Construction progress plan review method and device, computer equipment and storage medium

Abstract

The invention relates to the technical field of artificial intelligence, and discloses a construction progress plan review method and device, computer equipment and a storage medium, and the method comprises the steps: carrying out the recognition processing of a to-be-reviewed bidding file, and determining the first work content contained in a progress plan crossover chart in the bidding file; determining first unit engineering information in the engineering quantity list file; inputting the first work content and the first unit engineering information into a preset review large model, and determining a matching relationship between the first work content and the first unit engineering information according to an output result of the review large model; and generating a review report of the construction progress plan according to the matching relationship. The working efficiency of progress plan review can be remarkably improved, human errors and subjective deviations are reduced, and the accuracy and consistency of progress plan review can be ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, in particular to a construction progress plan review method and device, a computer device and a storage medium. BACKGROUND

[0002] In the bid evaluation process of construction organization design, the completeness of work content in the progress plan Gantt chart is one of the dimensions that the bid evaluation committee focuses on. The completeness of work content directly affects the judgment of the feasibility and reasonableness of the bid scheme. By checking whether the progress plan Gantt chart covers all construction stages, key processes and professional engineering required in the bidding document or construction contract, important work content is avoided.

[0003] In the bid evaluation of engineering projects, the progress plan generally adopts the traditional manual bid evaluation method, and the progress plan and the like are judged subjectively by professional evaluators, which not only consumes a long time, but also the evaluation scales may not be consistent. SUMMARY

[0004] Therefore, the present application provides a construction progress plan review method and device, a computer device and a storage medium to solve the problem of low and inconsistent construction progress review efficiency.

[0005] In a first aspect, the present application provides a construction progress plan review method, comprising:

[0006] identifying and processing the bid document to be evaluated to determine the first work content contained in the progress plan Gantt chart in the bid document;

[0007] determining the first unit engineering information in the bill of quantities file;

[0008] inputting the first work content and the first unit engineering information into a pre-set review large model, and determining the matching relationship between the first work content and the first unit engineering information according to the output result of the review large model;

[0009] generating a review report of the construction progress plan according to the matching relationship.

[0010] In some optional embodiments, the identifying and processing of the bid document to be evaluated to determine the first work content contained in the progress plan Gantt chart in the bid document comprises:

[0011] converting the bid document to be evaluated into an original image;

[0012] image enhancement is performed on the original image to generate a target image;

[0013] detecting the target image to extract the progress plan Gantt chart;

[0014] content recognition is performed on the progress plan Gantt chart to determine first work content contained in the progress plan Gantt chart.

[0015] In some optional embodiments, the content recognition performed on the progress plan Gantt chart to determine first work content contained in the progress plan Gantt chart comprises:

[0016] table line detection is performed on the progress plan Gantt chart to determine table line information in the progress plan Gantt chart;

[0017] According to the table line information, corresponding cells are located;

[0018] content recognition is performed on each of the cells to extract the first work content contained in the progress plan Gantt chart.

[0019] In some optional embodiments, the training process of the review large model comprises:

[0020] According to a professional knowledge base in the field of construction, a preset large language model is pre-trained;

[0021] According to the constructed question and answer pair data, the large language model is fine-tuned; the question and answer pair data is used to represent a matching relationship between work content and unit engineering information;

[0022] The fine-tuned large language model is used as a review large model for reviewing construction progress plans.

[0023] In some optional embodiments, the process of constructing the question and answer pair data comprises:

[0024] determining second work content and second unit engineering information, and constructing a question for determining a matching relationship between the second work content and the second unit engineering information;

[0025] inputting the question into a preset question and answer large model, and determining an answer representing a matching relationship between the second work content and the second unit engineering information according to an output result of the question and answer large model; the number of model parameters of the question and answer large model is greater than the number of model parameters of the review large model;

[0026] According to the question and the answer, corresponding question and answer pair data is constructed.

[0027] In some optional embodiments, the inputting of the first work content and the first unit engineering information into a preset review large model comprises:

[0028] For each piece of the first unit project information, a corresponding prompt word template is constructed; the prompt word template includes the first work content, the first unit project information, and a task instruction for instructing the review model to determine a matching relationship between the first work content and the first unit project information;

[0029] The prompt word templates corresponding to each of the first unit project information are respectively input into the preset review model to respectively determine the matching relationship between each of the first unit project information and the first work content.

[0030] In some optional implementations, generating a review report of the construction schedule based on the matching relationship includes:

[0031] Determine the first unit project information missing from the bidding document and / or the first newly added work content according to the matching relationship;

[0032] Generate a review report of the construction progress plan; the review report includes the missing first unit project information and / or the newly added first work content.

[0033] In a second aspect, the present invention provides a construction schedule review device, comprising:

[0034] A content identification module is used to identify and process the bidding document to be reviewed, and determine the first work content included in the progress plan bar chart in the bidding document;

[0035] An information extraction module, configured to determine the first unit engineering information in the bill of quantities file;

[0036] a review module, configured to input the first work content and the first unit project information into a preset review model, and determine a matching relationship between the first work content and the first unit project information based on an output result of the review model;

[0037] A report module is used to generate a review report of the construction schedule based on the matching relationship.

[0038] In a third aspect, the present invention provides a computer device comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the construction schedule review method of the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0039] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the construction schedule review method of the first aspect or any corresponding embodiment thereof.

