Engineering investigation exploration point intelligent arrangement method and system based on AI dialogue interaction

Abstract

The invention relates to an engineering investigation exploration point intelligent arrangement method and system based on AI dialogue interaction, and the method comprises the steps that a user input module receives an investigation boundary polygon and building structure side lines described by voice / text, and uploads an investigation task book; a DeepSeek-R1 analyzer extracts boundary vertex coordinates and building and building side boundary coordinates through entity recognition; the document identification module analyzes the task book through a CNN-BiLSTM hybrid model, and outputs a reconnaissance type and a site complexity level; the dynamic interval generator calculates according to a standard library, executes according to a double-constraint optimization engine, displays a point distribution scheme according to a visual interaction interface and receives a dragging instruction of a user, the real-time compliance detection module verifies whether an adjustment point is met or not, and the output module generates a CAD point distribution diagram and a section line relation table. According to the method, a complete technical closed loop from intelligent analysis and dynamic planning to man-machine cooperation is constructed, and the industrial defects that a traditional method depends on experience, adjustment is tedious, and section lines are redundant are overcome.

Description

Technical Field

[0001] This application relates to the interdisciplinary field of artificial intelligence and geological engineering, and in particular to a method and system for intelligent layout of engineering exploration points based on AI dialogue interaction. Background Technology

[0002] The current layout of engineering survey points has three major technical defects:

[0003] High reliance on manual labor: Engineers need to manually consult the "Engineering Survey Specifications" to determine the spacing, which takes an average of 4.6 hours for complex sites;

[0004] Insufficient software intelligence: CAD-aided tools cannot dynamically adapt to changes in site complexity;

[0005] Redundant profile lines: Manual placement of points leads to a 30%-50% over-limit in the number of profile lines, increasing exploration costs. Summary of the Invention

[0006] The purpose of this application is to provide an intelligent layout method and system for engineering survey and exploration points based on AI dialogue interaction. It constructs a complete technical closed loop from intelligent analysis and dynamic planning to human-machine collaboration, solving industry pain points such as reliance on experience, cumbersome adjustments, and redundant profile lines in traditional methods.

[0007] To achieve the above objectives, this application provides the following technical solution:

[0008] In a first aspect, embodiments of this application provide a method for intelligent layout of engineering survey and exploration points based on AI dialogue interaction, comprising the following steps:

[0009] The user input module receives voice / text descriptions of the survey boundary polygons and building boundaries, and uploads the survey task book;

[0010] The DeepSeek-R1 parser extracts the coordinates of boundary vertices and the coordinates of building edges through entity recognition.

[0011] The document recognition module parses the task book using a CNN-BiLSTM hybrid model and outputs the survey type and site complexity level.

[0012] The dynamic spacing generator calculates based on the specification library:

[0013] Recommended spacing = max(5, base spacing × (1 + 0.1 × (complexity level - 1))),

[0014] Flexible interval = [5m, recommended spacing × 1.1];

[0015] Dual-constraint optimization engine execution:

[0016] a) Generate orthogonal meshes according to the recommended spacing.

[0017] b) Map boundary points to grid intersections.

[0018] c) Increase the density of data points in the building area.

[0019] d) Optimize the connection using the profile line minimization algorithm;

[0020] A visual interactive interface displays the site layout plan and receives drag-and-drop commands from the user.

[0021] The real-time compliance detection module verifies whether the adjustment points are met.

[0022] The moving distance should be ≤0.5 × recommended spacing.

[0023] New spacing ∈ [5m, recommended spacing × 1.1],

[0024] The new location is at the intersection of the grid lines;

[0025] The output module generates CAD layout diagrams and cross-sectional line relationship tables.

[0026] The DeepSeek-R1 parser extracts the coordinates of boundary vertices and building edges through entity recognition, specifically including...

[0027] (1) Serialization extraction of vertex coordinates of the boundary polygon.

[0028] Serialized coordinate expression format: The extracted coordinates adopt a normalized sequence format: [(x1,y1), (x2,y2), ..., (x n ,y n Ensure that the vertices are arranged in a clockwise or counterclockwise order to form a closed polygon;

[0029] Serialized coordinates are directly used as input data for the dual-constraint optimization engine;

[0030] (2) The topological relationship between the building outline and the project boundary is separated.

