Design compliance evaluation method based on BIM

Abstract

The invention relates to the technical field of building information models, and discloses a BIM-based design compliance evaluation method, which comprises the following steps: S1, importing and preparing a BIM model of a design project, ensuring that the BIM model contains design data of a building structure, facility arrangement and material selection, and carrying out data preprocessing; s2, according to the specific requirements of a design project, a nested algorithm automatically selects an applicable rule set from a plurality of provided compliance rule libraries; s3, starting an automatic compliance checking function of the system, and checking each design content in the BIM model one by one by the system according to the configured rule base; and S4, automatically generating a detailed compliance check report. A self-learning mechanism is utilized, the result and correction measures of each compliance check are recorded, a rule base and detection parameters are automatically optimized based on the data, the check accuracy and efficiency are continuously improved, and it is ensured that compliance check always keeps high precision in constantly changing design requirements.

Description

Technical Field

[0001] The present invention relates to the technical field of building information modeling, and in particular to a design compliance evaluation method based on BIM. Background Art

[0002] With the development of the construction industry, Building Information Modeling (BIM) technology has gradually become an important tool for modern building design, construction and management. BIM technology integrates various building information through digital three-dimensional models, realizing full-process management from design to construction. However, although BIM technology has significant advantages in improving building design efficiency and collaborative work, there are still some challenges and shortcomings in the field of design compliance inspection.

[0003] Deficiencies of existing technology:

[0004] Inefficient manual inspection: In the traditional design compliance inspection process, a lot of work depends on manual review. This is not only time-consuming, but also easily affected by human factors, making it difficult to ensure the accuracy and consistency of the inspection results. In addition, with the increasing complexity of architectural design, the workload of manual inspection has increased significantly, further extending the design cycle.

[0005] Delayed code updates: Existing compliance checking methods often have difficulty adapting to the latest design codes and regulatory changes. The rule bases used by many checking tools have not been updated in a timely manner, resulting in non-compliance in the design after the code is updated, increasing the cost and risk of later corrections. In addition, due to differences in building codes in different regions, existing tools are difficult to effectively manage and apply multiple code rules.

[0006] Lack of intelligence and adaptive capabilities: Traditional compliance inspection methods are usually based on fixed rule sets and lack intelligent self-learning and optimization mechanisms. This means that the inspection tools cannot optimize themselves based on historical inspection results and project experience, and cannot effectively reduce false positives and missed negatives. In addition, when the content of the rule base or design specifications change, the existing methods are difficult to automatically adapt and update, resulting in reduced reliability of the inspection results.

[0007] To this end, those skilled in the art provide a design compliance evaluation method based on BIM to solve the problems raised in the above background technology. Summary of the invention

[0008] In view of the shortcomings of the prior art, the present invention provides a design compliance evaluation method based on BIM, which solves the problems of low efficiency of manual inspection, lagging specification update and lack of intelligence and self-adaptation in the prior art.

[0009] To achieve the above objectives, the present invention is implemented through the following technical solutions: A design compliance evaluation method based on BIM, comprising the following steps:

[0010] S1. Import and prepare the BIM model of the design project, ensure that it contains the design data of building structure, facility layout, and material selection, and pre-process the design data of data format, coordinate system consistency, and connection and dependency relationships between components;

[0011] S2. According to the specific requirements of the design project, the algorithm automatically selects the applicable rule set from the various compliance rule libraries provided;

[0012] S3. Start the system's automated compliance check function. The system checks each design content in the BIM model one by one according to the configured rule base, and automatically identifies potential non-compliant items in the model through the set algorithm, and highlights them at the same time;

[0013] S4. Automatically generate detailed compliance inspection reports;

[0014] S5. After correcting the BIM model according to the report published in step S3, re-enter steps S1-S4 for compliance check until all non-compliance items have been resolved;

[0015] S6. During the S1-S4 cycle, the system records the results of each compliance check and the corresponding corrective measures as the basic data for the self-learning mechanism. Then, the algorithm automatically optimizes the rule base and adjusts the detection parameters to reduce false positives and false negatives in future checks. When the rules are updated or the design specifications change, the system automatically identifies and introduces new rules to ensure that the compliance check is consistent with the latest standards.

[0016] S7. Through multiple rounds of compliance checks and model corrections, the non-compliant items in the model are gradually reduced until all compliance requirements are met. After completing the final compliance check, a comprehensive compliance assessment report is automatically generated to meet all applicable design specifications and regulatory requirements;

[0017] S8. Archive the final compliance assessment report and the compliant BIM model, and feed back new findings, special circumstances and correction experiences in this project into the system's knowledge base through data comparison.

