A construction briefing method based on BIM and generative model technology
By constructing a generative model on BIM data, generating and evaluating the construction briefing plan, the construction briefing process in the existing technology is solved, and the rapid and accurate generation of the construction plan is achieved.
Patent Information
- Application Number
- CN202410519226.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-04-28
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-04-28
AI Technical Summary
The existing construction plan briefing process relies on manual operations, which is time-consuming and labor-intensive and error-prone. How to quickly generate feasible construction plans in massive BIM data is a challenge.
The construction briefing method based on BIM and generative model technology is adopted, and the historical BIM data is obtained for preprocessing and training, a generative model is constructed, multiple construction briefing plans are generated, and the optimal solution is selected through feasibility assessment.
It improves the efficiency and accuracy of the formulation of construction plans and can quickly output the optimal construction briefing plan.
Smart Images

Figure CN118427928B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of construction disclosure, and in particular to a construction disclosure method based on BIM and generative model technology. Background Art
[0002] BIM technology is a "digital model" that uses various relevant information data of construction projects as the basis for the design, construction and operation of construction projects throughout their life cycle. BIM technology is a digital tool used in engineering planning, design, construction and management. It integrates various project information through models, and shares and transmits information throughout the life cycle of project planning, design, construction, operation and maintenance, so that all parties can correctly understand and respond to the information, playing an important role in improving production efficiency, saving costs and shortening construction periods. Generative models refer to large-scale neural network models that can generate, understand and reason about natural language in an end-to-end manner; currently, ChatGPT is the most popular. ChatGPT is a natural language processing tool driven by artificial intelligence technology, which has language understanding and text generation capabilities. In particular, it trains models by connecting a large number of corpora to generate meaningful text.
[0003] Construction briefing is a construction method. Technical briefing in construction companies refers to the technical briefing given by relevant professional and technical personnel to the personnel involved in the construction before the start of a unit project or a sub-project. The existing construction plan briefing process usually relies on manual operation to convert information such as architectural design drawings and construction drawings into construction plans. This process is time-consuming, labor-intensive, and prone to errors. With the continuous development of Building Information Modeling (BIM) technology, more and more construction projects have begun to adopt BIM technology for design and construction. BIM technology can provide comprehensive and detailed information for the construction process, but how to quickly generate feasible construction plans from massive data remains a challenge.
[0004] Therefore, there is an urgent need to provide a construction briefing method based on BIM and generative model technology to improve the efficiency and accuracy of construction plan formulation compared with existing technologies. Summary of the invention
[0005] The present invention solves the technical problems existing in the prior art and provides a construction briefing method based on BIM and generative model technology.
[0006] To achieve the above purpose, the technical solution adopted by the present invention is as follows:
[0007] A construction disclosure method based on BIM and generative model technology, comprising the following steps:
[0008] S1. Establish a generative model based on BIM data, which specifically includes the following steps:
[0009] S101, obtaining historical BIM data of N buildings;
[0010] S102, preprocessing the data obtained in step S101;
[0011] S103. Building generative models through deep learning techniques;
[0012] S104, training the generative model constructed in step S103 according to the N historical BIM data processed in step S102, to obtain a generative model based on BIM data;
[0013] S2. Input the BIM data of the building to be constructed into a generative model based on BIM data to generate multiple construction briefing plans;
[0014] S3, conducting feasibility assessment on multiple construction disclosure plans generated in step S2, and selecting the optimal construction disclosure plan;
[0015] S4. Output the optimal construction briefing plan.
[0016] Furthermore, the structure of the generative model constructed in step S103 includes an encoder, a generator and a discriminator. The encoder is used to encode the input BIM data into a low-dimensional vector, the generator is used to receive the low-dimensional vector output by the encoder and generate candidate construction briefing plans through multiple mapping layers, and the discriminator is used to determine whether the construction briefing plan generated by the generator is true.
[0017] Furthermore, S104 specifically includes the following steps:
[0018] S1041, obtaining the briefing files corresponding to the N historical BIM data obtained in step S101, forming N training groups, each training group including a historical BIM data processed in step S102 and its corresponding briefing file;
[0019] S1042, passing the historical BIM data processed by step S102 through an encoder and then using it as the input of the generator, and then using the discriminator to compare the output of the generator with the handover file corresponding to the BIM data input to the generator, and modifying the generator according to the comparison result;
[0020] S1043. Using the method in step S1042, apply N training groups to the generative model constructed in step S103 in sequence to obtain a generative model based on BIM data through training.