[0040] In a fifth aspect, the present application provides a computer program product comprising computer instructions for causing a computer to perform the construction progress plan review method of the first aspect or any of its corresponding embodiments.

[0041] The present application identifies the work content related to the construction progress plan from the progress plan Gantt chart of the bid document, extracts the unit project information from the bill of quantities document, uses the unit project information as the review standard, and automatically performs correlation analysis on the work content using a large language model with review capability, thereby realizing the review of the construction progress plan. The use of information extraction technology and large language model technology can significantly improve the work efficiency of progress plan review. Moreover, the use of objective large language models to judge and analyze the progress plan reduces human error and subjective bias, ensuring the accuracy and consistency of progress plan review. The automatically generated review report not only provides accurate review results for review experts quickly, but also assists review experts in making more comprehensive and accurate decisions. BRIEF DESCRIPTION OF DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the specific embodiments or related art, the following will briefly introduce the drawings needed to be used in the specific embodiments or related art descriptions. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0043] Figure 1 is a flowchart of a construction progress plan review method according to an embodiment of the present application;

[0044] Figure 2 is a process diagram of review by a review large model according to an embodiment of the present application;

[0045] Figure 3 is a flowchart of another construction progress plan review method according to an embodiment of the present application;

[0046] Figure 4 is a schematic diagram of a progress plan Gantt chart according to an embodiment of the present application;

[0047] Figure 5 is a schematic diagram of a review report according to an embodiment of the present application;

[0048] Figure 6 is a structural block diagram of a construction progress plan review device according to an embodiment of the present application;

[0049] Figure 7Fig. 1 is a schematic diagram of a hardware structure of a computer device according to an embodiment of the present application. DETAILED DESCRIPTION

[0050] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of the present application.

[0051] For the convenience of description, first, some terms related to the present embodiment are explained. These explanations are to make the embodiments of the present application easier to be understood, and should not be regarded as the limitation on the protection scope required by the present application.

[0052] (1) Tender documents: The tender documents are the outline of the tender engineering construction, the guiding documents prepared and publicly published by the tenderer for the engineering project tender, and provide all the information required for the bidding units to participate in the bidding. The tender documents include the tender announcement, the tenderer's guide, the evaluation method and the engineering quantity list documents, etc.

[0053] (2) Engineering quantity list documents: The engineering quantity list documents are a part of the tender documents, which include the detailed list of each unit project in the proposed engineering, and mark the characteristic description, the measurement unit and the engineering quantity of each unit project, and are the important component of the tender documents.

[0054] (3) Unit project: It is the core division unit of the engineering quantity list documents, and is usually divided according to the specialty or function (such as the civil engineering, the mechanical and electrical engineering, the outdoor engineering, etc.), and contains the detailed engineering quantity of the sub-part item project, the measure item, etc. The bidding unit needs to reflect the construction time, the sequence and the resource allocation of the unit project (or further refined into the sub-part item project) in the schedule plan.

[0055] (4) Bidding documents: The bidding documents refer to the responsive documents prepared by the bidding unit according to the requirements of the tender documents, and are generally composed of the commercial documents, the technical documents and other parts. The technical documents (technical bid) contain the content related to the construction schedule plan, and are attached with the detailed schedule plan Gantt chart.

[0056] (5) Schedule plan Gantt chart: It can also be called the schedule plan Gantt chart, and is a kind of chart tool for directly representing the start time, the duration, the end time and the relationship (simple relationship) between activities (tasks) of the project on the time coordinate axis by using the horizontal bar chart. In the field of building engineering, it is one of the most widely used construction schedule plan expression forms.

[0057] In the process of evaluating the construction organization design, the completeness of the work content in the progress plan Gantt chart is one of the dimensions that the evaluation committee focuses on. The completeness of the work content directly affects the judgment of the feasibility and reasonableness of the bidding scheme. By checking whether the progress plan Gantt chart covers all construction stages, key processes and professional engineering required in the bidding document or construction contract, important work content is avoided. Missing key processes may lead to inaccurate duration estimates or conflicts in subsequent construction arrangements; in addition, if the work content is incomplete, it may imply the risk of insufficient understanding, lack of experience or casual attitude of the bidding unit.

[0058] The current progress plan generally adopts the traditional manual evaluation method, which is subjectively evaluated by professional evaluators according to the substantive requirements in the bidding document, so the more bidders there are, the longer the evaluation time is, and the progress plan review process will consume a lot of time and human resources. And the progress plan review is largely dependent on the responsibility of the evaluation experts and the level of their knowledge of construction engineering technology and equipment and their management methods. Evaluation experts need to make different judgments according to different project evaluation standards, and the completeness of the progress plan is completely judged by the evaluation experts manually, and the evaluation experts are easily influenced by historical experience, habits and subjective feelings, and the evaluation standards for the same type of project may not be consistent, and even there may be wrong evaluation and missed evaluation.

[0059] The construction progress plan evaluation method provided by the embodiment of the present application determines the corresponding unit engineering information by using the bill of quantities file, and performs correlation matching on the unit engineering information and the work content extracted from the progress plan Gantt chart by using a pre-set evaluation model, so that the evaluation result of the construction progress plan can be automatically obtained. This process does not require human intervention, is not only efficient, but also has uniform evaluation standards.

