Construction scheme auditing method and device, equipment and storage medium

Abstract

The invention provides a construction scheme auditing method and device, equipment and a storage medium, and relates to the technical field of building construction. The method comprises the steps of obtaining a to-be-audited construction scheme, and determining a scheme format of the to-be-audited construction scheme; determining a target analysis strategy the same as the scheme format; performing analysis processing on the to-be-audited construction scheme by using the target analysis strategy to obtain an analysis scheme corresponding to the to-be-audited construction scheme; determining a target scheme auditing strategy based on the target scheme identifier of the analysis scheme; and performing auditing processing on the analysis scheme by using the target scheme auditing strategy to obtain an auditing result corresponding to the to-be-audited construction scheme. According to the embodiment, the defects that in the prior art, manual construction scheme auditing depends on, efficiency is low, and accuracy is low are overcome, the construction scheme to be audited is automatically audited through the target auditing strategy in the auditing strategy library, the auditing result is automatically generated, and auditing efficiency and auditing accuracy are greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of building construction, and in particular to a construction plan review method, device, equipment and storage medium. Background Art

[0002] With the rapid development of the construction industry, the quality and efficiency of construction plans have become the focus of enterprises. Traditional construction plan review relies on manual work, which has problems such as low efficiency, high cost, poor consistency, and low review accuracy.

[0003] In recent years, artificial intelligence technology has made breakthrough progress in fields such as text analysis and image recognition, providing new ideas for the intelligent review of construction plans. Summary of the invention

[0004] The present invention provides a construction plan review method, device, equipment and storage medium, which are used to solve the defects of relying on manual review of construction plans in the prior art, which are low in efficiency and low in accuracy, and realize automatic review of the construction plan to be reviewed by using the target review strategy in the review strategy library, and automatically generate the review results, which greatly improves the review efficiency and the review accuracy.

[0005] In a first aspect, the present invention provides a method for reviewing a construction plan, comprising the following steps: Obtaining the construction plan to be reviewed and determining the format of the construction plan to be reviewed; Determine a target parsing strategy having the same format as the solution; Analyzing the construction plan to be reviewed using the target analysis strategy to obtain an analysis solution corresponding to the construction plan to be reviewed; Based on the target scheme identifier of the analytical scheme, the target scheme audit strategy is determined from a preset audit strategy library; wherein the audit strategy library is obtained by training the neural network multiple times using a large number of construction scheme audit samples and construction scheme to be audited samples; The target solution review strategy is used to review the analytical solution to obtain a review result corresponding to the construction solution to be reviewed.

[0006] Preferably, according to the construction plan review method provided by the present invention, the target parsing strategy at least includes: a text parsing strategy, a table parsing strategy, a picture parsing strategy, and a named entity parsing strategy; The text parsing strategy is obtained by training the natural language processing model multiple times using a large number of construction text samples and construction knowledge graphs; The table parsing strategy is obtained by training the table recognition model multiple times using a large number of construction table samples; The image parsing strategy is obtained by training the image recognition model multiple times using a large number of construction image samples; The named entity resolution strategy is obtained by training the named entity recognition model multiple times using a large number of construction entity samples.

[0007] Preferably, according to the construction plan review method provided by the present invention, the target analysis strategy is used to analyze the construction plan to be reviewed to obtain an analysis solution corresponding to the construction plan to be reviewed, including: In the case where the target parsing strategy indicates the text parsing strategy, the text parsing strategy is used to perform text feature extraction processing on the text content of the construction plan to be reviewed to obtain a text parsing solution; the text features include at least titles, keywords, and paragraphs; When the target parsing strategy indicates the table parsing strategy, the table parsing strategy is used to extract table features from the table content of the construction plan to be reviewed to obtain a table parsing solution; the table features at least include cell location, table text information, table merging relationship, and table stacking relationship; When the target parsing strategy indicates the image parsing strategy, the image parsing strategy is used to perform image feature extraction processing on the image content of the construction plan to be reviewed to obtain an image parsing solution; the image features include at least image type, construction drawings, and construction site photos; When the target resolution strategy indicates the named entity resolution strategy, the named entity resolution strategy is used to perform named entity feature extraction processing on the named entity content of the construction plan to be reviewed to obtain a named entity resolution plan; the named entity features include at least the number of construction entities, the size of the construction entities, the parameters of the construction entities, and the data of the construction entities.

[0008] Preferably, the construction scheme review method provided by the present invention includes, before the step of identifying the target scheme based on the analytical scheme and determining the target scheme review strategy from a preset review strategy library: Get an identification pair consisting of a feature identification and an audit identification; The feature identifier is added to each feature element of the analysis solution, and the audit identifier is added to each solution audit strategy in the audit strategy library; the feature identifier is used to identify the feature element of the corresponding analysis solution, and the audit identifier is used to identify the corresponding solution audit strategy.

[0009] Preferably, according to the construction scheme review method provided by the present invention, the target scheme identification at least includes a target feature identification; The target solution identifier based on the analytical solution, and determining the target solution audit strategy from a preset audit strategy library, comprises: Based on the target feature identifier of the analysis solution, determine a target audit identifier corresponding to the target feature identifier from a plurality of audit identifiers in a preset audit identifier library; Based on the target audit identifier, the corresponding target solution audit policy is determined from a plurality of solution audit policies in the audit policy library.

[0010] Preferably, according to the construction plan review method provided by the present invention, the review result at least includes a first review result and a second review result; After the step of using the target solution review strategy to review the analytical solution and obtaining a review result corresponding to the construction solution to be reviewed, the method includes: Generate a visual audit report on a preset display interface according to the construction plan to be audited, the audit results and a preset report generation template; The first review result indicates that the construction plan to be reviewed has passed the review, and the second review result indicates that the construction plan to be reviewed has not passed the review; And, when the second audit result indicates that the construction plan to be audited has not passed the audit, the reason for the failure of the construction plan to be audited, the location of the failure, and the type of the failure are added to the visual audit report.

