Earthwork intelligent allocation optimization method and system

Abstract

The application discloses a kind of earthwork intelligent deployment optimization method and system, comprising: obtaining the original block degree spatial distribution point cloud of to-be-excavated blast area and generating blast area original gradation curve set;The original gradation curve is similar to the target gradation curve in the envelope line library of engineering design based on Hausdorff distance mapping, and the movable gradation curve subset is obtained by screening;For movable gradation curve subset, determine the optimal processing procedure vector and corresponding processing cost coefficient that meet the target gradation requirement, and predict its compaction and mechanical properties by discrete element-finite element coupling, only keep the qualified alternative material source curve with performance up to standard;Based on qualified alternative material source curve, construct a deployment model that integrates processing and transportation costs, and output the earthwork deployment scheme with maximum comprehensive utilization rate and minimum total cost by global optimization solution. The application effectively solves the technical problems of low raw material utilization, conservative deployment scheme and lack of mechanical property verification support in the prior art.

Description

Technical Field

[0001] This invention relates to the field of allocation and optimization technology, and in particular to an intelligent allocation and optimization method and system for earthwork. Background Technology

[0002] In the construction of large-scale geotechnical engineering projects such as pumped storage projects, high-face rockfill dams, and integrated road-rail projects in mountainous areas, there is often a demand for various engineering materials within the same construction area, including filter media, transition media, A / B group roadbed materials, and building-grade crushed stone. These materials have different and specific design requirements for particle size distribution, throughput, and mechanical properties. Therefore, it is necessary to rationally allocate the earth and rock materials obtained from blasting excavation during construction to improve material utilization efficiency and reduce project costs.

[0003] Existing earthwork allocation methods generally employ discrete matching models based on the engineering design gradation envelope, simplifying the matching relationship between "material source and receiving point" into a binary decision: when the gradation curve of the excavated material source falls completely within the design envelope range corresponding to the receiving point, the material source is determined to be usable directly; otherwise, it is determined to be unusable or discarded. Although this type of method can ensure that the allocation results formally meet the design specifications, the following technical problems that have not yet been effectively solved have gradually emerged in practical engineering applications.

[0004] First, there is a lack of an effective mechanism for utilizing the actual gradation continuity of blasted excavated materials.

[0005] The gradation curves of materials obtained from on-site blasting excavation are usually continuously distributed, while the engineering design envelope often has a narrow allowable range. Existing methods only use whether the material falls completely within the envelope as a criterion, which cannot identify near-gradation materials that are "close to the design requirements but slightly deviate". As a result, a large amount of material with potential engineering value is directly judged as waste, and the comprehensive utilization rate of materials in the area has been at a low level for a long time.

[0006] Second, there is a lack of quantitative description of the relationship between the degree of gradation deviation and processing feasibility.

[0007] Existing blending models lack a continuous measurement method for the degree of difference between gradation curves and do not incorporate conventional processing methods such as screening, washing, crushing, and mixing into the blending decision-making process. When there are slight oversize or mud content deviations in the material gradation, the model cannot determine whether reasonable processing can make it meet engineering requirements, thus directly excluding such materials at the blending stage. This results in overly conservative blending schemes, making it difficult to achieve the engineering optimization goal of "replacing high transportation costs with processing costs".

[0008] Third, there is a lack of mechanical performance constraints and verification mechanisms during the adjustment phase.

[0009] Existing technologies mostly use particle size distribution as the criterion for blending, without establishing a correlation model between gradation differences and compaction and mechanical properties, nor do they predict and constrain permeability, deformation performance, and shear resistance during the blending stage. As a result, the blending optimization results cannot answer whether the alternative materials will still meet the engineering safety requirements such as dam stability and roadbed deformation control after being put into use, and there is a risk of relying on post-experimental tests or experience corrections.

[0010] In summary, existing earthwork allocation methods have not yet solved the problem of how to simultaneously achieve continuous quantification of gradation substitutability, pre-assessment of processing and migration feasibility, and simultaneous verification of mechanical performance and safety in complex engineering scenarios where blasted excavation materials are the main source, so as to improve the comprehensive utilization rate of materials and reduce the overall allocation cost while ensuring engineering safety.

[0011] To address the aforementioned issues, a new intelligent earthwork allocation method is urgently needed to overcome the shortcomings of existing technologies in terms of material utilization, allocation flexibility, and engineering safety verification. Summary of the Invention

[0012] To address the problem that existing models exhibit lag and non-adaptability when processing new data after surgery, making it difficult to update prediction results in real time, this invention provides an intelligent earthwork allocation and optimization method and system.

[0013] To achieve the above objectives, the technical solution adopted by the present invention is as follows: On one hand, this invention discloses an intelligent allocation and optimization method for earthwork, comprising the following steps: Step 1: Obtain the original block size spatial distribution point cloud of the blasting area to be excavated, and generate the original gradation curve set of the blasting area based on the original block size spatial distribution point cloud. Step 2: For each target gradation curve in the engineering design envelope library, insert a gradation similarity measurement function driven by Hausdorff distance between the upper and lower envelopes of the target gradation curve, map the original gradation curve set of the blasting area to the target gradation curve point by point, and output the transferable gradation curve subset and its initial similarity value. Step 3: Using the transferable gradation curve subset as input, generate the optimal processing operation vector and corresponding processing cost coefficient required for the transferable gradation curve subset to fall into the envelope of the target gradation curve. Step 4: Substitute the optimal processing procedure vector into the discrete element-finite element coupled performance predictor and perform a rolling forecast of compaction performance and mechanical properties on the processed transferable gradation curve subset. If the predicted permeability coefficient, compression modulus, and shear strength all meet the engineering design threshold, then mark the transferable gradation curve subset as a qualified alternative material source curve. Step 5: Construct a three-dimensional preference cost matrix using qualified alternative material source curves, processing cost coefficients, and transportation distances, and directly write the three-dimensional preference cost matrix into the objective function coefficient term of the linear earthwork allocation model to form a quantitative allocation model of substitutability. Step 6: Call the branch-price solver to perform a global optimal solution on the substitutability quantification allocation model, and output an earthwork allocation scheme that simultaneously maximizes the comprehensive utilization rate of the area, minimizes the total cost, and ensures that the mechanical performance indicators of all material receiving points meet the standards.

[0014] Furthermore, step 1 includes: A laser-image joint scanning system, consisting of a lidar and an industrial camera mounted on a fixed support above the blasting excavation face, performs millimeter-resolution scanning of the blasting area to be excavated before blasting. This acquires the original block size spatial distribution point cloud of the blasting area, including three-dimensional coordinates, echo intensity, and synchronous image information. The original block size spatial distribution point cloud is then input into a pre-trained particle size recognition neural network. This neural network uses voxelized point cloud as the input layer and a block size particle size frequency distribution vector as the output layer. It extracts block size edge features through convolution kernels and statistically analyzes the percentage of points within each particle size interval, generating a set of original gradation curves for the blasting area consisting of 50 sets of gradation samples under different blasting parameters. The set of original gradation curves is then encapsulated in XML format as a structured data package containing three elements: particle size interval, pass rate, and non-uniformity coefficient. This structured data package serves as the sole input gradation curve set for all subsequent operations in steps 2 to 6, and the introduction of gradation curve data from any other source is prohibited in steps 2 to 6.

