Machine-readable constructional engineering standard digital processing method and device

Abstract

The embodiment of the invention discloses a machine-readable building engineering standard digital processing method and a machine-readable building engineering standard digital processing device. A specific embodiment of the method comprises the steps of generating text sequence information according to a first preset processing mode; generating layout information and bibliography and reference relation information according to preset layout annotation information, the layout level identification model and preset bibliography annotation information; text information is generated according to preset text labeling information and a paragraph extraction model; according to a pre-trained formula identification model, generating special format text information; determining the text sequence information, the layout information, the bibliography and reference relation information, the text information and the special format text information as construction standard control information; and generating control parameter information according to the construction standard control information, and controlling the construction robot to perform construction processing according to the control parameter information. According to the embodiment, the accuracy of obtaining the building engineering standard information is improved, the construction time consumption is shortened, and the waste of equipment resources is reduced.

Description

Technical Field

[0001] The embodiments disclosed herein relate to the field of building engineering standards and computer technology, specifically to a method and apparatus for digitizing building engineering standards in a machine-readable manner. Background Technology

[0002] The application of building engineering standards in the field of intelligent construction, such as residential construction robots, has high requirements for construction efficiency, accuracy, and safety. Based on building engineering standards, key construction parameters can be analyzed, and then the construction robot can be controlled according to these parameters. Current methods for controlling construction robots based on building engineering standards involve: first, parsing the unstructured building engineering standards using a PDF parsing tool; and then, controlling the construction robot's operation based on the parsed building engineering standards.

[0003] However, in practice, it has been found that when using the above methods to control the operation of construction robots, the following technical problems often arise: The operating parameters of construction robots must strictly adhere to building engineering standards. PDF parsing tools struggle to accurately identify the multi-level nested hierarchical structure within these standards, failing to effectively correlate parameter limits with safety requirements. This results in low accuracy in extracting information from building engineering standards, making the mechanical structure prone to collisions and deformations due to incorrect control parameters. Furthermore, PDF parsing tools exhibit errors in recognizing special formats such as tables and formulas within building engineering standards, including cell misalignment, misreading formula symbols, and parameter value deviations. Repeated correction of erroneous data leads to prolonged preparation time, resulting in extended construction time. Additionally, the sustained standby time of the equipment consumes significant energy resources.

[0004] The information disclosed in this background section is only intended to enhance the understanding of the background of the inventive concept, and therefore may contain information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The summary portion of this disclosure is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description portion. This summary portion is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0006] Some embodiments of this disclosure provide methods, apparatus, electronic devices, and computer-readable media for digitizing building engineering standards in a machine-readable manner to address one or more of the technical problems mentioned in the background section above.

[0007] In a first aspect, some embodiments of this disclosure provide a machine-readable digital processing method for building engineering standards. The method includes: in response to detecting construction request information corresponding to a construction robot, performing the following steps based on the acquired building engineering standard information corresponding to the construction robot: generating text sequence information corresponding to the building engineering standard information according to a first preset processing method; generating page layout information and bibliographic and citation relationship information corresponding to the building engineering standard information according to preset page layout annotation information, a page layout hierarchy recognition model, and preset bibliographic annotation information; generating main text information corresponding to the building engineering standard information according to preset text annotation information, a paragraph extraction model, and the page layout information; generating special format text information corresponding to the building engineering standard information according to a pre-trained formula recognition model; determining the text sequence information, page layout information, bibliographic and citation relationship information, main text information, and special format text information as construction standard control information; generating control parameter information corresponding to the construction robot based on the construction standard control information; and controlling the construction robot to perform construction processing based on the control parameter information.

[0008] Secondly, some embodiments of this disclosure provide a machine-readable digital processing device for building engineering standards, including an execution unit configured to, in response to detecting construction request information corresponding to a construction robot, perform the following steps based on the acquired building engineering standard information corresponding to the construction robot: generating text sequence information corresponding to the building engineering standard information according to a first preset processing method; generating page layout information and bibliographic and citation relationship information corresponding to the building engineering standard information according to preset page layout annotation information, page layout hierarchy recognition model, and preset bibliographic annotation information; generating body text information corresponding to the building engineering standard information according to preset text annotation information, paragraph extraction model, and the page layout information; generating special format text information corresponding to the building engineering standard information according to a pre-trained formula recognition model; determining the text sequence information, page layout information, bibliographic and citation relationship information, body text information, and special format text information as construction standard control information; generating control parameter information corresponding to the construction robot based on the construction standard control information; and controlling the construction robot to perform construction processing based on the control parameter information.

[0009] Thirdly, some embodiments of this disclosure provide an electronic device, including: one or more processors; and a storage device having one or more programs stored thereon, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation of the first aspect above.

[0010] Fourthly, some embodiments of this disclosure provide a computer-readable medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method described in any of the implementations of the first or second aspect.

[0011] The various embodiments disclosed above have the following beneficial effects: Through the machine-readable digital processing method for building engineering standards provided by some embodiments of this disclosure, the accuracy of acquiring building engineering standard information can be improved, the time spent on preliminary preparation can be shortened, and the accuracy of the acquired building engineering standard information in the construction field can be improved, as well as automation efficiency, decision-making efficiency, and collaboration efficiency. For example, in the application of construction robots, construction time can be shortened, and waste of equipment resources can be reduced. The low accuracy of obtaining construction engineering standard information, the long preparation time, the long construction time, and the significant waste of equipment resources are due to the following reasons: the operating parameters of construction robots must strictly follow construction engineering standards. PDF parsing tools have difficulty accurately identifying the multi-level nested hierarchical structure in construction engineering standards and cannot effectively associate parameter limits with safety requirements, resulting in low accuracy of information extraction from construction engineering standards. Incorrect control parameters can easily cause collisions and deformations of mechanical structures. Furthermore, PDF parsing tools have errors in recognizing special formats such as tables and formulas in construction engineering standards, such as cell misalignment, misreading formula symbols, and parameter value deviations. Repeated correction of erroneous data is required, leading to long preparation time, long construction time, and significant energy consumption due to prolonged equipment standby. Based on this, some embodiments of the present disclosure provide a machine-readable digital processing method for construction engineering standards, taking the application of construction robots as an example: First, in response to detecting the construction request information of the corresponding construction robot, based on the obtained construction engineering standard information corresponding to the construction robot, the following steps are performed: According to a first preset processing method, a text sequence information corresponding to the aforementioned construction engineering standard information is generated. Thus, the text sequence of the aforementioned construction engineering standard information can be obtained. Then, based on the preset page layout annotation information, page layout hierarchy recognition model, and preset bibliographic annotation information, the page layout information and bibliographic and citation relationship information corresponding to the aforementioned architectural engineering standard information are generated. This yields the physical structure, logical structure, and bibliographic and citation relationships of the aforementioned architectural engineering standard information. Next, based on the preset text annotation information, paragraph extraction model, and the aforementioned page layout information, the main text information corresponding to the aforementioned architectural engineering standard information is generated. This yields the main text content of the aforementioned architectural engineering standard information. Then, based on the pre-trained formula recognition model, special-format text information corresponding to the aforementioned architectural engineering standard information is generated. This yields the special-format content (e.g., tables, formulas, and images) of the aforementioned architectural engineering standard information. Finally, the aforementioned text sequence information, page layout information, bibliographic and citation relationship information, main text information, and special-format text information are determined as construction standard control information. This yields structured architectural engineering standards. Finally, based on the aforementioned construction standard control information, control parameter information corresponding to the aforementioned construction robot is generated, and based on the aforementioned control parameter information, the aforementioned construction robot is controlled to perform construction processing.Because the identification and extraction of building engineering standards are not performed using PDF parsing tools, but rather by first pre-processing unstructured building engineering standards to obtain text sequence information, the following steps are taken: First, pre-training is performed on a layout hierarchy recognition model and a paragraph extraction model using pre-set layout annotation information and pre-set text annotation information. This allows the models to accurately identify and extract information from the building engineering standards. Next, a pre-trained formula recognition model extracts special formats from the building engineering standards, thus obtaining text information with special formats. Furthermore, because the control parameters of the construction robot are determined based on the construction standard control information, and the robot is controlled according to these parameters, the accuracy of obtaining building engineering standard information is improved, the time spent on preliminary preparation is shortened, construction time is reduced, and equipment resource waste is minimized. Attached Figure Description

[0012] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0013] Figure 1 This is a flowchart of some embodiments of the machine-readable digital processing method for building engineering standards according to the present disclosure; Figure 2 This is a schematic diagram of the structure of some embodiments of the machine-readable digital processing apparatus for building engineering standards according to the present disclosure; Figure 3 This is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. Detailed Implementation

[0014] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0015] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0016] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0017] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0018] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0019] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] Figure 1 A flow 100 of some embodiments of a machine-readable digital processing method for building engineering standards according to the present disclosure is shown. This machine-readable digital processing method for building engineering standards includes the following steps: Step 101: In response to detecting the construction request information of the corresponding construction robot, the following steps are performed based on the obtained construction engineering standard information of the corresponding construction robot: Step 1011: Generate text sequence information corresponding to the building engineering standard information according to the first preset processing method.