[0011] In a second aspect, the present invention further provides a construction plan review device, comprising the following modules: A scheme format determination module is used to obtain the construction scheme to be reviewed and determine the scheme format of the construction scheme to be reviewed; A target parsing strategy determination module is used to determine a target parsing strategy having the same format as the solution; Determine an analysis solution module, which is used to analyze the construction plan to be reviewed using the target analysis strategy to obtain an analysis solution corresponding to the construction plan to be reviewed; A module for determining a target solution audit strategy is used to determine a target solution audit strategy from a preset audit strategy library based on the target solution identifier of the analytical solution; wherein the audit strategy library is obtained by training a neural network multiple times using a large number of construction solution audit samples and construction solution samples to be audited; The audit result determination module is used to audit the analytical solution using the target solution audit strategy to obtain an audit result corresponding to the construction solution to be audited.

[0012] In a third aspect, the present invention further provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, a method for reviewing a construction plan as described above is implemented.

[0013] In a fourth aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a method for reviewing a construction plan as described in any one of the above.

[0014] In a fifth aspect, the present invention further provides a computer program product, comprising a computer program, which, when executed by a processor, implements a method for reviewing a construction plan as described in any one of the above.

[0015] The construction plan audit method, device, equipment and storage medium provided by the present invention obtain the construction plan to be audited, and determine the plan format of the construction plan to be audited; determine the target parsing strategy with the same plan format; use the target parsing strategy to parse the construction plan to be audited, and obtain the parsing plan corresponding to the construction plan to be audited; based on the target plan identifier of the parsing plan, determine the target plan audit strategy from a preset audit strategy library; wherein the audit strategy library is obtained by training the neural network multiple times using a large number of construction plan audit samples and construction plan samples to be audited; use the target plan audit strategy to audit the parsing plan, and obtain the audit result corresponding to the construction plan to be audited. It is used to solve the defects of the prior art that the construction plan is manually audited, which is inefficient and has low accuracy, and realizes the automatic audit of the construction plan to be audited using the target audit strategy in the audit strategy library, and automatically generates the audit result, which greatly improves the efficiency and accuracy of the audit. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0017] Figure 1 It is one of the flow charts of the construction scheme review method provided by the present invention.

[0018] Figure 2 This is the second flow chart of the construction plan review method provided by the present invention.

[0019] Figure 3 It is a structural schematic diagram of the construction plan review device provided by the present invention.

[0020] Figure 4 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] First, several nouns involved in the present invention are analyzed: Pre-trained language model (Bidirectional Encoder Representations from Transformers, BERT): is a pre-trained language model based on the Transformer architecture for natural language processing (NLP) tasks.

[0023] Natural language processing technology based on deep learning (Generative Pre-trained Transformer, GPT): It is pre-trained on large-scale text data through unsupervised learning, and then can be fine-tuned on various natural language processing tasks to adapt to specific application scenarios.

[0024] At the heart of the GPT model is the Transformer architecture, a powerful neural network structure that can process sequence data and capture long-range dependencies. Transformer achieves this through a self-attention mechanism, allowing the model to take into account all other elements in the sequence when processing each element. This architecture is particularly suitable for natural language processing tasks because words and sentences in language often have complex contextual dependencies.

[0025] Optical Character Recognition (OCR) is a technology that converts text data in image files into electronic text. OCR obtains text information on images through optical input methods such as scanning and video recording, uses various pattern recognition algorithms to analyze the morphological characteristics of text, determines the standard code of text, and stores it in a text file in a universal format. The OCR system usually consists of three parts: preprocessing, recognition, and post-processing. The preprocessing part is responsible for inputting the text to be recognized into the system through a scanning device and performing digital image processing; the recognition part is responsible for normalizing the separated text graphics, extracting the geometric features and statistical characteristics of the text, and sending them to the recognizer for recognition; the post-processing part comprehensively considers the recognition results and certain factors of the preprocessing part, generates a recognition result with a certain format, and performs a linguistic check on the entire recognition result to correct the misrecognition components.

[0026] JavaScript Object Notation (JSON): is a lightweight data exchange format that is easy for people to read and write, and easy for machines to parse and generate.

[0027] Natural Language Processing (NLP): This model is an important branch of artificial intelligence and computer science that is dedicated to enabling computers to understand and generate human language.

[0028] Named Entity Recognition Model: Named Entity Recognition (NER) is a basic and key task in NLP, which aims to identify and classify entities with specific meanings from text. These entities can include names of people, places, organizations, dates, times, currencies, etc. Rule-based methods: Early NER systems were mainly based on rules to build vocabulary, syntactic semantic analysis, question-answering, chat, and machine translation systems. This method relies on expert knowledge and a lot of manual coding work. Statistical methods: With the development of the Internet and the emergence of a large amount of text data, statistical-based machine learning has become popular, and many NER tasks have begun to adopt machine learning algorithms. Common algorithms include conditional random fields (CRF), support vector machines (SVM), and recurrent neural networks (RNN).

[0029] Deep learning methods: In recent years, deep learning technology has achieved remarkable results in the field of NER. In particular, the introduction of the BERT model has greatly promoted the development of NER technology. BERT learns the bidirectional contextual information of language from a large-scale text corpus through pre-training, which can capture long-distance dependencies and thus improve the accuracy of NER. In addition, there are implementation methods based on models such as BiLSTM+CRF, which combine the advantages of recurrent neural networks and conditional random fields to further improve the effect of NER.

[0030] Table recognition model: is an intelligent technology based on deep learning, which is used to automatically extract and recognize table data from documents or images. TableMaster is an advanced table recognition model that simultaneously trains the position of text blocks in cells and the table structure through a multi-task learning model. The model uses a combination of convolutional neural networks (CNN) and recurrent neural networks (RNN) to achieve automatic recognition and information extraction of tables. TableStructureRec is a table structure recognition inference library that integrates multiple table recognition algorithm models, including PaddleOCR, Alibaba Optical Reading Wired and Wireless Table Recognition Algorithm Models, etc. The library improves the pre- and post-processing code of table recognition and integrates it with the OCR recognition model to ensure that the output result is a complete table and corresponding content.

[0031] In the related art, the existing intelligent review method of construction plan has the following problems: The analysis is not sophisticated enough: It is unable to effectively identify complex structures in the construction plan, such as stacked items in a table, text in a picture, etc.

