Assembly type floor heating wiring diagram automatic generation algorithm based on LORA large language model fine tuning

Abstract

The invention relates to the technical field of floor heating winding displacement, in particular to an assembly type floor heating winding displacement diagram automatic generation algorithm based on LORA large language model fine tuning, and the assembly type floor heating winding displacement diagram automatic generation algorithm comprises the steps: 1, inputting a building base map, and appointing a floor heating module laying area; 2, inputting floor heating parameters in an appointed floor heating module laying area of the building base map; 3, determining a plane bounding box range according to the laying area; 4, determining the maximum point, the minimum point and the span of the bounding box; 5, creating grids in a blocking manner; and 6, determining coordinates of four corners of each unit of the grid, and generating floor heating module unit occupation coordinates. According to the assembly type floor heating wiring diagram automatic generation algorithm based on LORA large language model fine tuning, the problems that floor heating construction drawings are time-consuming and high in error rate are solved, the efficiency and accuracy of floor heating system design are greatly improved, the labor cost and the error rate are reduced, and revolutionary changes are brought to design and construction of the assembly type floor heating system.

Description

Technical Field

[0001] The present invention relates to the technical field of floor heating wiring, and in particular to an automatic generation algorithm for assembled floor heating wiring diagrams based on LORA large language model fine-tuning. Background Art

[0002] The traditional process of generating wiring diagrams for prefabricated floor heating usually relies on manual design and calculation, which is not only time-consuming and labor-intensive, but also prone to errors. Designers need to manually measure the building base map, determine the laying area of ​​the floor heating module, and perform complex calculations based on the size of the floor heating module and the room layout to plan a reasonable layout of the floor heating module and the heating pipe path. During this process, designers need to frequently adjust and optimize the design plan to ensure the performance and efficiency of the floor heating system.

[0003] With the rise of building information modeling (BIM) technology and the rapid development of large language model algorithms (such as LoRA) in the field of artificial intelligence, new solutions have been provided for the automatic generation of prefabricated floor heating wiring diagrams. BIM technology can integrate various information of construction projects, including building base maps, floor heating module dimensions, heating pipe parameters, etc., providing strong data support for the design and optimization of floor heating systems. The LoRA algorithm can quickly adapt to the needs of specific tasks, such as searching and optimizing heating pipeline paths, by fine-tuning the pre-trained basic model. Therefore, it is necessary to design an automatic generation algorithm for prefabricated floor heating wiring diagrams based on fine-tuning of the LoRA large language model. Summary of the invention

[0004] The purpose of the present invention is to provide an automatic generation algorithm for prefabricated floor heating wiring diagrams based on LORA large language model fine-tuning, so as to solve the problem raised by the above background technology that the prefabricated floor heating wiring diagrams currently on the market rely on manual design and calculation, which is not only time-consuming and labor-intensive, but also prone to errors.

[0005] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: an automatic generation algorithm of prefabricated floor heating wiring diagram based on LORA large language model fine-tuning, the steps of the automatic generation algorithm of prefabricated floor heating wiring diagram include: 1. Input the building base map and specify the floor heating module laying area; 2. Input the floor heating parameters in the specified floor heating module laying area of ​​the building base map; 3. Determine the plane bounding box range according to the laying area; 4. Determine the maximum point and minimum point and span of the bounding box; 5. Create a grid in blocks; 6. Determine the coordinates of the four corners of each unit of the grid, and generate the floor heating module unit placeholder coordinates; 7. Import the floor heating module family model corresponding to the floor heating parameters into the BIM software; 7. Define the feasible laying path of the heating pipes on the floor heating module family; 8. According to the placeholder coordinates, use BIM software to fill the floor heating module family model into the laying area to complete the plane laying of the floor heating module and the creation of the three-dimensional model; 9. Introduce low-rank matrices A and B; 10. Import the pre-trained large language model as the basis; 11. Keep the LoRA low-rank matrix for training; 12. Train the LoRA layer, and use the back propagation algorithm to determine the A and B parameters; 13. The search path is set to an inaccessible path; 14. All heating pipe loop path planning is completed; 15. The heating pipe loops between the room and the living room are connected; 16. Optimize the BIM model; 17. Export the two-dimensional construction plan.

