to ensure that the network parameters are optimized in the direction of high precision. The final output structural parameters can be directly used for dredging area identification, significantly improving the model structure adaptability and generalization ability, and providing robust and accurate recognition support for the automatic accounting system.

[0131] In this embodiment, S6 specifically includes:

[0132] S61. Based on the optimized residual neural network model, perform layer-by-layer convolution operations on the triangulated ground model of the desilting area to obtain the classification results of each pixel point and output a binary mask image. Each pixel position only has two labels: desilting area or non-desilting area.

[0133] S62: extract all pixel points marked as desilting areas in the binary mask image, match them with corresponding coordinate points in the ground point cloud data, and construct a desilting area spatial point set. The desilting area spatial point set consists of the three-dimensional coordinates corresponding to all points determined to be desilting areas in the binary mask image;

[0134] S63, dividing the three-dimensional space with fixed side lengths, constructing a regular voxel grid, projecting and mapping the spatial point set of the desilting area onto the voxel grid, determining whether each voxel contains at least one valid spatial point, and counting the number of all occupied voxel units;

[0135] S64. Calculate the volume of each voxel based on the side length of the voxel unit, and sum the volumes of all occupied voxel units to obtain the three-dimensional total volume of the desilting area:

[0136] ;

[0137] in, is the total three-dimensional volume of the desilting area, K is the total number of all divided spatial voxels, is the spatial partition accuracy threshold, is the index of a valid spatial point in the spatial voxel unit k, k is the kth spatial voxel unit obtained by dividing the space, is the kth spatial voxel The volume weighting factor of each point, is the spatial volume of the k-th spatial voxel unit, is the number of valid points contained in the k-th spatial voxel, is the position of spider individual p in generation t+1;

[0138] S65, extracting the boundary contour of the desilting area based on the edge pixel points of the desilting area in the binary mask image, and corresponding the contour to the coordinates of the three-dimensional space point set to form a boundary map of the desilting area;

[0139] S66: Arrange the identified binary mask image, the boundary image of the desilting area, and the three-dimensional total volume of the desilting area into a unified data structure, and output it as the calculation result of the desilting engineering quantity.

[0140] The present invention performs layer-by-layer convolution on the triangulated ground model through the optimized residual neural network, outputs a binary mask map, and accurately identifies the scope of the dredging area. The system matches the dredging labels in the mask map with the point cloud coordinates one by one, constructs a three-dimensional dredging point set, and realizes the integration of image recognition and spatial modeling. By dividing the three-dimensional space with a fixed side length, constructing a regular voxel grid, mapping the point set to the voxel unit, counting the number of occupied voxels and calculating the total volume in combination with the side length, the structural accuracy and computational efficiency of the volume estimation are improved. The three-dimensional coordinates corresponding to the edge pixels are further extracted to generate a boundary map to realize the spatial visualization expression of the dredging area. Finally, the mask map, boundary map and volume data are integrated into a unified accounting result file to ensure the integrity of the results and the readability of the platform, and significantly improve the calculation accuracy of the dredging project volume and the engineering adaptability.

[0141] In this embodiment, the S7 specifically includes:

[0142] S71. Outputting the binary mask map in the calculation result of the desilting engineering quantity into two-dimensional image data. Specifically, encoding the binary mask map into a layer image format file, and performing spatial registration and color overlay with the image of the drone aerial survey of the highway construction area before desilting, to generate visual two-dimensional image data.

[0143] S72, encapsulating the boundary line map of the desilting area, the three-dimensional total volume of the desilting area, and the visualized two-dimensional image data into a structured data result file;

[0144] S73. Generate a unique task identification number for each set of data in the structured data result file, bind the structured data result file to the task identification number, and upload it to a remote database;

[0145] S74. Indexing the task identification number in the remote database, loading the structured data result file on the user interface, displaying the visualized two-dimensional image data as the background image, overlaying the boundary map of the desilting area in wireframe form, and displaying the three-dimensional total volume statistics of the desilting area on the numerical panel. The three together constitute a visualization display structure for the results;

[0146] S75: Generate a result preview number for each set of data in the structured data result file based on the desilting identification time, encapsulate the visualized two-dimensional image data, the boundary map of the desilting area, and the three-dimensional total volume of the desilting area together with the corresponding task identification number and result preview number to generate a result preview file;

[0147] S76. The results preview file is displayed in a unified format in the accounting interface, and supports online viewing, downloading and archiving of the results preview file and task archiving operations through the results preview number.