[0032] Single review rules: There is a lack of differentiated review rules for different types of construction plans, which cannot meet the needs of different scenarios.

[0033] The audit method is simple: It mainly relies on keyword matching, lacks understanding of semantics and similarity judgment, resulting in a high misjudgment rate.

[0034] Combine the following Figure 1-Figure 4 The present invention describes a construction plan review method, device, equipment and storage medium, which are used to solve the defects of the prior art that construction plans rely on manual review, which is inefficient and has low accuracy. The target review strategy in the review strategy library is used to automatically review the construction plan to be reviewed, and the review results are automatically generated, which greatly improves the review efficiency and the review accuracy.

[0035] Figure 1 FIG. 1 is a flow chart of a method for reviewing a construction plan provided by the present invention. Figure 1 As shown, the method may include but is not limited to steps S100 to S500, which may specifically be: S100, obtaining a construction plan to be reviewed, and determining the format of the construction plan to be reviewed; S200, determining a target parsing strategy having the same format as the solution; S300, using the target parsing strategy to parse the construction plan to be reviewed, and obtain a parsing solution corresponding to the construction plan to be reviewed; S400, based on the target solution identifier of the analytical solution, determining the target solution audit strategy from a preset audit strategy library; wherein the audit strategy library is obtained by training a neural network multiple times using a large number of construction solution audit samples and construction solution to be audited samples; S500, utilizing the target solution review strategy to review the analysis solution, and obtaining a review result corresponding to the construction solution to be reviewed.

[0036] In step S100 of some embodiments, a construction plan to be reviewed is obtained, and a plan format of the construction plan to be reviewed is determined.

[0037] It is understandable that when the user inputs the construction plan to be reviewed into the construction plan review system, the construction plan review system automatically obtains the construction plan to be reviewed, performs file format detection on the construction plan to be reviewed, and identifies the format of the construction plan to be reviewed, such as wold format, PDF format, and picture format.

[0038] In step S200 of some embodiments, a target parsing strategy having the same format as the solution is determined.

[0039] It can be understood that after executing step S100, the specific execution steps may be: first determine the target parsing strategy for parsing the construction plan to be reviewed from the parsing strategy library, and use the determined target parsing strategy to parse the construction plan to be reviewed, and then obtain the parsing plan parsed by the parsing strategy.

[0040] Furthermore, the target parsing strategy includes at least: text parsing strategy, table parsing strategy, image parsing strategy, and named entity parsing strategy.

[0041] The text parsing strategy is obtained by training the natural language processing model multiple times using a large number of construction text samples and construction knowledge graphs.

[0042] The table parsing strategy is obtained by training the table recognition model multiple times using a large number of construction table samples.

[0043] The image parsing strategy is obtained by training the image recognition model multiple times using a large number of construction image samples.

[0044] The named entity resolution strategy is obtained by training the named entity recognition model multiple times using a large number of construction entity samples.

[0045] Furthermore, the text parsing strategy: Use pre-trained natural language processing models, such as BERT, GPT, etc., to parse the text content and extract information such as titles, keywords, paragraphs, tables, etc. The natural language processing model can recognize professional terms in the construction field, such as "beam", "column", "slab", "foundation", etc., and use it as one of the audit elements.

[0046] Table parsing strategy: Use the pre-trained table recognition model to parse the table, identify cell location, merge relationship, text content and other information, and convert it into a structured data format, such as JSON format. The table recognition model can recognize the complex table structures commonly seen in the architectural field, such as stacked items, merged cells, irregular layout, etc., and convert them into a format that is easy to process.

[0047] Image parsing strategy: Use a pre-trained image recognition (OCR) model to recognize text in images and convert them into text format. The image recognition model can recognize common image types in the construction field, such as construction drawings, site photos, etc., and extract text information from them, such as material specifications, equipment models, construction requirements, etc.

[0048] Named entity resolution strategy: Use the pre-trained named entity recognition model to perform named entity recognition on the text content, identify the professional terms in it, and use it as one of the review factors. For example, identify the number, type, size and other information of the "beam" mentioned in the construction plan, and compare it with the specification requirements.

[0049] In step S300 of some embodiments, the construction plan to be reviewed is analyzed using the target analysis strategy to obtain an analysis solution corresponding to the construction plan to be reviewed.

[0050] It is understandable that after executing step S200, the specific execution steps may be: When the target parsing strategy indicates the text parsing strategy, the text parsing strategy is used to perform text feature extraction processing on the text content of the construction plan to be reviewed to obtain a text parsing solution.

[0051] Furthermore, the text features include at least title, keywords, and paragraphs.

[0052] When the target parsing strategy indicates the table parsing strategy, the table parsing strategy is used to perform table feature extraction processing on the table content of the construction plan to be reviewed to obtain a table parsing plan.

[0053] Furthermore, the table features include at least cell location, table text information, table merging relationship, and table stacking relationship.

[0054] When the target analysis strategy indicates the image analysis strategy, the image analysis strategy is used to perform image feature extraction processing on the image content of the construction plan to be reviewed to obtain an image analysis solution.

[0055] Furthermore, the picture features include at least picture type, construction drawings, and construction site photos.

[0056] When the target resolution strategy indicates the named entity resolution strategy, the named entity resolution strategy is used to perform named entity feature extraction processing on the named entity content of the construction plan to be reviewed to obtain a named entity resolution solution.

[0057] Furthermore, the named entity features include at least the number of construction entities, the size of the construction entities, the parameters of the construction entities, and the data of the construction entities.

[0058] In some embodiments of the present invention, the steps of parsing the construction plan to be reviewed using the target parsing strategy may specifically be: During the analysis phase of the construction plan documents to be reviewed: When parsing text content, in addition to extracting information such as titles, paragraphs, tables, and pictures, it also extracts element identifiers, such as "structural components," "construction technology," and "material lists."

[0059] The parsed elements and their identifiers are stored in structured data, such as JSON format, for subsequent rule matching and similarity comparison.