[0021] In some embodiments, the execution entity (e.g., a computing device) of the machine-readable digital processing method for building engineering standards can generate text sequence information corresponding to the aforementioned building engineering standard information according to a first preset processing method. The construction robot can be a robot used for engineering construction. Here, the specific type of the construction robot is not limited. For example, the construction robot can be a residential construction robot. The construction request information can represent a request to start the construction robot for construction. For example, the construction request information can be to start the construction robot for masonry work. The first preset processing method can be the entire process of OCR extraction of the aforementioned building engineering standard information. The entire OCR extraction process can include preprocessing, region segmentation, core recognition, and post-processing. The preprocessing can be: first, using Gaussian filtering and median filtering to denoise the aforementioned building engineering standard information (PDF image); then, using Hough transform to perform tilt detection and correction on the denoised aforementioned building engineering standard information; finally, using adaptive binarization technology to enhance contrast; and applying morphological operations to perform character separation and concatenation on the tilt-detected and corrected aforementioned building engineering standard information. For example, the morphological operation can be opening operation. The aforementioned region segmentation can be characterized by segmenting the preprocessed building engineering standard information into regions using connected component analysis and edge detection techniques. The aforementioned core recognition can be characterized by recognizing the segmented building engineering standard information using a recognition architecture incorporating convolutional recurrent neural networks (CRNN). The aforementioned post-processing can be characterized by verifying the core-recognized building engineering standard information through a combination of professional dictionary matching, n-gram language model correction, and rule verification. The aforementioned building engineering standard information can be characterized by unstructured data corresponding to the building engineering standards. The aforementioned engineering construction standard information can be PDF images of the engineering construction standards. Here, the specific content of the aforementioned building engineering standards and the aforementioned unstructured data is not limited. For example, the building engineering standard can be the "Code for Design of Building Structures" (GB 50009), and the unstructured data can be PDF images. The aforementioned text sequence information can be characterized by structured text corresponding to the aforementioned building engineering standard information. It should be noted that the generated text sequence information corresponding to the aforementioned building engineering standard information can provide a standard basis for construction robots to carry out construction, and instructions to control the construction robot to carry out construction can be generated based on the text sequence information.

[0022] It should be noted that the application scenarios of the machine-readable digital processing method and device for building engineering standards disclosed herein can be any scenario involving the digital processing and application of building engineering standards. Here, the specific type of application scenario for the machine-readable digital processing method and device for building engineering standards disclosed herein is not limited and can be adjusted according to actual needs. For example, application scenarios can include various practical applications such as construction robots, compliance review, digital design and collaboration, cities, and digital twins.

[0023] In some optional implementations of certain embodiments, the aforementioned execution entity may generate text sequence information corresponding to the aforementioned building engineering standard information through the following steps according to a first preset processing method: First, according to the first preset processing method described above, perform the following steps: First, the aforementioned building engineering standard information is modified to obtain the modified building engineering standard information as the building engineering standard information. In practice, the implementing entity can modify the aforementioned building engineering standard information according to the preprocessing methods included in the first preset processing method, and obtain the preprocessed building engineering standard information as the building engineering standard information.

[0024] Then, based on the aforementioned building engineering standard information, region division information is generated. This region division information can represent text region information or non-text region information. The text region information can represent the main text-related information and auxiliary text information within the aforementioned building engineering standard information. The main text-related information can represent the content of the main text within the aforementioned building engineering standard information. The auxiliary text information can represent characters used to assist users in understanding the main text-related information. Here, the specific content of the main text-related information and auxiliary text information is not limited; for example, the main text-related information could be "Foundation treatment should comply with the following regulations:…", and the auxiliary text information could be "Table 5.3.2 Foundation Bearing Capacity Correction Coefficient". The non-text region information can represent images, tables, formula symbols, and other non-text elements in the aforementioned building engineering standard information. Here, the specific content of the other non-text elements is not limited; for example, other non-text elements could be the watermark on the cover of the building engineering standard information. In practice, the executing entity can segment the pre-processed building engineering standard information according to the region segmentation method included in the first preset processing method to obtain the region division information.

[0025] Finally, based on the aforementioned region division information, recognition result information is generated. This recognition result information can represent the character or text sequence identified by the executing entity from the aforementioned region division information. Here, the specific content of the characters and text sequences is not limited; for example, the character could be "3.1.5", and the text sequence could be "reinforcing steel protective layer thickness 1.2m". In practice, the executing entity can perform recognition processing on the aforementioned building engineering standard information after region segmentation according to the core recognition method included in the first preset processing method, to obtain the recognition result information.

[0026] The second step involves modifying the aforementioned recognition result information to obtain the modified recognition result information as text sequence information. In practice, the executing entity can use the post-processing methods included in the first preset processing method to verify the core-recognized building engineering standard information to obtain text sequence information.

[0027] Step 1012: Based on the preset page labeling information, page hierarchy recognition model, and preset bibliographic labeling information, generate the page information and bibliographic and citation relationship information of the corresponding building engineering standard information.

[0028] In some embodiments, the aforementioned execution entity can generate layout information and bibliographic and citation relationship information corresponding to the aforementioned architectural engineering standard information based on preset layout annotation information, layout hierarchy recognition model, and preset bibliographic annotation information. The preset layout annotation information can represent a pre-defined dataset. The data in the dataset can represent architectural engineering standards with both physical and logical structures annotated. Here, the number of data points in the dataset is not limited; for example, the number of data points in the dataset can be 100. The layout hierarchy recognition model can represent a pre-trained model based on the ViT architecture. For example, the layout hierarchy recognition model can be ViT-LayoutLM. The layout hierarchy recognition model can include a visual branch based on a multi-layer Transformer encoder, a text branch based on a multi-layer Transformer encoder, and a multimodal fusion branch responsible for feature interaction. Each layer of the multi-layer Transformer encoder can include a multi-head self-attention layer, two residual connection layers, two normalization layers, and a feedforward neural network layer. The training method for the layout hierarchy recognition model can be batch training. The aforementioned page layout hierarchy recognition model can be a model that takes the aforementioned text sequence information as input and outputs element classification information. The aforementioned element classification information can represent the location of the physical structural elements of the aforementioned architectural engineering standard information. Here, the specific content of the aforementioned element classification information is not limited; for example, the element classification information can be "Element: Title; Location: Top center area of ​​the document's first page". The aforementioned page layout information can represent the physical and logical structure of the aforementioned architectural engineering standard information. The aforementioned preset bibliographic annotation information can represent the dataset obtained by annotating the bibliographic information, cited chapter location information, and in-text citation information of each architectural engineering standard in the first preset number of architectural engineering standards. The aforementioned bibliographic information can represent the metadata of the architectural engineering standard. The aforementioned cited chapter location information can represent the location of the chapter that cites all other architectural engineering standards. The aforementioned in-text citation information can represent other cited architectural engineering standards or other clauses within the current architectural engineering standard. Here, the specific content of the aforementioned bibliographic information, the aforementioned cited chapter location information, and the aforementioned in-text citation information is not limited; for example, the bibliographic information can be "Standard Number: GB 50007-2011". The referenced chapter location information can be "Location: Chapter 2: Normative References; Referenced Building Engineering Standard Number: GB 50007-2011; ...; GB 50010...".The cited information within the text can be, for example, "Reference point: According to Article 5.2.4 of the 'Code for Design of Building Foundations' GB 50007-2011, the characteristic value of the foundation bearing capacity can be comprehensively determined by load tests or other in-situ tests, formula calculations, and combined with engineering practice experience," or "Reference point: The seismic grade shall comply with the provisions of Article 3.1.2 of this standard." The above bibliographic information and citation relationships characterize the bibliographic information and citation relationships of the aforementioned building engineering standard information. Here, the specific content of the above bibliographic information and citation relationships is not limited. For example, the bibliographic information could be "Standard Name: 'Standard for Seismic Design of Buildings'," and the citation relationship could be, "Reference point: The standard value of the strength of reinforcing steel shall comply with the provisions of 'GB / T 1499.2-2018'; Citation relationship: The current building engineering standard information cites the external building engineering standard 'GB / T 1499.2-2018'." The above citation points characterize the content of the building engineering standard information that cites other building engineering standard information. It should be noted that external building engineering standards can represent building engineering standards that differ from current building engineering standard information.

[0029] In some optional implementations of certain embodiments, the aforementioned execution entity can generate page information and bibliographic and citation relationship information corresponding to the aforementioned architectural engineering standard information through the following steps: based on preset page annotation information, page hierarchy recognition model, and preset bibliographic annotation information. The first step is to generate page information corresponding to the aforementioned architectural engineering standard information based on the preset page labeling information and page hierarchy recognition model.

[0030] The second step is to generate bibliographic and citation relationship information corresponding to the above-mentioned architectural engineering standard information based on the preset bibliographic annotation information and the above-mentioned layout information.

