An AI-based intelligent mine design process point-surface fusion method

Abstract

This invention discloses an AI-based intelligent mine design process point-to-surface fusion method, belonging to the field of artificial intelligence mine design technology. The method includes: automatically assembling candidate process fragments according to upstream and downstream interface rules, and simultaneously expanding process parameters to generate a parameterized candidate process flow chain set; using the parameterized candidate process flow chain set to map process units and related objects as heterogeneous graph nodes, forming a heterogeneous graph node set, and establishing process associations, spatial associations, constraint associations, and result reference associations between nodes to generate a point-to-surface fused heterogeneous graph; using the point-to-surface fused heterogeneous graph to perform state propagation and association analysis on various types of nodes, identifying conflict points between processes, equipment, buildings, transportation, and specifications, forming a conflict point set, and performing local optimization on relevant process gene fragments and related nodes to generate a preferred point-to-surface fused design diagram. This invention achieves unified relationship modeling between multiple types of objects in mine design.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence mine design technology, and in particular to a point-to-surface fusion method for intelligent mine design process based on AI. Background Technology

[0002] In recent years, mine engineering design has gradually shifted from being dominated by manual experience to digital and intelligent development. Especially in the design of mineral processing and related engineering projects, a technical path has been formed that combines parametric design, knowledge base retrieval, and large-scale model-assisted writing. This path enables data archiving, index table generation, equipment table generation, and auxiliary output of quantity flow diagrams, building connection diagrams, water balance diagrams, and feasibility study documents.

[0003] However, existing technologies still have two shortcomings: First, there is a lack of a unified process-level linkage mechanism between process objectives, process constraints, equipment capabilities, site conditions, and tailings return water requirements, making it difficult to automatically select and assemble suitable process flows based on input conditions; Second, drawings, tables, and texts are often generated in a scattered manner, lacking unified relationship modeling and conflict identification capabilities, which can easily lead to inconsistencies in the relationships between processes, equipment, buildings, transportation, and specifications, affecting the overall coordination and traceability of design results. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an AI-based intelligent mine design process point-to-surface fusion method to solve the problems of difficult linkage construction of process flow and insufficient coordination and traceability of design results in the existing technology.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides an AI-based intelligent mine design process point-to-surface fusion method, which includes: acquiring multi-source data of a mine project and extracting information such as processing scale, ore properties, target indicators, tailings water return requirements, and site constraints to generate a standardized input set for the project; based on the standardized input set, using AI to identify the process objectives and constraints corresponding to each process step, generating a process constraint vector, and selecting suitable process gene fragments from a preset process gene fragment library to generate a candidate process fragment set; based on the candidate process fragment set, automatically assembling them according to upstream and downstream interface rules, and simultaneously expanding process parameters to generate a parameterized candidate process flow chain set; Using a parameterized candidate process flow chain set, process units and associated objects are mapped to heterogeneous graph nodes, forming a heterogeneous graph node set. Process associations, spatial associations, constraint associations, and result reference associations are established between each node, generating a point-surface fusion heterogeneous graph. The point-surface fusion heterogeneous graph is used to perform state propagation and association analysis on various types of nodes, identify conflict points between processes, equipment, buildings, transportation, and specifications, form a conflict point set, and perform local optimization on relevant process gene fragments and associated nodes to generate a preferred point-surface fusion design graph. Based on the preferred point-surface fusion design graph, integrated design results of diagrams, tables, and text are generated in a coordinated manner to obtain the point-surface fusion results of the mine design process.

[0007] As a preferred embodiment of the AI-based intelligent mine design process point-surface fusion method described in this invention, the specific steps for generating the standardized input set for the project are as follows: Acquire multi-source data from mining projects, and classify, catalog, analyze, and unify the units of the multi-source data to generate a standardized analysis set; Using a standardized parsing set, field location, semantic extraction, and multi-source fusion are performed on processing scale, ore properties, target indicators, tailings water return requirements, and site constraints to generate a fusion parameter set; The site constraint comprehensive coefficient is calculated based on the fusion parameter set, and the processing scale, ore properties, target indicators, tailings water return requirements and site constraint information are uniformly encapsulated to generate a standardized input set for the project.

[0008] As a preferred embodiment of the AI-based intelligent mine design process point-surface fusion method described in this invention, the specific steps for generating the process constraint vector are as follows: The standardized input set of the project is used to map the processing scale, ore properties, target indicators, tailings water return requirements and site constraints into a unified feature vector. The unified feature vector is then input into the AI ​​recognition model and matched with the prototype of the preset process link to form the initial activation vector of each process link. Based on the initial activation vector, and combined with the target indicators and ore property information, the process objectives and process constraints corresponding to each process step are identified, forming a set of target constraints for each process step. Based on the target constraint set, cross-stage propagation and consistent fusion are performed along the process sequence of crushing, screening, grinding, classification, beneficiation, concentration, filtration and tailings water recycling to form a process constraint vector.

[0009] As a preferred embodiment of the AI-based intelligent mine design process point-surface fusion method described in this invention, the specific steps for generating the candidate process segment set are as follows: Based on the process constraint vector, the constraint information corresponding to each process step is broken down into step-level search targets; The process-level search targets are matched with the process gene fragment description information in the process gene fragment library, and the process-level preferred fragment sequences are selected. Based on the optimal sequence of process segments at the stage level, and combined with the input-output interface adaptation relationship between adjacent process stages, a pre-compatibility screening is performed to form a set of candidate process segments.

[0010] As a preferred embodiment of the AI-based intelligent mine design process point-surface fusion method described in this invention, the specific steps for generating a parameterized candidate process flow chain set are as follows: The candidate process segments are reconstructed into typed process segments with input ports, output ports, status ports and resource ports respectively using the candidate process segment set, forming a typed process segment ledger; Based on the typified process segment ledger, a hierarchical assembly diagram is constructed according to the process sequence, and forward expansion is performed according to the input and output interface rules between adjacent process links to form a partial process flow chain. Based on a partial process flow chain, calculate the process parameter vector for each process segment, and use the process parameters to filter out process flow chains that can be further expanded. The process parameters, spatial resources, and results references are modified using an expandable process chain, and the chain continues to expand along the process sequence to the tailings water recycling stage, forming a set of parameterized candidate process chains.