[0060] In some embodiments of the present invention, before the step of identifying the target solution based on the parsing solution and determining the target solution audit policy from a preset audit policy library, the following steps are included: Get an identification pair consisting of a feature identification and an audit identification; The feature identifier is added to each feature element of the analysis solution, and the audit identifier is added to each solution audit strategy in the audit strategy library; the feature identifier is used to identify the feature element of the corresponding analysis solution, and the audit identifier is used to identify the corresponding solution audit strategy.

[0061] For example, the feature identifier is M1, and the audit identifier is N1. The feature identifier M1 and the audit identifier N1 are associated with each other, and one of the identifiers can be used to identify the other.

[0062] The feature identifier is added to each feature element of the parsing solution in advance, and the audit identifier is added to each solution audit policy in the audit policy library.

[0063] In some embodiments of the present invention, the results of the analysis scheme are classified into categories such as title, paragraph, picture, table, etc., and a label is added to each element to identify the chapter and review point to which it belongs. For example, the identified "beam" is labeled as a "structural component" category and associated with the "structural component" review point.

[0064] For example, before determining the target audit strategy, first select the audit rules related to the element from the rule library based on the parsed element identifier.

[0065] It should be noted that the rules and policies in the embodiments of the present invention have the same meaning, for example, the meaning of the audit policy is equivalent to the meaning of the audit rule.

[0066] For example, for the element "beam", the system will filter out all the review rules related to "structural components" from the rule base, such as "beam size", "beam material", "beam construction technology", etc.

[0067] The parsed element content is then compared with the screened audit rules to determine whether it meets the regulatory requirements.

[0068] This association method ensures that each element is reviewed according to the corresponding rules, thereby improving the accuracy and comprehensiveness of the review.

[0069] For example: First, the system parses the construction plan document, extracts the "beam" element and its identifier "structural component", and stores it in structured data. Secondly, based on the "structural component" identifier, the system filters out all the review rules related to "structural components" from the rule library, such as "beam size", "beam material", "beam construction process", etc. Finally, the system compares the parsed "beam" element content with the filtered review rules to determine whether it meets the specification requirements, such as checking whether the beam size meets the specification requirements, whether the material meets the specification requirements, whether the construction process meets the specification requirements, etc.

[0070] In step S400 of some embodiments, based on the target solution identifier of the analytical solution, a target solution audit strategy is determined from a preset audit strategy library; wherein the audit strategy library is obtained by training a neural network multiple times using a large number of construction solution audit samples and construction solution samples to be audited.

[0071] It is understandable that after executing step S300, the specific execution steps may be: Based on the target feature identifier of the analytical solution, a target audit identifier corresponding to the target feature identifier is determined from multiple audit identifiers in a preset audit identifier library. Based on the target audit identifier, a corresponding target solution audit strategy is determined from multiple solution audit strategies in the audit strategy library.

[0072] It should be noted that the target scheme identifier at least includes a target feature identifier.

[0073] It can be understood that, based on the target feature identifier of the parsing solution, the target audit identifier in the policy library can be identified, and then the target solution audit strategy can be determined based on the target audit identifier.

[0074] The audit policy library contains multiple program audit policies, each of which corresponds to an audit identifier, which is used to identify the corresponding program audit policy according to the audit identifier.

[0075] In step S500 of some embodiments, the target solution review strategy is used to review the analysis solution to obtain a review result corresponding to the construction solution to be reviewed.

[0076] It is understandable that after executing step S400, the specific execution step may be to audit the analysis solution using the target solution audit strategy to obtain a corresponding audit result.

[0077] Further, expert system construction: Build an expert system to formulate differentiated audit strategies based on the preparation specifications and audit requirements of different types of construction plans. The audit strategy library includes strategies in terms of laws and regulations, construction technology, material lists, construction risks, etc. For example, for scaffolding plans, formulate audit strategies in terms of scaffolding erection specifications, material specifications, safety measures, etc.; for concrete plans, formulate audit strategies in terms of concrete mix ratio, pouring technology, maintenance measures, etc.

[0078] Strategy matching: Compare the results of the solution analysis with the solution review strategy, and make a preliminary judgment on each element, such as whether it meets the format requirements, whether it contains keywords, whether it meets the construction specifications, etc. For example, check whether the correct laws and regulations are cited in the construction plan, whether the cited version is the latest, and whether the construction content meets the requirements of laws and regulations; check whether the key steps are described in the construction plan, whether safety measures are adopted, and whether appropriate construction equipment is used.

[0079] Similarity comparison: For elements that fail, further similarity comparisons are performed, such as word vector similarity, paragraph similarity, etc., to improve fault tolerance. For example, word vector models such as Word2Vec and GloVe are used to convert the text content in the construction plan into word vectors, and similarity comparisons are performed with word vectors in the professional vocabulary library to determine whether they are professional terms. For example, for the description of "beam", even if different expressions such as "beam slab" and "floor slab" are used, it can be identified that it belongs to the category of "structural component" and compared with its corresponding audit strategy. Stress value database construction: Build a stress value database, which contains the standard range of stress values ​​for different materials and structures.

[0080] The stress value range in the database is set according to the standards and experience in the construction field, such as the "Concrete Structure Design Code", "Steel Structure Design Code", etc.

[0081] Stress value identification: Extract stress value information from the construction plan, such as the stress values ​​of beams, columns, plates and other components. Extraction methods include text recognition, table parsing, etc., and can identify different types of stress values, such as compressive strength, tensile strength, etc.

[0082] Reasonable judgment of stress value: Use the pre-trained deep learning language model to analyze the extracted stress value to determine whether it is within a reasonable range and analyze the possible reasons. For example, determine whether the force on the beam exceeds its bearing capacity and analyze the possible reasons, such as insufficient material strength, too small cross-sectional size, etc.

[0083] In some embodiments of the present invention, the audit result includes at least a first audit result and a second audit result; After the step of using the target solution review strategy to review the analytical solution and obtaining a review result corresponding to the construction solution to be reviewed, the method includes: A visual audit report is generated on a preset display interface according to the construction plan to be audited, the audit results and a preset report generation template.

[0084] And, when the second audit result indicates that the construction plan to be audited has not passed the audit, the reason for the failure of the construction plan to be audited, the location of the failure, and the type of the failure are added to the visual audit report.