[0060] According to the embodiment of the present application, a construction progress plan evaluation method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a group of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0061] In this embodiment, a construction progress plan evaluation method is provided, which can be applied to devices with processing capabilities, such as personal computers or mobile terminals. Figure 1 The flowchart of the construction progress plan evaluation method according to the embodiment of the present application is shown in FIG. Figure 1 As shown in the figure, the flowchart includes the following steps.

[0062] Step S101, identifying and processing the bidding document to be evaluated to determine the first work content contained in the progress plan Gantt chart in the bidding document.

[0063] In this embodiment, in the bid evaluation stage of the construction project, the bid document that needs to be reviewed can be determined, and the progress plan Gantt chart in the bid document can be identified, which can directly represent the relevant information of the construction progress plan. By identifying the progress plan Gantt chart, the work content contained therein can be determined. For ease of distinction, the work content identified at this time is referred to as first work content.

[0064] The work content is the content related to the construction plan determined by the bidder according to the bidding requirements, and it is the content that needs to be reviewed. For example, the work content can include construction preparation, earthwork excavation, reinforcement installation, exterior decoration engineering, door and window installation engineering, etc., which is based on the actual situation.

[0065] In step S102, the first unit engineering information in the bill of quantities document is determined.

[0066] The bidding document corresponding to the above bid document contains the bill of quantities document, and the bill of quantities document contains the relevant information of each unit engineering, i.e., unit engineering information. For ease of distinction, the unit engineering information is referred to as first unit engineering information.

[0067] The unit engineering information can specifically include the name of the unit engineering, and can also include cost information, etc. Since the bill of quantities document generally adopts a standard format, such as XML format, etc., the first unit engineering information can be easily extracted from the standard format bill of quantities document.

[0068] For example, the bill of quantities document can include civil decoration engineering, water supply and drainage engineering, heating engineering, fire protection engineering, etc. The specific unit engineering is based on the actual situation.

[0069] The bill of quantities document is generally mainly used for cost accounting. In this embodiment, the unit engineering information in the bill of quantities document is extracted, which is complete and easy to extract, and is used as a review standard. The implementation is simple, and it is also conducive to subsequent comprehensive review of the construction progress plan of the bid document to avoid omissions.

[0070] In step S103, the first work content and the first unit engineering information are input into a pre-set review large model, and the matching relationship between the first work content and the first unit engineering information is determined according to the output result of the review large model.

[0071] In this embodiment, a large language model (LLM) for reviewing the construction progress plan in the bid document is pre-set, i.e., a review large model.

[0072] After determining the first work content and the first unit engineering information, the first work content and the first unit engineering information are input as input data of the review large model into the review large model, and the review large model can evaluate the correlation between the first work content and the first unit engineering information, and further determine the matching relationship therebetween.

[0073] For example, the matching relationship between the first work content and the first unit engineering information can represent that the two can match or the two cannot match.

[0074] In step S104, a review report of the construction progress plan is generated according to the matching relationship.

[0075] In this embodiment, the progress plan Gantt chart generally contains multiple work contents, that is, the number of first work contents is generally multiple; similarly, the unit engineering in the bill of quantities file is also generally multiple, that is, multiple first unit engineering information can be identified. According to the matching relationship between each first work content and the first unit engineering information, a detailed review report representing the construction progress plan can be generated, and based on the review report, it can be determined whether each unit engineering in the bill of quantities file is satisfied, thereby realizing the review of the construction progress plan in the bid document.

[0076] Figure 2 A process schematic diagram of the review based on the review large model is shown. As shown in Figure 2 As shown, each unit engineering, such as unit engineering 1, unit engineering 2, and unit engineering 3, can be identified from the bill of quantities file, and each work content, such as work content a, work content b, and work content c, can be identified from the progress plan Gantt chart; the review large model is used to review these unit engineering and work content, so as to determine the corresponding matching relationship, and finally generate a review report.

[0077] The review method of the construction progress plan provided in this embodiment identifies the work content related to the construction progress plan from the progress plan Gantt chart of the bid document, extracts the unit engineering information from the bill of quantities file, uses the unit engineering information as the review standard, and uses the large language model with review capability to automatically analyze the correlation of the work content, thereby realizing the review of the construction progress plan. Using information extraction technology and large language model technology can significantly improve the work efficiency of progress plan review; moreover, using objective large language model to judge and analyze the progress plan can reduce human errors and subjective bias, thereby ensuring the accuracy and consistency of the progress plan review; the automatically generated review report can not only provide accurate review results for the review experts, but also assist the review experts to make more comprehensive and accurate decisions.

[0078] The construction progress plan review method provided in the embodiment can be applied to a device with processing capability, such as a personal computer or a mobile terminal. Figure 3 is a flowchart of the construction progress plan review method according to an embodiment of the present application, as shown in the figure, the flow includes the following steps. Figure 3

[0079] In step S301, the bidding document to be reviewed is subjected to identification processing, and first work content contained in the progress plan Gantt chart in the bidding document is determined.

[0080] For details, refer to step S101 of the embodiment shown in Figure 1 , which will not be described here again.

[0081] In some optional embodiments, the above step S301 “the bidding document to be reviewed is subjected to identification processing, and first work content contained in the progress plan Gantt chart in the bidding document is determined” can include steps A1 to A4.

[0082] In step A1, the bidding document to be reviewed is converted into an original image.