[0015] Furthermore, step 2 includes: The target gradation curves of filter material, transition material, A / B group roadbed material, and building graded crushed stone, classified according to nominal maximum particle size, are read one by one from the engineering design envelope library. A continuous gradation coordinate point series is established between the upper and lower envelopes of each target gradation curve with a step size of 0.1 mm. Taking each original gradation curve of the blasting area as a reference, the gradation similarity measurement function, which is a weighted synthesis of Hausdorff distance and Frechet distance, is called to perform a point-by-point mapping operation on the gradation coordinate point series of the original gradation curve of the blasting area and each target gradation curve to obtain an initial similarity value reflecting the deviation of the curve spatial shape. The original gradation curves of the blasting area with an initial similarity value less than or equal to the preset threshold of 0.15 are collectively included in the transferable gradation curve subset, and the corresponding initial similarity value is written into the metadata header of the transferable gradation curve subset to form a transferable gradation curve subset file directly associated with the subsequent step 3 operation.

[0016] Furthermore, step 3 includes: The transferable gradation curve subset is input line by line into the gradation offset processing response surface model, which has been pre-calibrated through a three-factor, five-level orthogonal experiment. The response surface model uses the particle size offset removed by screening and washing, the oversize correction by impact crushing, and the mud content reduction by water mixing as independent variables, and the sum of squares of the percentage differences between each particle size point between the transferable gradation curve subset and the corresponding target gradation curve as the response value. An explicit functional relationship between the independent variables and the response value is established through quadratic polynomial regression. With minimizing the response value as the optimization objective, a sequential quadratic programming algorithm is called to solve the problem under the physical boundary constraints of the independent variables. The output is the optimal processing step vector composed of screening and washing time, impact crushing discharge port width, and water mixing volume. At the same time, the processing cost coefficient obtained by accumulating equipment shifts, water and electricity consumption, and labor quota is calculated. The optimal processing step vector and the processing cost coefficient are written into the same-name field of the transferable gradation curve subset to form the transferable gradation curve subset with cost annotation required for the mechanical performance verification in the subsequent step 4.

[0017] Furthermore, step 4 includes: The cost-labeled subset of transferable gradation curves, along with the optimal processing step vector, is fed into a discrete-element-finite-element coupled performance predictor. The predictor uses discrete-element analysis to generate three-dimensional particle packing samples of the processed transferable gradation curve subset, and applies designed compaction work to the samples using finite-element analysis. It outputs a real-time rolling forecast of compaction performance and mechanical properties, including permeability coefficient, compressive modulus, and shear strength. If the permeability coefficient is less than 1×10⁻⁴ cm / s, the compressive modulus is greater than 80 MPa, and the shear strength is greater than 35° in the rolling forecast, a qualified identifier is written into the metadata header of the transferable gradation curve subset, forming a qualified alternative material source curve, and the corresponding processing cost coefficient is locked and remains unchanged. If any indicator fails to reach the threshold, the current transferable gradation curve subset is discarded, and it is prohibited from entering the three-dimensional preference cost matrix in subsequent step 5, ensuring a one-to-one correspondence between the qualified alternative material source curve set and the mechanical performance verification.

[0018] Furthermore, step 5 includes: Using the processing cost coefficient corresponding to the qualified alternative material source curve as the Z-axis, the transportation cost calculated based on the shortest transportation distance from the excavation point to the receiving point as the Y-axis, and the blasting area number in the original gradation curve set of the blasting area from which the qualified alternative material source curve originates as the X-axis, a three-dimensional preference cost matrix containing only qualified alternative material source curves is constructed. Each cell stores the preference cost value of the sum of a unique processing cost coefficient and transportation cost. The three-dimensional preference cost matrix is ​​expanded into a two-dimensional sparse coefficient vector and directly written into the objective function coefficient term of the linear earthwork allocation model, replacing the unit transportation cost coefficient in the traditional 0-1 matching matrix. This allows the decision variables of the linear earthwork allocation model to simultaneously reflect the economic trade-off between transportation paths and processing procedures, forming a substitutable quantitative allocation model. In the substitutable quantitative allocation model, the primary objective is to maximize the comprehensive utilization rate of the area, and the secondary objective is to minimize the total cost. The preference cost value of the three-dimensional preference cost matrix is ​​embedded into the single linear objective function of the substitutable quantitative allocation model through objective weighting, completing the coefficient initialization of the substitutable quantitative allocation model.

[0019] Furthermore, step 6 includes: Using the substitutability quantification allocation model as the sole input, a branch-price solver based on a branch-price framework is invoked. In the main problem of the branch-price solver, a column generation method is used to generate columns only for the transportation-processing joint variables corresponding to the qualified alternative material source curves. In the subproblems, the preference cost value of the three-dimensional preference cost matrix is ​​used to reduce the cost coefficient, and the pricing loop is completed through the shortest path label algorithm. When the reduced cost coefficient can no longer be negative, the branch-price solver terminates column generation and enters the branching stage to perform binary branching on the blast zone number variable in the original blast zone gradation curve set until the difference between the relaxation solution and the integer solution of all nodes is less than 0.01%. The output is the global optimal solution result composed of the qualified alternative material source curve, the corresponding transportation path, and the corresponding optimal processing procedure vector. The global optimal solution result is written into the intelligent earthwork allocation scheme file. The file simultaneously records the comprehensive utilization rate of the area, the total cost, and the rolling forecast index of compaction performance-mechanical performance of all receiving points, thus completing the entire execution of the intelligent earthwork allocation optimization method.

[0020] On the other hand, the present invention also discloses an intelligent earthwork allocation and optimization system, comprising: Gradation acquisition module: acquires the original block size spatial distribution point cloud of the blasting area to be excavated, and generates the original gradation curve set of the blasting area based on the original block size spatial distribution point cloud of the blasting area; Gradation screening module: For each target gradation curve in the engineering design envelope library, insert a gradation similarity measurement function driven by Hausdorff distance between the upper and lower envelopes of the target gradation curve, map the original gradation curve set of the blasting area to the target gradation curve point by point, and output the transferable gradation curve subset and its initial similarity value. Process determination module: Taking the transferable gradation curve subset as input, it generates the optimal processing operation vector and the corresponding processing cost coefficient required for the transferable gradation curve subset to fall into the envelope of the target gradation curve. Performance verification module: Substitute the optimal processing procedure vector into the discrete element-finite element coupled performance predictor, and perform rolling forecast of compaction performance and mechanical properties on the processed transferable gradation curve subset. If the predicted permeability coefficient, compression modulus and shear strength all meet the engineering design threshold, then the transferable gradation curve subset is marked as a qualified alternative material source curve. Allocation Modeling Module: Construct a three-dimensional preference cost matrix using qualified alternative material source curves, processing cost coefficients, and transportation distances. Then, directly write the three-dimensional preference cost matrix into the objective function coefficients of the linear earthwork allocation model to form a substitutable quantitative allocation model. Optimization Solver: Calls the branch-price solver to perform global optimal solution on the substitutability quantification allocation model, and outputs an earthwork allocation scheme that simultaneously maximizes the comprehensive utilization rate of the area, minimizes the total cost, and ensures that the mechanical performance indicators of all material receiving points meet the standards.