[0018] Preferably, in step S2, the rule base includes but is not limited to building structure specifications, fire safety specifications, energy-saving and environmental protection standards, and the algorithm in step S2 is specifically a logistic regression algorithm, which recommends the rule set in the rule base according to the geographical location and building use factors of the project, and the specific calculation formula is:

[0019]

[0020]

[0021]

[0022] Among them, z is the result of linear combination, β0 is the bias term, β i is the weight of the i-th feature, x i is the i input feature value, n is the total number of features, and finally through x1,x2,…,x n In the case of , the probability p that the feature combination meets a certain rule set is calculated, so as to automatically match the most suitable rule set.

[0023] Preferably, the S3 step uses a graph neural model to automatically identify model non-compliant items, and the calculation steps are:

[0024] S31, initializing a feature vector for each node (i.e., a component in the BIM model) to represent the initial state of the node;

[0025] S32. In each layer of propagation, a node receives “messages” from its neighboring nodes and updates its own feature representation. The node feature update formula is:

[0026]

[0027] in, is the feature representation of node v at layer k, is the set of neighbor nodes of node v, f message is the message passing function used to combine the features of neighbor nodes;

[0028] S33, after several layers of propagation, the final feature representation of the node It is used to determine whether there are any non-compliant items at the node, and the output is displayed through the classifier:

[0029]

[0030] Among them, W class is the weight matrix of the classifier, y v is the probability distribution of node v belonging to different categories (compliant / non-compliant).

[0031] Preferably, the S4 step uses Pandoc and LaTeX to generate a compliance check report, which includes but is not limited to the non-compliant items found, the corresponding regulatory clauses, the severity and modification suggestions.

[0032] Preferably, the step S4 further includes using a risk scoring formula to quantitatively rate the non-compliance items found, including three severity scores and three occurrence probability recommendations, specifically including:

[0033] Severity Rating:

[0034] 1: Minor impact, no emergency treatment required;

[0035] 3: Moderate impact, needs to be dealt with in the short term;

[0036] 5: Severe impact, must be dealt with immediately;

[0037] Probability of occurrence:

[0038] 1: Rarely occurs;

[0039] 3: Occasionally;

[0040] 5: Occurs frequently;

[0041] The specific risk scoring formula is: R = S × P, where S is the severity score, P is the probability of occurrence, and R is the risk score.

[0042] Preferably, the self-learning mechanism in step S6 includes the following calculation steps:

[0043] S61. Calculation of false alarm and missed alarm rates:

[0044] False Positive Rate:

[0045] Among them, False Positives are items that the system mistakenly marks as non-compliant, and True Negatives are items that the system correctly marks as compliant;

[0046] False negative rate:

[0047] Among them, False Negatives are non-compliant items that the system failed to detect, and True Positives are items that the system correctly marked as non-compliant;

[0048] S62, parameter adjustment; including gradient descent method and Bayesian update formula:

[0049] Gradient Descent:

[0050] Among them, θ t is the current detection parameter, η is the learning rate, is the gradient of the loss function with respect to the parameters;

[0051] Bayesian update formula:

[0052] Among them, P(H|D) is the confidence of the updated rule, P(D|H) is the probability of new data under the current rule, P(H) is the prior probability of the rule, and P(D) is the marginal probability of the data;

[0053] S63, Rule Importance Score: Evaluate the importance of rules to decide whether to retain or introduce new rules. The calculation formula is:

[0054] Among them, I j It is rule R j Importance score, Gain(R j ,D i ) is the rule R j Right D i The information gain of n is the total amount of data;

[0055] S64, Identification and introduction of new rules: The introduction of new rules is determined by threshold, the formula is: IfI j >Threshold, introduce R j into the rule set,

[0056] Here, Threshold is the preset importance threshold of the introduced rule.

[0057] Preferably, after obtaining the rule base, detection parameters, and false positive and false negative rate data, the step S6 is optimized by deep reinforcement learning, and the calculation formula is:

[0058]

[0059] Here, θ is the policy parameter and J(θ) is the expected cumulative reward.