[0083] In step A2, the original image is subjected to image enhancement to generate a target image.

[0084] In step A3, the target image is detected to extract the progress plan Gantt chart.

[0085] In step A4, the progress plan Gantt chart is subjected to content identification to determine the first work content contained in the progress plan Gantt chart.

[0086] In the embodiment, the bidding document to be reviewed is first subjected to data preprocessing. The bidding document is generally a PDF file. To facilitate identification of the progress plan Gantt chart and the work content therein, the bidding document (PDF) can be uniformly converted into a high-resolution image (such as a PNG or JPG image), i.e., an original image.

[0087] The original image is subjected to image enhancement to generate an enhanced image, i.e., a target image. For example, the watermark interference in the original image can be eliminated, and the original image is subjected to image enhancement using a contrast enhancement algorithm, a brightness correction algorithm, etc., so that the accuracy of content identification of the progress plan Gantt chart can be improved. The bidding document contains a large amount of content, which is converted into multiple original images. For example, each page of the bidding document is converted into a corresponding original image and subjected to image enhancement.

[0088] ​For the target image obtained after preprocessing, whether the progress plan horizontal graph is contained is detected respectively, and then the progress plan horizontal graph is extracted. For example, a detection model of a YOLO (You Only Look Once, i.e. only look once) series can be used to detect the target image to determine whether the progress plan horizontal graph is contained; if the progress plan horizontal graph is not detected, a prompt of lacking data information is returned, and the bidding document does not meet the checking requirements.

[0089] After the progress plan horizontal graph is determined, content recognition is performed on the progress plan horizontal graph, for example, visual recognition can be performed based on an OCR recognition technology, so as to extract each work content contained in the progress plan horizontal graph, i.e. the first work content. After the progress plan horizontal graph is determined, the progress plan horizontal graph can be separately subjected to image enhancement again to ensure the subsequent recognition effect, for example, the accuracy of subsequent text OCR recognition and the accuracy of table line detection can be improved.

[0090] Optionally, the step A4 of “performing content recognition on the progress plan horizontal graph to determine the first work content contained in the progress plan horizontal graph” can specifically include steps A41 to A43.

[0091] The step A41 performs table line detection on the progress plan horizontal graph to determine table line information in the progress plan horizontal graph.

[0092] The step A42 positions the corresponding cells according to the table line information.

[0093] The step A43 performs content recognition on each cell to extract the first work content contained in the progress plan horizontal graph.

[0094] In this embodiment, the work content in the progress plan horizontal graph is divided by table lines. When the content recognition is performed on the progress plan horizontal graph, the table line detection is performed first, for example, a TTFNet (Table Transformer Network) model can be used to predict the position and direction of the table lines in the progress plan horizontal graph, so as to extract the corresponding table line information. The table line information can specifically include the position coordinates (including the start point coordinates and the end point coordinates) of each horizontal line and vertical line in the progress plan horizontal graph.

[0095] According to the detected table line information, the intersection of the table lines can be determined, and then a plurality of cells are divided, each cell corresponding to a part of the rectangular region in the progress plan horizontal graph, which can contain the work content.

[0096] By performing content recognition on each cell, such as optical character recognition (OCR), the text information contained in the progress bar chart can be extracted. This information can be used as the first task content. The information contained in multiple cells in each row of the progress bar chart is extracted on a row-by-row basis to determine the first task content. Each first task content corresponds to a row in the progress bar chart.

[0097] Figure 4 A schematic diagram of a progress plan bar chart is shown in Figure 1. Figure 4 As shown, the table in the progress plan bar chart contains the corresponding work content, that is, the content corresponding to the work name column. The work content specifically includes: construction preparation, earthwork excavation, earthwork and foundation engineering, foundation acceptance, reinforcement fabrication and installation, underground first floor structure, one to three floor structures, secondary structure and roof engineering, scaffolding erection, structure acceptance and other work contents, which need to be reviewed later.

[0098] In addition, if Figure 4 As shown, the progress plan bar chart also includes construction time information such as the start and end time of each work content, construction time (i.e., duration), etc. The construction time information of each work content can be identified from the progress plan bar chart and these construction time information can also be reviewed.

[0099] In this embodiment, the image corresponding to the bidding document is enhanced, and the enhanced target image is used to detect the progress bar chart, which can ensure the accuracy of subsequent content recognition of the progress bar chart; the table lines in the progress bar chart are first detected, and then the content of the progress bar chart is recognized based on the cells determined by the table lines. The first work content corresponding to each row of cells can be quickly and accurately extracted based on visual recognition technologies such as OCR.

[0100] Step S302: Determine the first unit project information in the bill of quantities file.

[0101] For details, please see Figure 1 Step S102 of the illustrated embodiment will not be described in detail here.

[0102] Step S303: input the first work content and the first unit project information into a preset review model, and determine the matching relationship between the first work content and the first unit project information according to the output result of the review model.

[0103] Specifically, the above step S303 "inputting the first work content and the first unit project information into the preset review model" includes steps S3031 to S3032.

[0104] In step S3031, for each first unit engineering information, a corresponding prompt word template is constructed; the prompt word template includes the first work content, the first unit engineering information, and a task instruction for instructing the review large model to determine the matching relationship between the first work content and the first unit engineering information.