[0021] The technological advancements achieved by this invention compared to existing technologies are as follows: This invention uses the original spatial distribution point cloud of the blast zone's block size as the sole data source. It generates a set of original gradation curves for the blast zone through a particle size recognition neural network, avoiding information bias caused by manual selection or empirical assumptions about gradation, thus achieving objectivity and uniqueness in gradation input. By introducing a gradation similarity metric function weighted by Hausdorff distance and Frechet distance into the target gradation envelope, the morphological deviation between the original gradation curve of the blast zone and the engineering design gradation is continuously quantified, breaking through the traditional binary judgment method and providing a calculable basis for the substitutability between gradations. By constructing a gradation offset processing response surface model, the impact of processing steps such as screening and washing, crushing correction, and water mixing on gradation migration is explicitly characterized, enabling a quantitative assessment of processing feasibility before blending decisions, thereby significantly improving the engineering usability of near-gradation materials. By substituting the optimal processing step vector into a discrete-element-finite-element coupled performance predictor, the mechanical properties of permeability coefficient, compressibility modulus, and shear strength are pre-verified during the allocation stage. Only gradations meeting the mechanical performance standards are allowed to enter the allocation model, thus technically preventing alternative solutions that do not meet engineering safety requirements. By constructing a three-dimensional preference cost matrix that integrates processing and transportation costs, and combining it with a branch-price solver for global optimization, the system achieves synergistic optimization that maximizes the comprehensive utilization rate of the area and minimizes the total allocation cost, while ensuring that the mechanical performance indicators of all receiving points meet the standards.

[0022] In summary, this invention effectively solves the technical problems of low raw material utilization, overly conservative formulation schemes, and lack of mechanical performance verification support in the prior art, and has significant, stable technical effects that cannot be directly derived from the prior art. Attached Figure Description

[0023] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0024] In the attached diagram: Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0025] The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of the present invention will now be described with reference to the accompanying drawings.

[0026] Example 1 like Figure 1 As shown, this invention discloses an intelligent earthwork allocation and optimization method, comprising: Step 1: Obtain the original block size spatial distribution point cloud of the blasting area to be excavated, and generate the original gradation curve set of the blasting area based on the original block size spatial distribution point cloud. Step 2: For each target gradation curve in the engineering design envelope library, insert a gradation similarity measurement function driven by Hausdorff distance between the upper and lower envelopes of the target gradation curve, map the original gradation curve set of the blasting area to the target gradation curve point by point, and output the transferable gradation curve subset and its initial similarity value. Step 3: Using the transferable gradation curve subset as input, generate the optimal processing operation vector and corresponding processing cost coefficient required for the transferable gradation curve subset to fall into the envelope of the target gradation curve. Step 4: Substitute the optimal processing procedure vector into the discrete element-finite element coupled performance predictor and perform a rolling forecast of compaction performance and mechanical properties on the processed transferable gradation curve subset. If the predicted permeability coefficient, compression modulus, and shear strength all meet the engineering design threshold, then mark the transferable gradation curve subset as a qualified alternative material source curve. Step 5: Construct a three-dimensional preference cost matrix using qualified alternative material source curves, processing cost coefficients, and transportation distances, and directly write the three-dimensional preference cost matrix into the objective function coefficient term of the linear earthwork allocation model to form a quantitative allocation model of substitutability. Step 6: Call the branch-price solver to perform a global optimal solution on the substitutability quantification allocation model, and output an earthwork allocation scheme that simultaneously maximizes the comprehensive utilization rate of the area, minimizes the total cost, and ensures that the mechanical performance indicators of all material receiving points meet the standards.

[0027] Specifically, step 1 includes: Before blasting, a fixed support is installed above the working face of the blasting area to be excavated. A lidar and industrial camera are rigidly integrated into a laser-image joint scanning system. This joint scanning system simultaneously acquires spatial geometric information and texture image information at the same time reference, performing a full-coverage scan of the blasting area to be excavated. The lidar acquires point cloud data of the blasting area surface with millimeter-level spatial resolution. Each laser echo point includes at least the following information: three-dimensional spatial coordinates. echo intensity value The industrial camera acquires synchronous image information corresponding one-to-one with the point cloud space while performing laser scanning. This image information is used to enhance the recognizability of block boundaries, edges, and shadow changes. Through time stamp synchronization and spatial coordinate registration, the point cloud data acquired by the lidar and the image information acquired by the industrial camera are fused to form the original block size spatial distribution point cloud of the blasting area to be excavated, which includes three types of information: three-dimensional coordinates, echo intensity, and image texture. This point cloud completely preserves the spatial location, shape outline, and surface features of each rock block within the blasting area at the data level, without any manual screening or prior gradation assumptions.

[0028] To facilitate unified perception of block size features by the subsequent neural network, the original spatial distribution point cloud of the block size of the blasting area to be excavated is converted into a regular voxel representation. Specifically, the blasting area space is divided into three-dimensional voxel units of fixed size, and the following statistics are collected within each voxel unit: point density, mean echo intensity, and image texture gradient features.

[0029] This voxelization process maps unstructured point cloud data into regular 3D tensor inputs, enabling the neural network to stably extract spatial neighborhood relationships, block edge continuity, and scale changes.

[0030] The voxelized point cloud of the original block size spatial distribution of the blasting area to be excavated is input into a pre-trained particle size recognition neural network, which satisfies the following functional characteristics: 1. Input layer Using the voxelized 3D tensor as input, each voxel unit carries spatial location, echo intensity, and image fusion features.

[0031] 2. Feature Extraction Layer By performing sliding calculations on the neighborhood of voxels using 3D convolution kernels, the following block-level related features are extracted: block edge continuity; block gap distribution; and block surface undulations and edge variations.

[0032] The above features collectively characterize the spatial scale and morphological features of a single rock block.

[0033] 3. Particle size discrimination layer The extracted block size features are mapped to a predefined set of particle size intervals, and the probability of each voxel region belonging to the particle size interval is determined.

[0034] 4. Statistical Output Layer Global statistics are performed on the particle size discrimination results of all voxels within the entire explosion zone, and the output is a particle size frequency distribution vector representing the proportion of different particle size intervals.

[0035] The particle size frequency distribution vector reflects the overall particle size structure characteristics of the blasting zone under the current blasting parameters.

[0036] To eliminate the influence of gradation randomness under a single blasting parameter condition on subsequent allocation decisions, the particle size recognition neural network uses the original block size spatial distribution point cloud corresponding to multiple sets of blasting parameter conditions as input. Under each set of blasting parameter conditions, the above voxelization and particle size recognition process is repeated to generate the corresponding block size and particle size frequency distribution vector. The particle size interval is mapped to its cumulative pass rate to form an original gradation curve for the blasting area. Finally, a set of original gradation curves for the blasting area, consisting of 50 sets of original gradation curves generated under different blasting parameter conditions, is obtained. This set comprehensively describes the possible gradation morphology distribution of the blasting area to be excavated within a reasonable range of blasting parameter variations.