[0031] In some optional implementations of certain embodiments, the aforementioned execution entity can generate page information corresponding to the aforementioned architectural engineering standard information through the following steps based on preset page annotation information and page hierarchy recognition model: The first step involves modifying the aforementioned layout hierarchy recognition model based on the building engineering standard information and the preset layout annotation information, resulting in a modified layout hierarchy recognition model. In practice, the executing entity can pre-train the initial layout hierarchy recognition model using the preset layout annotation information to obtain a pre-trained layout hierarchy recognition model. The structure of the initial layout hierarchy recognition model is the same as that of the aforementioned layout hierarchy recognition model, and will not be described again here.

[0032] The second step involves generating element classification information based on the aforementioned layout hierarchy recognition model and text sequence information. In practice, the executing entity can input the text sequence information into the layout hierarchy recognition model to obtain element classification information.

[0033] The third step is to generate text feature information based on the aforementioned element classification information. This text feature information can include semantic features, visual features, and topological features. In practice, the executing entity can use the BERT model to extract and process the aforementioned text sequence information, obtaining the semantic features, visual features (font size, thickness, position, etc.), and topological features (relative positional relationships between elements) corresponding to the aforementioned architectural engineering standard information.

[0034] The fourth step involves generating hierarchical information corresponding to the aforementioned building engineering standard information based on the textual feature information described above. This hierarchical information represents the hierarchical relationships within the logical structure of the building engineering standard information. The specific type of this hierarchical information is not limited here; for example, it could be "First level: Chapter 1 General Provisions; Second level: Section 1 General Regulations". In practice, the executing entity can use a combination of Conditional Random Fields (CRF) and Recurrent Neural Networks (RNN) to parse the building engineering standard information and obtain the corresponding hierarchical relationships.

[0035] The fifth step is to determine the layout information of the above-mentioned element classification information and the above-mentioned hierarchical information as the layout information of the above-mentioned building engineering standard information.

[0036] In some optional implementations of certain embodiments, the aforementioned executing entity can generate bibliographic and citation relationship information for corresponding architectural engineering standards based on preset bibliographic annotation information and layout information through the following steps: The first step is to generate bibliographic named entity information based on a preset named entity extraction method and the aforementioned layout information. The preset named entity extraction method can be a sequence labeling method based on a pre-trained model. The bibliographic named entity information can represent entities in the aforementioned architectural engineering standard information. Here, the specific content of the bibliographic named entity information is not limited; for example, the bibliographic named entity information can be "standard name". In practice, firstly, the executing entity can generate a contextual representation of the aforementioned layout information using a BERT encoder. Then, a bidirectional LSTM layer is used to capture the reference relationships of the aforementioned architectural engineering standard information. Finally, a CRF layer is used to constrain the rule design for extracting bibliographic information from the aforementioned architectural engineering standard information, thus obtaining the bibliographic named entity information.

[0037] The second step is to generate location information based on the aforementioned pre-set bibliographic annotation information. This location information can include bibliographic location information and citation location information. The bibliographic location information indicates the position of the bibliographic information within the aforementioned architectural engineering standard information. The citation location information indicates the position (page number or chapter) of the citation point within the aforementioned architectural engineering standard information. Here, the specific content of the bibliographic location information and the citation location information is not limited. For example, the bibliographic location information could be "Bibliographic location: top of the first page," and the citation location information could be "Citation point: The material properties of the reinforcing steel should comply with the provisions of GB / T 1499 'Steel for Reinforced Concrete'; Location of the citation point: Page 42, paragraph 3" or "Citation point: The seismic resistance grade should comply with the provisions of Article 3.1.2 of this standard; Location of the citation point: Chapter 3." In practice, firstly, the executing entity can pre-train the relation extraction model based on the aforementioned pre-set bibliographic annotation information to obtain the pre-trained relation extraction model. Then, the aforementioned page layout information is input into the relation extraction model to obtain the location information. The aforementioned relation extraction model can represent the UIE (Universal Information Extraction) model. The UIE model can include a Transformer encoder layer, a prefix guidance layer, and a span prediction layer. It should be noted that the structure of the Transformer mentioned above is the same as that included in the layout hierarchy recognition model, and will not be repeated here. The aforementioned relation extraction model can take the layout information as input and location information as output, or take the generated construction standard control information as input and output the relevant clause information and triplet information corresponding to the construction standard control information. The training method for the aforementioned relation extraction model can be batch training.

[0038] The third step is to generate reference type information based on the preset entity classification method and the aforementioned location information. The preset entity classification method can be a method that determines the type of reference relationship using a relationship classifier. The reference type information can characterize the type of reference relationship of the aforementioned building engineering standard information. The reference type information can be external reference information or internal reference information. External reference information can characterize the reference relationship of the aforementioned building engineering standard information as referencing external building engineering standard information. Internal reference information can characterize the reference relationship of the aforementioned building engineering standard information as referencing content within the aforementioned building engineering standard information. In practice, the executing entity can process the aforementioned location information according to the preset entity classification method to obtain the reference type information.

[0039] The fourth step is to generate reference link information based on the aforementioned reference type and reference location information. The reference link information represents the reference path corresponding to the aforementioned reference type information. The specific content of the reference path is not limited here; for example, the reference path could be "Reference type: External reference information; Reference point: The standard value of the strength of the reinforcing steel should comply with the provisions of GB 50007; Name of the referenced building engineering standard information: Code for Design of Building Foundations" or "Reference type: Internal reference information; Reference point: Should comply with the requirements of Article 3.2 of this code; Internal content of the reference: Article 3.2". In practice, firstly, in response to determining that the aforementioned reference type information is external reference information, based on the aforementioned reference location information, the number of the external building engineering standard included in the aforementioned reference location information is matched and associated with the standard entity in the standard library to obtain the reference link information corresponding to the aforementioned external reference information. The aforementioned standard library can represent a database storing building engineering standards. The aforementioned standard entity can represent the complete data included in any building engineering standard in the aforementioned standard library. In response to determining that the above-mentioned reference type information is internal reference information, the reference link information corresponding to the above-mentioned internal reference information is determined based on the clauses that reference the current building engineering standard information included in the above-mentioned reference location information.

[0040] The fifth step is to determine the bibliographic name entity information and the citation link information mentioned above as the bibliographic and citation relationship information corresponding to the above-mentioned building engineering standard information.

[0041] Step 1013: Generate the main text information of the corresponding building engineering standard information based on the preset text annotation information, paragraph extraction model and layout information.

[0042] In some embodiments, the execution entity can generate the main text information corresponding to the aforementioned building engineering standard information based on preset text annotation information, a paragraph extraction model, and the aforementioned layout information. The preset text annotation information can represent a dataset obtained by annotating the special text element information within each of the first preset number of building engineering standards. The main text information can include paragraph structure information and special text element information. The paragraph structure information can represent the paragraphs included in the aforementioned building engineering standard information. The special text element information can represent the special text elements included in the aforementioned building engineering standard information. The special text elements can represent non-main text content within the aforementioned building engineering standard information. Here, the specific content of the special text elements is not limited; for example, a special text element can be a comment.

[0043] In some optional implementations of certain embodiments, the aforementioned execution entity can generate the main text information corresponding to the aforementioned architectural engineering standard information through the following steps based on preset text annotation information, paragraph extraction model, and the aforementioned layout information: The first step is to generate visual feature differentiation information based on the aforementioned layout information and the layout hierarchy recognition model. This visual feature differentiation information may include paragraph boundary information and visual feature marker information. The paragraph boundary information can characterize the start and end positions and separation boundaries of different paragraphs in the aforementioned architectural engineering standard information. The specific content of these separation boundaries is not limited here; for example, a separation boundary can be indentation. The visual feature marker information can characterize the visual attributes of different paragraphs in the aforementioned architectural engineering standard information. These visual attributes can characterize the font size in the aforementioned architectural engineering standard information. In practice, the executing entity can input the aforementioned layout information into the aforementioned layout hierarchy recognition model to obtain the visual feature differentiation information.

[0044] The second step involves generating precise paragraph boundary information based on the aforementioned visual feature discrimination information and the pre-trained semantic segmentation model. The pre-trained semantic segmentation model can represent the YOLOv8-Seg model. The YOLOv8-Seg model may include a backbone network, a neck network, and a segmentation head network. The pre-trained semantic segmentation model can take the aforementioned visual feature discrimination information as input and output the precise paragraph boundary information. The pre-trained semantic segmentation model can be trained in batches. The precise paragraph boundary information can represent the corrected visual feature discrimination information. In practice, the executing entity can input the aforementioned visual feature discrimination information into the pre-trained semantic segmentation model to obtain the precise paragraph boundary information.