[0021] Furthermore, S3 specifically includes the following steps:
[0022] S301, construct a construction briefing comparison model;
[0023] S302, constructing an architectural term base, the architectural term base comprising a first term base, a second term base, a third term base, a fourth term base and a fifth term base;
[0024] S303, extracting the professional terms for construction briefing in the comparison model through natural language processing technology, and matching the extracted professional terms for construction briefing with the first term base, the second term base, the third term base, the fourth term base, and the fifth term base, respectively, to obtain a first matching quantity set, denoted as A;
[0025] S304, extracting the professional terms for construction briefing in the multiple construction briefing plans generated in step S2 by natural language processing technology, matching the extracted professional terms for construction briefing with the first term base, the second term base, the third term base, the fourth term base, and the fifth term base, respectively, to obtain multiple second matching quantity sets, denoted as B1, ..., B k , k is the number of construction briefing plans generated in step S2;
[0026] S305, respectively calculating correlation coefficients of the first matching quantity set and a plurality of second matching quantity sets;
[0027] S306. Compare the calculated correlation coefficient with the first threshold and the second threshold. If the first threshold is greater than the second threshold, when the first setting condition is met, output the construction briefing plan with the largest correlation coefficient. When the first setting condition is not met, perform the setting method and output the construction briefing plan with the largest correlation coefficient.
[0028] Furthermore, S301 specifically includes the following steps:
[0029] S3011, obtaining the number of problems existing in the building acceptance work corresponding to the N historical BIM data obtained in step S101;
[0030] S3012. Select the briefing file corresponding to the historical BIM data with the least number of problems as the construction briefing comparison model.
[0031] Furthermore, the correlation coefficient in step S305 is calculated by the following formula:
[0032]
[0033] In the above formula, p is the correlation coefficient, A j represents the jth number in the first matching quantity set, is the mean of the first matching data set, is the mean of the hth second matching data set, Represents the jth number in the hth second matching data set, 1≤h≤k, and h is an integer.
[0034] Furthermore, the setting method in step S306 includes two cases, namely:
[0035] (1) When the second setting condition is met, the generative model is iteratively trained using the method in step S104, the number of iterations being the first setting number, to form a new generative model based on BIM data, and then steps S2-S3 are performed until the first setting condition is met;
[0036] (2) When the second set condition is not met, the generative model is iteratively trained using the method of step S104, the number of iterations being the second set number, to form a new generative model based on BIM data, and then steps S2-S3 are performed until the first set condition is met.
[0037] Furthermore, the first set number of times and the second set number of times satisfy the following formula:
[0038]
[0039] In the above formula, b is the second set number of times, a is the first set number of times, N is the number of training groups, and α is a positive number.
[0040] Furthermore, the first setting condition is that the correlation coefficients of more than half of the construction briefing plans are greater than the first threshold; the second setting condition is that the correlation coefficients of more than half of the construction briefing plans are greater than the second threshold and do not meet the first setting condition.
[0041] Furthermore, the historical BIM data includes geometric information, attribute information, time information, cost information and design conflict information; in step S302, the professional terms for briefing of geometric information, attribute information, time information, cost information and design conflict information are collected to form a building terminology library, wherein the first terminology library stores the professional terms for briefing of geometric information, the second terminology library stores the professional terms for briefing of attribute information, the third terminology library stores the professional terms for briefing of time information, the fourth terminology library stores the professional terms for briefing of cost information, and the fifth terminology library stores the professional terms for briefing of design conflict information.
[0042] Compared with the prior art, the present invention has the following beneficial effects:
[0043] The present invention combines BIM data with a generative model to form a BIM-based generative model, and inputs BIM data of the building to be constructed into the BIM-based generative model, so as to quickly output multiple construction briefing plans, and then analyzes the feasibility of the multiple construction briefing plans to output the optimal construction briefing plan, thereby improving the efficiency and accuracy of formulating construction briefing plans. BRIEF DESCRIPTION OF THE DRAWINGS
[0044] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION
[0045] The technical solution of the present invention will be clearly described below in conjunction with the accompanying drawings. Obviously, the described embodiments are not all embodiments of the present invention, and all other embodiments obtained by ordinary technicians in the field without making creative work are within the protection scope of the present invention.