[0037] To ensure the uniqueness and traceability of the gradation information source in subsequent steps, the set of original gradation curves for the blasting zone is structured and encapsulated. Specifically, each original gradation curve for the blasting zone is described in XML format, including at least the following three core elements: particle size range, percentage of passage for each particle size range, and non-uniformity coefficient. All original gradation curves for the blasting zone are combined into a structured data package with a unified data structure, and are logically defined as the unique set of input gradation curves upon which all calculation, matching, optimization, and prediction processes in subsequent steps 2 to 6 depend.

[0038] During the calculation process from steps 2 to 6, no other gradation curves from any other source are allowed to be introduced. The original gradation curve set of the blasting area is not allowed to be replaced, resampled, or manually modified. All transferability analysis, processing decisions, mechanical property predictions, and blending optimizations are strictly based on this unique input set.

[0039] The above constraints ensure that the entire intelligent allocation and optimization method for earthwork has consistency, traceability, and closed-loop at the gradation data level.

[0040] Specifically, step 2 includes: In step 2, the pre-established engineering design envelope library is first invoked. The engineering design envelope library is classified and stored according to the nominal maximum particle size in accordance with the requirements of engineering design specifications, and includes at least the following types of target gradation curves: target gradation curve for filter media, target gradation curve for transition media, target gradation curve for subgrade material of group A, target gradation curve for subgrade material of group B, and target gradation curve for building grade crushed stone.

[0041] For each target gradation curve in the engineering design envelope library, there is an upper envelope and a lower envelope, which are used to limit the allowable gradation variation range of this type of engineering material. To achieve continuous morphological comparison between the original gradation curve and the target gradation curve in the blasting zone, a continuous gradation coordinate point series is established between the upper and lower envelopes of each target gradation curve, with particle size as the independent variable. The continuous gradation coordinate point series is discretely sampled across the entire particle size range with a particle size step size of 0.1 mm, so that the target gradation envelope forms a high-resolution continuous curve band in the particle size-percentage passing two-dimensional space. Through this processing, each target gradation curve is represented as an ordered set of two-dimensional coordinate points to support subsequent refined curve mapping and similarity calculation.

[0042] The set of original gradation curves for the blast zone generated in step 1 serves as the only permitted input gradation information in step 2. For each original gradation curve in the set, its particle size—expressed as a percentage—is considered as a continuous curve in a two-dimensional particle size space. Subsequently, for each target gradation curve in the engineering design envelope library, the gradation similarity metric function is invoked to perform a point-by-point mapping operation on the continuous gradation coordinate points of the original gradation curve and the target gradation curve. This point-by-point mapping operation follows the following principles: using the particle size axis as a unified reference dimension; calculating the geometric deviation between the original gradation curve and the target gradation envelope interval at each particle size sampling point; and not introducing any artificial adjustments or interpolation corrections.

[0043] To comprehensively characterize the similarity of the two gradation curves in terms of spatial morphology, the gradation similarity measurement function is a weighted synthesis of the Hausdorff distance and the Frechet distance. The Hausdorff distance is used to measure the maximum local deviation between the envelope of the original gradation curve and the target gradation curve in the blasting area, which reflects the spatial difference between the two curves at the most unfavorable point.

[0044] Let the original gradation curve of the blasting zone be a set. The set of continuous gradation coordinate points of the target gradation curve is denoted as a series. Then the Hausdorff distance is expressed as: in, This represents a specific gradation point on the original gradation curve of the blast zone; This represents a specific allowed gradation point within the target gradation envelope. This represents the distance between the explosion zone gradation point and the target gradation point.

[0045] Frechet distance is used to describe the consistency of the overall shape and evolution trend of two curves, emphasizing the overall similarity of the curves as they advance along the grain size axis.

[0046] By weighting and combining the two distances mentioned above, a gradation similarity metric function is constructed for this method. : in, Preset weighting coefficients are used to balance the impact of local extreme value deviations and overall shape deviations on similarity evaluation; Representing the Frechet distance, it characterizes the overall curve shape difference; The Hausdorff distance represents the maximum local deviation. The output value of this gradation similarity metric function is defined as the initial similarity value, used to quantify the spatial morphological deviation between the original gradation curve and the target gradation curve in the blast zone.

[0047] After calculating the gradation similarity between the original gradation curve of the blasting area and each target gradation curve, the initial similarity value corresponding to each original gradation curve of the blasting area is compared with a preset threshold. When the initial similarity value of an original gradation curve of a blasting area relative to any target gradation curve satisfies: If the original gradation curve of the blasting area is found to have an acceptable similarity in geometric shape to the corresponding target gradation curve, it is considered to have the potential to achieve gradation migration through subsequent processing steps.

[0048] The original gradation curves of the blasting area that meet the above conditions are collectively classified into the transferable gradation curve subset, and their corresponding initial similarity values ​​are written into the metadata header of the transferable gradation curve subset, which are used as prior information for the optimization of the processing procedures in the subsequent step 3.

[0049] If the initial similarity value of the original gradation curve of a certain blasting area to all target gradation curves in the engineering design envelope library is greater than 0.15, then the original gradation curve of the blasting area is determined to be a non-transferable gradation curve and will not enter the subsequent processing and performance verification process.

[0050] Step 2 completes the initial screening of the original gradation curve set of the explosion zone generated in Step 1 at the geometric level. Without introducing any processing assumptions, it identifies the original gradation curves with gradation migration potential and assigns a clear and quantifiable initial similarity value to each transferable gradation curve. The resulting subset of transferable gradation curves becomes the direct input to the gradation offset processing response surface model in Step 3, ensuring that subsequent processing decisions are based on clear gradation similarity.

[0051] Specifically, step 3 includes: The subset of transferable gradation curves selected in step 2 includes the original gradation curves of the blasting area that meet the initial similarity threshold requirements and their corresponding initial similarity values. In step 3, this subset of transferable gradation curves is strictly used as the sole source of input gradation information, without introducing any other gradation data or manual correction results. For each transferable gradation curve in the subset, its corresponding target gradation curve type and its engineering design envelope range are clearly defined, serving as a constraint basis for subsequent processing optimization.

[0052] The gradation offset processing response surface model is a pre-calibrated explicit function model used to characterize the influence of changes in processing parameters on the degree of gradation curve morphology migration. This response surface model uses the following three types of processing parameters as independent variables: 1. Screening and rinsing to remove particle size deviation. Used to characterize the adjustment intensity of fine particle size distribution segments through screening and washing; 2. Impact crusher corrects oversize Used to characterize the degree of correction to the proportion of oversized particles through the crushing process; 3. Add water and mix to reduce mud content. It is used to characterize the effect of adding water and mixing on the adhesion of fine materials and the mud content.