[0031] Identify and separate the spatial relationship between the building outline and the project boundary to ensure that exploration points maintain a safe distance from buildings.

[0032] The document recognition module parses the task book using a CNN-BiLSTM hybrid model, outputting the survey type and site complexity level. Specifically, it extracts image features from the task book, performs CNN convolution to obtain local feature vectors, and samples key visual features through an attention mechanism. Simultaneously, it uses OCR text and BiLSTM encoding to obtain semantic feature vectors. After fusing visual and semantic features across modalities, it processes them in parallel through a dual-branch classifier: the first branch outputs the survey type {detailed survey, preliminary survey, construction survey}, and the second branch outputs the complexity level {level one, level two, level three}. A joint loss function L = 0.6L_type + 0.4L_complex is used for simultaneous optimization to ensure the model accurately identifies survey parameters to drive subsequent intelligent site deployment decisions.

[0033] The profile line minimization algorithm includes:

[0034] Constructing an exploration point location matrix ;

[0035] Establish a candidate set of profile lines ;

[0036] Find the minimum covering subset ,satisfy: make and Connect,

[0037] The definitions of each character are as follows: : Exploration point location matrix, where: Matrix representation of the set of exploration point locations Number of exploration points Number of exploration points Any two exploration points in the matrix, , : The set of all possible profile lines connecting exploration points : No. A cross-sectional line, , This represents the total number of candidate profile lines.

[0038] , indicating the connection point and points line segments,

[0039] : Minimal covering subset, where: : a subset of

[0040] Subset The number of profile lines included.

[0041] :make The optimization objective is to obtain the minimum value.

[0042] Connectivity condition: ,

[0043] Universal quantifier, meaning "for all".

[0044] : Existential quantifier, indicating "existence"

[0045] For the position matrix Any two exploration points and There is at least one cross-sectional line in each. Make these two points pass through Direct or indirect connection.

[0046] The real-time compliance detection includes dynamic visual prompts:

[0047] Green marker: Simultaneously satisfies movement distance, spacing, and grid constraints;

[0048] Red flashing: The spacing exceeds the elastic range;

[0049] Yellow border: Position off-center from grid intersection.

[0050] Secondly, embodiments of this application provide an intelligent layout system for engineering survey and exploration points based on AI dialogue interaction, including:

[0051] User input module: Supports multimodal input interfaces for voice, text, and documents;

[0052] DeepSeek-R1 parser: Geographic coordinate extraction unit integrating NER model;

[0053] Document recognition module: includes ResNet-50 image channels and BiLSTM text channels;

[0054] Dynamic Spacing Generator: Stores spacing calculation units according to GB50021 standard;

[0055] Dual-constraint optimization engine: a parallel structure of mesh aligner and profile optimizer;

[0056] Visual interactive interface: 3D terrain rendering area, profile line network area, adjustment control area;

[0057] Real-time compliance detection module: movement distance monitor, spacing interval verifier, grid alignment detector;

[0058] Output module: CAD / GIS dual-format output unit.

[0059] The mesh aligner of the dual-constraint optimization engine performs the following:

[0060] grid_size = recommended spacing × rand(0.9, 1.1) / / Allow ±10% fluctuation.

[0061] The visual interactive interface includes:

[0062] Terrain rendering area: Overlaying DEM elevation model and exploration point heat map;

[0063] Adjust control area: Displays the current spacing's flexible range in real time;

[0064] History stack: Supports multi-step undo / redo.

[0065] Compared with existing technologies, the beneficial effects of this invention are as follows: It is the first to deeply integrate AI dialogue interaction with large language models into the field of engineering surveying, automatically parsing boundary information and task specifications through natural language commands, overturning the traditional manual input mode and improving efficiency by more than 5 times; It innovatively proposes flexible spacing intervals and dual-constraint optimization algorithms, minimizing the number of profile lines while ensuring compliance with standards, reducing exploration costs by more than 30%; It constructs a complete technical closed loop from intelligent analysis and dynamic planning to human-machine collaboration, solving industry pain points such as reliance on experience, cumbersome adjustments, and redundant profile lines in traditional methods. Attached Figure Description

[0067] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0068] Figure 1 This is a schematic diagram of the method flow of the present invention.