[0046] like Figure 1 As shown, the present invention provides a construction disclosure method based on BIM and generative model technology, comprising the following steps:
[0047] S1. Establish a generative model based on BIM data, which specifically includes the following steps:
[0048] S101. Obtain historical BIM data of N buildings. The BIM data is data in BIM format exported from Building Information Modeling (BIM) software. The BIM data includes geometric information, attribute information, time information, cost information and design conflict information. The geometric information includes the appearance of the building and the geometric shape and location information of the internal structure, pipelines and equipment of the building. The attribute information includes the material, size, weight and color of the building. The time information includes the construction progress, engineering plan and equipment maintenance cycle. The cost information includes the cost of building materials, construction costs and equipment maintenance costs. The design conflict information includes the conflict between pipelines and structures and the conflict of equipment layout.
[0049] S102, preprocessing the data obtained in step S101, specifically including the following steps:
[0050] S1021. Data cleaning: Remove invalid, duplicate and inconsistent data from BIM data to ensure data quality;
[0051] S1022. Data normalization: standardize the geometric information, attribute information, time information, cost information and design conflict information in BIM data to facilitate subsequent training and calculation;
[0052] S1023. Data dimensionality reduction: For high-dimensional BIM data, principal component analysis is used to reduce the dimensionality of the data and reduce the computational complexity.
[0053] S103, build a generative model: use the BIM data preprocessed in step S2 to build a generative model through deep learning technology. The structure of the generative model includes an encoder, a generator and a discriminator. The encoder is used to encode the input BIM data into a low-dimensional vector and input the processed low-dimensional vector into the generator. The generator is used to receive the low-dimensional vector output by the encoder and generate candidate construction plans through multiple mapping layers. The discriminator is used to determine whether the construction plan generated by the generator is true through a binary classification task.
[0054] S104: training the generative model constructed in step S103 to obtain a generative model based on BIM data, specifically including the following steps:
[0055] S1041. Obtain the briefing files corresponding to the N historical BIM data obtained in step S101 to form N training groups, each of which includes a historical BIM data processed in step S102 and its corresponding briefing file.
[0056] S1042. The historical BIM data processed by step S102 is passed through an encoder and then used as the input of the generator. The output of the generator is then compared with the briefing file corresponding to the BIM data input to the generator through a discriminator, and the generator is corrected according to the comparison result.
[0057] S1043. Using the method in step S1042, apply N training groups to the generative model constructed in step S103 in sequence to obtain a generative model based on BIM data through training.
[0058] S2. Generate multiple construction briefing plans through the generative model based on BIM data established in step S1, which specifically includes the following steps:
[0059] S201, obtaining BIM data of the building to be constructed;
[0060] S202: pre-process the BIM data of the building to be constructed. The pre-processing method in this step is the same as the pre-processing method in step S102.
[0061] S203, inputting the BIM data processed in step S202 into a generative model based on BIM data, and outputting a plurality of construction briefing plans.
[0062] S3, conduct feasibility assessment on multiple construction disclosure plans generated in step S2, and select the optimal construction disclosure plan, which specifically includes the following steps:
[0063] S301. Construct a construction briefing comparison model, including:
[0064] S3011. Obtain the number of problems existing in the building acceptance work corresponding to the N historical BIM data obtained in step S101.
[0065] S3012. Select the briefing file corresponding to the historical BIM data with the least number of problems as the construction briefing comparison model.
[0066] S302, construct an architectural terminology library, collect professional terms for briefing on geometric information, attribute information, time information, cost information and design conflict information, and form an architectural terminology library. The formed architectural terminology library includes a first terminology library, a second terminology library, a third terminology library, a fourth terminology library and a fifth terminology library. The first terminology library stores professional terms for briefing on geometric information, the second terminology library stores professional terms for briefing on attribute information, the third terminology library stores professional terms for briefing on time information, the fourth terminology library stores professional terms for briefing on cost information, and the fifth terminology library stores professional terms for briefing on design conflict information.
[0067] S303. Extract the professional terms for briefing in the construction briefing comparison model through natural language processing technology, and match the extracted professional terms for briefing with the first term base, the second term base, the third term base, the fourth term base, and the fifth term base respectively to obtain a first matching number set, denoted as A=[A1, A2, A3, A4, A5], where A1 is the number of matches between the professional terms for briefing in the extracted construction briefing comparison model and the first term base, A2 is the number of matches between the professional terms for briefing in the extracted construction briefing comparison model and the second term base, A3 is the number of matches between the professional terms for briefing in the extracted construction briefing comparison model and the third term base, A4 is the number of matches between the professional terms for briefing in the extracted construction briefing comparison model and the fourth term base, and A5 is the number of matches between the professional terms for briefing in the extracted construction briefing comparison model and the fifth term base.