[0105] In step S3032, the prompt word templates corresponding to the respective first unit engineering information are input into the pre-set review large model to determine the matching relationship between each first unit engineering information and the first work content.

[0106] In this embodiment, based on the first work content and the first unit engineering information, a prompt word template (Prompt) suitable for the review large model is constructed, and the prompt word template is input into the review large model, so that the review task of each first work content can be realized by using the review large model. If all the first work content and the first unit engineering information are input into the review large model, the correlation between the two may not be accurately determined by the review large model. Since the first unit engineering information extracted from the bill of quantities file is the review standard, i.e., each first unit engineering information is determined, therefore, in this embodiment, for each first unit engineering information, a corresponding prompt word template is constructed.

[0107] Specifically, for a certain first unit engineering information, a prompt word template corresponding to the first unit engineering information is generated by combining all the first work content extracted from the progress plan Gantt chart, and the prompt word template specifically includes the first unit engineering information and all the first work content to be reviewed; and a task instruction for instructing the review large model to determine the matching relationship between the first work content and the first unit engineering information can be constructed and added to the prompt word template.

[0108] For example, for the unit engineering of “civil engineering decoration and renovation engineering”, the prompt word template configured therefor can be as follows, which specifically includes the role, the “progress plan work content”, the “unit engineering”, and the related content of the task instruction.

[0109] Role: You are an engineering construction field technical specification construction organization design content evaluation expert, mainly responsible for house building construction field. Now you need to judge whether the following “progress plan work content” involves the work described in the “unit engineering”.

[0110] The “progress plan work content” is as follows:

[0111] Training base building floor decoration engineering; outdoor supporting facility earthwork engineering; training base other decoration engineering; training base prefabricated engineering; outdoor supporting facility road engineering.

[0112] The “unit engineering” is “civil engineering decoration and renovation engineering”.

[0113] The task instructions are as follows:

[0114] The "Progress Plan Work Content" listed is a plurality of possible related work content. It is determined whether it involves the related content described in the "Civil Engineering Decoration and Renovation Project". It does not require direct mention or complete match, and the reason is given.

[0115] The related work content described in the "Progress Plan Work Content" does not require direct mention or complete match. The content may contain or be related to the content described in the "Unit Project", that is, it is determined to involve related work content and returns "1". As long as there is any related item in the "Progress Plan Work Content", it is determined to involve related work content and returns "1". The specific content does not need to be further verified.

[0116] The return result is in the standard JSON format: field name: response_result; type: string type; description: "0" indicates that it does not involve related work content, and "1" indicates that it involves related work content.

[0117] Field name: reason; type: string; description: specific reason for involving / not involving related work content.

[0118] The following is an output format example. No other redundant information is output:

[0119] {"response_result":"1","reason":"specific reason"}

[0120] {"response_result":"0","reason":"specific reason"}

[0121] The above prompt word template is input into the review large model. Based on the output result of the review large model, it is determined whether the work content listed in the "Progress Plan Work Content" matches the unit project "Civil Engineering Decoration and Renovation Project", so as to obtain the matching relationship between the two, and the specific reason for whether the two match.

[0122] It can be understood that for other unit projects, suitable prompt word templates can also be constructed based on the above method, which is not described here.

[0123] In this embodiment, a corresponding prompt word template is constructed for each first unit project information, so that each first unit project information can be reviewed based on all work content. Each time, accurate information can be input to the review large model, which can ensure the accuracy and reliability of the review large model processing result each time, and will not miss the unit project, which can ensure the integrity of the review.

[0124] In some optional embodiments, the training process of the review large model can specifically include steps B1 to B3.

[0125] Step B1, pre-training the preset large language model according to the professional knowledge base in the field of construction.

[0126] Step B2, fine-tuning the large language model according to the constructed question and answer pair data; the question and answer pair data is used to represent the matching relationship between the work content and the unit engineering information.

[0127] Step B3, taking the fine-tuned large language model as the review large model for reviewing the construction progress plan.

[0128] In this embodiment, an open source large language model can be used as a base model, and the professional knowledge base in the field of construction is used for pre-training. The professional knowledge base may, for example, be a knowledge base related to the policies and regulations, standards and specifications, and professional examination data in the construction industry. Pre-training the preset large language model using the professional knowledge base in the field of construction can enable the review large model trained to have the ability to accurately identify the named entities and professional vocabulary in the field of construction, and the ability to accurately extract professional information in the field of construction.

[0129] Furthermore, question and answer pair data capable of matching the work content and the unit engineering information is constructed, the question and answer pair data includes a question and a corresponding answer, the question includes the work content and the unit engineering information, and the answer includes the matching relationship between the two, i.e., whether the two match. Fine-tuning the large language model using the constructed question and answer pair data can improve the review capability of the large model and ensure that the review large model can accurately review the construction progress plan.

[0130] Specifically, using the fine-tuned review large model to analyze the relevance of the work content in the progress plan bar chart and the unit engineering in the bidding bill of quantities file can infer whether the work content in the progress plan involves the relevant work described by the unit engineering.

[0131] Optionally, the process of constructing the question and answer pair data can include steps C1 to C3.

[0132] Step C1, determining a second work content and a second unit engineering information, and constructing a question for determining the matching relationship between the second work content and the second unit engineering information.