[0006] Preferably, the floor heating parameters in step 2 include: size parameters of the assembled floor heating module, starting and ending points of the heating pipes in each room, material, diameter and spacing of the heating pipes, etc.

[0007] Preferably, step 9 introduces two low-rank matrices A and B based on the starting point and end point position parameters of the heating pipes in each room and the living room input in step 2 as parameters that need to be fine-tuned in the basic model.

[0008] Preferably, the starting point and the end point of the heating pipe specifically include the following two aspects:

[0009] 1) The supply and return connection points where the supply and return pipes of each room and the living room are connected;

[0010] 2) The supply and return connection points where the living room heating pipes are connected to the manifold.

[0011] Preferably, the laying area in step 3 uses an AABB bounding box algorithm to determine the range of a plane bounding box.

[0012] Preferably, step 4 finds the maximum and minimum points of the bounding box and calculates their distance spans in the X and Y directions, and divides the bounding box plane into blocks according to the obtained distance span and floor heating module size, so that step 5 can divide X and Y into blocks to create grids respectively.

[0013] Preferably, in step 6, the values ​​obtained by segmentation in step 5 are used as X and Y input values ​​to obtain the four-point coordinates of each unit of the grid.

[0014] Preferably, the floor heating module family model in step 7 is automatically filled with the assistance of a secondary development plug-in based on BIM software.

[0015] Preferably, the large language model imported in step 10 is a Wenxin large model or an iFlytek large model, etc. By loading the pre-trained basic model and freezing most of its parameters, step 11 only retains the low-rank matrix part used for LoRA for training.

[0016] Preferably, step 12 sets the hyperparameters in the training process, uses the training data to train the LoRA layer, updates the parameters of the low-rank matrices A and B through the back propagation algorithm, and fixes the parameters of the basic model to ensure that they remain unchanged throughout the training process, thereby realizing the heating pipeline path search.

[0017] Compared with the prior art, the present invention has the following beneficial effects:

[0018] 1. The automatic generation algorithm of prefabricated floor heating wiring diagrams based on the fine-tuning of the LORA large language model solves the problem of time-consuming and high error rate in the production of floor heating construction drawings. It not only greatly improves the efficiency and accuracy of floor heating system design, but also reduces labor costs and error rates, bringing revolutionary changes to the design and construction of prefabricated floor heating systems.

[0019] 2. The automatic generation algorithm of prefabricated floor heating wiring diagram based on LORA large language model fine-tuning utilizes the building information and floor heating parameters provided by BIM technology, combined with the powerful capabilities of the LoRA large language model algorithm, to achieve automatic planning of floor heating module layout and automatic search and optimization of heating pipeline paths. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A flowchart of an algorithm for automatically generating a wiring diagram for assembled floor heating based on a fine-tuning of a LORA large language model;

[0021] Figure 2 The prefabricated floor heating module laying area in the architectural drawing of the present invention is based on the automatic generation algorithm of the prefabricated floor heating wiring diagram fine-tuned by the LORA large language model;

[0022] Figure 3 It is a bounding box plane division grid diagram of an automatic generation algorithm of a prefabricated floor heating wiring diagram based on LORA large language model fine-tuning of the present invention;

[0023] Figure 4The present invention discloses a prefabricated floor heating heating pipe loop path planning diagram based on an automatic generation algorithm of a prefabricated floor heating wiring diagram fine-tuned based on the LORA large language model. DETAILED DESCRIPTION