[0068] S304, extract the professional terms for construction briefing in the multiple construction briefing plans generated in step S2 by using natural language processing technology, match the terms extracted from the multiple construction briefing plans with the first term base, the second term base, the third term base, the fourth term base, and the fifth term base, respectively, to obtain multiple second matching quantity sets, which are respectively denoted as B1=[B 11 , B 12 , B 13 , B 14 , B 15 ], …, B k =[B k1 , B k2 , B k3 , B k4 , B k5 ], where k is the number of construction briefing plans generated in step S2, B1 represents the first second matching number set, and B11 represents the number of matches between the professional terms extracted from the first construction briefing plan and the first terminology library, B 12 represents the number of matches between the professional terms extracted from the first construction briefing plan and the second terminology library, B 13 represents the number of matches between the professional terms extracted from the first construction briefing plan and the third terminology library, B 14 represents the number of matches between the professional terms extracted from the first construction briefing plan and the fourth terminology library, B 15 represents the number of matches between the professional terms extracted from the first construction briefing plan and the fifth terminology library; B k represents the kth second matching quantity set, B k1 represents the number of matches between the professional terms extracted from the kth construction briefing plan and the first terminology library, B k2 represents the number of matches between the professional terms extracted from the kth construction briefing plan and the second terminology library, B k3 represents the number of matches between the professional terms extracted from the k-th construction briefing plan and the third terminology library, B k4 represents the number of matches between the professional terms extracted from the k-th construction briefing plan and the fourth terminology library, B k5 It represents the number of matches between the professional terms extracted from the k-th construction briefing plan and the fifth terminology library.
[0069] S305, respectively calculating the correlation coefficients between the first matching quantity set and the plurality of second matching quantity sets, which are specifically calculated by the following formula:
[0070]
[0071] In the above formula, p is the correlation coefficient, A j represents the jth number in the first matching quantity set, is the mean of the first matching data set, is the mean of the hth second matching data set, Represents the jth number in the hth second matching data set, 1≤h≤k, and h is an integer.
[0072] S306. Compare the correlation coefficients of multiple construction briefing plans with a first threshold and a second threshold, where the first threshold is greater than the second threshold; when the first setting condition is met, output the construction briefing plan with the largest correlation coefficient; when the first setting condition is not met, perform the setting method, and then output the construction briefing plan with the largest correlation coefficient.
[0073] There are two setting methods:
[0074] (1) When the second set condition is met, the generative model is iteratively trained using the method in step S104, the number of iterations being the first set number, to form a new generative model based on BIM data, and then steps S2-S3 are performed until the first set condition is met.
[0075] (2) When the second set condition is not met, the generative model is iteratively trained using the method of step S104, the number of iterations being the second set number, to form a new generative model based on BIM data, and then steps S2-S3 are performed until the first set condition is met.
[0076] The first setting condition is that the correlation coefficients of more than half of the construction briefing plans are greater than the first threshold; the second setting condition is that the correlation coefficients of more than half of the construction briefing plans are greater than the second threshold and do not meet the first setting condition.
[0077] The first set number of times and the second set number of times satisfy the following formula:
[0078]
[0079] In the above formula, b is the second set number of times, a is the first set number of times, N is the number of training groups, and α is a positive number.
[0080] S4. Output the final construction briefing plan and present it to the construction personnel through one or more methods including reports, graphical displays, and training.
[0081] The present invention combines BIM data with a generative model to form a BIM-based generative model, and inputs BIM data of the building to be constructed into the BIM-based generative model, so as to quickly output multiple construction briefing plans, and then analyzes the feasibility of the multiple construction briefing plans to output the optimal construction briefing plan, thereby improving the efficiency and accuracy of formulating construction briefing plans.
[0082] Finally, it should be noted that the above content is only used to illustrate the technical solution of the present invention, rather than to limit the scope of protection of the present invention. Simple modifications or equivalent substitutions of the technical solution of the present invention by ordinary technicians in this field do not deviate from the essence and scope of the technical solution of the present invention.