[0045] The third step involves generating specific text element information based on the aforementioned precise paragraph boundary information, the aforementioned preset text annotation information, and the aforementioned paragraph extraction model. The aforementioned paragraph extraction model can represent a CNN-BiLSTM-CRF sequence annotation model based on an attention mechanism. This model takes the aforementioned precise paragraph boundary information as input and the specific text element information as output. The paragraph extraction model may include an input layer, a convolutional layer (CNN), a bidirectional long short-term memory network layer (Bi-LSTM), an attention layer, a fully connected layer, and a conditional random field layer (CRF). The aforementioned convolutional layer (CNN) may include multiple convolutional kernels, a pooling layer, and an activation function layer. The aforementioned bidirectional long short-term memory network layer (Bi-LSTM) may include multiple forward LSTM layers and multiple backward LSTM layers. The aforementioned attention layer may include a weight calculation layer and a feature weighting layer. The training method for the aforementioned paragraph extraction model can be batch training. In practice, firstly, the aforementioned execution entity can pre-train the aforementioned paragraph extraction model based on the aforementioned preset text annotation information to obtain the pre-trained paragraph extraction model. Then, the precise paragraph boundary information is input into the pre-trained paragraph extraction model to obtain special text element information.

[0046] The fourth step is to determine the above-mentioned special text element information and the above-mentioned precise paragraph boundary information as the main text information corresponding to the above-mentioned building engineering standard information.

[0047] Step 1014: Generate special format text information corresponding to the building engineering standard information based on the pre-trained formula recognition model.

[0048] In some embodiments, the executing entity can generate specially formatted text information corresponding to the aforementioned building engineering standard information based on a pre-trained formula recognition model. This specially formatted text information can represent specially formatted content within the aforementioned building engineering standard information. This specially formatted content can represent non-textual content. For example, the specially formatted content can be a table.

[0049] In addressing the technical problems mentioned above, and considering the application scenario of controlling construction robots based on building structure design standards, the building structure design standard documents contain numerous tables containing mechanical properties, dimensional specifications, and material parameters. Accurate identification of cross-page tables and formulas within these documents is crucial for controlling the construction robot. However, this often leads to the following technical problem: due to page breaks, cross-page tables are prone to header and content breaks and misaligned row and column relationships, resulting in poor accuracy in identifying table structure information. Furthermore, formulas within tables are not extracted separately and are mixed with paragraph structure information, easily causing misreading or omission of formula symbols. This results in low accuracy in setting construction parameters based on formula calculations, potentially leading to collisions and deformations of the mechanical structure due to incorrect control parameters. To address the following requirements for this application scenario: adapting to table breaks caused by pagination in engineering standard documents, adapting to nested tables and formulas, and adapting to the large number of parameters within engineering standard documents, we have decided to adopt the following solution: In some optional implementations of certain embodiments, the aforementioned execution entity can generate special format text information corresponding to the aforementioned building engineering standard information by recognizing the model according to a pre-trained formula through the following steps: The first step involves generating table region information based on the aforementioned building engineering standard information and a pre-trained table extraction model. The pre-trained table extraction model can represent YOLOv8. This model takes the aforementioned building engineering standard information as input and outputs the table region information. It may include a backbone network, a neck network, and a detection head network. The pre-trained model can be trained in batches. The table region information represents the position of the tables within the aforementioned building engineering standard information.

[0050] The second step is to generate special format text information corresponding to the above-mentioned building engineering standard information in response to determining that the above-mentioned table area information is the first type of table area information.

[0051] In some optional implementations of certain embodiments, in response to determining that the above-mentioned table area information is first type of table area information, the execution entity can generate special format text information corresponding to the above-mentioned building engineering standard information: The first step is to determine the target table structure information corresponding to the aforementioned first type of table area information. The aforementioned first type of table area information can represent tables that do not span multiple pages in the aforementioned architectural engineering standard information. The aforementioned table structure information can represent the row and column structure of the tables included in the aforementioned architectural engineering standard information. In practice, firstly, the executing entity can input the aforementioned first type of table area information into the table structure recognition model to obtain the table structure information corresponding to the aforementioned first type of table area information. Then, the table structure information corresponding to the aforementioned first type of table area information is formatted to obtain Markdown format table structure information as the target table structure information corresponding to the aforementioned first type of table area information. The aforementioned table structure recognition model can represent an Attention-based TableTransformer (ATT) model. The aforementioned table structure recognition model can take the aforementioned table area information as input and the table structure information as output. The aforementioned table structure recognition model can include an input layer, a Transformer encoder layer, and an output layer. It should be noted that the structure of the aforementioned Transformer is the same as the structure of the Transformer included in the aforementioned layout hierarchy recognition model, and will not be repeated here. The training method for the above table structure recognition model can be batch training.

[0052] The second step involves, in response to the determination that the aforementioned table area information is of the second type, generating target table structure information corresponding to the second type of table area information according to a preset cross-page merging method. This preset cross-page merging method can be based on table style consistency analysis and content continuity analysis, automatically identifying and connecting tables across pages. For example, table style consistency analysis can involve comparing the border styles and font sizes of cross-page tables to see if they are consistent. Content continuity analysis can be used to determine whether the numerical sequences and text keywords of the first and last rows of the cross-page table are continuous. The aforementioned second type of table area information can represent cross-page tables in the aforementioned architectural engineering standard information. In practice, in response to the determination that the aforementioned table area information is of the second type of table area information, firstly, the executing entity can perform fusion processing on the aforementioned second type of table area information according to the preset cross-page merging method to obtain table structure information corresponding to the aforementioned second type of table area information. Then, the table structure information corresponding to the aforementioned second type of table area information is formatted using LaTeX code to obtain Markdown format table structure information as the target table structure information corresponding to the aforementioned second type of table area information.

[0053] The third step involves generating target formula region information based on the aforementioned architectural engineering standard information and the pre-trained formula recognition model. This target formula region information represents the location of the formula within the architectural engineering standard information. In practice, the executing entity first inputs the architectural engineering standard information into the visual detection model to obtain formula region information. Then, it inputs this formula region information into the pre-trained formula recognition model to obtain the target formula region information. The visual detection model can take the architectural engineering standard information as input and the formula region information as output. This visual detection model can be a DETR (Detection Transformer) model. It can include a backbone network (CNN), an encoder (Transformer), a decoder (Transformer), and a prediction head. It should be noted that the structure of the Transformer is the same as that included in the layout hierarchy recognition model, and will not be elaborated further here. The training method for the visual detection model can be batch training. The pre-trained formula recognition model can represent the MinerU model. This pre-trained formula recognition model can include an encoder (DenseNet) and a decoder (Transformer). It should be noted that the structure of the included decoder (Transformer) is the same as that of the Transformer included in the above-mentioned layout hierarchy recognition model, and will not be repeated here. The pre-trained formula recognition model can take the formula region information output by the above-mentioned visual detection model as input and the target formula region information as output. The training method of the pre-trained formula recognition model can be batch training.

[0054] The fourth step involves generating formula symbol information corresponding to the table area information based on the pre-trained formula recognition model, in response to the determination that the table area information meets preset constraints. These preset constraints can be the presence of formulas or special symbols in the tables of the architectural engineering standard information. The formula symbol information can represent mathematical operators, engineering-specific symbols, and variable / parameter symbols present in the architectural engineering standard information. In practice, in response to the determination that the table area information meets preset constraints, the executing entity first extracts and processes the table area information using a layout analysis algorithm to obtain table area information that meets the preset constraints. Then, the table area information that meets the preset constraints is input into the pre-trained formula recognition model to obtain the formula symbol information corresponding to the table area information.

[0055] The fifth step involves updating the formula symbol information to obtain the updated formula symbol information as the target formula symbol information. In practice, the executing entity can use LaTeX code to convert the formula symbol information to Markdown format as the target formula symbol information.

[0056] The sixth step is to determine the target table structure information, the target formula area information, and the target formula symbol information as special format text information corresponding to the above-mentioned building engineering standard information.

[0057] The above-described technical solution, as an inventive point of this disclosure, solves technical problem two: "poor accuracy in recognizing table structure information and low accuracy in setting construction parameters based on formula calculations." The reasons for this poor accuracy are as follows: Cross-page tables, due to page segmentation, are prone to header and content breaks and row / column correspondence errors, leading to poor accuracy in recognizing table structure information. Furthermore, formulas within the table are not extracted separately and are mixed with paragraph structure information, easily causing misreading or omission of formula symbols, resulting in low accuracy in setting construction parameters based on formula calculations. This, in turn, can easily cause collisions and deformations in mechanical structures due to incorrect control parameters. Solving these factors can improve the accuracy of table structure information recognition and enhance the accuracy of setting construction parameters based on formula calculations. To achieve this effect, the machine-readable digital processing method for building engineering standards disclosed in this disclosure first extracts and processes the building engineering standard information according to the above-described page hierarchy recognition model and the above-described preset target detection method. This yields table area information. Then, when the table area information is detected as first-type table area information, it is extracted according to the table structure recognition model. This yields the target table structure information corresponding to the first-type table area information. When the table area information is detected as second-type table area information, it is processed according to a preset page merging method. This yields the target table structure information corresponding to the second-type table area information. Next, the table area information is extracted using a layout analysis algorithm. This yields table area information that meets preset constraints. Then, the table area information that meets preset constraints is extracted using the pre-trained formula recognition model. This yields formula symbol information corresponding to the table area information. Therefore, the accuracy of table structure information recognition and the accuracy of construction parameter settings based on formula calculations can be improved, thereby reducing the occurrence of mechanical structure collisions and deformations caused by incorrect control parameters.