[0011] As a preferred embodiment of the AI-based intelligent mine design process point-surface fusion method described in this invention, the specific steps for forming a heterogeneous graph node set are as follows: The parameterized candidate process flow chain set is used to decompose and classify the process units, equipment objects, building objects, transportation objects, auxiliary objects and result anchors in each process flow chain to form a source object set; Based on the source object set, the corresponding equipment objects, building objects, transportation objects, auxiliary objects and result anchors are gathered around each process unit, and the correspondence between objects is established to obtain the node projection list; Based on the node projection list, process units and associated objects are written into the point-surface fusion heterogeneous graph, and a mapping relationship between the source object and the nodes of the heterogeneous graph is established to form a heterogeneous graph node set.

[0012] As a preferred embodiment of the point-surface fusion method for AI-based intelligent mine design process described in this invention, the specific steps for generating the point-surface fusion heterogeneous map are as follows: Based on the heterogeneous graph node set, corresponding equipment nodes, building nodes, transportation nodes, auxiliary nodes, standard nodes and result nodes are gathered around each process node to form candidate related units; Based on candidate association units, process associations, spatial associations, constraint associations, and result reference associations between nodes are identified according to process sequence, spatial location, constraint scope, and result anchor reference path, forming an effective association set; By utilizing the effective association set, process associations, spatial associations, constraint associations, and result reference associations are written into the heterogeneous graph node set to generate a point-surface fused heterogeneous graph.

[0013] As a preferred embodiment of the AI-based intelligent mine design process point-surface fusion method described in this invention, the specific steps for forming the conflict point set are as follows: Based on the point-to-surface fusion heterogeneous graph, equipment nodes, building nodes, transportation nodes, auxiliary nodes and specification nodes are gathered and associated around each process node to form a state propagation unit; Based on the state propagation unit, state propagation and correlation analysis are performed along process correlation, spatial correlation, constraint correlation and result reference correlation to form candidate conflict units; Use candidate conflicting units for process, equipment, construction, and transportation: As a preferred embodiment of the AI-based intelligent mine design process point-surface fusion method described in this invention, the specific steps for generating the preferred point-surface fusion design map are as follows: Based on the set of conflict points, process gene fragments, equipment nodes, building nodes, transportation nodes, auxiliary nodes, standard nodes and result nodes corresponding to each conflict point are extracted from the point-surface fusion heterogeneous graph to form local conflict slice units. The relevant process gene fragments are replaced using local conflict slicing units, and the associated nodes are redirected to form a local repair candidate set; Based on the local repair candidate set, the target local repair scheme is selected and written back to the point-surface fusion heterogeneous graph to generate the optimal point-surface fusion design graph.

[0014] As a preferred embodiment of the AI-based intelligent mine design process point-to-surface fusion method described in this invention, the obtained mine design process point-to-surface fusion result is as follows: Semantic information corresponding to process nodes, equipment nodes, building nodes, transportation nodes, auxiliary nodes, standard nodes, and result anchor points is extracted from the optimized point-surface integrated design drawing to form the result source ledger; By utilizing the source ledger of results, the main process line, engineering objects, standard basis and result citation relationship are mapped into drawing tasks, table tasks and text tasks respectively, forming a multi-view compilation dependency network; Based on the multi-view compilation dependency network, a quantity flow diagram, a building connection diagram, and a water balance diagram are generated, and the corresponding information is written into a table task and a text task to obtain the point-to-surface fusion result of the mine design process.

[0015] The beneficial effects of this invention are as follows: By reconstructing candidate process segments into typified process segments and automatically assembling them according to upstream and downstream interface rules, the automatic construction of discrete process segments into continuous process flow chains is realized, enabling candidate process flow chains to directly participate in subsequent equipment mapping, spatial constraint verification, computer-aided design, and result generation; By mapping process units and related objects into point-surface fusion heterogeneous graph nodes and establishing the association relationship between nodes, unified relationship modeling between multiple types of objects in mine design is realized, thereby enabling local repair and the formation of optimal design drawings without overturning the global design. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart of a point-to-surface fusion method for AI-based intelligent mine design.

[0018] Figure 2 A flowchart for generating a standardized input set for the project.

[0019] Figure 3 A flowchart for generating a parameterized candidate process flow chain set.

[0020] Figure 4 A flowchart for generating the point-to-surface integration results of the mine design process.

[0021] Figure 5 Relationship between site constraint comprehensive coefficient and process chain assembly success rate.

[0022] Figure 6 Scatter plot of the impact range of local repair and residual conflict after repair. Detailed Implementation

[0023] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0024] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0025] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0026] Reference Figures 1-6 This is one embodiment of the present invention, which provides an AI-based intelligent mine design process point-surface fusion method, including the following steps: S1. Obtain multi-source data for the mining project and extract information on processing scale, ore properties, target indicators, tailings water return requirements, and site constraints to generate a standardized input set for the project.

[0027] Acquire multi-source data from mining projects, and classify, catalog, analyze, and unify the units of these multi-source data to generate a standardized analysis set.

[0028] The specific process includes receiving the entrustment letter, mineral processing test report, geological survey data, topographic map, equipment data, and specification data from the project collaboration platform; reading electronic documents, scanned copies, forms, and drawings from the data upload directory; and registering each item according to data name, source, creation time, and corresponding profession. After registration, the construction scale and target requirements in the entrustment letter are read, the ore properties and index data in the mineral processing test report are extracted, the elevation, boundary, and site area information in the topographic map are extracted, the specification parameters in the equipment data are extracted, and the professional clauses in the specification data are extracted. The units of measurement, field names, and expression forms in different data are uniformly processed to form a standardized parsing set.

[0029] By using a standardized parsing set, field location, semantic extraction, and multi-source fusion of information on processing scale, ore properties, target indicators, tailings water return requirements, and site constraints are performed to generate a fusion parameter set.

[0030] The specific process includes: using the text content, table content, and drawing annotation information in the standardized parsing set, searching for the field names, field locations, and field contexts corresponding to the processing scale, ore properties, target indicators, tailings return water requirements, and site constraints information item by item to complete field location; based on field location, combining the professional terminology, sentence semantics, and table meanings in the data, performing semantic extraction on the processing scale, ore properties, target indicators, tailings return water requirements, and site constraints information; based on semantic extraction, comparing, verifying, and multi-source fusing parameters with consistent content or the same meaning from different data sources to eliminate differences in expression and information duplication, and generating a fused parameter set.

[0031] It should be noted that tailings water return requirements refer to the constraints imposed on the utilization method, volume, quality, and proportion of return water during the tailings treatment process; site constraint information refers to the limitations imposed on the process layout and engineering location in terms of the area, elevation difference, slope, boundaries, and buildable scope of the mining area.

[0032] The site constraint comprehensive coefficient is calculated based on the fusion parameter set, and the processing scale, ore properties, target indicators, tailings water return requirements and site constraint information are uniformly encapsulated to generate a standardized input set for the project.