[0069] Figure 2 This is an application effect diagram of an embodiment of the present invention. Detailed Implementation

[0071] The technical solutions of the embodiments of this application will now be described with reference to the accompanying drawings. It should be noted that similar reference numerals and letters in the following drawings indicate similar items; therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0072] The terms “comprising,” “including,” or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase “comprising one…” does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0073] The terms “first,” “second,” etc., are used only to distinguish one entity or operation from another, and should not be construed as indicating or implying relative importance, nor as requiring or implying any such actual relationship or order between these entities or operations.

[0074] like Figure 1 and Figure 2 As shown, a method for intelligent layout of engineering survey and exploration points based on AI dialogue interaction includes the following steps:

[0075] The user input module receives voice / text descriptions of the survey boundary polygons and building boundaries, and uploads the survey task book;

[0076] The DeepSeek-R1 parser extracts the coordinates of boundary vertices and the coordinates of building edges through entity recognition.

[0077] The document recognition module parses the task book using a CNN-BiLSTM hybrid model and outputs the survey type and site complexity level.

[0078] The dynamic spacing generator calculates based on the specification library:

[0079] Recommended spacing = max(5, base spacing × (1 + 0.1 × (complexity level - 1))),

[0080] Flexible interval = [5m, recommended spacing × 1.1];

[0081] Dual-constraint optimization engine execution:

[0082] a) Generate orthogonal meshes according to the recommended spacing.

[0083] b) Map boundary points to grid intersections.

[0084] c) Increase the density of data points in the building area.

[0085] d) Optimize the connection using the profile line minimization algorithm;

[0086] A visual interactive interface displays the site layout plan and receives drag-and-drop commands from the user.

[0087] The real-time compliance detection module verifies whether the adjustment points are met.

[0088] The moving distance should be ≤0.5 × recommended spacing.

[0089] New spacing ∈ [5m, recommended spacing × 1.1],

[0090] The new location is at the intersection of the grid lines;

[0091] The output module generates CAD layout diagrams and cross-sectional line relationship tables.

[0092] The DeepSeek-R1 parser extracts the coordinates of boundary vertices and building edges through entity recognition, specifically including...

[0093] (1) Serialization extraction of vertex coordinates of the boundary polygon.

[0094] Serialized coordinate expression format: The extracted coordinates adopt a normalized sequence format: [(x1,y1), (x2,y2), ..., (x n ,y n Ensure that the vertices are arranged in a clockwise or counterclockwise order to form a closed polygon;

[0095] Serialized coordinates are directly used as input data for the dual-constraint optimization engine;

[0096] (2) The topological relationship between the building outline and the project boundary is separated.

[0097] Identify and separate the spatial relationship between the building outline and the project boundary to ensure that exploration points maintain a safe distance from buildings.

[0098] The document recognition module parses the task book using a CNN-BiLSTM hybrid model, outputting the survey type and site complexity level. Specifically, it extracts image features from the task book, performs CNN convolution to obtain local feature vectors, and samples key visual features through an attention mechanism. Simultaneously, it uses OCR text and BiLSTM encoding to obtain semantic feature vectors. After fusing visual and semantic features across modalities, it processes them in parallel through a dual-branch classifier: the first branch outputs the survey type {detailed survey, preliminary survey, construction survey}, and the second branch outputs the complexity level {level one, level two, level three}. A joint loss function L = 0.6L_type + 0.4L_complex is used for simultaneous optimization to ensure the model accurately identifies survey parameters to drive subsequent intelligent site deployment decisions.

[0099] The mathematical model of the CNN-BiLSTM hybrid model is expressed as follows:

[0100] Forward propagation formula:

[0101]

[0102] Loss function:

[0103]

[0104] The profile line minimization algorithm includes:

[0105] Constructing an exploration point location matrix ;

[0106] Establish a candidate set of profile lines ;

[0107] Find the minimum covering subset ,satisfy: make and Connect,

[0108] The definitions of each character are as follows: : Exploration point location matrix, where: Matrix representation of the set of exploration point locations Number of exploration points Number of exploration points Any two exploration points in the matrix, ,

[0109] : The set of all possible profile lines connecting exploration points

[0110] : No. A cross-sectional line, , This represents the total number of candidate profile lines.