[0053] The above three types of independent variables together constitute the processing procedure parameter vector, which is used to describe a complete set of processing procedures.

[0054] The response value of a response surface model is defined as the degree of gradation deviation between the processed transferable gradation curve and the corresponding target gradation curve.

[0055] To quantify the degree of matching between the processed gradation curve and the target gradation curve, the response value is measured by the sum of squares of the percentage differences at each particle size point.

[0056] set up: The gradation curve after processing is ; The corresponding target gradation curve is ; These are the particle size sampling points in the continuous coordinate point column of the target gradation.

[0057] Then the response value Represented as: This response value reflects the effect of the processing steps on the overall shape of the gradation curve. The smaller the value, the closer the gradation curve is to the center region of the target gradation curve envelope after processing.

[0058] By performing regression analysis on the results of a three-factor, five-level orthogonal experiment, an explicit functional relationship between processing parameters and response values ​​is established using a quadratic polynomial. The general form of the response surface model is expressed as follows: in: Indicates the first Each processing step is an independent variable; These are regression coefficients, used to characterize the weighted influence of changes in processing parameters on the degree of gradation offset.

[0059] Based on the response surface methodology, with the minimization of the response value as the optimization objective, and under the physical boundary constraints of the independent variables of each processing step, a sequential quadratic programming algorithm is used to jointly optimize the processing step parameters. The optimization objective can be expressed as: Through this optimization process, a set of processing parameters that minimizes the response value is obtained.

[0060] This parameter combination is defined as the optimal processing step vector corresponding to the transferable gradation curve, and it includes at least: screening and washing time; impact crusher discharge port width; and water addition and mixing volume. This optimal processing step vector ensures that the processed gradation curve falls within the engineering design envelope of the corresponding target gradation curve.

[0061] While determining the optimal processing step vector, the corresponding processing cost coefficient is calculated based on the resource consumption corresponding to the processing step vector. The processing cost coefficient is composed of the following cost items: processing equipment shift consumption; water and electricity consumption; and labor quota consumption. The processing cost coefficient, as an economic attribute indicator corresponding one-to-one with the transferable gradation curve, is generated synchronously with the optimal processing step vector.

[0062] The optimal processing step vector and its corresponding processing cost coefficient obtained from the solution are written into the corresponding fields of the transferable gradation curves, forming a subset of transferable gradation curves with cost annotations. This subset of transferable gradation curves with cost annotations undergoes the following expansions at the data structure level: retaining the original gradation curve information and initial similarity value; adding an optimal processing step vector field; and adding a processing cost coefficient field. The resulting data object becomes the direct input object for the compaction performance-mechanical performance prediction and verification in step 4.

[0063] Specifically, step 4 includes: The subset of transferable gradation curves with cost annotations formed in step 3 includes, for each data object: the transferable gradation curve itself; the corresponding target gradation curve type; the optimal processing step vector; and the processing cost coefficient.

[0064] In step 4, the cost-labeled, transferable gradation curve subset is strictly used as the input object. This subset, along with the corresponding optimal processing vector, is fed into the discrete-element-finite-element coupled performance predictor without introducing any external mechanical parameters or empirical correction coefficients. The discrete-element-finite-element coupled performance predictor is used to numerically simulate the particle rearrangement and macroscopic mechanical response process of the processed gradation material under design compaction conditions. It consists of the following two core sub-modules: 1. Discrete Element Particle Stacking Generation Module Based on the transferable gradation curve and its corresponding optimal processing step vector, a particle size distribution reflecting the gradation morphology after processing is constructed, and a particle packing sample consistent with the gradation phase is generated in three-dimensional space.

[0065] Discrete element model (DEM) accurately reflects the packing structure and pore distribution characteristics of particles of different sizes after processing by analyzing particle contact relationships and spatial arrangement.

[0066] 2. Finite Element Compaction and Response Calculation Module Based on the three-dimensional particle packing sample generated by discrete element method, the finite element method is introduced to apply the compaction work specified by the engineering design to the sample, and to simulate the overall deformation and stress evolution of the particle system during the compaction process.

[0067] By using the finite element method, the changes in macroscopic physical quantities of the particle pack during the compaction process can be obtained in real time.

[0068] During the discrete element-finite element coupled calculation process, the performance predictor performs continuous calculations of the compaction process in terms of time or step, and synchronously outputs rolling forecast results of compaction performance and mechanical properties. The rolling forecast results include at least the following three key indicators: 1. Permeability coefficient Used to characterize the internal pore connectivity and permeability of granular aggregates; 2. Compression modulus Used to characterize the overall stiffness and deformation resistance of a material under compaction conditions; 3. Shear strength Used to characterize the stability and load-bearing capacity of materials under shear stress.

[0069] The above indicators are all automatically calculated by the coupled model during the compaction process based on the particle contact state and stress-strain response.

[0070] For each transferable gradation curve, the rolling forecast result is used to extract its stable value at the design compaction state, and then compared item by item with the engineering design threshold, which is defined as follows: in: Indicates the permeability coefficient; Indicates the compressibility modulus; Indicates shear strength.

[0071] When the rolling forecast result corresponding to a certain transferable gradation curve simultaneously meets all the above mechanical property thresholds, it is determined that the transferable gradation curve, after being processed and compacted by the optimal processing procedure, can meet the mechanical property safety requirements of the corresponding engineering part. When this determination is true: Write a qualification flag in the metadata header of the transferable gradation curve; This gradation curve is formally defined as a qualified alternative material source curve. Lock in the corresponding processing cost coefficient so that it will not be adjusted in subsequent steps.

[0072] This locking mechanism ensures that the economic parameters used in the subsequent allocation and optimization phases maintain a one-to-one correspondence with the mechanical performance verification results.

[0073] If any mechanical performance index of a transferable gradation curve fails to meet the engineering design threshold in the rolling forecast results, then the gradation curve is deemed unusable as an alternative material source under the premise of ensuring engineering safety. In this case: the transferable gradation curve is immediately discarded; it is prohibited from entering the three-dimensional preference cost matrix constructed in subsequent step 5; and it no longer participates in any blending optimization calculations. Through this strict rejection rule, it is ensured that the final set of qualified alternative material source curves entering the blending model is strictly consistent with the mechanical performance verification results at the data level.

[0074] Step 4 further maps the geometric similarity of the gradation and processing feasibility to mechanical performance safety, completing the pre-verification of the mechanical properties of alternative material sources before blending optimization, eliminating the uncertainty of engineering safety that cannot be determined solely by gradation matching. The resulting set of qualified alternative material source curves provides a unique and reliable data foundation for the construction of the three-dimensional preference cost matrix in step 5.

[0075] Specifically, step 5 includes: After completing step 4, the system has obtained a set of qualified alternative material source curves that have undergone rigorous mechanical property verification. Each qualified alternative material source curve is associated with at least the following information in its data structure: the original gradation curve number of the blasting zone from which it originates; the target gradation curve type of the corresponding receiving point; the locked processing cost coefficient; and verification through compaction performance-mechanical property rolling forecast. Step 5, based on this, no further gradation or mechanical level judgments are made. Instead, the conclusion of whether it is substitutable is transformed into an economic quantification form that can be directly used in blending optimization.