[0033] The specific process includes: using the processing scale, ore properties, target indicators, tailings water return requirements, and site constraint information from the fusion parameter set to quantify the effective construction area, site elevation difference, and slope constraints, forming a comprehensive site constraint coefficient that can characterize the site's carrying capacity and the degree of restriction on process layout; and encapsulating and organizing the processing scale, ore properties, target indicators, tailings water return requirements, and site constraint information according to unified field names, unified unit formats, and unified data order, so that the key input content of the same project forms a standardized project input set with consistent structure, consistent expression, and direct callability.

[0034] The expression for calculating the site constraint comprehensive coefficient using the fused parameter set is as follows: ; in, This represents the overall coefficient of site constraints. Indicates the effective construction area. Indicates the total area of ​​the site. Indicates the elevation difference of the site. This represents the normalized baseline value for elevation difference. This indicates the average slope of the site. This represents the normalized baseline value for slope.

[0035] Furthermore, before calculating the comprehensive coefficient of site constraints, the effective construction area, total site area, site elevation difference, average site slope, normalized elevation difference benchmark value, and normalized slope benchmark value have been unified in terms of dimensions using a unified unit conversion and dimensionless processing method. Specifically, the effective construction area and total site area are expressed in the same area unit, the site elevation difference and the normalized elevation difference benchmark value are expressed in the same length unit, and the average site slope and the normalized slope benchmark value are expressed in the same slope characterization method, so that each ratio term in the expression is a dimensionless quantity when it is entered into the calculation.

[0036] It should be noted that, It was obtained by normalizing and calibrating statistical samples of site elevation differences from similar mining projects. It was obtained by normalizing and calibrating statistical samples of slope at sites of similar mining projects.

[0037] S2. Based on the standardized input set of the project, use AI to identify the process objectives and constraints corresponding to each process step, generate process constraint vectors, and select suitable process gene fragments from the preset process gene fragment library to generate a candidate process fragment set.

[0038] The standardized input set of the project is used to map the processing scale, ore properties, target indicators, tailings water return requirements and site constraints into a unified feature vector. The unified feature vector is then input into the AI ​​recognition model and matched with the preset process prototype to form the initial activation vector of each process.

[0039] The specific process includes, when using the standardized input set for the project, reading the field content from the processing scale, ore properties, target indicators, tailings reclaimed water requirements, and site constraints, uniformly encoding the text fields, numerical fields, and category fields, and arranging them according to a unified field order, so that the processing scale, ore properties, target indicators, tailings reclaimed water requirements, and site constraints form a unified feature vector in the same vector space; inputting the unified feature vector into the AI ​​recognition model, which compares the unified feature vector with the preset process prototype item by item, and outputs the matching strength for each process step of crushing, screening, grinding, grading, beneficiation, concentration, filtration, and tailings reclaimed water, forming the initial activation vector for each process step.

[0040] It should be noted that the prototype of the process steps is based on the typical input conditions, typical output characteristics and typical constraint information of each process step in historical mining projects, including crushing, screening, grinding, classification, beneficiation, concentration, filtration and tailings water recycling. It is pre-formed after being classified, labeled and standardized.

[0041] The AI ​​recognition model is a recognition network used to identify the initial activation vectors of each process step in crushing, screening, grinding, grading, beneficiation, concentration, filtration, and tailings recycling based on a standardized input set of the project. The AI ​​recognition model is constructed from a feature encoding layer and a process step prototype matching layer. The feature encoding layer is used to uniformly represent the processing scale, ore properties, target indicators, tailings recycling requirements, and site constraints. The process step prototype matching layer is used to match the uniform feature vectors with the preset process step prototypes and output the matching strength of each process step, thereby forming the initial activation vector of each process step.

[0042] Furthermore, the training process of the AI ​​recognition model includes collecting condition authorization letters, ore beneficiation test reports, topographic data, equipment data, and specification data from historical mining projects, and completing sample labeling based on processing scale, ore properties, target indicators, tailings return water requirements, and site constraints; cleaning, deduplication, unit unification, and unified coding of text fields, numerical fields, and category fields in the samples to form training sample sets, validation sample sets, and test sample sets; inputting the training sample set into the AI ​​recognition model for supervised training, adjusting parameters in conjunction with the validation sample set, and then using the test sample set for verification to obtain the AI ​​recognition model.

[0043] Based on the initial activation vector, and combined with the target indicators and ore property information, the process objectives and process constraints corresponding to each process step are identified, forming a set of target constraints for each process step.

[0044] The specific process includes, based on the response intensity of each process step in the initial activation vector, combined with the concentrate grade requirements, recovery rate requirements, processing scale requirements, and tailings water recovery requirements in the target indicators, as well as the particle size composition, mineral composition, dispersive characteristics, grindability, and beneficiation characteristics in the ore property information, identifying each process step of crushing, screening, grinding, classification, beneficiation, concentration, filtration, and tailings water recovery item by item, determining the process objectives corresponding to each process step as product particle size control objectives, separation effect objectives, concentration control objectives, and water recovery utilization objectives, determining the process constraints corresponding to each process step as processing capacity constraints, particle size boundary constraints, concentration range constraints, and water recovery boundary constraints, and merging and organizing them according to the process steps to form the target constraint set corresponding to each process step.

[0045] Based on the target constraint set, cross-stage propagation and consistent fusion are performed along the process sequence of crushing, screening, grinding, classification, beneficiation, concentration, filtration and tailings water recycling to form a process constraint vector.

[0046] The specific process includes, based on the target constraint set, sequentially transmitting the target constraint information of each process step (crushing, screening, grinding, classification, beneficiation, concentration, filtration, and tailings reclaimed water) according to the process sequence. The processing capacity constraints, particle size boundary constraints, concentration range constraints, and reclaimed water boundary constraints in the previous process step are transmitted to the next process step. The target constraint information of the next process step is compared, corrected, and fused to ensure consistency, so that the target constraint information of each process step (crushing, screening, grinding, classification, beneficiation, concentration, filtration, and tailings reclaimed water) forms a continuous connection relationship, forming a process constraint vector.

[0047] It should be noted that the processing capacity constraint refers to the range of material processing scale that each process step is allowed to handle; the particle size boundary constraint refers to the range of input and output particle sizes that each process step is allowed to meet; the concentration range constraint refers to the range of slurry concentration or solid-liquid ratio that each process step is allowed to meet; and the reclaimed water boundary constraint refers to the limited conditions that each process step is allowed to meet regarding reclaimed water volume, reclaimed water method, and reclaimed water utilization conditions.