[0024] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] See also Figure 1-4 The present invention provides a technical solution: an automatic generation algorithm of assembled floor heating wiring diagram based on LORA large language model fine-tuning, the steps of the automatic generation algorithm of assembled floor heating wiring diagram include: 1. Input the building base map and specify the floor heating module laying area; 2. Input the floor heating parameters in the specified floor heating module laying area of ​​the building base map; 3. Determine the plane bounding box range according to the laying area; 4. Determine the maximum point and minimum point and span of the bounding box; 5. Create a grid by block; 6. Determine the coordinates of the four corners of each unit of the grid, and generate the floor heating module unit placeholder coordinates; 7. Import the floor heating module family model corresponding to the floor heating parameters into the BIM software, and define the floor heating Feasible laying paths for heating pipes on the module family; 8. According to the placeholder coordinates, use BIM software to fill the floor heating module family model into the laying area to complete the plane laying of the floor heating module and the creation of a three-dimensional model; 9. Introduce low-rank matrices A and B; 10. Import the pre-trained large language model as the basis; 11. Keep the LoRA low-rank matrix for training; 12. Train the LoRA layer, and use the back propagation algorithm to determine the A and B parameters; 13. The search path is set to an inaccessible path; 14. All heating pipe loop path planning is completed; 15. The heating pipe loops between the room and the living room are connected; 16. Optimize the BIM model; 17. Export the two-dimensional construction plan.

[0026] The floor heating parameters in step 2 include: the size parameters of the prefabricated floor heating module, the starting and ending points of the heating pipes in each room, the material, diameter and spacing of the heating pipes, etc. This step provides a reliable reference for the prefabricated floor heating wiring diagram, so as to generate a high-precision prefabricated floor heating wiring diagram.

[0027] Step 9 introduces two low-rank matrices A and B as parameters that need to be fine-tuned in the basic model based on the starting and ending position parameters of the heating pipes in each room and the living room entered in step 2. In this step, a low-rank matrix refers to a matrix whose rank is less than the smaller value of its number of rows and columns. The rank of a matrix refers to the maximum number of its linearly independent rows or columns, which can effectively represent the structure of the data.

[0028] The starting point and end point of the heating pipe specifically include the following two aspects:

[0029] 1) The supply and return connection points where the supply and return pipes of each room and the living room are connected;

[0030] 2) The supply and return connection points where the living room heating pipes are connected to the manifold.

[0031] The laying area in step 3 uses the AABB (Axis-Aligned Bounding Box) bounding box algorithm to determine the plane bounding box range. The AABB (Axis-Aligned Bounding Box) bounding box algorithm in this step is a technology commonly used in computer graphics and collision detection. AABB refers to a bounding box aligned with the coordinate axis. A rectangle or cube is usually used to represent the boundary of an object. The main feature of this bounding box is that its edges are parallel to the coordinate axis.

[0032] Step 4 finds the maximum and minimum points of the bounding box and calculates their distance spans in the X and Y directions, and divides the bounding box plane into blocks using the distance span and floor heating module size, so that step 5 can divide the X and Y blocks to create grids respectively.

[0033] Step 6 uses the values ​​obtained by segmentation in step 5 as X and Y input values ​​to obtain the four-point coordinates of each cell in the grid.

[0034] The floor heating module family model in step 7 is automatically filled with the secondary development plug-in based on BIM software to assist in implementation.

[0035] The large language model imported in step 10 is the Wenxin large model or the iFlytek large model. By loading the pre-trained basic model and freezing most of its parameters, step 11 only retains the low-rank matrix part for LoRA for training. This step uses large language models such as the Wenxin large model or the iFlytek large model to pre-train with massive text data, learn the statistical characteristics and contextual relationships of the language, and fine-tune on specific tasks to improve the performance of the model in specific application areas.

[0036] Step 12 sets the hyperparameters in the training process, uses the training data to train the LoRA layer, updates the parameters of the low-rank matrices A and B through the back-propagation algorithm, and fixes the parameters of the basic model to ensure that they remain unchanged throughout the training process, thereby realizing the heating pipeline path search.