[0085] It should be further explained that the first review result indicates that the construction plan to be reviewed has passed the review, and the second review result indicates that the construction plan to be reviewed has not passed the review.

[0086] It is understandable that the audit report is generated: based on the construction plan to be audited, the audit results and the preset report generation template, an audit report of the same format is generated, including information such as the audit point name, audit results, error location, etc.

[0087] Visual display: The audit report is output to a visual interface for users to view and analyze.

[0088] Furthermore, a new construction plan may be generated based on the reasons for failure to pass the review, the location of failure to pass the review, and the type of failure to pass the review. After the new construction plan is generated, it may be reviewed again until it passes the review.

[0089] Construction plan generation: Using the pre-trained deep learning language model, the model automatically generates a construction plan that meets the requirements of the specification based on the construction plan template and the information entered by the user. The model can generate a complete construction plan including a catalog of chapters, construction process description, material list, construction schedule, etc. according to different construction plan types. For example, for scaffolding plans, the model can automatically generate scaffolding erection plans, including scaffolding type, erection height, safety measures, etc.; for concrete plans, the model can automatically generate concrete pouring plans, including concrete mix ratio, pouring process, maintenance measures, etc.

[0090] In some embodiments of the present invention, construction plan type recognition: Use a pre-trained deep learning model to identify the type of construction plan and display the corresponding review rules and templates. For example, identify whether the construction plan is a scaffolding plan, a concrete plan, or a steel structure plan, and display the corresponding review rules and templates, such as scaffolding erection specifications, concrete pouring technology, etc.

[0091] Audit result visualization: The audit results are presented to the user in a visual way, such as using different colors or icons to represent different audit statuses, and providing detailed error descriptions and modification suggestions. For example, green is used to indicate a passed audit, and red is used to indicate a failed audit, and detailed error descriptions and modification suggestions are provided.

[0092] Knowledge graph visualization: Build a knowledge graph in the construction field and visualize it to users, such as showing the relationship between materials, equipment, construction technology, etc. used in the construction plan, and provide relevant reference materials. For example, display information such as concrete mix ratio, pouring technology, maintenance measures, and provide relevant specifications and standards.

[0093] Audit result summary: Summarize the audit results, including the audit status of each element, reasons for failure, and other information.

[0094] Audit report generation: Based on the audit results, an audit report is generated, including an overview of the audit results, a list of failed elements, modification suggestions, etc. The audit report can clearly show the audit status of the construction plan and provide detailed modification suggestions, such as modification directions and reference specifications for failed elements.

[0095] Audit report visualization: The audit report is presented to the user in a visual way, such as using charts and tables to display the audit results, and providing interactive functions, such as viewing detailed error information, exporting audit reports, etc. For example, a bar chart is used to display the audit status of different types of elements, and a table is used to display the elements that failed, their reasons, and modification suggestions.

[0096] In some embodiments of the present invention, a construction plan review method may include but is not limited to the following embodiments: Example 1: Document analysis of construction plans to be reviewed: Identify document type: Use the large model to automatically identify the type of uploaded construction plan documents, such as reinforcement plan, concrete plan, etc.

[0097] Parsing document content: Perform preliminary analysis on the document to identify and extract different types of content such as titles, keywords, paragraphs, tables, pictures, etc.

[0098] Identify the extracted content, such as title, keyword, paragraph, table, image, etc.

[0099] Parse complex tables, identify cells stacked one above the other, and identify them using coordinates for machine understanding.

[0100] Example 2: Matching of audit strategy (rules) of construction plans to be audited: Build a rule library: Industry experts and R&D personnel will compile specifications and project experience based on the construction plan to build a review rule library for different types of plans.

[0101] Rule matching: According to the identification of the parsing result, the content is matched with the corresponding review rules.

[0102] Classify the matching results into different levels, for example: Level 1: Mandatory requirements of laws and regulations, which need to be strictly reviewed, such as the completeness and standardization of the basis for preparation.

[0103] Level 2: Basic project information, which requires general review, such as project name, location, duration, etc.

[0104] Level 3: Construction technology and methods, which require detailed review, such as the amount of steel bars used, concrete strength grade, etc.

[0105] Level 4: Other content, which requires a rough review, such as the order of the table of contents, etc.

[0106] Example 3: Content review of the construction plan to be reviewed: Short text review: For short text content such as titles and keywords, a rule-based review is first conducted to check, for example, whether keywords are missing and whether the title complies with regulations.

[0107] For content that fails the rule review, a synonym similarity comparison is performed to improve fault tolerance.

[0108] Long text review: For long text contents such as paragraphs and tables, a rule-based review is first conducted, for example, whether the construction process description is complete and whether the table data is accurate.

[0109] For content that fails the rule review, a deep learning model is used to perform semantic similarity comparison to improve accuracy and fault tolerance.

[0110] Complex form review: For complex tables, a large model is used to parse and identify the coordinates and relationships of cells.

[0111] Based on the analysis results, conduct rule-based audits, such as whether the data is within a reasonable range and whether the table structure is correct.

[0112] Example 4: Generation and visualization of audit results: Generate audit report: Based on the audit results, generate an audit report in a consistent format, including information such as audit point name, audit results, error location, etc.

[0113] Visual display: The audit report is output to a visual interface for users to view and analyze.

[0114] Example 5: Extension of the construction plan review system: Rule base expansion: With the accumulation of project experience and the advancement of technology, the rule base is continuously updated and improved, and new audit rules and types are added.

[0115] Model training: Regularly train the deep learning model to improve the accuracy of semantic understanding and similarity comparison.

[0116] Furthermore, in some embodiments of the present invention, in addition to extracting text content, element identifiers are also extracted during the document parsing stage to provide a basis for subsequent rule matching. Specifically: When parsing text content, the system will classify the parsed elements, such as titles, paragraphs, tables, pictures, etc.

[0117] Each feature is given a unique identifier to distinguish different feature types and contents.

[0118] For example, a title element can use "title" as an identifier, a paragraph element can use "paragraph" as an identifier, a table element can use "table" as an identifier, and an image element can use "image" as an identifier. In addition to the type identifier, other identifiers can be added, such as chapter identifiers and review point identifiers, to more accurately locate the chapter and review point to which the element belongs.