[0048] Based on the process constraint vector, the constraint information corresponding to each process step is broken down into step-level search targets.

[0049] The specific process includes, based on the constraint status information of each process step (crushing, screening, grinding, classification, beneficiation, concentration, filtration, and tailings reclaimed water) in the process constraint vector, breaking down the constraints of processing capacity, particle size boundary, concentration range, reclaimed water boundary, and site adaptability into each step. The overall constraint content mixed in the process constraint vector is restored into independent constraint entries corresponding to each process step. The independent constraint entries are then merged and organized according to the process step category, so that each process step (crushing, screening, grinding, classification, beneficiation, concentration, filtration, and tailings reclaimed water) forms a step-level retrieval target that can be directly used for subsequent process gene fragment retrieval and matching.

[0050] The process-level search targets are matched with the process gene fragment description information in the process gene fragment library, and then selected to form the process-level preferred fragment sequences.

[0051] The specific process includes comparing the processing capacity constraints, particle size boundary constraints, concentration range constraints, reclaimed water boundary constraints, and site adaptation constraints in the stage-level search targets with the process gene fragment description information in the process gene fragment library item by item. Adaptability matching is performed on the process gene fragment description information corresponding to each process stage, such as crushing, screening, grinding, classification, beneficiation, concentration, filtration, and tailings reclaimed water. Process gene fragments that meet the constraints are retained as candidates, while process gene fragments that do not meet the process boundary conditions are screened out. The retained process gene fragments are then sorted according to the degree of matching to form the stage-level preferred fragment sequence corresponding to each process stage.

[0052] It should be noted that the process gene fragment library is formed by breaking down and classifying the crushing, screening, grinding, grading, beneficiation, concentration, filtration and tailings water recycling process units in historical mining projects, and recording the applicable conditions, processing capacity, interface characteristics, site occupation requirements and tailings water recycling boundaries; process boundary conditions refer to the limiting conditions that each process link is allowed to meet in terms of processing capacity, input particle size, output particle size, slurry concentration, water recycling requirements and site adaptability.

[0053] Based on the optimal sequence of process segments at the stage level, and combined with the input-output interface adaptation relationship between adjacent process stages, a pre-compatibility screening is performed to form a set of candidate process segments.

[0054] The specific process includes, based on the sorting of process gene fragments corresponding to each process step in the optimized fragment sequence at the stage level (crushing, screening, grinding, classification, beneficiation, concentration, filtration, and tailings reclaiming), comparing the input-output interface adaptation relationship between adjacent process steps item by item, checking whether the output particle size, output flow rate, output slurry concentration, and output reclaiming conditions of the process gene fragment of the previous process step can match the input requirements of the process gene fragment of the next process step, retaining process gene fragments that can maintain the connection between the previous and subsequent steps, and removing process gene fragments with interface mismatches, thus forming a set of candidate process fragments.

[0055] S3. Based on the set of candidate process segments, automatically assemble them according to the upstream and downstream interface rules, and simultaneously expand the process parameters to generate a parameterized candidate process flow chain set.

[0056] The candidate process fragment set is used to reconstruct each candidate process fragment into a typed process fragment with input ports, output ports, status ports and resource ports respectively, forming a typed process fragment ledger.

[0057] The specific process includes reorganizing the process gene fragment description information, input / output interface information, process status information, and resource occupancy information using each candidate process fragment in the candidate process fragment set. The information on receiving upstream materials, slurry, and return water conditions is assigned to the input port; the information on output particle size, flow rate, slurry concentration, and return water conditions is assigned to the output port; the information on process link category, operating stage, and constraint status is assigned to the status port; and the information on site occupancy, auxiliary needs, and resource consumption is assigned to the resource port. All of these are recorded uniformly according to the process sequence, process link category, and fragment number to form a categorized process fragment ledger.

[0058] Based on the typified process segment ledger, a hierarchical assembly diagram is constructed according to the process sequence, and forward expansion is performed according to the input and output interface rules between adjacent process links to form a partial process flow chain.

[0059] The specific process includes, based on the process sequence information, input port information, output port information, status port information, and resource port information in the typified process segment ledger, typified process segments corresponding to each process link of crushing, screening, grinding, classification, beneficiation, concentration, filtration, and tailings water recycling are arranged in layers to form a hierarchical assembly diagram unfolded according to the process sequence. Then, according to the interface rules of the input and output ports between adjacent process links, the output content of the previous process link and the input requirements of the next process link are connected item by item and extended forward to form a partial process flow chain.

[0060] Based on a partial process flow chain, the process parameter vectors of each process segment are calculated, and the process parameters are used to filter out process flow chains that can be further expanded.

[0061] The specific process includes, based on the sequence of crushing, screening, grinding, classification, beneficiation, concentration, filtration, and tailings reclaiming processes already connected in a partial process flow chain, reading the particle size, flow rate, slurry concentration, reclaiming conditions, and resource usage information output from the previous process segment, and calculating the corresponding input ports, status ports, and resource port requirements of the next process segment to form a process parameter vector for each process segment. The process parameter vector is then compared with the input / output interface conditions, processing capacity conditions, site usage conditions, and tailings reclaiming boundaries of the process segment to be connected, retaining the process flow chains that meet the requirements for continued connection, thus forming a process flow chain that can be further expanded.

[0062] It should be noted that the continuity requirement refers to the requirement that the input particle size, input flow rate, slurry concentration, return water conditions, processing capacity, and site occupancy meet the connection requirements of the previous process segment.

[0063] The expression for calculating the process parameter vector for each process segment is: ; in, Indicates the first A vector of process parameters for each process segment. Indicates the index of a process segment. Indicates the first The output flow rate of each process segment Indicates the first Characteristic granularity parameters of each process segment Indicates the first Slurry concentration parameters for each process segment Indicates the first Recovery and separation efficiency parameters for each process segment Indicates the first The output flow rate of each process segment Indicates the first The loss coefficient of each process segment Indicates the first Additional flow rate for each process segment No. Characteristic granularity parameters of each process segment This represents the natural exponential function. Indicates the first The particle size variation coefficient of each process segment Indicates the first The duration of each process segment Indicates the first The corresponding ore hardness for each process segment Indicates the first Solid mass fraction coefficient of each process segment Indicates the first The initial water volume of each process segment Indicates the first Water replenishment amount for each process segment Indicates the first The amount of water removed in each process segment Indicates the first The separation intensity coefficient of each process segment Indicates the first The useful component action coefficient of each process segment Indicates the first Interference correction coefficient for each process segment.