[0037] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An automatic generation algorithm for assembled floor heating wiring diagram based on LORA large language model fine-tuning, characterized by: The steps of the automatic generation algorithm of the assembled floor heating wiring diagram include:

1. Input the building base map and specify the floor heating module laying area; 2. Input the floor heating parameters in the specified floor heating module laying area of ​​the building base map; 3. Determine the plane bounding box range according to the laying area; 4. Determine the maximum point and minimum point and span of the bounding box; 5. Create a grid in blocks; 6. Determine the coordinates of the four corners of each unit of the grid, and generate the floor heating module unit placeholder coordinates; 7. Import the floor heating module family model corresponding to the floor heating parameters into the BIM software, and define the feasible laying path of the heating pipe on the floor heating module family; 8. Determine the layout of the floor heating module family according to the placeholder coordinates.

10. Use BIM software to fill the floor heating module family model into the laying area to complete the plane laying of the floor heating module and the creation of the three-dimensional model; 11. Introduce low-rank matrices A and B; 12. Import the pre-trained large language model as the basis; 13. Keep the LoRA low-rank matrix for training; 14. Train the LoRA layer, and use the back propagation algorithm to determine the A and B parameters; 15. The search path is set to an inaccessible path; 16. All heating pipe loop path planning is completed; 17. The heating pipe loops of the room and the living room are connected; 18. Optimize the BIM model; 19. Export the two-dimensional construction plan.

2. According to claim 1, an automatic generation algorithm of assembled floor heating wiring diagram based on LORA large language model fine-tuning is characterized by: The floor heating parameters in step 2 include: size parameters of the assembled floor heating module, starting and ending points of the heating pipes in each room, material, diameter and spacing of the heating pipes, etc.

3. According to claim 2, an automatic generation algorithm of assembled floor heating wiring diagram based on LORA large language model fine-tuning is characterized by: The step 9 introduces two low-rank matrices A and B as parameters that need to be fine-tuned in the basic model based on the starting point and end point position parameters of the heating pipes in each room and the living room input in step 2.

4. According to claim 3, an automatic generation algorithm of assembled floor heating wiring diagram based on LORA large language model fine-tuning is characterized by: The starting point and end point of the heating pipe specifically include the following two aspects: 1) The supply and return connection points where the supply and return pipes of each room and the living room are connected; 2) The supply and return connection points where the living room heating pipes are connected to the manifold.

5. According to claim 1, an automatic generation algorithm of assembled floor heating wiring diagram based on LORA large language model fine-tuning is characterized by: The laying area in step 3 uses the AABB bounding box algorithm to determine the plane bounding box range.

6. According to claim 1, an automatic generation algorithm of assembled floor heating wiring diagram based on LORA large language model fine-tuning is characterized by: The step 4 finds the maximum and minimum points of the bounding box and calculates their distance spans in the X and Y directions, and divides the bounding box plane into blocks using the obtained distance span and floor heating module size, so that step 5 can divide the X and Y blocks to create grids respectively.

7. The automatic generation algorithm of the assembled floor heating wiring diagram based on LORA large language model fine-tuning according to claim 6 is characterized by: The step 6 uses the values ​​obtained by segmentation in step 5 as X and Y input values ​​to obtain the four-point coordinates of each unit of the grid.

8. According to claim 1, an automatic generation algorithm of assembled floor heating wiring diagram based on LORA large language model fine-tuning is characterized by: The floor heating module family model in step 7 is automatically filled in with the assistance of a secondary development plug-in based on BIM software.

9. The automatic generation algorithm of the assembled floor heating wiring diagram based on LORA large language model fine-tuning according to claim 1 is characterized by: The large language model imported in step 10 is a Wenxin large model or an iFlytek large model, etc. By loading the pre-trained basic model and freezing most of its parameters, step 11 only retains the low-rank matrix part used for LoRA for training.

10. The automatic generation algorithm of the assembled floor heating wiring diagram based on LORA large language model fine-tuning according to claim 1 is characterized by: The step 12 sets the hyperparameters in the training process, uses the training data to train the LoRA layer, updates the parameters of the low-rank matrices A and B through the back propagation algorithm, and fixes the parameters of the basic model to ensure that they remain unchanged throughout the training process, thereby realizing the heating pipeline path search.

Images