[0058] In addressing the aforementioned technical problems in the application scenario—specifically, when controlling construction robots based on building structure design standards—and accurately identifying cross-page tables within building structure design standard documents containing a large amount of multi-source heterogeneous data, the rapid and precise location of these tables often presents the following technical problem: Existing methods do not differentiate between the sources of building engineering standards (e.g., editable PDFs with complete underlying structural information versus scanned documents based entirely on image recognition). For editable PDFs containing usable metadata, the inherent metadata is not prioritized for reuse; instead, a large-scale target detection model is directly used for full analysis, resulting in high computational resource consumption and long processing time, thus extending construction time. For scanned images, the same method lacks sufficient robustness to address image quality issues, leading to poor accuracy in locating table area information boundaries. Considering the following requirements for this application scenario: adaptability to differentiated processing of building engineering standards containing a large amount of multi-source heterogeneous data and adaptability to high-precision construction requirements within a short timeframe, we have decided to adopt the following solution: In some optional implementations of certain embodiments, the aforementioned execution entity can generate table region information based on the aforementioned building engineering standard information and a pre-trained table extraction model through the following steps: The first step, in response to determining that the above-mentioned building engineering standard information is Class I building engineering standard information, is to perform the following steps: The first sub-step involves generating candidate metadata regions based on a preset metadata extraction method. Here, the aforementioned first type of building engineering standard information can be characterized as an editable PDF format building engineering standard. The preset metadata extraction method can be a method of data extraction and processing of the aforementioned first type of building engineering standard information using a Python library. The aforementioned candidate metadata regions can represent the coordinates of the upper left and lower right corners of the tables in the aforementioned building engineering standard information. In practice, firstly, the executing entity can use the preset metadata extraction method to extract and process the aforementioned first type of building engineering standard information, obtaining the corresponding metadata and chapter titles. Here, the specific content of the aforementioned metadata and chapter titles is not limited. For example, the metadata can be the identity identifier of the table object in the aforementioned first type of building engineering standard information (such as the " / Type / Table" tag) and its corresponding location range (such as the boundary coordinates of " / BBox [x1 y1 x2 y2]"). The chapter title can be "3.2 Construction Parameter Table". Then, the location range of the table is parsed from the aforementioned metadata, and a correlation is established between the coordinates of the aforementioned metadata and the coordinates of the aforementioned chapter titles through location proximity. Finally, the aforementioned location range is determined as candidate metadata region information.

[0059] The second sub-step generates detection result information based on the aforementioned metadata candidate region information and line detection method. The line detection method can be an edge detection algorithm used to detect lines in the image portion corresponding to the aforementioned metadata candidate region information. For example, the edge detection algorithm could be the Canny algorithm. The detection result information can represent the candidate region information to be tested or the page information to be tested. The candidate region information to be tested can represent the result obtained when the aforementioned metadata candidate region information meets preset detection conditions. The page information to be tested can represent the result obtained when the aforementioned metadata candidate region information does not meet the preset detection conditions. The preset detection conditions can be the presence of engineering table grid lines in the aforementioned metadata candidate region information. In practice, the executing entity can use the aforementioned line detection method to detect the aforementioned metadata candidate region information and obtain the detection result information.

[0060] The third sub-step generates the first region information to be corrected based on the aforementioned detection results, preset semantic verification method, preset verification constraints, and the pre-trained table extraction model. The preset semantic verification method can be a multi-dimensional semantic matching method. The preset verification constraints can be that the chapter title contains the target keyword and the proportion of preset combined words in the detection results is greater than or equal to a preset proportion threshold. The preset combined words can represent a combination of numbers and units (mathematical or physical units). The target keyword can represent the name of a table in a building engineering standard. Here, the specific content of the preset combined words and the target keyword is not limited; for example, the preset combined word could be 25MPa, and the target keyword could be a parameter table. The preset proportion threshold can be 60%. The first region information to be corrected can represent the detection results that satisfy the preset verification constraints. In practice, in response to determining the detection results as candidate region information, the executing entity can determine the candidate region information that satisfies the preset verification constraints as the first region information to be corrected. In response to determining that the above detection result information is the page information to be tested, the above execution entity can input the page information to be tested into the pre-trained table extraction model to obtain the first region information to be corrected.

[0061] The fourth sub-step involves updating the information of the first region to be corrected according to the first preset multi-dimensional correction method, resulting in the updated information of the first region to be corrected, which is then used as the information of the first table region. The first preset multi-dimensional correction method can be as follows: First, identify missing rows by detecting abrupt changes in row pixel grayscale; extract the normal row height of the preset number of rows above and below the missing row position and calculate the average; then, complete the y-axis coordinate of the missing row based on this average. Next, perform pixel-level detection on the cell borders for boundary correction. This pixel-level detection method can be as follows: when the pixel continuity of the cell border is lower than a preset threshold, it is determined to be a half-frame cell; the complete column width of the column containing the half-frame cell is extracted and the average is calculated; then, the x-axis coordinate of the half-frame cell is corrected based on this average. Here, the specific values ​​of the preset number of rows and the preset threshold are not limited and can be adjusted according to the actual application scenario. For example, the preset threshold can be 80%, and the preset number of rows can be 3. The information of the first table region can represent the corrected information of the first region to be corrected. In practice, firstly, the executing entity can correct the first area information to be corrected according to the first preset multi-dimensional correction method to obtain the corrected first area information to be corrected. Then, semantic verification is performed on the corrected first area information to be corrected using semantic matching to obtain the first table area information.

[0062] The second step, in response to determining that the above-mentioned building engineering standard information is Category II building engineering standard information, is to perform the following steps: The first sub-step involves generating denoising alignment region information according to the aforementioned first preset processing method. Here, the aforementioned second type of building engineering standard information can represent a scanned copy (e.g., a scanned PDF) of the building engineering standard. The aforementioned denoising alignment region information can represent the aforementioned second type of building engineering standard information after denoising processing. In practice, firstly, the executing entity can preprocess the aforementioned second type of building engineering standard information according to the aforementioned first preset processing method to obtain preprocessed second type of building engineering standard information. Then, a Gaussian filtering method is used to denoise the preprocessed second type of building engineering standard information to obtain denoised second type of building engineering standard information. Finally, a Hough transform is used to correct skewed pages in the denoised second type of building engineering standard information to obtain denoising alignment region information.

[0063] The second sub-step involves generating second region information to be corrected based on the pre-trained table extraction model and the denoised alignment region information. This second region information represents the location of the table extracted by the pre-trained table extraction model. In practice, the executing entity can input the denoised alignment region information into the pre-trained table extraction model to obtain the second region information to be corrected.

[0064] The third sub-step involves updating the second region information to be corrected according to the first preset multi-dimensional correction method, resulting in the updated second region information as the second table region information. The second table region information represents the corrected second region information to be corrected. In practice, firstly, the executing entity can correct the second region information to be corrected according to the first preset multi-dimensional correction method, obtaining the corrected second region information to be corrected. Then, semantic verification is performed on the corrected second region information to be corrected using semantic matching to obtain the second table region information.

[0065] The third step is to determine the information in the first table area and the information in the second table area as the table area information.

[0066] The above technical solution, as an inventive point of this disclosure, solves technical problem three: "leading to excessive computational resource consumption, long processing time, long construction time, and poor accuracy in locating table area information boundaries." The reasons for excessive computational resource consumption, long processing time, long construction time, and poor accuracy in locating table area information boundaries are as follows: Existing methods do not distinguish between the source of building engineering standards (e.g., editable PDFs with complete underlying structural information and scanned documents based entirely on image recognition). For editable PDF documents containing usable metadata, they fail to prioritize reusing their inherent metadata, instead directly using large-scale target detection models for full analysis, resulting in excessive computational resource consumption, long processing time, and long construction time. For scanned images, the same method lacks sufficient robustness to address image quality issues, leading to poor accuracy in locating table area information boundaries. Solving these factors can reduce computational resource consumption, shorten processing time, and improve the accuracy of boundary location. To achieve this effect, the disclosed machine-readable digital processing method for architectural engineering standards first processes the aforementioned first type of architectural engineering standard information using the aforementioned preset metadata extraction method and line detection method. Second, it verifies the detection results using a preset semantic verification method and preset verification constraints. Next, it introduces the aforementioned first preset multi-dimensional correction method to correct the verified first area information to be corrected, obtaining first table area information. Then, for the aforementioned second type of architectural engineering standard information, it preprocesses it using the aforementioned first preset processing method. Next, it extracts the preprocessed noise-reduced and aligned area information using the aforementioned pre-trained table extraction model. Then, it corrects the extracted second area information to be corrected using the aforementioned first preset multi-dimensional correction method, obtaining second table area information. Therefore, it can reduce computational resources and shorten processing time, thereby shortening construction time and improving the accuracy of positioning boundaries.