[0064] Furthermore, before calculating the process parameter vectors for each process segment, a unified unit conversion and dimensionless processing method was used to achieve dimensional unification. Specifically, output flow rate, replenishment flow rate, and water volume-related parameters were uniformly converted to the same flow rate unit or mass flow rate unit; characteristic particle size parameters were uniformly converted to the same length unit; action time was uniformly converted to the same time unit; slurry concentration parameters and recovery separation efficiency parameters were uniformly processed into dimensionless proportional quantities; and loss coefficient, particle size variation coefficient, solid mass fraction coefficient, useful component action coefficient, interference correction coefficient, and ore hardness parameters were uniformly processed into dimensionless coefficients. This ensures that the addition and subtraction terms, fractional terms, and exponential terms in the expression have a consistent dimensional basis when entering the calculation.

[0065] It should be noted that, It is obtained by statistically analyzing the material loss ratio of similar process segments in historical mining projects, and then correcting it based on the ore properties, processing conditions and equipment operating parameters corresponding to the current process segment; It is obtained by statistically analyzing the particle size variation patterns of similar process segments under historical projects, and then calibrating them in conjunction with the crushability of the ore, the intensity of equipment operation, and the residence conditions corresponding to the current process segment. This was obtained by statistically analyzing the proportion of solid mass to total mass in slurry under historical projects for similar process segments and correcting for it with the current feed concentration. It was obtained by calibrating the separation efficiency, action time, and equipment operating intensity of similar process segments in historical projects; It is obtained by comprehensively converting the grade, distribution characteristics and beneficiation test results of the target useful components in the ore; The effect of interference factors such as gangue content, impurity composition, particle size distribution fluctuation and return water quality on separation effect was statistically corrected.

[0066] The process parameters, spatial resources, and results references are modified using an expandable process chain, and the chain continues to expand along the process sequence to the tailings water recycling stage, forming a set of parameterized candidate process chains.

[0067] The specific process includes using an expandable process flow chain to correct the process parameters, spatial resources, and result reference associations corresponding to the process segments that have formed connections item by item. The particle size, flow rate, slurry concentration, and reclaimed water conditions output by the previous process segment are compared again with the input requirements of the next process segment. The site occupancy, auxiliary needs, and result anchor reference paths are adjusted synchronously. Based on the corrected connection relationship, the process continues to expand along the process sequence of crushing, screening, grinding, classification, beneficiation, concentration, filtration, and tailings reclaimed water to form a set of parameterized candidate process flow chains.

[0068] like Figure 5 The graph illustrates the relationship between the site constraint comprehensive coefficient and the process chain assembly success rate. The horizontal axis represents the site constraint comprehensive coefficient, and the vertical axis represents the process chain assembly success rate. The three curves correspond to control group A, control group B, and the experimental group, respectively. Control group A represents the scheme that only performs link-level matching without strictly implementing input / output interface adaptation and process parameter vector screening. Control group B represents the scheme that performs basic interface adaptation and pre-compatibility screening but does not form a complete typified process segment ledger and deep local repair mechanism. The experimental group represents the scheme that adopts the complete method of this invention. As the site constraint comprehensive coefficient increases, the process chain assembly success rate of all three groups improves. However, the curve of the experimental group is generally higher than that of control groups A and B. This indicates that by reconstructing candidate process segments into typified process segments and automatically assembling them according to upstream and downstream interface rules, this invention can more stably form a set of parameterized candidate process chains under different site constraints. This makes it easier for candidate process chains to directly participate in subsequent equipment mapping, spatial constraint verification, and result generation.

[0069] S4. Use the parameterized candidate process flow chain set to map the process unit and associated objects into heterogeneous graph nodes, and establish process associations, spatial associations, constraint associations and result reference associations between nodes to generate a point-surface fused heterogeneous graph.

[0070] The parameterized candidate process flow chain set is used to decompose and classify the process units, equipment objects, building objects, transportation objects, auxiliary objects and result anchors in each process flow chain to form a source object set.

[0071] The specific process includes using a parameterized candidate process flow chain set, and reading each process unit, equipment object, building object, transportation object, auxiliary object, and result anchor point in each process flow chain according to the determined process sequence, connection relationship, and parameter record. Then, based on the object's category, functional attributes, and corresponding position in the process flow chain, the process unit is classified into a process object, and the equipment object, building object, transportation object, auxiliary object, and result anchor point are classified into their respective object categories. Finally, the objects are uniformly organized according to the process flow chain number and object category number to form a source object set.

[0072] Based on the source object set, the corresponding equipment objects, building objects, transportation objects, auxiliary objects and result anchors are gathered around each process unit, and the correspondence between objects is established to obtain the node projection list.

[0073] The specific process includes, based on the source object set, taking each process unit as the center, correspondingly aggregating the equipment objects, building objects, transportation objects, auxiliary objects, and result anchors in the source object set that have process connection, spatial bearing, transportation connection, public auxiliary support, and result reference relationships with each process unit, and establishing a one-to-one correspondence between objects according to the connection order in the process flow chain, object category affiliation, parameter association position, and result reference position, and listing each process unit and its corresponding equipment objects, building objects, transportation objects, public auxiliary objects, and result anchors in a unified numbering method to obtain the node projection list.

[0074] Based on the node projection list, process units and associated objects are written into the point-surface fusion heterogeneous graph, and a mapping relationship between the source object and the nodes of the heterogeneous graph is established to form a heterogeneous graph node set.

[0075] The specific process includes assigning node categories, node numbers, and node attribute information to the process units and related objects in the point-to-surface fusion heterogeneous diagram based on the process units, equipment objects, building objects, transportation objects, auxiliary objects, and result anchor points listed in the node projection list. Then, the process units and related objects are written into the heterogeneous diagram item by item, so that the process units and related objects form corresponding nodes in the point-to-surface fusion heterogeneous diagram. At the same time, a one-to-one mapping relationship between the source objects and the nodes of the point-to-surface fusion heterogeneous diagram is established according to the source records in the node projection list. The nodes are then uniformly organized according to node categories and node numbers to form a heterogeneous diagram node set.

[0076] Based on the heterogeneous graph node set, corresponding equipment nodes, building nodes, transportation nodes, auxiliary nodes, standard nodes, and result nodes are gathered around each process node to form candidate related units.