[0111] , indicating the connection point and points line segments,

[0112] : Minimal covering subset, where: : a subset of

[0113] Subset The number of profile lines included.

[0114] :make The optimization objective is to obtain the minimum value.

[0115] Connectivity condition: ,

[0116] Universal quantifier, meaning "for all".

[0117] : Existential quantifier, indicating "existence"

[0118] For the position matrix Any two exploration points and There is at least one cross-sectional line in each. Make these two points pass through Direct or indirect connection.

[0119] The real-time compliance detection includes dynamic visual prompts:

[0120] Green marker: Simultaneously satisfies movement distance, spacing, and grid constraints;

[0121] Red flashing: The spacing exceeds the elastic range;

[0122] Yellow border: Position off-center from grid intersection.

[0123] This application provides an intelligent layout system for engineering survey and exploration points based on AI dialogue interaction, including:

[0124] User input module 101: Supports multimodal input interfaces for voice / text / document;

[0125] DeepSeek-R1 Parser 102: Geographic Coordinate Extraction Unit with Integrated NER Model;

[0126] Document recognition module 103: includes ResNet-50 image channels and BiLSTM text channels;

[0127] Dynamic Spacing Generator 104: Stores spacing calculation units according to GB50021 standards;

[0128] Dual-constraint optimization engine 105: Parallel structure of mesh aligner and profile optimizer;

[0129] Visual interactive interface 106: 3D terrain rendering area, profile line network area, adjustment control area;

[0130] Real-time compliance detection module 107: movement distance monitor, spacing interval verifier, grid alignment detector;

[0131] Output module 108: CAD / GIS dual-format output unit.

[0132] Example: A detailed exploration project

[0133] 1. Input Phase

[0134] Input boundaries: A(120,80)-B(320,80)-C(320,280)-D(120,280)......n(x,y)

[0135] Task description recognition result: {Detailed exploration, Complexity level 3}

[0136] 2. Decision-making stage

[0137] Basic spacing 15m → Recommended spacing = 15 × (1 + 0.1 × 2) = 18m

[0138] Elastic range = [5m, 19.8m]

[0139] 3. Optimization Phase

[0140] Generate an 18m floating mesh (±10%)

[0141] The containment area is reinforced to a spacing of 9m.

[0142] Profile line optimization: 28 lines → 16 lines

[0143] 4. Adjust the user drag point P7(216,152)→(220,150) for verification:

[0144] The moving distance was 4.47m < 0.5 × 18m

[0145] New spacing 17.2m ∈ [5,19.8]

[0146] The new location is the intersection of the grid lines → Compliance

[0147] 5. Output Results

[0148] CAD layout diagram (including encrypted area markings);

[0149] Profile line connection matrix;

[0150] Spacing compliance certificate;

[0151] Comparison of experimental data

[0152]

[0153] The above description is merely an embodiment of the present invention and is not intended to limit the scope of protection of the present invention. For those skilled in the art, the present invention can have various modifications and variations. 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 present invention.

Claims

1. A method for intelligent layout of engineering survey and exploration points based on AI dialogue interaction, characterized in that, Includes the following steps: The user input module receives voice / text descriptions of the survey boundary polygons and building boundaries, and uploads the survey task book; The DeepSeek-R1 parser extracts the coordinates of boundary vertices and the coordinates of building edges through entity recognition. The document recognition module parses the task book using a CNN-BiLSTM hybrid model and outputs the survey type and site complexity level. The dynamic spacing generator calculates based on the specification library: Recommended spacing = max(5, base spacing × (1 + 0.1 × (complexity level - 1))), Flexible interval = [5m, recommended spacing × 1.1]; Dual-constraint optimization engine execution: a) Generate orthogonal meshes according to the recommended spacing. b) Map boundary points to grid intersections. c) Increase the density of data points in the building area. d) Optimize the connection using the profile line minimization algorithm; A visual interactive interface displays the site layout plan and receives drag-and-drop commands from the user. The real-time compliance detection module verifies whether the adjustment points are met. The moving distance should be ≤0.5 × recommended spacing. New spacing ∈ [5m, recommended spacing × 1.1], The new location is at the intersection of the grid lines; The output module generates CAD layout diagrams and cross-sectional line relationship tables.