[0076] To reflect the combined impact of processing costs and transportation route costs in the allocation model, a three-dimensional preference cost matrix is ​​introduced as a unified cost representation carrier.

[0077] 1. Definition of three-dimensional coordinate axes The three-dimensional preference cost matrix is ​​constructed in the following three dimensions: X-axis: Explosion zone number dimension The blasting zone number in the set of original gradation curves of the blasting zone from which the qualified alternative material source curve originates is used to distinguish different excavation sources; Y-axis: Transportation cost dimension The shortest transportation distance between the excavation point and the receiving point is calculated as the transportation cost based on the predetermined transportation unit price. Z-axis: Machining cost dimension The corresponding qualified alternative material source curve is the processing cost coefficient locked in step 3.

[0078] Through the above three-dimensional definition, the economic relationship between the material source, the receiving material, and the processing material is mapped to a unified three-dimensional cost space.

[0079] In the three-dimensional preference cost matrix, each valid cell corresponds to the comprehensive cost of a qualified alternative material source curve under specific transportation routes and processing conditions. The preference cost of this cell is defined as the linear superposition of processing costs and transportation costs, expressed as: in: The third in the three-dimensional preference cost matrix The explosion zone, corresponding to the first... The first receiving point, using the first The preferred cost when a qualified alternative material source curve is available; This refers to the processing cost coefficient corresponding to the qualified alternative material source curve; To get from the explosion zone To the receiving point Transportation costs.

[0080] A cell in the matrix is ​​filled only when a qualified alternative material source curve that satisfies mechanical performance verification exists for a given blast zone-receiving point combination; otherwise, the cell remains empty. The resulting three-dimensional preference cost matrix naturally exhibits sparsity and includes only engineering-feasible alternative paths.

[0081] To enable the three-dimensional preference cost matrix to be directly used for earthwork allocation optimization, it is expanded in dimensions and mapped to coefficients.

[0082] 1. Sparse coefficient vector expansion The three-dimensional preference cost matrix is ​​expanded into a two-dimensional sparse coefficient vector according to the combination of the explosion zone number, receiving point and qualified alternative material source curve, so that each non-empty cell corresponds to a transportation-processing joint decision variable in the allocation model.

[0083] 2. Replace the traditional 0-1 matching coefficient In traditional linear earthwork allocation models, the objective function coefficients typically only include unit transportation costs, and the decision variables are limited to 0-1 gradation matching relationships.

[0084] In this method, the aforementioned sparse coefficient vector is directly written into the objective function coefficient term of the linear earthwork allocation model, completely replacing the unit transportation cost coefficient in the traditional model. This allows the model's objective function to simultaneously reflect: the processing differences between different material sources; the distance differences between different transportation routes; and the comprehensive economic preference among different alternatives.

[0085] This leads to a quantitative allocation model for substitutability, where the decision variables are no longer simply available / unavailable, but rather implicitly represent a continuous economic judgment on whether or not a substitution is worthwhile.

[0086] In the substitutability quantitative allocation model, a multi-objective allocation concept is introduced, unifying engineering utilization efficiency and economic cost into a single linear objective function.

[0087] 1. Definition of primary and secondary objectives Main objective: to maximize the comprehensive utilization rate of the area, reflecting the priority pursuit of efficient utilization of raw material resources in the blasting area; Sub-objective: Minimize total cost, which is reflected in the economic optimization under the premise of meeting the utilization rate target.

[0088] 2. Weighted expression of the objective function By setting weight coefficients for the primary and secondary objectives, they are combined into a single linear objective function. The preference cost in the three-dimensional preference cost matrix is ​​embedded in the objective function as the core coefficient of the secondary objective. This objective function completes the comprehensive encoding of transportation distance, processing cost, and alternative feasibility during the coefficient initialization stage, so that no nonlinear or logical judgments are required in the subsequent solution stage.

[0089] By implementing step 5, the gradation, processing, and mechanical property judgment results formed in steps 1-4 are uniformly transformed into economic parameters that can be directly used in the blending model. This eliminates the crude assumption of compliance equals price in the traditional 0-1 matching model, enabling the blending model to express the ability to exchange processing for transportation and near-gradation for high utilization in terms of mathematical structure. This completes the key leap from material substitutability judgment to the computability of blending optimization.

[0090] Specifically, step 6 includes: Step 6 uses the substitutability quantitative allocation model constructed in Step 5 as the sole input. This model has the following significant characteristics: the decision variables are joint transportation-processing variables, corresponding only to qualified alternative material source curves; the objective function coefficients are derived from the three-dimensional preference cost matrix, comprehensively reflecting processing and transportation costs; the constraints reflect the conservation relationships of material supply, receiving demand, and engineering allocation; all variables entering the model have undergone mechanical performance verification in Step 4, and safety assessments are no longer required in the solution phase.

[0091] Because the model has the characteristics of large variable scale, sparse rows and columns, and linear but complex objective function, a solution strategy based on the branch-and-price framework is adopted to calculate the global optimal solution.

[0092] The branch-price solver consists of the following two core stages: 1. Column generation stage (price stage) Used to efficiently generate transport-processing joint variables with potential improvement value under continuous relaxation; 2. Branching Phase It is used to perform branch search on key discrete decision variables under integer constraints until the globally optimal integer solution is obtained.

[0093] The two stages described above operate collaboratively within a unified framework to ensure the global optimality and integer feasibility of the solution results.

[0094] In the main problem of the branch-price solver, a column generation method is used to construct the relaxed solution of the allocation model. Only transport-processing joint variable columns corresponding to qualified alternative material source curves are allowed to be generated in the main problem; any gradation scheme that fails the mechanical property verification in step 4 is not allowed to enter the main problem. Each column corresponds to: a qualified alternative material source curve; a determined transport path; and a set of locked optimal processing operation vectors. The main problem is initialized with only a minimum set of columns that satisfy basic feasibility to ensure model solvability, not optimality.

[0095] During the column generation phase, the existence of new columns with improvement potential is determined by solving subproblems.

[0096] 1. Definition of simplified cost The subproblem uses the preference cost in the three-dimensional preference cost matrix as the core cost term to construct the reduced cost of the corresponding transportation-processing path. For candidate columns, the reduced cost is... Represented as: in: The value of the preference corresponding to this column; The dual variables of each constraint in the main problem; For this column in constraints The coefficients in.

[0097] 2. Pricing Solution The subproblem employs the shortest path labeling algorithm, searching for paths with negative reduced costs within the implicitly constructed transport-processing network. When a path with negative reduced cost is found... When adding a new column, add that column to the main problem and re-solve the relaxed solution of the main problem.

[0098] 3. Column generation termination condition When no column with negative reduction cost can be found in the subproblem, it is determined that the relaxation solution of the current main problem has reached the optimal in the existing column space, and the column generation phase terminates.

[0099] After the column generation phase ends, if the current solution is an integer solution, it is directly output as the optimal solution; otherwise, the process proceeds to the branching phase.