[0067] Step 1015: The text sequence information, page layout information, bibliographic and citation relationship information, main text information, and special format text information are identified as construction standard control information.

[0068] In some embodiments, the aforementioned executing entity may determine the aforementioned text sequence information, page layout information, bibliographic and citation relationship information, main text information, and specially formatted text information as construction standard control information. The aforementioned construction standard control information can represent the structured data of the corresponding building engineering standard. It should be noted that the steps for generating construction standard control information using the aforementioned building engineering standard information demonstrate that the aforementioned building engineering standard information has undergone digital processing.

[0069] Step 1016: Generate control parameter information for the corresponding construction robot based on the construction standard control information, and control the construction robot to perform construction processing based on the control parameter information.

[0070] In some embodiments, the execution entity can generate control parameter information corresponding to the construction robot based on the construction standard control information, and control the construction robot to perform construction processing based on the control parameter information. The control parameter information can characterize the control parameters of the construction robot during the construction process. Here, the specific content of the control parameters is not limited; for example, the control parameter information can be the movement speed of the construction robot's robotic arm.

[0071] In addressing the technical problems mentioned above, and considering the application scenario—intelligent operation control of construction robots in high-altitude engineering construction projects—the harsh terrain of high-altitude areas often presents the following technical problem: directly applying static parameters from engineering construction documents is difficult to adapt to the dynamic and uncertain real-world operating environment, resulting in poor safety in controlling the construction robot and an inability to systematically coordinate and optimize multiple conflicting objectives (e.g., low energy consumption and high efficiency), requiring repeated trial and error adjustments, leading to longer construction times and higher resource consumption during robot standby. To meet the following requirements of this application scenario—adapting to rapid and accurate collaborative decision-making under multi-objective conflicts—we have decided to adopt the following solution: In some optional implementations of certain embodiments, the execution entity may generate control parameter information corresponding to the construction robot based on the construction standard control information, and control the construction robot to perform construction processing based on the control parameter information, through the following steps: The first step is to generate semantic construction standard control information based on the aforementioned construction standard control information. This semantic construction standard control information can represent the semantics of the structured data corresponding to the aforementioned building engineering standard information. In practice, the implementing entity can first input the aforementioned construction standard control information into the aforementioned relation extraction model to obtain clause statement-related information and triple information. The clause statement-related information can represent the semantic content of the clauses included in the aforementioned building engineering standard information. Here, the specific content of the clauses is not limited; for example, the clause could be "Before the self-climbing system is lifted, the synchronization control module should be calibrated." The semantic content can include the clause type, clause content information, and strictness information. The clause content information can represent the statements included in the clause. The strictness information can represent the compliance level of the clause. Here, neither the clause type nor the specific content of the clause content information is limited; for example, the clause type could be a precondition, and the clause content information could be "Before the self-climbing system is lifted." The aforementioned compliance levels can be categorized as Level 1 (most stringent compliance), Level 2 (strict compliance), Level 3 (slight differences from the requirements of the clause are permissible), or Level 4 (the clause can be referenced, but compliance is not mandatory). The aforementioned triplet information can represent structured semantics presented in the form of subject-verb-object (SPO) triples. For example, the triplet information could be (concrete leveling robot, control, walking speed). Finally, the relevant information from the aforementioned clause statements and the aforementioned triplet information are determined as semantic construction standard control information.

[0072] The second step involves generating comprehensive, multi-dimensional correlated data information based on the aforementioned semantic construction standard control information. This comprehensive correlated data information may include construction task parameter data, operational environment parameter data, and performance testing data. The construction task parameter data characterizes the operational control parameters of the construction robot. These operational control parameters characterize various parameters controlling the construction robot's construction operations. These operational control parameters may include, but are not limited to: robotic arm joint angles, moving platform speed, end effector force, material supply rate, load range, spatial throughput range, and stability range. The spatial throughput range characterizes the probability that the construction robot can pass through the narrowest point in the construction environment. The stability range characterizes the operational stability of the construction robot. The operational environment parameter data characterizes the construction environment parameters. These construction environment parameters characterize the temperature and humidity during construction. The performance testing data characterizes the performance testing parameters of the construction robot. These performance testing parameters characterize the engineering standards after the construction robot's work. These performance testing parameters may include, but are not limited to: the range of flatness error, the range of verticality deviation, and the range of mortar fullness. In practice, firstly, the aforementioned implementing entities can use a classification extraction method to extract and process the aforementioned construction standard control information to obtain multi-dimensional related data information.

[0073] The third step involves updating the aforementioned all-dimensional correlation data information using the box plot removal method and linear interpolation method, obtaining the updated all-dimensional correlation data information as the target all-dimensional correlation data information. The box plot removal method can be the box plot method itself, and the linear interpolation method can be linear interpolation. In practice, firstly, the executing entity can use the box plot method to remove outliers from the aforementioned all-dimensional correlation data information. Then, using linear interpolation, missing values ​​are filled into the outlier-removed all-dimensional correlation data information to obtain the target all-dimensional correlation data information.

[0074] Fourth, based on the aforementioned comprehensive data related to the target, perform the following steps: The first sub-step involves generating a task accuracy function based on a generalized regression optimization method. This generalized regression optimization method can be a modeling approach that optimizes a generalized regression neural network (GRNN) using simulated annealing (SA) algorithm. The task accuracy function characterizes the mapping relationship between the construction robot's task control parameters and task accuracy indicators (such as wall flatness error and brick joint alignment deviation). The GRNN can take the construction robot's task control parameters as input and the task accuracy function as output. The GRNN may include an input layer, a pattern layer, a summation layer, and an output layer. The training of the GRNN can be batch training. In practice, firstly, the execution entity can initialize the initial GRNN using a non-parametric modeling method based on radial basis functions (RBF). Then, the smoothing factor of the initialized GRNN is optimized using the simulated annealing (SA) algorithm. Next, based on the optimized smoothing factor, the initialized GRNN is trained to obtain the trained GRNN. Next, the aforementioned job control parameters are input into the trained generalized regression neural network to obtain the initial job accuracy function. It should be noted that the initial generalized regression neural network has the same structure, input, and output as the previous generalized regression neural network, and will not be elaborated upon here. In response to the determination coefficient of the initial job accuracy function satisfying the preset verification condition, the initial job accuracy function is determined as the job accuracy function. In response to the determination coefficient of the initial job accuracy function not satisfying the preset verification condition, the smoothing factor of the initialized generalized regression neural network is further optimized using the simulated annealing (SA) algorithm, and the initialized generalized regression neural network is trained based on the optimized smoothing factor. The preset verification condition can be that the determination coefficient is greater than or equal to a preset threshold. For example, the preset threshold can be 0.89.

[0075] The second sub-step involves generating a wear function for key components based on the wear coefficient prediction method. This method can be a backpropagation (BP) neural network used to predict the wear coefficient (K) in the Archard wear model. The key component wear function characterizes the relationship between the operation control parameters of the construction robot and the average wear of the key components. The specific type of the key component is not limited; for example, it could be a transmission gear of the construction robot. The BP neural network takes the corresponding operating parameters of the construction robot as input and the wear coefficient as output. These operating parameters may include, but are not limited to, component material hardness and component movement speed. Component material hardness characterizes the hardness of the key component, and component movement speed characterizes the operating speed of the key component during construction. The BP neural network may include an input layer, a hidden layer, and an output layer. In practice, the execution entity first initializes the initial BP neural network using random initialization. Then, the initialized BP neural network is trained to obtain a trained BP neural network. Finally, the operating parameters are input into the trained BP neural network to obtain the initial wear coefficient. In response to the determination that the initial wear coefficient meets the preset wear verification condition, the initial wear coefficient is input into the Archard wear model to obtain the wear function of the key component. In response to the determination that the initial wear coefficient does not meet the preset wear verification condition, the network parameters of the BP neural network (e.g., number of network nodes, learning rate) are adjusted, and the BP neural network is trained again using the adjusted network parameters. The preset wear verification condition can be that the determination coefficient of the initial wear coefficient is greater than or equal to a preset wear threshold. For example, the preset wear threshold can be 0.9. It should be noted that the construction robot can be integrated with an incremental photoelectric encoder to detect the movement speed of the robot's components in real time.

[0076] The third sub-step involves determining the task energy efficiency function and the unit workload energy consumption function. The task energy efficiency function characterizes the relationship between the construction robot's operational control parameters and its energy efficiency. The unit workload energy consumption function characterizes the relationship between the construction robot's operational control parameters and the energy consumed to complete a unit of construction task. In practice, the executing entity can use Support Vector Regression (SVR) modeling to determine the task energy efficiency function and the unit workload energy consumption function based on the aforementioned operational control parameters.