[0119] For example, for the "beam" element, you can add the "structural member" identifier and associate it with the "structural member" review point.

[0120] Specific implementations of the parsing identifier and the review identifier in the identifier pair include: (1) Set the parsing flag: Title: {"title", "Project Overview"} Paragraph: {"paragraph", "construction technology"} Table: {"table", "Bill of Materials"} Image: {"image", "on-site photo"} (2) Set the audit mark: Rule 1: {“Audit point”, “Structural component”} Rule 2: {“Audit point”, “Construction process”} Rule 3: {“Audit Points”, “Bill of Materials”} (3) Specific ways to associate elements with audit identifiers when conducting a policy audit: {"title", "project overview"} is associated with {"review point", "structural component"} {"paragraph", "construction technology"} is associated with {"review point", "construction technology"} {"table", "bill of materials"} is associated with {"review point", "bill of materials"} In some embodiments of the present invention, during the rule base construction phase, an expert system is constructed to formulate differentiated audit rules based on the compilation specifications and audit requirements of different types of building construction plans. The rule base includes rules on laws and regulations, construction processes, material lists, construction risks, etc., and is stored and managed in a unified format. Each rule will be added with a unique identifier to distinguish different rule types and contents.

[0121] For example, the review rules for scaffolding plans can use "scaffolding erection specifications" as a label, and the review rules for concrete plans can use "concrete pouring technology" as a label.

[0122] Through the embodiments provided above, the present invention can at least achieve the following technical effects: Fine parsing: It can effectively identify complex structures in construction plans, such as stacked items in tables, text in pictures, etc., to improve the accuracy of parsing.

[0123] Differentiated audit rules: Differentiated audit rules are formulated for different types of construction plans to meet the needs of different scenarios.

[0124] Diversified audit methods: Combining multiple methods such as rule audit, similarity comparison and large model audit to improve the accuracy and fault tolerance of the audit.

[0125] High efficiency: Automated review process saves labor costs and improves review efficiency.

[0126] In some embodiments, the audit rules are divided into multiple levels, such as mandatory requirements of laws and regulations, basic project information, construction technology and methods, and other content, and different audit methods and levels of strictness are adopted according to different levels.

[0127] When parsing the document content, each content is identified and matched with the audit rules of the corresponding level to achieve refined auditing.

[0128] In some embodiments, complex table parsing: For complex tables commonly seen in documents in the architectural field, such as cells stacked up and down, merged cells, etc., a large model is used to parse them, identify the coordinates and relationships of the cells, and use the coordinates to mark them for machine understanding.

[0129] Based on the analysis results, conduct rule-based audits, such as whether the data is within a reasonable range and whether the table structure is correct.

[0130] In some embodiments, the (stress value) database reviews: Build a (stress value) database to review the rationality of the stress values ​​in the construction plan, such as whether they are within a reasonable range and whether they are consistent with the design parameters.

[0131] The database can be expanded and updated according to different types of construction schemes and structure types.

[0132] In some embodiments of the present invention, Figure 2 The second schematic diagram of the construction plan review method provided by the embodiment of the present invention is shown. The construction plan review method may at least include but is not limited to the following parts: Part I: Classification of construction plans to be reviewed, that is, classification of the types of construction plans to be reviewed, which may at least include but are not limited to high formwork plans, scaffolding plans, and concrete plans.

[0133] Part II: The review rules for the construction plans to be reviewed are graded, also known as the review strategy grading, which can at least include but are not limited to strict review, loose review, and threshold review. Furthermore, strict review can include but are not limited to legal and regulatory strategy review, and standard strategy review; loose review can include but are not limited to construction plan strategy review, and schedule strategy review; threshold strategy review can include but are not limited to stress value strategy review and steel bar quantity strategy review.

[0134] Part III: Classification of elements of the construction plan to be reviewed, including but not limited to titles, keywords, paragraphs, tables, and charts.

[0135] Part 4: Review process, which may include but is not limited to short text review, long text review, table review, image review and other review processes.

[0136] Furthermore, the short text review process of the construction plan to be reviewed first conducts a rule review, and then conducts a word vector similarity comparison review to review the short text of the construction plan to be reviewed.

[0137] The long text review process of the construction plan to be reviewed first conducts a rule review, followed by a semantic understanding review and a similarity comparison review to review the long text of the construction plan to be reviewed.

[0138] The process of table review of construction plans to be reviewed is to first conduct a rule review and then a large-scale analytical review to review the tables of construction plans to be reviewed.

[0139] The image review process of the construction plan to be reviewed first conducts rule review, then uses the target detection model to identify the target image, and then realizes the review of the images in the construction plan to be reviewed.

[0140] Part 5: Generate audit results, which may include but are not limited to the generation of audit reports, the screening of content that has passed the audit, the screening of content that has failed the audit, and the display of the location of errors in the audit steps and the causes of the errors.

[0141] The construction plan audit method, device, equipment and storage medium provided by the present invention obtain the construction plan to be audited, and determine the plan format of the construction plan to be audited; determine the target parsing strategy with the same plan format; use the target parsing strategy to parse the construction plan to be audited, and obtain the parsing plan corresponding to the construction plan to be audited; based on the target plan identifier of the parsing plan, determine the target plan audit strategy from a preset audit strategy library; wherein the audit strategy library is obtained by training the neural network multiple times using a large number of construction plan audit samples and construction plan samples to be audited; use the target plan audit strategy to audit the parsing plan, and obtain the audit result corresponding to the construction plan to be audited. It is used to solve the defects of the prior art that the construction plan is manually audited, which is inefficient and has low accuracy, and realizes the automatic audit of the construction plan to be audited using the target audit strategy in the audit strategy library, and automatically generates the audit result, which greatly improves the efficiency and accuracy of the audit.

[0142] The review device for the construction plan provided by the present invention is described below. The review device for the construction plan described below and the review method for the construction plan described above can be referenced to each other.