Claims
1. A construction disclosure method based on BIM and generative model technology, characterized in that: The following steps are involved: S1. Establish a generative model based on BIM data, which specifically includes the following steps: S101, obtaining historical BIM data of N buildings; S102, preprocessing the data obtained in step S101; S103. Building generative models through deep learning techniques; S104, training the generative model constructed in step S103 according to the N historical BIM data processed in step S102, to obtain a generative model based on BIM data; S2. Input the BIM data of the building to be constructed into a generative model based on BIM data to generate multiple construction briefing plans; S3, conduct feasibility assessment on multiple construction disclosure plans generated in step S2, and select the optimal construction disclosure plan; specifically, the following steps are included: S301, construct a construction briefing comparison model; S302, constructing an architectural term base, the architectural term base comprising a first term base, a second term base, a third term base, a fourth term base and a fifth term base; S303, extracting the professional terms for construction briefing in the comparison model through natural language processing technology, and matching the extracted professional terms for construction briefing with the first term base, the second term base, the third term base, the fourth term base, and the fifth term base, respectively, to obtain a first matching quantity set, denoted as A; S304, extracting the professional terms for construction briefing in the multiple construction briefing plans generated in step S2 by using natural language processing technology, matching the extracted professional terms for construction briefing with the first term base, the second term base, the third term base, the fourth term base, and the fifth term base, respectively, to obtain multiple second matching quantity sets, recorded as , k is the number of construction briefing plans generated in step S2; S305, respectively calculating correlation coefficients of the first matching quantity set and a plurality of second matching quantity sets; S306, comparing the calculated correlation coefficient with a first threshold and a second threshold, the first threshold being greater than the second threshold, and when the first setting condition is met, outputting a construction briefing scheme with the largest correlation coefficient; when the first setting condition is not met, performing a setting method, and outputting a construction briefing scheme with the largest correlation coefficient; The setting method in step S306 includes two cases, namely: (1) When the second setting condition is met, the generative model is iteratively trained using the method in step S104, the number of iterations being the first setting number, to form a new generative model based on BIM data, and then steps S2-S3 are performed until the first setting condition is met; (2) When the second set condition is not met, the generative model is iteratively trained using the method in step S104, the number of iterations being the second set number of times, to form a new generative model based on BIM data, and then steps S2-S3 are performed until the first set condition is met; The first set number of times and the second set number of times satisfy the following formula: ; In the above formula, b is the second set number of times, a is the first set number of times, and N is the number of sets in the training group. is a positive number; The first setting condition is that the correlation coefficients of more than half of the construction briefing plans are greater than the first threshold value; the second setting condition is that the correlation coefficients of more than half of the construction briefing plans are greater than the second threshold value and do not meet the first setting condition; S4. Output the optimal construction briefing plan.
2. A construction disclosure method based on BIM and generative model technology according to claim 1, characterized in that: The structure of the generative model constructed in step S103 includes an encoder, a generator and a discriminator. The encoder is used to encode the input BIM data into a low-dimensional vector, the generator is used to receive the low-dimensional vector output by the encoder and generate candidate construction briefing plans through multiple mapping layers, and the discriminator is used to determine whether the construction briefing plan generated by the generator is true.
3. A construction disclosure method based on BIM and generative model technology according to claim 2, characterized in that: S104 specifically includes the following steps: S1041, obtaining the briefing files corresponding to the N historical BIM data obtained in step S101, forming N training groups, each training group including a historical BIM data processed in step S102 and its corresponding briefing file; S1042, passing the historical BIM data processed by step S102 through an encoder and then using it as the input of the generator, and then using the discriminator to compare the output of the generator with the handover file corresponding to the BIM data input to the generator, and modifying the generator according to the comparison result; S1043. Using the method in step S1042, apply N training groups to the generative model constructed in step S103 in sequence to obtain a generative model based on BIM data through training.
4. The construction disclosure method based on BIM and generative model technology according to claim 3 is characterized in that: S301 specifically includes the following steps: S3011, obtaining the number of problems existing in the building acceptance work corresponding to the N historical BIM data obtained in step S101; S3012. Select the briefing file corresponding to the historical BIM data with the least number of problems as the construction briefing comparison model.
5. The construction disclosure method based on BIM and generative model technology according to claim 3 is characterized in that: The correlation coefficient in step S305 is calculated by the following formula: In the above formula, p is the correlation coefficient, represents the jth number in the first matching quantity set, is the mean of the first matching data set, is the mean of the hth second matching data set, represents the jth number in the hth second matching data set, , h is an integer.
6. The construction disclosure method based on BIM and generative model technology according to claim 3 is characterized in that: Historical BIM data includes geometric information, attribute information, time information, cost information and design conflict information; in step S302, professional terms for briefing of geometric information, attribute information, time information, cost information and design conflict information are collected to form a building terminology library, wherein the first terminology library stores professional terms for briefing of geometric information, the second terminology library stores professional terms for briefing of attribute information, the third terminology library stores professional terms for briefing of time information, the fourth terminology library stores professional terms for briefing of cost information, and the fifth terminology library stores professional terms for briefing of design conflict information.