[0077] The fifth step is to determine the constraint information based on the aforementioned comprehensive data. This constraint information characterizes the construction robot's maximum load, minimum spatial passability, and minimum stability threshold. In practice, firstly, the executing entity can extract the construction robot's load range, spatial passability range, and stability range from the comprehensive data using multi-source data fusion analysis. Then, using the aforementioned operational accuracy function and key component wear function, the constraint information is determined based on the extracted load range, spatial passability range, and stability range.

[0078] Step 6: Based on a preset hierarchical analysis method, perform importance scoring on the aforementioned operational accuracy function, key component wear function, task energy efficiency function, and unit workload energy consumption function to obtain weight vector information. The preset hierarchical analysis method can be the Analytic Hierarchy Process (AHP). The weight vector information represents the weights of the aforementioned operational accuracy function, key component wear function, task energy efficiency function, and unit workload energy consumption function in the comprehensive evaluation. In practice, firstly, the executing entity can use the analytic hierarchy process (e.g., the 1-9 scale method) to determine the weight vector information corresponding to the aforementioned operational accuracy function, key component wear function, task energy efficiency function, and unit workload energy consumption function.

[0079] Step 7: Based on the aforementioned operational accuracy function, key component wear function, task energy efficiency function, unit workload energy consumption function, constraint information, weight vector information, and a preset hierarchical genetic optimization method, generate an optimal solution set. The preset hierarchical genetic optimization method can be a hierarchical genetic algorithm. The optimal solution in the optimal solution set can represent the operational control parameter set of the construction robot that satisfies the aforementioned constraint information. Here, the specific content of the operational control parameter set of the construction robot that satisfies the aforementioned constraint information is not limited; for example, the operational control parameter set can be "robotic arm movement speed: 0.5 m / s; end effector pressure: 150 N; spraying flow rate: 300 ml / min". In practice, firstly, the aforementioned execution entity can, based on MATLAB, use a linear weighting method to fuse the aforementioned operational accuracy function, key component wear function, task energy efficiency function, and unit workload energy consumption function into a comprehensive fitness function according to the aforementioned weight vector information. Then, based on the aforementioned constraint information, randomly generate a second preset number of operational control parameter sets as the initial population. For example, the second preset number can be 50 sets. Secondly, using the aforementioned comprehensive fitness function, the fitness value of each job control parameter group in the initial population is determined. For job control parameter groups that do not meet the aforementioned constraint information, a penalty function method is used to reduce their fitness value. Finally, based on the fitness value of each job control parameter group in the initial population, a genetic algorithm is used to iterate the initial population for a predetermined number of generations to obtain the optimal solution set. For example, the predetermined number of iterations can be 100.

[0080] Step 8: Based on the aforementioned optimal solution set, generate control parameter information, and control the construction robot to perform construction processing according to the aforementioned control parameter information. In practice, the aforementioned executing entity can determine the operation control parameter group with the highest fitness value in the aforementioned optimal solution set as the control parameter information. Then, the aforementioned control parameter information is sent to the controller of the aforementioned construction robot to control the construction robot to perform construction processing according to the aforementioned control parameter information. Here, the type of the aforementioned controller is not limited and can be adjusted according to actual needs. For example, the controller type can be a PLC-based integrated controller. It should be noted that the connection method between the aforementioned executing entity and the aforementioned construction robot can be a communication connection. It should be pointed out that the aforementioned communication connection can include, but is not limited to, 3G / 4G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other currently known or future developed communication methods.

[0081] The above-described technical solution, as an inventive point of this disclosure, solves technical problem four: "poor safety in controlling construction robots, resulting in long construction times and excessive resource consumption during robot standby." The reasons for this are as follows: directly applying static parameters from engineering construction documents makes it difficult to adapt to dynamic and uncertain real-world operating environments, leading to poor safety in controlling construction robots. Furthermore, it prevents the systematic optimization of multiple conflicting objectives (e.g., low energy consumption and high efficiency), requiring repeated trial and error adjustments, resulting in long construction times and excessive resource consumption during robot standby. Solving these factors can improve the safety of controlling construction robots and shorten construction time. To achieve this, the machine-readable digital processing method for building engineering standards disclosed in this disclosure constructs the above-described operation accuracy function, key component wear function, task energy efficiency function, and unit workload energy consumption function through the generalized regression optimization method. The constructed function is evaluated for importance using the aforementioned hierarchical analysis method, and then globally optimized using the aforementioned hierarchical genetic algorithm to generate collaboratively optimal control parameters. These parameters are then used to control the construction robot for construction tasks. Therefore, the safety of controlling the construction robot during operation can be improved, construction time can be shortened, and the resources consumed by the construction robot during standby can be reduced.

[0082] The various embodiments disclosed above have the following beneficial effects: Through the machine-readable digital processing method for building engineering standards provided by some embodiments of this disclosure, the accuracy of acquiring building engineering standard information can be improved, the time spent on preliminary preparation can be shortened, and the accuracy of the acquired building engineering standard information in the construction field can be improved, as well as automation efficiency, decision-making efficiency, and collaboration efficiency. For example, in the application of construction robots, construction time can be shortened, and waste of equipment resources can be reduced. The low accuracy of obtaining construction engineering standard information, the long preparation time, the long construction time, and the significant waste of equipment resources are due to the following reasons: the operating parameters of construction robots must strictly follow construction engineering standards. PDF parsing tools have difficulty accurately identifying the multi-level nested hierarchical structure in construction engineering standards and cannot effectively associate parameter limits with safety requirements, resulting in low accuracy of information extraction from construction engineering standards. Incorrect control parameters can easily cause collisions and deformations of mechanical structures. Furthermore, PDF parsing tools have errors in recognizing special formats such as tables and formulas in construction engineering standards, such as cell misalignment, misreading formula symbols, and parameter value deviations. Repeated correction of erroneous data is required, leading to long preparation time, long construction time, and significant energy consumption due to prolonged equipment standby. Based on this, some embodiments of the present disclosure provide a machine-readable digital processing method for construction engineering standards, taking the application of construction robots as an example: First, in response to detecting the construction request information of the corresponding construction robot, based on the obtained construction engineering standard information corresponding to the construction robot, the following steps are performed: According to a first preset processing method, a text sequence information corresponding to the aforementioned construction engineering standard information is generated. Thus, the text sequence of the aforementioned construction engineering standard information can be obtained. Then, based on the preset page layout annotation information, page layout hierarchy recognition model, and preset bibliographic annotation information, the page layout information and bibliographic and citation relationship information corresponding to the aforementioned architectural engineering standard information are generated. This yields the physical structure, logical structure, and bibliographic and citation relationships of the aforementioned architectural engineering standard information. Next, based on the preset text annotation information, paragraph extraction model, and the aforementioned page layout information, the main text information corresponding to the aforementioned architectural engineering standard information is generated. This yields the main text content of the aforementioned architectural engineering standard information. Then, based on the pre-trained formula recognition model, special-format text information corresponding to the aforementioned architectural engineering standard information is generated. This yields the special-format content (e.g., tables, formulas, and images) of the aforementioned architectural engineering standard information. Finally, the aforementioned text sequence information, page layout information, bibliographic and citation relationship information, main text information, and special-format text information are determined as construction standard control information. This yields structured architectural engineering standards. Finally, based on the aforementioned construction standard control information, control parameter information corresponding to the aforementioned construction robot is generated, and based on the aforementioned control parameter information, the aforementioned construction robot is controlled to perform construction processing.Because the identification and extraction of building engineering standards are not performed using PDF parsing tools, but rather by first pre-processing unstructured building engineering standards to obtain text sequence information, the following steps are taken: First, pre-training is performed on a layout hierarchy recognition model and a paragraph extraction model using pre-set layout annotation information and pre-set text annotation information. This allows the models to accurately identify and extract information from the building engineering standards. Next, a pre-trained formula recognition model extracts special formats from the building engineering standards, thus obtaining text information with special formats. Furthermore, because the control parameters of the construction robot are determined based on the construction standard control information, and the robot is controlled according to these parameters, the accuracy of obtaining building engineering standard information is improved, the time spent on preliminary preparation is shortened, construction time is reduced, and equipment resource waste is minimized.

[0083] Further reference Figure 2 As an implementation of the methods shown in the above figures, this disclosure provides some embodiments of a machine-readable digital processing method for building engineering standards. These device embodiments are similar to... Figure 1 Corresponding to the method embodiments shown, the device can be specifically applied to various electronic devices.

[0084] like Figure 2 As shown, a machine-readable digital processing device 200 for building engineering standards in some embodiments includes an execution unit 201. The execution unit 201 is configured to, in response to detecting construction request information corresponding to a construction robot, perform the following steps based on the acquired building engineering standard information corresponding to the construction robot: generating text sequence information corresponding to the building engineering standard information according to a first preset processing method; generating page layout information and bibliographic and citation relationship information corresponding to the building engineering standard information according to preset page layout annotation information, page layout hierarchy recognition model, and preset bibliographic annotation information; generating main text information corresponding to the building engineering standard information according to preset text annotation information, paragraph extraction model, and the page layout information; generating special format text information corresponding to the building engineering standard information according to a pre-trained formula recognition model; determining the text sequence information, page layout information, bibliographic and citation relationship information, main text information, and special format text information as construction standard control information; generating control parameter information corresponding to the construction robot according to the construction standard control information; and controlling the construction robot to perform construction processing according to the control parameter information.