[0100] 1. Selection of branch variables The branching phase uses the blast zone number variable in the original gradation curve set of the blast zone as the branching object and performs binary branching on it to ensure that the allocation decision satisfies the integer constraint at the blast zone level.

[0101] 2. Branch Search and Boundary Control In each branch node, the column generation process is repeated to update the node's relaxed solution, and the objective function values ​​of the integer solutions and relaxed solutions are compared. The branch search terminates when the following convergence condition is met: in: The target value is the current optimal integer solution. This corresponds to the objective value of the relaxation solution.

[0102] Once the branch-price solver has reached convergence, it outputs the globally optimal solution, which consists of three types of information: the set of selected qualified alternative material source curves; the optimal transportation path corresponding to each qualified alternative material source curve; and the optimal processing step vector matching it. This combined solution mathematically satisfies the following simultaneously: maximum comprehensive utilization rate of the area; minimum total cost; and compliance of mechanical performance indicators at all material receiving points.

[0103] The above-mentioned globally optimal solution results are written into the intelligent earthwork allocation scheme file. This file records at least the following: the final selected qualified alternative material source curve and its source blasting zone number; the corresponding transportation path and processing step vector; the comprehensive utilization rate of the area; the total allocation cost; and the rolling forecast index of compaction performance-mechanical properties corresponding to all material receiving points. This completes the closed-loop execution from initial blasting zone gradation perception, transferability judgment, processing and mechanical verification, to allocation optimization and global solution.

[0104] Example 2 This embodiment discloses an intelligent earthwork allocation and optimization system, including: Gradation acquisition module: acquires the original block size spatial distribution point cloud of the blasting area to be excavated, and generates the original gradation curve set of the blasting area based on the original block size spatial distribution point cloud of the blasting area; Gradation screening module: For each target gradation curve in the engineering design envelope library, insert a gradation similarity measurement function driven by Hausdorff distance between the upper and lower envelopes of the target gradation curve, map the original gradation curve set of the blasting area to the target gradation curve point by point, and output the transferable gradation curve subset and its initial similarity value. Process determination module: Taking the transferable gradation curve subset as input, it generates the optimal processing operation vector and the corresponding processing cost coefficient required for the transferable gradation curve subset to fall into the envelope of the target gradation curve. Performance verification module: Substitute the optimal processing procedure vector into the discrete element-finite element coupled performance predictor, and perform rolling forecast of compaction performance and mechanical properties on the processed transferable gradation curve subset. If the predicted permeability coefficient, compression modulus and shear strength all meet the engineering design threshold, then the transferable gradation curve subset is marked as a qualified alternative material source curve. Allocation Modeling Module: Construct a three-dimensional preference cost matrix using qualified alternative material source curves, processing cost coefficients, and transportation distances. Then, directly write the three-dimensional preference cost matrix into the objective function coefficients of the linear earthwork allocation model to form a substitutable quantitative allocation model. Optimization Solver: Calls the branch-price solver to perform global optimal solution on the substitutability quantification allocation model, and outputs an earthwork allocation scheme that simultaneously maximizes the comprehensive utilization rate of the area, minimizes the total cost, and ensures that the mechanical performance indicators of all material receiving points meet the standards.

[0105] The modules in Embodiment 2 are used to implement the functions in Embodiment 1. This embodiment can be implemented by a system including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement the intelligent earthwork allocation and optimization method according to Embodiment 1 of this application. The system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface. Their settings and functions are known in the art and will not be described in detail here.

[0106] In this application, the aforementioned memory can be any tangible medium containing or storing a program that can be used or combined with an instruction execution system, system, or device. For example, a computer-readable storage medium can be any suitable magnetic or magneto-optical storage medium, such as Resistive Random Access Memory (RRAM), Dynamic Random Access Memory (DRAM), Static Random Access Memory (SRAM), Enhanced Dynamic Random Access Memory (EDRAM), High-Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), etc., or any other medium that can be used to store required information and can be accessed by an application, module, or both. Any such computer storage medium can be part of a device or accessible to or connected to a device. Any application or module described in this application can be implemented using computer-readable / executable instructions that can be stored or otherwise retained by such a computer-readable medium.

[0107] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A method for intelligent allocation and optimization of earthwork, characterized in that, Includes the following steps: Step 1: Obtain the original block size spatial distribution point cloud of the blasting area to be excavated, and generate the original gradation curve set of the blasting area based on the original block size spatial distribution point cloud. Step 2: For each target gradation curve in the engineering design envelope library, insert a gradation similarity measurement function driven by Hausdorff distance between the upper and lower envelopes of the target gradation curve, map the original gradation curve set of the blasting area to the target gradation curve point by point, and output the transferable gradation curve subset and its initial similarity value. Step 3: Using the transferable gradation curve subset as input, generate the optimal processing operation vector and corresponding processing cost coefficient required for the transferable gradation curve subset to fall into the envelope of the target gradation curve. Step 4: Substitute the optimal processing procedure vector into the discrete element-finite element coupled performance predictor and perform a rolling forecast of compaction performance and mechanical properties on the processed transferable gradation curve subset. If the predicted permeability coefficient, compression modulus, and shear strength all meet the engineering design threshold, then mark the transferable gradation curve subset as a qualified alternative material source curve. Step 5: Construct a three-dimensional preference cost matrix using qualified alternative material source curves, processing cost coefficients, and transportation distances, and directly write the three-dimensional preference cost matrix into the objective function coefficient term of the linear earthwork allocation model to form a quantitative allocation model of substitutability. Step 6: Call the branch-price solver to perform a global optimal solution on the substitutability quantification allocation model, and output an earthwork allocation scheme that simultaneously maximizes the comprehensive utilization rate of the area, minimizes the total cost, and ensures that the mechanical performance indicators of all material receiving points meet the standards.

2. The intelligent allocation and optimization method for earthwork as described in claim 1, characterized in that, Step 1 includes: A laser-image joint scanning system composed of LiDAR and an industrial camera is used to perform millimeter-resolution scanning of the area to be excavated before blasting, acquiring the original block size spatial distribution point cloud of the area to be excavated, including three-dimensional coordinates, echo intensity, and synchronous image information. The original block size spatial distribution point cloud of the area to be excavated is input into a particle size recognition neural network. The particle size recognition neural network uses voxelization of the point cloud as the input layer and the block size particle size frequency distribution vector as the output layer. It extracts block size edge features through convolution kernels and counts the proportion of points within the particle size interval, generating a set of original gradation curves of the blasting area composed of gradation samples under different blasting parameters. The set of original gradation curves of the blasting area is encapsulated in XML format as a structured data package including three elements: particle size interval, pass percentage, and non-uniformity coefficient. The structured data package is then used as the input to the gradation curve set in subsequent calculation steps.