[0143] like Figure 3 The figure is a schematic diagram of the structure of the construction plan review device provided by the present invention, which includes the following modules: The scheme format determination module 310 is used to obtain the construction scheme to be reviewed and determine the scheme format of the construction scheme to be reviewed; A target resolution strategy determination module 320 is used to determine a target resolution strategy having the same format as the solution; A determination analysis solution module 330 is used to analyze the construction solution to be reviewed using the target analysis strategy to obtain an analysis solution corresponding to the construction solution to be reviewed; The target scheme audit strategy determination module 340 is used to determine the target scheme audit strategy from a preset audit strategy library based on the target scheme identifier of the analytical scheme; wherein the audit strategy library is obtained by training the neural network multiple times using a large number of construction scheme audit samples and construction scheme to be audited samples; The audit result determination module 350 is used to audit the analytical solution using the target solution audit strategy to obtain an audit result corresponding to the construction solution to be audited.

[0144] According to the construction plan review device provided by the present invention, the target parsing strategy at least includes: text parsing strategy, table parsing strategy, image parsing strategy, and named entity parsing strategy; The text parsing strategy is obtained by training the natural language processing model multiple times using a large number of construction text samples and construction knowledge graphs; The table parsing strategy is obtained by training the table recognition model multiple times using a large number of construction table samples; The image parsing strategy is obtained by training the image recognition model multiple times using a large number of construction image samples; The named entity resolution strategy is obtained by training the named entity recognition model multiple times using a large number of construction entity samples.

[0145] According to the construction plan review device provided by the present invention, the analysis plan module 330 is specifically used to extract text features from the text content of the construction plan to be reviewed using the text analysis strategy to obtain a text analysis plan when the target analysis strategy indicates that the text analysis strategy is the text analysis strategy; the text features include at least titles, keywords, and paragraphs; When the target parsing strategy indicates the table parsing strategy, the table parsing strategy is used to extract table features from the table content of the construction plan to be reviewed to obtain a table parsing solution; the table features at least include cell location, table text information, table merging relationship, and table stacking relationship; When the target parsing strategy indicates the image parsing strategy, the image parsing strategy is used to perform image feature extraction processing on the image content of the construction plan to be reviewed to obtain an image parsing solution; the image features include at least image type, construction drawings, and construction site photos; When the target resolution strategy indicates the named entity resolution strategy, the named entity resolution strategy is used to perform named entity feature extraction processing on the named entity content of the construction plan to be reviewed to obtain a named entity resolution plan; the named entity features include at least the number of construction entities, the size of the construction entities, the parameters of the construction entities, and the data of the construction entities.

[0146] According to the construction plan review device provided by the present invention, before the step of identifying the target plan based on the analytical plan and determining the target plan review strategy from a preset review strategy library, it is specifically used to obtain an identification pair consisting of a feature identification and an audit identification; The feature identifier is added to each feature element of the analysis solution, and the audit identifier is added to each solution audit strategy in the audit strategy library; the feature identifier is used to identify the feature element of the corresponding analysis solution, and the audit identifier is used to identify the corresponding solution audit strategy.

[0147] According to the construction plan review device provided by the present invention, the target plan identifier at least includes a target feature identifier; The target scheme audit strategy determination module 340 is specifically used to determine the target audit identifier corresponding to the target feature identifier from a plurality of audit identifiers in a preset audit identifier library based on the target feature identifier of the analytical scheme; Based on the target audit identifier, the corresponding target solution audit policy is determined from a plurality of solution audit policies in the audit policy library.

[0148] According to the construction plan review device provided by the present invention, the review result includes at least a first review result and a second review result; After the step of using the target scheme audit strategy to audit the analytical scheme and obtaining an audit result corresponding to the construction scheme to be audited, the device is specifically used to generate a visual audit report on a preset display interface according to the construction scheme to be audited, the audit result and a preset report generation template; The first review result indicates that the construction plan to be reviewed has passed the review, and the second review result indicates that the construction plan to be reviewed has not passed the review; And, when the second audit result indicates that the construction plan to be audited has not passed the audit, the reason for the failure of the construction plan to be audited, the location of the failure, and the type of the failure are added to the visual audit report.

[0149] The construction plan audit method, device, equipment and storage medium provided by the present invention obtain the construction plan to be audited, and determine the plan format of the construction plan to be audited; determine the target parsing strategy with the same plan format; use the target parsing strategy to parse the construction plan to be audited, and obtain the parsing plan corresponding to the construction plan to be audited; based on the target plan identifier of the parsing plan, determine the target plan audit strategy from a preset audit strategy library; wherein the audit strategy library is obtained by training the neural network multiple times using a large number of construction plan audit samples and construction plan samples to be audited; use the target plan audit strategy to audit the parsing plan, and obtain the audit result corresponding to the construction plan to be audited. It is used to solve the defects of the prior art that the construction plan is manually audited, which is inefficient and has low accuracy, and realizes the automatic audit of the construction plan to be audited using the target audit strategy in the audit strategy library, and automatically generates the audit result, which greatly improves the efficiency and accuracy of the audit.

[0150] Figure 4 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 4As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430 and a communication bus 440, wherein the processor 410, the communication interface 420 and the memory 430 communicate with each other through the communication bus 440. The processor 410 may call the logic instructions in the memory 430 to execute the construction plan review method, which includes: obtaining the construction plan to be reviewed and determining the plan format of the construction plan to be reviewed; determining the target parsing strategy with the same plan format; parsing the construction plan to be reviewed using the target parsing strategy to obtain a parsing solution corresponding to the construction plan to be reviewed; determining the target plan review strategy from a preset review strategy library based on the target plan identifier of the parsing solution; wherein the review strategy library is obtained by training the neural network multiple times using a large number of construction plan review samples and construction plan samples to be reviewed; and auditing the parsing solution using the target plan review strategy to obtain a review result corresponding to the construction plan to be reviewed.