[0085] It is understandable that the units and references recorded in the machine-readable building engineering standard digital processing device 200 Figure 1The steps in the described method correspond accordingly. Therefore, the operations, features, and beneficial effects described above for the method also apply to the machine-readable digital processing device for building engineering standards 200 and the units contained therein, and will not be repeated here.

[0086] The following is for reference. Figure 3 It shows a schematic diagram of the structure of an electronic device 300 (e.g., a computing device) suitable for implementing some embodiments of the present disclosure. Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of this disclosure.

[0087] like Figure 3 As shown, the electronic device 300 may include a processing unit 301 (e.g., a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage device 308 into a random access memory (RAM) 303. The RAM 303 also stores various programs and data required for the operation of the electronic device 300. The processing unit 301, ROM 302, and RAM 303 are interconnected via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0088] Typically, the following devices can be connected to I / O interface 305: input devices 306 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 307 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 308 including, for example, magnetic tapes, hard disks, etc.; and communication devices 309. Communication device 309 allows electronic device 300 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 3 An electronic device 300 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively. Figure 3 Each box shown can represent a device or multiple devices as needed.

[0089] In particular, according to some embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, some embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication device 309, or installed from storage device 308, or installed from ROM 302. When the computer program is executed by processing device 301, it performs the functions defined in the methods of some embodiments of this disclosure.

[0090] It should be noted that, in some embodiments of this disclosure, the computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In some embodiments of this disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In some embodiments of this disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0091] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0092] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable medium carries one or more programs. When the electronic device executes one or more of these programs, the electronic device, in response to detecting a construction request information corresponding to a construction robot, performs the following steps based on the acquired construction engineering standard information corresponding to the construction robot: generating text sequence information corresponding to the construction engineering standard information according to a first preset processing method; generating page layout information and bibliographic and citation relationship information corresponding to the construction engineering standard information according to preset page layout annotation information, page layout hierarchy recognition model, and preset bibliographic annotation information; generating body text information corresponding to the construction engineering standard information according to preset text annotation information, paragraph extraction model, and the page layout information; generating special format text information corresponding to the construction engineering standard information according to a pre-trained formula recognition model; determining the text sequence information, page layout information, bibliographic and citation relationship information, body text information, and special format text information as construction standard control information; generating control parameter information corresponding to the construction robot based on the construction standard control information; and controlling the construction robot to perform construction processing based on the control parameter information.

[0093] Computer program code for performing operations of some embodiments of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0094] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0095] The units described in some embodiments of this disclosure can be implemented in software or in hardware. The described units can also be located in a processor, for example, and can be described as: an execution unit. In some cases, the names of these units do not constitute a limitation on the unit itself. For example, the execution unit can also be described as "a unit that, in response to detecting construction request information of a corresponding construction robot, performs the following steps based on the acquired construction engineering standard information corresponding to the construction robot: generating text sequence information corresponding to the construction engineering standard information according to a first preset processing method; generating page information and bibliographic and citation relationship information corresponding to the construction engineering standard information according to preset page annotation information, page hierarchy recognition model, and preset bibliographic annotation information; generating body text information corresponding to the construction engineering standard information according to preset text annotation information, paragraph extraction model, and the page information; generating special format text information corresponding to the construction engineering standard information according to a pre-trained formula recognition model; determining the text sequence information, page information, bibliographic and citation relationship information, body text information, and special format text information as construction standard control information; generating control parameter information corresponding to the construction robot based on the construction standard control information; and controlling the construction robot to perform construction processing based on the control parameter information."

[0096] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), and so on.

[0097] The above description is merely a selection of preferred embodiments of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A machine-readable digital processing method for building engineering standards, comprising: In response to the detection of a construction request information corresponding to the construction robot, the following steps are performed based on the obtained construction engineering standard information corresponding to the construction robot: According to the first preset processing method, a text sequence information corresponding to the building engineering standard information is generated; Based on the preset page layout annotation information, page layout hierarchy recognition model, and preset bibliographic annotation information, generate page layout information and bibliographic and citation relationship information corresponding to the architectural engineering standard information; Based on the preset text annotation information, paragraph extraction model and the page layout information, the main text information corresponding to the architectural engineering standard information is generated; Based on a pre-trained formula recognition model, special format text information corresponding to the building engineering standard information is generated; The text sequence information, the page layout information, the bibliographic and citation relationship information, the main text information, and the special format text information are identified as construction standard control information. Based on the construction standard control information, control parameter information corresponding to the construction robot is generated, and based on the control parameter information, the construction robot is controlled to perform construction processing.

2. The method according to claim 1, wherein, The step of generating page information and bibliographic and citation relationship information corresponding to the architectural engineering standard information based on preset page annotation information, page hierarchy recognition model, and preset bibliographic annotation information includes: Based on the preset page labeling information and page hierarchy recognition model, generate page information corresponding to the architectural engineering standard information; Based on the preset bibliographic annotation information and the layout information, bibliographic and citation relationship information corresponding to the architectural engineering standard information is generated.

3. The method according to claim 1, wherein, The step of generating text sequence information corresponding to the building engineering standard information according to the first preset processing method includes: According to the first preset processing method, the following steps are performed: The aforementioned building engineering standard information is modified to obtain the modified building engineering standard information, which is then used as the building engineering standard information. Based on the aforementioned building engineering standard information, region division information is generated, wherein the region division information is either text region information or non-text region information; Based on the region division information, generate recognition result information; The identification result information is modified to obtain the modified identification result information as text sequence information.

4. The method according to claim 2, wherein, The step of generating page information corresponding to the architectural engineering standard information based on preset page annotation information and page hierarchy recognition model includes: Based on the architectural engineering standard information and the preset layout annotation information, the layout hierarchy recognition model is modified to obtain the modified layout hierarchy recognition model as the layout hierarchy recognition model. Based on the layout hierarchy recognition model and the text sequence information, element classification information is generated; Based on the element classification information, text feature information is generated, wherein the text feature information includes semantic feature information, visual feature information, and topological feature information; Based on the text feature information, hierarchical information corresponding to the building engineering standard information is generated; The element classification information and the hierarchical information are determined as the layout information of the architectural engineering standard information.

5. The method according to claim 2, wherein, The step of generating bibliographic and citation relationship information corresponding to the architectural engineering standard information based on preset bibliographic annotation information and the layout information includes: Based on the preset named entity extraction method and the page layout information, generate bibliographic named entity information; Based on the preset bibliographic annotation information, location information is generated, wherein the location information includes bibliographic location information and citation location information; Based on the preset entity classification method and the location information, reference type information is generated; Based on the reference type information and the reference location information, reference link information is generated; The bibliographic named entity information and the reference link information are determined as the bibliographic and reference relationship information corresponding to the building engineering standard information.

6. The method according to claim 1, wherein, The main text information includes paragraph structure information and special text element information; as well as The step of generating the main text information corresponding to the architectural engineering standard information based on preset text annotation information, paragraph extraction model, and layout information includes: Based on the layout information and the layout hierarchy recognition model, visual feature differentiation information is generated, wherein the visual feature differentiation information includes paragraph boundary information and visual feature marking information; Based on the visual feature discrimination information and the pre-trained semantic segmentation model, accurate paragraph boundary information is generated; Based on the precise paragraph boundary information, the preset text annotation information, and the paragraph extraction model, special text element information is generated; The special text element information and the precise paragraph boundary information are determined as the main text information corresponding to the architectural engineering standard information.

7. A machine-readable digital processing device for building engineering standards, comprising: The execution unit is configured to respond to the detection of construction request information for the corresponding construction robot, and perform the following steps based on the acquired construction engineering standard information corresponding to the construction robot: generating text sequence information corresponding to the construction engineering standard information according to a first preset processing method; generating page information and bibliographic and citation relationship information corresponding to the construction engineering standard information based on preset page annotation information, page hierarchy recognition model and preset bibliographic annotation information; Based on preset text annotation information, paragraph extraction model, and page layout information, generate main text information corresponding to the architectural engineering standard information; based on pre-trained formula recognition model, generate special format text information corresponding to the architectural engineering standard information; determine the text sequence information, page layout information, bibliographic and citation relationship information, main text information, and special format text information as construction standard control information; based on the construction standard control information, generate control parameter information corresponding to the construction robot, and control the construction robot to perform construction processing based on the control parameter information.

8. An electronic device, comprising: One or more processors; A storage device on which one or more programs are stored; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1 to 6.

9. A computer-readable medium having a computer program stored thereon, wherein, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 6.

Images