[0077] The specific process includes, based on the process nodes, equipment nodes, building nodes, transportation nodes, auxiliary nodes, standard nodes, and result nodes already written in the heterogeneous graph node set, taking each process node as the center, and according to the connection relationship, spatial location relationship, transportation connection relationship, auxiliary support relationship, standard constraint relationship, and result reference relationship in the process flow chain, the corresponding equipment nodes, building nodes, transportation nodes, auxiliary nodes, standard nodes, and result nodes are aggregated and organized, and all nodes associated with the same process node are merged into a candidate association unit to form a candidate association unit.

[0078] Based on candidate associated units, process associations, spatial associations, constraint associations, and result reference associations between nodes are identified according to process sequence, spatial location, constraint scope, and result anchor reference path, forming an effective association set.

[0079] The specific process includes comparing the connection order, location relationship, constraint coverage relationship, and result reference link between nodes according to the process nodes, equipment nodes, building nodes, transportation nodes, auxiliary nodes, specification nodes, and result nodes contained in the candidate associated units. Nodes that conform to the process flow sequence are identified as process associations, nodes that conform to the spatial carrying capacity and location correspondence are identified as spatial associations, nodes that are within the same constraint range are identified as constraint associations, and nodes that have a result anchor reference path correspondence relationship are identified as result reference associations. The identified association relationships are then merged and organized to form a valid association set.

[0080] It should be noted that process association refers to the relationship between different process nodes based on material flow, processing sequence, and parameter transmission. For example, when the output slurry from the grinding node enters the classification node, a process association is formed between the grinding node and the classification node. Spatial association refers to the relationship between process nodes and equipment nodes, building nodes, transportation nodes, or auxiliary nodes based on spatial placement, installation load, and location correspondence. For example, when flotation equipment is arranged in a flotation workshop, a spatial association is formed between the flotation process node and the flotation workshop building node. Constraint association refers to the restrictive relationship formed by specification nodes or constraints on process nodes, equipment nodes, building nodes, and transportation nodes. For example, when the underflow concentration requirement corresponding to the thickening node is constrained by relevant specification clauses, a constraint association is formed between the thickening node and the specification node. Result reference association refers to the reference relationship formed between process nodes, equipment nodes, building nodes, or specification nodes and quantity flow diagrams, equipment tables, building connection diagrams, and text chapters. For example, when the processing capacity parameter corresponding to the grinding node is written into the equipment table, a result reference association is formed between the grinding node and the equipment table.

[0081] By utilizing the effective association set, process associations, spatial associations, constraint associations, and result reference associations are written into the heterogeneous graph node set to generate a point-surface fused heterogeneous graph.

[0082] The specific process includes using the process associations, spatial associations, constraint associations, and result reference associations already identified in the effective association set to connect the process nodes, equipment nodes, building nodes, transportation nodes, auxiliary nodes, specification nodes, and result nodes in the heterogeneous graph node set item by item. Process associations are written into the process connection relationships between corresponding nodes, spatial associations are written into the spatial connection relationships between corresponding nodes, constraint associations are written into the constraint connection relationships between corresponding nodes, and result reference associations are written into the result reference relationships between corresponding nodes. The nodes are then organized according to a unified relationship type to generate a point-surface fused heterogeneous graph.

[0083] It should be noted that the point-surface fusion heterogeneous diagram can uniformly express the various objects and their corresponding relationships in the mine design process, thereby improving the overall coordination and consistency of process analysis, conflict identification, and the coordinated generation of results.

[0084] S5. Utilize point-to-surface fusion heterogeneous graphs to perform state propagation and correlation analysis on various nodes, identify conflict points between processes, equipment, buildings, transportation, and standards, and perform local optimization on relevant process gene fragments and associated nodes to generate optimal point-to-surface fusion design graphs.

[0085] Based on the point-to-surface fusion heterogeneous graph, equipment nodes, building nodes, transportation nodes, auxiliary nodes, and specification nodes are gathered and associated around each process node to form a state propagation unit.

[0086] The specific process includes, based on the established process associations, spatial associations, constraint associations, and result reference associations in the point-surface fusion heterogeneous diagram, retrieving, verifying, and centrally aggregating the equipment nodes, building nodes, transportation nodes, auxiliary nodes, and standard nodes connected to each process node item by item, merging and organizing them according to process connection position, spatial bearing position, transportation connection position, auxiliary support position, and standard function position, and unifying the equipment nodes, building nodes, transportation nodes, auxiliary nodes, and standard nodes that can jointly represent the operating state, spatial state, constraint state, and association state of the same process node to form a state propagation unit.

[0087] Based on the state propagation unit, state propagation and association analysis are performed along process association, spatial association, constraint association and result reference association to form candidate conflict units.

[0088] The specific process includes, based on the existing process associations, spatial associations, constraint associations, and result reference associations among process nodes, equipment nodes, building nodes, transportation nodes, auxiliary nodes, and specification nodes in the state propagation unit, transmitting, comparing, and analyzing the process state, spatial state, constraint state, and reference state carried by each node along the association path, transmitting the state information of the previous node to the next node, and identifying the consistency, matching, and conflict of the transmitted state, screening out node combinations with mismatch risks or association anomalies to form candidate conflict units.

[0089] Candidate conflict units are used to identify the causes of mismatches between processes, equipment, buildings, transportation, and specifications, forming a set of conflict points.

[0090] The specific process includes, when using candidate conflict units, checking each of the process nodes, equipment nodes, building nodes, transportation nodes and specification nodes involved in the candidate conflict units, comparing the process status, equipment capacity, building load-bearing conditions, transportation connection conditions and specification constraints, identifying the source location, action location and influence location of the mismatch relationship, and merging and organizing abnormal nodes with the same mismatch source or the same influence range to form a set of conflict points.

[0091] Based on the set of conflict points, process gene fragments, equipment nodes, building nodes, transportation nodes, auxiliary nodes, standard nodes, and result nodes corresponding to each conflict point are extracted from the point-surface fusion heterogeneous graph to form local conflict slice units.

[0092] The specific process includes using a set of conflict points to conduct item-by-item retrieval, corresponding location, and scope division of process gene fragments, equipment nodes, building nodes, transportation nodes, auxiliary nodes, standard nodes, and result nodes that have process associations, spatial associations, constraint associations, and result reference associations with each conflict point in the point-surface fusion heterogeneous diagram. Process gene fragments, equipment nodes, building nodes, transportation nodes, auxiliary nodes, standard nodes, and result nodes within the influence range of the same conflict point are extracted in a centralized manner and merged and organized according to the conflict source relationship, action relationship, and influence relationship to form local conflict slice units.

[0093] The relevant process gene fragments are replaced using local conflict slicing units, and the associated nodes are redirected to form a local repair candidate set.