[0151] In addition, the logic instructions in the above-mentioned memory 430 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0152] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the construction plan review method provided by the above-mentioned methods, which method includes: obtaining a construction plan to be reviewed, and determining the plan format of the construction plan to be reviewed; determining a target parsing strategy with the same plan format; using the target parsing strategy to parse the construction plan to be reviewed, and obtain a parsing plan corresponding to the construction plan to be reviewed; based on the target plan identifier of the parsing plan, determining a target plan audit strategy from a preset audit strategy library; wherein the audit strategy library is obtained by multiple training of a neural network using a large number of construction plan review samples and construction plan samples to be reviewed; using the target plan audit strategy to audit the parsing plan, and obtain an audit result corresponding to the construction plan to be reviewed.

[0153] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a construction plan review method provided by the above-mentioned methods, the method comprising: obtaining a construction plan to be reviewed, and determining a plan format of the construction plan to be reviewed; determining a target parsing strategy with the same plan format; parsing the construction plan to be reviewed using the target parsing strategy to obtain a parsing plan corresponding to the construction plan to be reviewed; determining a target plan review strategy from a preset review strategy library based on a target plan identifier of the parsing plan; wherein the review strategy library is obtained by training a neural network multiple times using a large number of construction plan review samples and construction plan samples to be reviewed; and reviewing the parsing plan using the target plan review strategy to obtain a review result corresponding to the construction plan to be reviewed.

[0154] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0155] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0156] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A construction plan review method, characterized in that: include: Obtaining the construction plan to be reviewed and determining the format of the construction plan to be reviewed; Determine a target parsing strategy having the same format as the solution; Analyzing the construction plan to be reviewed using the target analysis strategy to obtain an analysis solution corresponding to the construction plan to be reviewed; Based on the target scheme identifier of the analytical scheme, the target scheme audit strategy is determined from a preset audit strategy library; wherein the audit strategy library is obtained by training the neural network multiple times using a large number of construction scheme audit samples and construction scheme to be audited samples; The target solution review strategy is used to review the analytical solution to obtain a review result corresponding to the construction solution to be reviewed.

2. The construction plan review method according to claim 1, characterized in that: The target parsing strategy includes at least: text parsing strategy, table parsing strategy, image parsing strategy, and named entity parsing strategy; The text parsing strategy is obtained by training the natural language processing model multiple times using a large number of construction text samples and construction knowledge graphs; The table parsing strategy is obtained by training the table recognition model multiple times using a large number of construction table samples; The image parsing strategy is obtained by training the image recognition model multiple times using a large number of construction image samples; The named entity resolution strategy is obtained by training the named entity recognition model multiple times using a large number of construction entity samples.

3. The construction plan review method according to claim 2, characterized in that: The method of using the target parsing strategy to parse the construction plan to be reviewed to obtain a parsing solution corresponding to the construction plan to be reviewed includes: In the case where the target parsing strategy indicates the text parsing strategy, the text parsing strategy is used to perform text feature extraction processing on the text content of the construction plan to be reviewed to obtain a text parsing solution; the text features include at least titles, keywords, and paragraphs; When the target parsing strategy indicates the table parsing strategy, the table parsing strategy is used to extract table features from the table content of the construction plan to be reviewed to obtain a table parsing solution; the table features at least include cell location, table text information, table merging relationship, and table stacking relationship; When the target parsing strategy indicates the image parsing strategy, the image parsing strategy is used to perform image feature extraction processing on the image content of the construction plan to be reviewed to obtain an image parsing solution; the image features include at least image type, construction drawings, and construction site photos; When the target resolution strategy indicates the named entity resolution strategy, the named entity resolution strategy is used to perform named entity feature extraction processing on the named entity content of the construction plan to be reviewed to obtain a named entity resolution plan; the named entity features include at least the number of construction entities, the size of the construction entities, the parameters of the construction entities, and the data of the construction entities.

4. The construction plan review method according to claim 1, characterized in that: Before the step of identifying the target solution based on the parsing solution and determining the target solution audit policy from a preset audit policy library, the method includes: Get an identification pair consisting of a feature identification and an audit identification; The feature identifier is added to each feature element of the analysis solution, and the audit identifier is added to each solution audit strategy in the audit strategy library; the feature identifier is used to identify the feature element of the corresponding analysis solution, and the audit identifier is used to identify the corresponding solution audit strategy.

5. The construction plan review method according to claim 4, characterized in that: The target scheme identification includes at least a target feature identification; The target solution identifier based on the analytical solution, and determining the target solution audit strategy from a preset audit strategy library, comprises: Based on the target feature identifier of the analysis solution, determine a target audit identifier corresponding to the target feature identifier from a plurality of audit identifiers in a preset audit identifier library; Based on the target audit identifier, the corresponding target solution audit policy is determined from a plurality of solution audit policies in the audit policy library.

6. The construction plan review method according to any one of claims 1 to 5, characterized in that: The audit results at least include a first audit result and a second audit result; After the step of using the target solution review strategy to review the analytical solution and obtaining a review result corresponding to the construction solution to be reviewed, the method includes: Generate a visual audit report on a preset display interface according to the construction plan to be audited, the audit results and a preset report generation template; The first review result indicates that the construction plan to be reviewed has passed the review, and the second review result indicates that the construction plan to be reviewed has not passed the review; And, when the second audit result indicates that the construction plan to be audited has not passed the audit, the reason for the failure of the construction plan to be audited, the location of the failure, and the type of the failure are added to the visual audit report.

7. A construction plan review device, characterized in that: include: A scheme format determination module is used to obtain the construction scheme to be reviewed and determine the scheme format of the construction scheme to be reviewed; A target parsing strategy determination module is used to determine a target parsing strategy having the same format as the solution; Determine an analysis solution module, which is used to analyze the construction plan to be reviewed using the target analysis strategy to obtain an analysis solution corresponding to the construction plan to be reviewed; A module for determining a target solution audit strategy is used to determine a target solution audit strategy from a preset audit strategy library based on the target solution identifier of the analytical solution; wherein the audit strategy library is obtained by training a neural network multiple times using a large number of construction solution audit samples and construction solution samples to be audited; The audit result determination module is used to audit the analytical solution using the target solution audit strategy to obtain an audit result corresponding to the construction solution to be audited.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the program, the review method of the construction plan as described in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for reviewing the construction plan as described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for reviewing the construction plan as described in any one of claims 1 to 6 is implemented.

Images