2. The intelligent layout method for engineering survey and exploration points based on AI dialogue interaction according to claim 1, characterized in that, The DeepSeek-R1 parser extracts the coordinates of boundary vertices and building edges through entity recognition, specifically including... (1) Serialization extraction of vertex coordinates of the boundary polygon. Serialized coordinate expression format: The extracted coordinates adopt a normalized sequence format: [(x1,y1), (x2,y2),..., (x n ,y n Ensure that the vertices are arranged in a clockwise or counterclockwise order to form a closed polygon; Serialized coordinates are directly used as input data for the dual-constraint optimization engine; (2) The topological relationship between the building outline and the project boundary is separated. Identify and separate the spatial relationship between the building outline and the project boundary to ensure that exploration points maintain a safe distance from buildings.

3. The intelligent layout method for engineering survey and exploration points based on AI dialogue interaction according to claim 1, characterized in that, The document recognition module parses the task book using a CNN-BiLSTM hybrid model, outputting the survey type and site complexity level. Specifically, it extracts image features from the task book, performs CNN convolution to obtain local feature vectors, and samples key visual features through an attention mechanism. Simultaneously, it uses OCR text and BiLSTM encoding to obtain semantic feature vectors. After fusing visual and semantic features across modalities, it processes them in parallel through a dual-branch classifier: the first branch outputs the survey type {detailed survey, preliminary survey, construction survey}, and the second branch outputs the complexity level {level one, level two, level three}. A joint loss function L = 0.6L_type + 0.4L_complex is used for simultaneous optimization to ensure the model accurately identifies survey parameters to drive subsequent intelligent site deployment decisions.

4. The method for intelligent layout of engineering survey and exploration points based on AI dialogue interaction according to claim 1, characterized in that, The profile line minimization algorithm includes: Constructing an exploration point location matrix ; Establish a candidate set of profile lines ; Find the minimum covering subset ,satisfy: make and Connect, The definitions of each character are as follows: : Exploration point location matrix, where: Matrix representation of the set of exploration point locations Number of exploration points Number of exploration points Any two exploration points in the matrix, , : The set of all possible profile lines connecting exploration points : No. A cross-sectional line, , This represents the total number of candidate profile lines. , indicating the connection point and points line segments, : Minimal covering subset, where: : a subset of Subset The number of profile lines included. :make The optimization objective is to obtain the minimum value. Connectivity condition: , Universal quantifier, meaning "for all". : Existential quantifier, indicating "existence". For the position matrix Any two exploration points and There is at least one cross-sectional line in each. Make these two points pass through Direct or indirect connection.

5. The intelligent layout method for engineering survey and exploration points based on AI dialogue interaction according to claim 1, characterized in that, The real-time compliance detection includes dynamic visual prompts: Green marker: Simultaneously satisfies movement distance, spacing, and grid constraints; Red flashing: The spacing exceeds the elastic range; Yellow border: Position off-center from grid intersection.

6. An intelligent layout system for engineering survey and exploration points based on AI dialogue interaction, characterized in that, include, User input module: Supports multimodal input interfaces for voice, text, and documents; DeepSeek-R1 parser: Geographic coordinate extraction unit integrating NER model; Document recognition module: includes ResNet-50 image channels and BiLSTM text channels; Dynamic Spacing Generator: Stores spacing calculation units according to GB50021 standard; Dual-constraint optimization engine: a parallel structure of mesh aligner and profile optimizer; Visual interactive interface: 3D terrain rendering area, profile line network area, adjustment control area; Real-time compliance detection module: movement distance monitor, spacing interval verifier, grid alignment detector; Output module: CAD / GIS dual-format output unit.

7. The intelligent layout system for engineering survey and exploration points based on AI dialogue interaction as described in claim 6, characterized in that, The mesh aligner of the dual-constraint optimization engine performs the following: grid_size = recommended spacing × rand(0.9, 1.1) / / Allow ±10% fluctuation.

8. The intelligent layout system for engineering survey and exploration points based on AI dialogue interaction as described in claim 6, characterized in that, The visual interactive interface includes: Terrain rendering area: Overlaying DEM elevation model and exploration point heat map; Adjust control area: Displays the current spacing's flexible range in real time; History stack: Supports multi-step undo / redo.

Images