[0094] The specific process includes, when using local conflict slice units, comparing each extracted process gene fragment, equipment node, building node, transportation node, auxiliary node, specification node, and result node in the local conflict slice unit, replacing the process gene fragment directly corresponding to the source of conflict with a process gene fragment that meets the current process constraints and interface requirements, and redetermining the connection direction and corresponding position of the equipment node, building node, transportation node, auxiliary node, specification node, and result node that have a connection relationship with the replaced process gene fragment, so that the association relationship in the local conflict slice unit forms a new repair combination, forming a local repair candidate set.

[0095] Based on the local repair candidate set, the target local repair scheme is selected and written back to the point-surface fusion heterogeneous graph to generate the optimal point-surface fusion design graph.

[0096] The specific process includes comparing and comprehensively judging each candidate in the local repair candidate set based on the process gene fragment replacement relationship, equipment node connection relationship, building node correspondence relationship, transportation node connection relationship, public auxiliary node support relationship, standard node role relationship and result node reference relationship corresponding to different candidates. The target local repair scheme that can eliminate conflict points and maintain process continuity, spatial adaptability, standard compliance and result reference integrity is determined. Then, the process gene fragment and associated nodes corresponding to the target local repair scheme are updated and written into the point-surface fusion heterogeneous diagram to generate the optimal point-surface fusion design diagram.

[0097] It should be noted that the preferred point-to-surface integrated design drawing can maintain the consistency of process flow, spatial relationships, standard relationships, and result reference relationships while eliminating local conflicts, thereby improving the coordination, feasibility, and traceability of mine design results.

[0098] like Figure 6The scatter plot of the impact range of local repair and the residual conflict after repair is shown. The horizontal axis represents the proportion of the impact range of local repair, and the vertical axis represents the number of residual conflicts after repair. The three types of scatter plots correspond to control group A, control group B, and experimental group, respectively. It can be seen that the scatter plots of the experimental group are generally concentrated in the lower left area, indicating that a lower number of residual conflicts after repair can be obtained under the condition of a smaller proportion of the impact range of local repair. In contrast, control groups A and B generally require a larger adjustment range to achieve conflict reduction, and still retain a lot of residual conflicts after repair. Therefore, it can be intuitively explained that the present invention, by mapping process units and related objects to nodes of a point-to-surface fusion heterogeneous diagram and establishing the relationship between nodes, can complete local repair and form an optimal point-to-surface fusion design diagram without overturning the global design.

[0099] S6. Based on the optimized point-surface fusion design drawing, generate integrated design results of diagrams, tables, and text to obtain the point-surface fusion results of the mine design process.

[0100] Semantic information corresponding to process nodes, equipment nodes, building nodes, transportation nodes, auxiliary nodes, standard nodes, and result anchor points is extracted from the optimized point-surface integrated design drawings to form the result source ledger.

[0101] The specific process includes reading the name, parameter, connection, constraint, and reference information of each process node, equipment node, building node, transportation node, auxiliary node, specification node, and result anchor point from the optimized point-surface fusion design drawing. It also involves organizing the existing process associations, spatial associations, constraint associations, and result reference associations among process nodes, equipment nodes, building nodes, transportation nodes, auxiliary nodes, specification nodes, and result anchor points, and summarizing them into a result source ledger that can support the generation of subsequent drawing tasks, table tasks, and text tasks in a unified recording method.

[0102] By utilizing the source ledger of results, the main process line, engineering objects, standard basis, and result citation relationships are mapped into drawing tasks, table tasks, and text tasks, respectively, forming a multi-view compilation dependency network.

[0103] The specific process includes using the process mainline information, engineering object information, standard basis information, and result citation relationships recorded in the result source ledger to decompose the content corresponding to process nodes, equipment nodes, building nodes, transportation nodes, auxiliary nodes, standard nodes, and result anchor points into tasks. Content suitable for generating quantity flow diagrams, building connection diagrams, and water balance diagrams is mapped as drawing tasks; content suitable for generating design index tables, equipment tables, and material tables is mapped as table tasks; and content suitable for generating explanatory content is mapped as text tasks. A multi-view compilation dependency network is formed based on the sequential dependencies between process mainlines, engineering objects, standard basis, and result citation relationships.

[0104] Based on the multi-view compilation dependency network, a quantity flow diagram, a building connection diagram, and a water balance diagram are generated, and the corresponding information is written into a table task and a text task to obtain the point-to-surface fusion result of the mine design process.

[0105] The specific process includes, based on the dependencies between the drawing tasks, table tasks, and text tasks already established in the multi-view compilation dependency network, retrieving, organizing, and processing the process main line information, engineering object information, standard basis information, and result reference information item by item, generating quantity flow diagrams, building connection diagrams, and water balance diagrams, and writing the name information, parameter information, connection information, and reference information in the quantity flow diagrams, building connection diagrams, and water balance diagrams into the table tasks and text tasks, so that the drawing content is consistent and corresponds with the table content and text content, and the point-to-surface integration result of the mine design process is obtained.

[0106] In summary, this invention achieves the automatic construction of a continuous process flow chain from discrete process segments by: reconstructing candidate process segments into typified process segments and automatically assembling them according to upstream and downstream interface rules. This enables the candidate process flow chain to directly participate in subsequent equipment mapping, spatial constraint verification, computer-aided design, and result generation. By mapping process units and related objects into point-surface fusion heterogeneous graph nodes and establishing the relationships between nodes, this invention achieves unified relationship modeling between multiple types of objects in mine design. This allows for local repair and the formation of an optimal design drawing without overturning the overall design.

[0107] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An AI-based intelligent mine design process point-surface fusion method, characterized in that, include: Acquire multi-source data for mining projects and extract information on processing scale, ore properties, target indicators, tailings water recovery requirements, and site constraints to generate a standardized input set for the project. Based on the standardized input set of the project, AI is used to identify the process objectives and process constraints corresponding to each process step, generate process constraint vectors, and select suitable process gene fragments from the preset process gene fragment library to generate a candidate process fragment set. Based on the set of candidate process segments, they are automatically assembled according to the rules of upstream and downstream interfaces, and the process parameters are expanded simultaneously to generate a parameterized candidate process flow chain set. Using a parameterized candidate process flow chain set, process units and associated objects are mapped to heterogeneous graph nodes to form a heterogeneous graph node set. Process associations, spatial associations, constraint associations, and result reference associations are established between each node to generate a point-surface fused heterogeneous graph. By using point-surface fusion heterogeneous graphs to perform state propagation and correlation analysis on various nodes, conflict points between processes, equipment, buildings, transportation and standards are identified, a set of conflict points is formed, and local optimization is performed on relevant process gene fragments and associated nodes to generate a preferred point-surface fusion design graph. Based on the optimized point-surface integrated design drawing, the integrated design results of diagrams, tables and text are generated in a coordinated manner, resulting in the point-surface integrated results of the mine design process.