[0133] Step C2, inputting the question into a preset question and answer large model, and determining an answer representing the matching relationship between the second work content and the second unit engineering information according to the output result of the question and answer large model; the number of model parameters of the question and answer large model is greater than the number of model parameters of the review large model.

[0134] Step C3, constructing a corresponding question and answer pair data according to the question and the answer.

[0135] In this embodiment, in order to automatically and efficiently generate a large amount of question and answer pair data and form a question and answer pair data set, a more performant question and answer large model is used to construct the question and answer pair data. The question and answer large model is also a large language model, and the number of model parameters of the question and answer large model is greater than the number of model parameters of the review large model; for example, the number of model parameters of the review large model is 32B (B represents billion), and the number of model parameters of the question and answer large model is 64B or even more, so that the question and answer large model has better performance and can more accurately determine the matching relationship between the work content and the unit engineering information.

[0136] Specifically, the work content and the unit engineering information used to generate the question and answer pair data, i.e., the second work content and the second unit engineering information, can be determined; for example, the second work content and the second unit engineering information can be extracted based on historical bid documents and corresponding bill of quantities documents. Based on the second work content and the second unit engineering information, a question for determining the matching relationship between the second work content and the second unit engineering information can be constructed, and the question includes the second work content and the second unit engineering information; for example, the question can be constructed based on the prompt word template described above.

[0137] The question is input into the more performant question and answer large model, and the matching relationship between the second work content and the second unit engineering information can be determined according to the output result of the question and answer large model, and a corresponding answer is generated; for example, the output result of the question and answer large model is directly used as the answer.

[0138] Based on the above question and answer, a question and answer pair can be formed, and a set of question and answer pair data can be formed. Subsequently, the question and answer pair data can be used to fine-tune the large language model, so that the review large model can learn the review ability of the question and answer large model, thereby improving the review effect of the review large model.

[0139] In this embodiment, a larger-scale question and answer large model is used to generate question and answer pair data, and then the review large model is trained. Subsequently, only the review large model with a small amount of model parameters is needed to complete the review task during the review, and the question and answer large model is not needed. This not only improves the model review speed, but also reduces the computing cost and saves computing resources.

[0140] Step S304, generating a review report of the construction progress plan according to the matching relationship.

[0141] For details, please refer to Figure 1 Step S104 of the embodiment shown in the figure will not be described here.

[0142] Optionally, the step S304 of generating the review report of the construction progress plan according to the matching relationship comprises steps D1 to D2.

[0143] In the step D1, the missing first unit engineering information and / or the newly added first work content in the bidding document are determined according to the matching relationship.

[0144] In the step D2, the review report of the construction progress plan is generated; the review report comprises the missing first unit engineering information and / or the newly added first work content.

[0145] In the embodiment, according to the matching relationship between the first work content and the corresponding first unit engineering information, it can be determined whether there is a situation of not matching with one or more first unit engineering information, if there is, it can be determined that the first unit engineering information is missing, that is, the progress plan Gantt chart does not cover all the key unit engineering in the BOQ file (such as missing mechanical and electrical installation engineering, etc.), at this time, it can be determined that the progress plan is incomplete, which will lead to technical bid deduction or even bid rejection.

[0146] Alternatively, after determining the matching relationship between each first work content and the first unit engineering information, it can also be determined which first work content is matched with the corresponding first unit engineering information, so that it can be further determined which first work content is not matched with the first unit engineering information, and these unmatched first work content is the newly added first work content.

[0147] If the progress plan Gantt chart adds work content other than the BOQ (such as temporary road construction not included in the list), the necessity needs to be explained, otherwise it may be regarded as "deviating from the bidding requirements".

[0148] After the automatic evaluation of the content integrity, a detailed review report can be generated, which can not only point out whether each unit engineering in the bidding BOQ file is satisfied, but also point out the missing first unit engineering information. For the newly added first work content, it can continue to be automatically reviewed, or the generated detailed review report will be used as an auxiliary tool for review experts to conduct manual review, helping them to complete the review work more efficiently and accurately.

[0149] In the embodiment, all the results analyzed can be integrated to form a detailed report about whether the bidding document meets the bidding requirement of the progress plan integrity. For example, a schematic diagram of the review report can be seen from Figure 5 .

[0150] The construction progress plan review method provided by the embodiment can, in the building engineering bid evaluation stage, perform semantic analysis and key information extraction through machine vision and large language model technology, reduce the human resources and processing time required for review, accelerate the entire review process, and ensure the accuracy and comprehensiveness of the parsed content, especially in the case of processing a large number of complex documents, the advantage is particularly obvious. The progress plan is judged and analyzed by using an objective large language model, thereby reducing human error and subjective bias and ensuring the accuracy and consistency of the review. The automatic evaluation technology can evaluate the completeness of the construction progress plan content in the technical bid document, thereby improving the efficiency and accuracy of the review.

[0151] By setting a fixed algorithm and a large model to standardize the review process, it is ensured that even if different types of projects and complex files are faced, the review standards applied can still maintain a high degree of consistency and standardization, avoiding the review uncertainty that may be introduced due to differences between different judges in the traditional method. The automatic identification of key issues in the progress plan and the generation of a detailed review report can include an analysis result of whether each construction stage, key process, and professional engineering unit engineering is satisfied, thereby providing a comprehensive, objective, and efficient analysis tool for the review experts and helping them make more comprehensive and accurate decisions to better complete the review task.