[0148] The present invention realizes the standardized output and visual management of dredging engineering quantity accounting results by constructing a unified result generation and display process. The system superimposes the binary mask map obtained by recognition with the drone image to generate an image layer, thereby enhancing the intuitive presentation of the spatial recognition results. The dredging boundary map and the volume statistics results are encapsulated as a structured data file, and bound to a unique task number and uploaded to the platform database to realize task-level archiving management of the results. The platform automatically indexes the task number, loads the layer, boundary line and numerical information, and constructs an integrated graphics and data result display interface. The result image generates a unique preview number based on the resolution and recognition time, and is uniformly included in the platform preview and download mechanism, supporting multi-task comparison, archiving and recall, and comprehensively improving the traceability, interactivity and engineering delivery efficiency of the dredging accounting results.

[0149] Example 1:

[0150] In order to verify the feasibility of the present invention in implementation, the present invention was applied to the accounting task of a dredging project in a certain section. Due to long-term water accumulation, sediment accumulation and slope scouring, this section has formed multiple ditches and undulating terrain areas, which seriously affect construction traffic and roadbed flatness. It is urgent to carry out dredging construction and simultaneously carry out dredging engineering quantity accounting and quality assessment. The traditional engineering quantity accounting method requires staff to be equipped with RTK equipment to go deep into the site to collect cross-section points. The points are sparse, dangerous and have a long processing cycle, which makes it difficult to meet the construction unit's dual requirements for data accuracy and accounting efficiency. Therefore, this project was selected as the field test demonstration section for the automatic accounting method of the present invention.

[0151] In this scenario, operators used a multi-rotor drone equipped with a lidar and high-resolution RGB image sensor to conduct two aerial surveys of the section, one before and one after dredging. Each flight was at an altitude of 80 meters, with an 80% overlap between routes, and each operation covered an area of approximately 1.6 square kilometers. Each flight collected approximately 48GB of raw data, including point cloud data, RGB imagery, flight path planning data, inertial measurement unit records, and acquisition timestamps. All data was automatically uploaded to the transportation construction accounting platform deployed in Province A after the flight, initiating data preprocessing and dredging area identification processes.

[0152] First, the raw data is converted into a standard .las format, and the point cloud is filtered, denoised, and ground points are extracted. The verification results of the check points show that the ground error of the point cloud is controlled within ±3cm, which meets the requirements of engineering measurement. Subsequently, the system extracts the point cloud within the scope and constructs a triangulated ground model based on the uploaded construction scope file, and performs cropping on the model boundary and triangle morphology optimization. By loading the residual neural network structure parameters optimized by the present invention, a layer-by-layer convolution operation is performed on the ground model, the dredging area is automatically identified, and an initial mask map is generated. The image is compared with the manually annotated samples, and its intersection-over-union ratio reaches 89.72%, which is significantly higher than the 76.85% of the traditional threshold method.

[0153] To further verify the effectiveness of the network structure, a spider colony algorithm was introduced to iteratively optimize the neural network parameters. Recognition accuracy steadily improved over 100 rounds of evolutionary iterations. The final structural parameter combination included a 5×5 convolution kernel, a residual module depth of 7, 96 feature channels, and a learning rate of 0.0008. This structure achieved a recognition accuracy of 91.08% on the validation set, an improvement of approximately 5.4% over the unoptimized network.

[0154] Finally, the desilting volume was calculated by aligning the mask map with the point cloud data, extracting the spatial point set of the desilting area, and projecting it onto the constructed 3D voxel grid. The calculated desilting volume was 4825.34 cubic meters. The boundary map was fused with the aerial survey imagery and displayed on the volume calculation platform. A preview file of the results was automatically generated and uploaded simultaneously to the project supervisor and design team.

[0155] Compared with traditional methods, this invention only takes 3 hours for data collection and 5 hours for data processing and identification calculation, shortening the overall accounting cycle to less than 8 hours. Traditional measurement methods require approximately 5 people and 3 days to complete, collecting less than 200 data points, resulting in low volume calculation accuracy and significant errors. To ensure operational safety, drones are used to complete all data collection, eliminating the need for personnel to enter silt areas, avoiding high-risk operational scenarios, and ensuring that data records throughout the entire process are traceable and auditable, greatly enhancing data transparency and credibility.

[0156] Table 1 Comparison of optimization effects of automatic calculation methods for dredging engineering quantities in highway construction

[0157] ;

[0158] Table 1 shows the automation, refinement, and structuring of the entire process of dredging engineering quantity accounting from data collection to results generation. It solves the problems of low efficiency, poor accuracy, and high safety risks of traditional quantity calculation methods, verifies the high adaptability and engineering feasibility of the present invention in complex terrain environments, and has good promotion and application value.