2. The AI-based intelligent mine design process point-surface fusion method of claim 1, wherein, The specific steps for generating the standardized input set for the project are as follows: Acquire multi-source data from mining projects, and classify, catalog, analyze, and unify the units of the multi-source data to generate a standardized analysis set; Using a standardized parsing set, field location, semantic extraction, and multi-source fusion are performed on processing scale, ore properties, target indicators, tailings water return requirements, and site constraints to generate a fusion parameter set; The site constraint comprehensive coefficient is calculated based on the fusion parameter set, and the processing scale, ore properties, target indicators, tailings water return requirements and site constraint information are uniformly encapsulated to generate a standardized input set for the project.

3. The AI-based intelligent mine design process point-surface fusion method of claim 1, wherein, The specific steps for generating the process constraint vector are as follows: The standardized input set of the project is used to map the processing scale, ore properties, target indicators, tailings water return requirements and site constraints into a unified feature vector. The unified feature vector is then input into the AI ​​recognition model and matched with the prototype of the preset process link to form the initial activation vector of each process link. Based on the initial activation vector, and combined with the target indicators and ore property information, the process objectives and process constraints corresponding to each process step are identified, forming a set of target constraints for each process step. Based on the target constraint set, cross-stage propagation and consistent fusion are performed along the process sequence of crushing, screening, grinding, classification, beneficiation, concentration, filtration and tailings water recycling to form a process constraint vector.

4. The AI-based intelligent mine design process point-surface fusion method of claim 1, wherein, The specific steps for generating the candidate process fragment set are as follows: Based on the process constraint vector, the constraint information corresponding to each process step is broken down into step-level search targets; The process-level search targets are matched with the process gene fragment description information in the process gene fragment library, and the process-level preferred fragment sequences are selected. Based on the optimal sequence of process segments at the stage level, and combined with the input-output interface adaptation relationship between adjacent process stages, a pre-compatibility screening is performed to form a set of candidate process segments.

5. The point-surface fusion method for AI-based intelligent mine design process as described in claim 1 or 4, characterized in that, The specific steps for generating the parameterized candidate process flow chain set are as follows: The candidate process segments are reconstructed into typed process segments with input ports, output ports, status ports and resource ports respectively using the candidate process segment set, forming a typed process segment ledger; Based on the typified process segment ledger, a hierarchical assembly diagram is constructed according to the process sequence, and forward expansion is performed according to the input and output interface rules between adjacent process links to form a partial process flow chain. Based on a partial process flow chain, calculate the process parameter vector for each process segment, and use the process parameters to filter out process flow chains that can be further expanded. The process parameters, spatial resources, and results references are modified using an expandable process chain, and the chain continues to expand along the process sequence to the tailings water recycling stage, forming a set of parameterized candidate process chains.

6. The point-surface fusion method for AI-based intelligent mine design process as described in claim 1, characterized in that, The specific steps for forming a heterogeneous graph node set are as follows: The parameterized candidate process flow chain set is used to decompose and classify the process units, equipment objects, building objects, transportation objects, auxiliary objects and result anchors in each process flow chain to form a source object set; Based on the source object set, the corresponding equipment objects, building objects, transportation objects, auxiliary objects and result anchors are gathered around each process unit, and the correspondence between objects is established to obtain the node projection list; Based on the node projection list, process units and associated objects are written into the point-surface fusion heterogeneous graph, and a mapping relationship between the source object and the nodes of the heterogeneous graph is established to form a heterogeneous graph node set.

7. The point-surface fusion method for AI-based intelligent mine design process as described in claim 1, characterized in that, The specific steps for generating the point-surface fused heterogeneous graph are as follows: Based on the heterogeneous graph node set, corresponding equipment nodes, building nodes, transportation nodes, auxiliary nodes, standard nodes and result nodes are gathered around each process node to form candidate related units; Based on candidate association units, process associations, spatial associations, constraint associations, and result reference associations between nodes are identified according to process sequence, spatial location, constraint scope, and result anchor reference path, forming an effective association set; By utilizing the effective association set, process associations, spatial associations, constraint associations, and result reference associations are written into the heterogeneous graph node set to generate a point-surface fused heterogeneous graph.

8. The point-surface fusion method for AI-based intelligent mine design process as described in claim 1 or 7, characterized in that, The specific steps for forming the set of conflict points are as follows: Based on the point-to-surface fusion heterogeneous graph, equipment nodes, building nodes, transportation nodes, auxiliary nodes and specification nodes are gathered and associated around each process node to form a state propagation unit; Based on the state propagation unit, state propagation and correlation analysis are performed along process correlation, spatial correlation, constraint correlation and result reference correlation to form candidate conflict units; Candidate conflict units are used to identify the causes of mismatches between processes, equipment, buildings, transportation, and specifications, forming a set of conflict points.

9. The point-surface fusion method for AI-based intelligent mine design process as described in claim 1, characterized in that, The specific steps for generating the optimal point-surface fusion design drawing are as follows: Based on the set of conflict points, process gene fragments, equipment nodes, building nodes, transportation nodes, auxiliary nodes, standard nodes and result nodes corresponding to each conflict point are extracted from the point-surface fusion heterogeneous graph to form local conflict slice units. The relevant process gene fragments are replaced using local conflict slicing units, and the associated nodes are redirected to form a local repair candidate set; Based on the local repair candidate set, the target local repair scheme is selected and written back to the point-surface fusion heterogeneous graph to generate the optimal point-surface fusion design graph.

10. The point-surface fusion method for AI-based intelligent mine design process as described in claim 1, characterized in that, The specific steps to obtain the point-to-surface fusion results of the mine design process are as follows: Semantic information corresponding to process nodes, equipment nodes, building nodes, transportation nodes, auxiliary nodes, standard nodes, and result anchor points is extracted from the optimized point-surface integrated design drawing to form the result source ledger; By utilizing the source ledger of results, the main process line, engineering objects, standard basis and result citation relationship are mapped into drawing tasks, table tasks and text tasks respectively, forming a multi-view compilation dependency network; Based on the multi-view compilation dependency network, a quantity flow diagram, a building connection diagram, and a water balance diagram are generated, and the corresponding information is written into a table task and a text task to obtain the point-to-surface fusion result of the mine design process.

Images