[0152] In the embodiment, a construction progress plan review device is also provided, which is used to implement the above-mentioned embodiments and preferred embodiments, and details are not repeated. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware, or a combination of software and hardware implementation is also possible and contemplated.

[0153] The embodiment provides a construction progress plan review device, as shown in Figure 6 The device includes:

[0154] A content identification module 601 is configured to identify and process a bid document to be reviewed, and determine first work content contained in a progress plan Gantt chart in the bid document.

[0155] An information extraction module 602 is configured to determine first unit engineering information in a bill of quantities file.

[0156] A review module 603 is configured to input the first work content and the first unit engineering information into a preset review large model, and determine a matching relationship between the first work content and the first unit engineering information according to an output result of the review large model.

[0157] A report module 604 is configured to generate a review report of the construction progress plan according to the matching relationship.

[0158] In some optional implementations, the content identification module 601 identifies the bid document to be reviewed and determines the first work content included in the progress plan bar chart in the bid document, including:

[0159] Convert the bidding documents to be reviewed into original images;

[0160] Performing image enhancement on the original image to generate a target image;

[0161] Detecting the target image and extracting a progress plan bar chart;

[0162] Content identification is performed on the progress plan bar chart to determine the first work content included in the progress plan bar chart.

[0163] In some optional implementations, the content identification module 601 performs content identification on the progress plan bar chart to determine the first work content included in the progress plan bar chart, including:

[0164] Performing table line detection on the progress plan bar chart to determine table line information in the progress plan bar chart;

[0165] Locate corresponding cells according to the table line information;

[0166] Content recognition is performed on each of the cells to extract the first work content included in the progress plan bar chart.

[0167] In some optional embodiments, the training process of the review model includes:

[0168] Pre-train a preset large language model based on the professional knowledge base in the field of architecture;

[0169] Fine-tuning the large language model based on the constructed question-answer pair data; the question-answer pair data is used to represent the matching relationship between work content and unit project information;

[0170] The fine-tuned large language model is used as the large review model for reviewing the construction schedule.

[0171] In some optional implementations, the process of constructing the question-answer pair data includes:

[0172] Determining a second work content and second unit project information, and constructing a question for determining a matching relationship between the second work content and the second unit project information;

[0173] input the question into a preset question and answer large model, and determine an answer representing a matching relationship between the second work content and the second unit engineering information according to an output result of the question and answer large model; a number of model parameters of the question and answer large model is greater than a number of model parameters of the review large model;

[0174] According to the question and the answer, a corresponding question and answer pair data is constructed.

[0175] In some optional embodiments, the review module 603 inputs the first work content and the first unit engineering information into a preset review large model, including:

[0176] For each of the first unit engineering information, a corresponding prompt word template is constructed; the prompt word template includes the first work content, the first unit engineering information, and a task instruction for instructing the review large model to determine a matching relationship between the first work content and the first unit engineering information.

[0177] Each of the first unit engineering information corresponding prompt word templates is input into a preset review large model to determine a matching relationship between each of the first unit engineering information and the first work content.

[0178] In some optional embodiments, the report module 604 generates a review report of the construction progress plan according to the matching relationship, including:

[0179] According to the matching relationship, determine the missing first unit engineering information and / or the added first work content in the bid document;

[0180] Generate a review report of the construction progress plan; the review report includes the missing first unit engineering information and / or the added first work content.

[0181] The further function description of each module and unit is the same as the above-mentioned corresponding embodiments, and will not be repeated here.

[0182] The review device of the construction progress plan in the embodiment is presented in the form of a functional unit, and the unit here refers to an ASIC (Application Specific Integrated Circuit, Application Specific Integrated Circuit) circuit, including a processor and a memory executing one or more software or fixed programs, and / or other devices that can provide the above functions.

[0183] The embodiment of the application also provides a computer device with the above-mentioned Figure 6 The review device of the construction progress plan.

[0184] Please refer to Figure 7, Figure 7 is a structural schematic diagram of a computer device provided by an optional embodiment of the present application. As shown in Figure 7 the computer device includes one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components communicate with each other using different buses, and can be mounted on a common mainboard or mounted in other manners as needed. The processor can process instructions executed within the computer device, including instructions stored in the memory or on the memory to display graphical information of a GUI on an external input / output device (such as a display device coupled to the interface). In some optional embodiments, multiple processors and / or multiple buses can be used together with multiple memories if needed. Similarly, multiple computer devices can be connected, each providing part of the necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 7 The processor 10 is taken as an example in the

[0185] The processor 10 can be a central processor, a network processor, or a combination thereof. The processor 10 can further include a hardware chip. The hardware chip can be an application specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device can be a complex programmable logic device, a field programmable logic device, a general array logic, or any combination thereof.

[0186] The memory 20 stores instructions executable by the at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiments.

[0187] The memory 20 can include a program storage area and a data storage area. The program storage area can store an operating system and application programs required by at least one function; the data storage area can store data created according to the use of the computer device, etc. In addition, the memory 20 can include a high-speed random access memory, and can also include a non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state memory device. In some optional embodiments, the memory 20 can optionally include a memory remotely arranged with respect to the processor 10, and these remote memories can be connected to the computer device through a network. Examples of the network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0188] The memory 20 can include a volatile memory, such as a random access memory; the memory can also include a non-volatile memory, such as a flash memory, a hard disk, or a solid state disk; and the memory 20 can further include a combination of the above kinds of memories.