[0159] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A method for automatically calculating the amount of dredging work in highway construction, characterized in that: The steps include: S1. Use drones to conduct aerial surveys of the highway construction area before and after desilting, collect raw data, and collect scope files defining the desilting area boundaries; S2. Preprocess the original data and perform range file verification to generate ground point cloud data of the desilting area; S3. Construct a triangulated ground model based on the range file and ground point cloud data, input the triangulated ground model into the residual neural network model, and output the initial mask map of the dredging area. Each residual module of the residual neural network model extracts the features of the dredging area through the convolution layer and retains the spatial boundary information through the skip connection technology; The S3 specifically includes: S31, read the specified area boundary data from the range file to determine the ground area range to be processed; S32, triangulating the processed ground point cloud data using a triangulation algorithm, connecting every three adjacent ground points into a triangle, and generating a triangulated ground model covering the ground; S33, clipping the triangulated ground model according to the boundary provided by the scope file, removing the portion outside the boundary, and retaining only the ground data within the desilting area; S34. For the cropped triangulated ground model, adjust the size and angle of the triangles to eliminate triangles with an area greater than 5 square meters and acute triangles with a minimum internal angle less than 20 degrees. S35, inputting the adjusted triangulated ground model into the residual neural network model, and outputting an initial mask map of the desilting area; S4. Using the spider colony algorithm, the search direction of the optimal parameters is guided according to the dredging area recognition accuracy and the initial mask map. In the spider colony algorithm, each spider individual corresponds to a set of residual network structure parameters. S5, the spider individual updates its position according to the optimal parameters of itself and neighboring spider individuals until the population converges, and outputs the optimized residual neural network model structure parameters; S6. Perform pixel-level segmentation of the desilting area based on the optimized residual neural network model, calculate the three-dimensional total volume of the desilting area based on the point cloud spatial coordinate information, and output the calculation results of the desilting engineering quantity; S7. Generate a result preview file based on the calculation results.

2. The automatic calculation method of dredging engineering quantity in highway construction according to claim 1 is characterized in that: The raw data specifically includes RGB images, lidar point cloud data, ground check point data, spatial coordinates, inertial measurement unit, flight path planning and timestamp.

3. The automatic calculation method of dredging engineering quantity in highway construction according to claim 1 is characterized in that: The desilting area characteristics specifically include height changes, regional undulations and desilting traces.

4. The automatic calculation method of dredging engineering quantity in highway construction according to claim 1 is characterized in that: The S2 specifically includes: S21. Convert the format of the original data into a universal .las ground point cloud data format; S22, performing a quality check on the ground point cloud data, using the collected ground check point data to perform point cloud precision check and calculation, and removing noise points and abnormal data through a filtering algorithm; S23. Based on the scope document, the construction unit shall verify whether the scope is qualified according to the set measurement criteria including area size, construction depth and volume calculation standards.

5. The automatic calculation method of dredging engineering quantity in highway construction according to claim 1 is characterized in that: The S35 specifically includes: S351, the triangulated ground model is passed into the residual neural network model as input data, and the convolution layer uses nodes to extract the desilting area features from the input triangulated ground model; ； in, is the desilting area feature of node i in the l-th layer residual neural network model, is the neighbor node set of node i, i is a node in the triangulated ground model, For the The convolution kernel weight matrix of the layer that weights nodes i and j, is the degree of node i, is the degree of node j, For the The desilting area characteristics of node j in the layer, For the The bias term of the layer; S352. Skip connection technology is used in the convolution operation to preserve the spatial boundary information of the dredging area. Skip connection allows input data to be directly transferred to the output data stream; S353, generating multiple feature maps through convolution operations, wherein the feature maps contain different spatial features of the desilting area, including the shape of the desilting area, the fineness of the boundary, and the depth change; S354, flattening the two-dimensional or three-dimensional feature map generated by the convolution operation, that is, converting the pixel values of each feature map into a one-dimensional vector in sequence, and then passing the one-dimensional vector as input to the fully connected layer; S355, the fully connected layer combines each input node received with the set weight matrix Perform matrix multiplication and generate an initial mask map of the desilting area.

6. The automatic calculation method of dredging engineering quantity in highway construction according to claim 1 is characterized in that: The S4 specifically includes: S41. Each spider individual corresponds to a set of residual network structure parameters, including convolution kernel size, number of residual modules, number of feature channels and learning rate; S42. The fitness value of each spider individual is determined by comparing the ratio of the intersection and union between the generated initial mask map and the manually marked dredging area. A ratio greater than 0.5 indicates that the recognition result of the residual neural network model is close to the actual dredging area and the fitness value is high; S43, the spider group simulates the vibration signal propagation behavior of social spiders based on the fitness value of each individual spider, so that the spider individuals with high fitness values guide the spiders to move to areas with high fitness values; S44. The spider individual guides the search direction of the optimal parameters by guiding the spider to move towards the area with high fitness value: ； in, is the vibration signal propagation state within the spider colony, In the propagation state The search guidance direction vector of the spider individual p is: is the set of all nodes in the graph structure of the residual neural network model, i is a node in the triangulated ground model, N(i) is the set of adjacent nodes of node i, For the The desilting area characteristics of node i in the layer residual neural network model, For the The desilting area feature of node j in the layer residual neural network model, j is a neighboring node adjacent to node i, For the propagation state The guidance weight of node j to node i.