3. The intelligent allocation and optimization method for earthwork as described in claim 1, characterized in that, Step 2 includes: The system reads the target gradation curves of filter material, transition material, A / B group roadbed material, and building graded crushed stone, which are classified according to the nominal maximum particle size, one by one from the engineering design envelope library. A continuous gradation coordinate point series is established between the upper and lower envelopes of each target gradation curve with a step size of 0.1 mm. Using each original gradation curve in the blasting area set as a reference, a gradation similarity measurement function weighted by Hausdorff distance and Frechet distance is called to perform a point-by-point mapping operation on the gradation coordinate point series of the original gradation curves in the blasting area and each target gradation curve, obtaining an initial similarity value reflecting the deviation of the curve's spatial shape. Original gradation curves in the blasting area with initial similarity values ​​less than or equal to a preset threshold are collectively included in the transferable gradation curve subset, and the corresponding initial similarity values ​​are written into the metadata header of the transferable gradation curve subset, forming a transferable gradation curve subset file.

4. The intelligent allocation and optimization method for earthwork as described in claim 1, characterized in that, Step 3 includes: The transferable gradation curve subset is input line by line into the gradation offset processing response surface model, which has been pre-calibrated through a three-factor, five-level orthogonal experiment. The response surface model uses the particle size offset removed by screening and washing, the oversize correction by impact crushing, and the mud content reduction by water mixing as independent variables, and the sum of squares of the percentage differences between each particle size point between the transferable gradation curve subset and the corresponding target gradation curve as the response value. An explicit functional relationship between the independent variables and the response value is established through quadratic polynomial regression. With minimizing the response value as the optimization objective, a sequential quadratic programming algorithm is called to solve the problem under the physical boundary constraints of the independent variables. The output is the optimal processing step vector composed of screening and washing time, impact crushing discharge port width, and water mixing volume. At the same time, the processing cost coefficient obtained by accumulating equipment shifts, water and electricity consumption, and labor quota is calculated. The optimal processing step vector and the processing cost coefficient are written into the same-name field of the transferable gradation curve subset to form a transferable gradation curve subset with cost annotation.

5. The intelligent allocation and optimization method for earthwork as described in claim 1, characterized in that, Step 4 includes: A cost-labeled subset of transferable gradation curves, along with the optimal processing step vector, is fed into a discrete-element-finite-element coupled performance predictor. The predictor uses discrete-element analysis to generate a three-dimensional particle packing sample of the processed transferable gradation curve subset, and applies the designed compaction work to the sample using finite-element analysis, outputting a real-time rolling forecast of compaction and mechanical properties, including permeability coefficient, compressive modulus, and shear strength. If the permeability coefficient in the rolling forecast of compaction and mechanical properties is less than 1 × 10⁻⁶, the prediction is considered complete. -4 If the compressive modulus is greater than 80 MPa and the shear strength is greater than 35°, then a qualified identifier is written into the metadata header of the migrated gradation curve subset to form a qualified alternative material source curve, and the corresponding processing cost coefficient is locked and will not be changed; if any of the indicators does not reach the threshold, then the current migrated gradation curve subset is discarded.

6. The intelligent allocation and optimization method for earthwork and rock excavation according to claim 1, characterized in that, Step 5 includes: Using the processing cost coefficient corresponding to the qualified alternative material source curve as the Z-axis, the transportation cost calculated based on the shortest transportation distance from the excavation point to the receiving point as the Y-axis, and the blasting area number in the original gradation curve set of the blasting area from which the qualified alternative material source curve originates as the X-axis, a three-dimensional preference cost matrix containing only qualified alternative material source curves is constructed. Each cell stores the preference cost value of the sum of a unique processing cost coefficient and transportation cost. The three-dimensional preference cost matrix is ​​expanded into a two-dimensional sparse coefficient vector and directly written into the objective function coefficient term of the linear earthwork allocation model, replacing the unit transportation cost coefficient in the traditional 0-1 matching matrix. This allows the decision variables of the linear earthwork allocation model to simultaneously reflect the economic trade-off between transportation paths and processing procedures, forming a substitutable quantitative allocation model. In the substitutable quantitative allocation model, the primary objective is to maximize the comprehensive utilization rate of the area, and the secondary objective is to minimize the total cost. The preference cost value of the three-dimensional preference cost matrix is ​​embedded into the single linear objective function of the substitutable quantitative allocation model through objective weighting, completing the coefficient initialization of the substitutable quantitative allocation model.

7. The intelligent allocation and optimization method for earthwork as described in claim 1, characterized in that, Step 6 includes: Using the substitutability quantification allocation model as the sole input, a branch-price solver based on a branch-price framework is invoked. In the main problem of the branch-price solver, a column generation method is used to generate columns only for the transportation-processing joint variables corresponding to the qualified alternative material source curves. In the subproblems, the preference cost value of the three-dimensional preference cost matrix is ​​used to reduce the cost coefficient, and the pricing loop is completed through the shortest path label algorithm. When the reduced cost coefficient can no longer be negative, the branch-price solver terminates column generation and enters the branching stage to perform binary branching on the blast zone number variable in the original blast zone gradation curve set until the difference between the relaxation solution and the integer solution of all nodes is less than 0.01%. The output is the global optimal solution result composed of the qualified alternative material source curve, the corresponding transportation path, and the corresponding optimal processing procedure vector. The global optimal solution result is written into the intelligent earthwork allocation scheme file. The file simultaneously records the comprehensive utilization rate of the area, the total cost, and the rolling forecast index of compaction performance-mechanical performance of all receiving points, thus completing the entire execution of the intelligent earthwork allocation optimization method.

8. An intelligent earthwork allocation and optimization system, characterized in that, include: Gradation acquisition module: acquires the original block size spatial distribution point cloud of the blasting area to be excavated, and generates the original gradation curve set of the blasting area based on the original block size spatial distribution point cloud of the blasting area; Gradation screening module: For each target gradation curve in the engineering design envelope library, insert a gradation similarity measurement function driven by Hausdorff distance between the upper and lower envelopes of the target gradation curve, map the original gradation curve set of the blasting area to the target gradation curve point by point, and output the transferable gradation curve subset and its initial similarity value. Process determination module: Taking the transferable gradation curve subset as input, it generates the optimal processing operation vector and the corresponding processing cost coefficient required for the transferable gradation curve subset to fall into the envelope of the target gradation curve. Performance verification module: Substitute the optimal processing procedure vector into the discrete element-finite element coupled performance predictor, and perform rolling forecast of compaction performance and mechanical properties on the processed transferable gradation curve subset. If the predicted permeability coefficient, compression modulus and shear strength all meet the engineering design threshold, then the transferable gradation curve subset is marked as a qualified alternative material source curve. Allocation Modeling Module: Construct a three-dimensional preference cost matrix using qualified alternative material source curves, processing cost coefficients, and transportation distances. Then, directly write the three-dimensional preference cost matrix into the objective function coefficients of the linear earthwork allocation model to form a substitutable quantitative allocation model. Optimization Solver: Calls the branch-price solver to perform global optimal solution on the substitutability quantification allocation model, and outputs an earthwork allocation scheme that simultaneously maximizes the comprehensive utilization rate of the area, minimizes the total cost, and ensures that the mechanical performance indicators of all material receiving points meet the standards.

Images