[0189] The computer device also comprises a communication interface 30 for communication of the computer device with other devices or communication networks.

[0190] The embodiments of the present application further provide a computer readable storage medium, and the method according to the embodiments of the present application can be implemented in hardware, firmware, or recorded in a storage medium, or stored in a remote storage medium or a non-transitory machine readable storage medium and downloaded from a network and stored in a local storage medium, so that the method described herein can be processed by such software on a storage medium using a general purpose computer, a special purpose processor, or programmable or special hardware. The storage medium can be a magnetic disk, an optical disk, a read-only memory, a random access memory, a flash memory, a hard disk, or a solid state disk, etc. Further, the storage medium can also include a combination of the above-mentioned memories. It can be understood that the computer, the processor, the microprocessor controller, or the programmable hardware includes a storage component that can store or receive software or computer code, when the software or computer code is accessed and executed by the computer, the processor, or the hardware, the method shown in the above embodiments is implemented.

[0191] Part of the present application can be applied as a computer program product, for example, computer program instructions, when executed by a computer, through the operation of the computer, the method and / or technical solutions according to the present application can be called or provided. Those skilled in the art should understand that the form of computer program instructions in a computer readable medium includes but is not limited to source files, executable files, installation package files, etc. Correspondingly, the way of computer program instructions executed by computer includes but is not limited to: the computer directly executes the instructions, or the computer compiles the instructions and then executes the corresponding compiled program, or the computer reads and executes the instructions, or the computer reads and installs the instructions and then executes the corresponding installed program. Here, the computer readable medium can be any available computer readable storage medium or communication medium accessible to the computer.

[0192] Although the embodiments of the present application are described in conjunction with the accompanying drawings, various modifications and changes can be made by those skilled in the art without departing from the spirit and scope of the present application, and such modifications and changes fall within the scope defined by the appended claims.

Claims

1. A construction progress plan review method, characterized in that: The method comprises: Identifying and processing the bid document to be reviewed, and determining the first work content included in the progress plan bar chart in the bid document; Determine the first unit project information in the bill of quantities document; Inputting the first work content and the first unit project information into a preset review model, and determining a matching relationship between the first work content and the first unit project information based on an output result of the review model; A review report of the construction schedule is generated based on the matching relationship.

2. The method according to claim 1, characterized in that The identifying and processing of the bid document to be reviewed and determining the first work content included in the progress plan bar chart in the bid document include: Convert the bidding documents to be reviewed into original images; Performing image enhancement on the original image to generate a target image; Detecting the target image and extracting a progress plan bar chart; Content identification is performed on the progress plan bar chart to determine the first work content included in the progress plan bar chart.

3. The method according to claim 2, characterized in that The performing content identification on the progress plan bar chart to determine the first work content included in the progress plan bar chart includes: Performing table line detection on the progress plan bar chart to determine table line information in the progress plan bar chart; Locate corresponding cells according to the table line information; Content recognition is performed on each of the cells to extract the first work content included in the progress plan bar chart.

4. The method according to claim 1, wherein The training process of the review model includes: Pre-train a preset large language model based on the professional knowledge base in the field of architecture; Fine-tuning the large language model based on the constructed question-answer pair data; the question-answer pair data is used to represent the matching relationship between work content and unit project information; The fine-tuned large language model is used as the large review model for reviewing the construction schedule.

5. The method according to claim 4, characterized in that The process of constructing the question-answer pair data includes: Determining a second work content and second unit project information, and constructing a question for determining a matching relationship between the second work content and the second unit project information; Inputting the question into a preset question-and-answer model, and determining an answer representing a matching relationship between the second work content and the second unit project information based on an output result of the question-and-answer model; the number of model parameters of the question-and-answer model is greater than the number of model parameters of the review model; According to the question and the answer, corresponding question-answer pair data is constructed.

6. The method according to claim 1, characterized in that The step of inputting the first work content and the first unit project information into a preset review model includes: For each piece of the first unit project information, a corresponding prompt word template is constructed; the prompt word template includes the first work content, the first unit project information, and a task instruction for instructing the review model to determine a matching relationship between the first work content and the first unit project information; The prompt word templates corresponding to each of the first unit project information are respectively input into the preset review model to respectively determine the matching relationship between each of the first unit project information and the first work content.

7. The method according to claim 1 or 6, characterized in that Generating a construction schedule review report based on the matching relationship includes: Determine the first unit project information missing from the bidding document and / or the first newly added work content according to the matching relationship; Generate a review report of the construction progress plan; the review report includes the missing first unit project information and / or the newly added first work content.

8. A construction progress plan review device, characterized in that: The device comprises: A content identification module is used to identify and process the bidding document to be reviewed, and determine the first work content included in the progress plan bar chart in the bidding document; An information extraction module, configured to determine the first unit engineering information in the bill of quantities file; a review module, configured to input the first work content and the first unit project information into a preset review model, and determine a matching relationship between the first work content and the first unit project information based on an output result of the review model; A report module is used to generate a review report of the construction schedule based on the matching relationship.

9. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the construction schedule review method according to any one of claims 1 to 7 by executing the computer instructions.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the construction schedule review method according to any one of claims 1 to 7.

Images