7. The automatic calculation method of dredging engineering quantity in highway construction according to claim 1 is characterized in that: The S5 specifically includes: S51. In each iteration, the individual spider compares its current fitness value with the fitness value of the best individual spider in the neighborhood, and performs a round of position update operation in the structural parameter space based on the current individual position and the guidance direction of the neighboring individuals; S52. The position of each spider individual represents a set of structural parameters of the residual neural network model. The spider individual updates its position based on the optimal fitness value of itself and its neighboring individuals: ； in, is the position of spider individual p in generation t, is the best position of spider individual q in history, is the global best position obtained by the entire spider population, 、 and is the acceleration constant, is the position of spider individual p in the t+1 generation, is the neighborhood individual set of spider individual p, p is the individual number in the spider group, g is the current global optimal spider individual number, is the coupling influence weight of spider individual q on p in the current iteration, To estimate the long-term average drift trend of historical position changes, is the position change of spider individual p in the i-th iteration, t is the number of the current iteration round, is the number of historical iterations; S53, the individual spider updates its position through multiple iterations and adjusts its search direction according to the recognition accuracy after each iteration until the population reaches the maximum number of iterations, 100; S54. During each iteration, the individual spider updates its position according to the dredging area recognition accuracy and the initial mask map, while monitoring the improvement of the dredging area recognition effect; S55. After the iteration is completed, the optimized residual neural network model structure parameters are output.

8. The automatic calculation method of dredging engineering quantity in highway construction according to claim 1 is characterized in that: The S6 specifically includes: S61. Based on the optimized residual neural network model, perform layer-by-layer convolution operations on the triangulated ground model of the desilting area to obtain the classification results of each pixel point and output a binary mask image. Each pixel position only has two labels: desilting area or non-desilting area. S62: extract all pixel points marked as desilting areas in the binary mask image, match them with corresponding coordinate points in the ground point cloud data, and construct a desilting area spatial point set. The desilting area spatial point set consists of the three-dimensional coordinates corresponding to all points determined to be desilting areas in the binary mask image; S63, dividing the three-dimensional space with fixed side lengths, constructing a regular voxel grid, projecting and mapping the spatial point set of the desilting area onto the voxel grid, determining whether each voxel contains at least one valid spatial point, and counting the number of all occupied voxel units; S64. Calculate the volume of each voxel based on the side length of the voxel unit, and sum the volumes of all occupied voxel units to obtain the three-dimensional total volume of the desilting area: ； in, is the total three-dimensional volume of the desilting area, K is the total number of all divided spatial voxels, is the spatial partition accuracy threshold, is the index of a valid spatial point in the spatial voxel unit k, k is the kth spatial voxel unit obtained by dividing the space, is the kth spatial voxel The volume weighting factor of each point, is the spatial volume of the k-th spatial voxel unit, is the number of valid points contained in the k-th spatial voxel, is the position of spider individual p in the t+1 generation; S65, extracting the boundary contour of the desilting area based on the edge pixel points of the desilting area in the binary mask image, and corresponding the contour to the coordinates of the three-dimensional space point set to form a boundary map of the desilting area; S66: Arrange the identified binary mask image, the boundary image of the desilting area, and the three-dimensional total volume of the desilting area into a unified data structure, and output it as the calculation result of the desilting engineering quantity.

9. The automatic calculation method of dredging engineering quantity in highway construction according to claim 1 is characterized in that: The S7 specifically includes: S71. Outputting the binary mask map in the calculation result of the desilting engineering quantity into two-dimensional image data. Specifically, encoding the binary mask map into a layer image format file, and performing spatial registration and color overlay with the image of the drone aerial survey of the highway construction area before desilting, to generate visual two-dimensional image data. S72, encapsulating the boundary map of the desilting area, the three-dimensional total volume of the desilting area, and the visualized two-dimensional image data into a structured data result file; S73. Generate a unique task identification number for each set of data in the structured data result file, bind the structured data result file to the task identification number, and upload it to a remote database; S74. Indexing the task identification number in the remote database, loading the structured data result file on the user interface, displaying the visualized two-dimensional image data as the background image, overlaying the boundary map of the desilting area in wireframe form, and displaying the three-dimensional total volume statistics of the desilting area on the numerical panel. The three together constitute a visualization display structure for the results; S75: Generate a result preview number for each set of data in the structured data result file based on the desilting identification time, encapsulate the visualized two-dimensional image data, the boundary map of the desilting area, and the three-dimensional total volume of the desilting area together with the corresponding task identification number and result preview number to generate a result preview file; S76. The results preview file is displayed in a unified format in the accounting interface, and supports online viewing, downloading and archiving of the results preview file and task archiving operations through the results preview number.

Images