[0060] A BIM-based design compliance evaluation system, used for a BIM-based design compliance evaluation method, comprising:

[0061] BIM data management module, used to import and pre-process BIM model data;

[0062] A rule base management module is used to manage and select applicable design specification rule sets and dynamically update the rule base;

[0063] A compliance checking module that automates compliance checking, identifies and marks potential non-compliance items in the model;

[0064] Report generation module, which automatically generates compliance check reports, including details of non-compliance items and risk scores;

[0065] Self-learning optimization module, which is used to record inspection results, automatically optimize the rule base and detection parameters, and reduce false positives and false negatives;

[0066] External data acquisition module: used to connect to external data sources, obtain the latest compliance requirements and design specification changes in real time, and synchronize this updated information to the system's rule base management module.

[0067] Preferably, the external data acquisition module includes but is not limited to crawling the real-time compliance rule library content through RESTful or GraphQL API and WebScraping, and updating it on a daily, weekly or monthly basis during the process.

[0068] Preferably, the BIM model data imported and preprocessed by the BIM data management module is passed to the compliance checking module; the rule base management module provides the applicable design specification rule set to the compliance checking module; the non-compliant items and related information identified and marked by the compliance checking module are passed to the report generation module; the inspection results recorded by the compliance checking module are passed to the self-learning optimization module; the compliance inspection report generated by the report generation module, including the risk score and modification suggestions for the non-compliant items, is provided as feedback to the self-learning optimization module; the self-learning optimization module updates the optimized rule base and detection parameters to the rule base management module; the external data acquisition module obtains the latest compliance requirements and design specification changes from the external data source and passes them to the rule base management module.

[0069] The present invention provides a design compliance evaluation method based on BIM. It has the following beneficial effects:

[0070] 1. The present invention uses a self-learning mechanism to record the results and corrective measures of each compliance check, and automatically optimizes the rule base and detection parameters based on these data. As the number of checks increases, the method can gradually reduce false positives and false negatives, continuously improve the accuracy and efficiency of the check, and ensure that the compliance check always maintains high accuracy in the ever-changing design requirements.

[0071] 2. Through multiple rounds of compliance checks and model corrections, the present invention gradually reduces non-compliant items in the model until the model fully complies with all regulatory requirements. The final generated compliance assessment report, combined with quantitative risk scores and modification suggestions, provides the design team with a clear direction for improvement, ensuring that the project can meet the highest standards in compliance, significantly reducing compliance risks in subsequent construction and operation stages.

[0072] 3. The present invention realizes compliance check of BIM model through automated process. From data import and preprocessing to the selection of compliance rules and the generation of inspection results, the whole process automation operation significantly improves the inspection efficiency. In addition, by obtaining the latest design specifications and regulatory changes in real time, the method ensures that each inspection is based on the current valid standards, avoiding design risks caused by lagging specifications. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 It is a schematic diagram of the method steps of the present invention;

[0074] Figure 2 It is a schematic diagram of the compliance evaluation system framework of the present invention. DETAILED DESCRIPTION

[0075] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0076] Embodiment 1:

[0077] Please see attached Figure 1 The embodiment of the present invention provides a design compliance evaluation method based on BIM, comprising the following steps:

[0078] S1. Import and prepare the BIM model of the design project, ensure that it contains the design data of building structure, facility layout, and material selection, and pre-process the design data of data format, coordinate system consistency, and connection and dependency between components;

[0079] Specifically, in step S1, the system first needs to import and prepare the BIM model of the design project. This process is the basis of the entire compliance evaluation process and determines the accuracy and effectiveness of subsequent inspections and analyses. To ensure the integrity of the BIM model and the consistency of the data, the system will conduct comprehensive preparations for the imported BIM model. The model must contain key design data such as building structure, facility layout, and material selection. These data constitute the basic elements of architectural design and are also an important basis for compliance inspections. During the data import process, the system will automatically detect and convert the data format to ensure that the model data is compatible with the system's internal processing mechanism.

[0080] In addition, the system will also perform a consistency check on the coordinate system and unify the coordinate systems of different components to avoid misjudgments due to coordinate differences. The system will use graph algorithms to analyze the connections and dependencies between components to ensure that the connection relationships between all components are reasonable and meet design specifications. During data preprocessing, the system will also perform a structured check on the model to identify and correct potential data missing, duplications, or outliers to improve the accuracy and reliability of the model. Through this series of detailed preparation and processing steps, the system lays a solid foundation for subsequent compliance checks, ensuring that each design element can be accurately analyzed and evaluated, effectively reducing compliance risks caused by data problems, and thus improving the quality and compliance of the overall project.

[0081] S2. According to the specific requirements of the design project, the algorithm automatically selects the applicable rule set from the various compliance rule libraries provided;

[0082] In step S2, the rule base includes but is not limited to building structure specifications, fire safety specifications, energy conservation and environmental protection standards, and the algorithm in step S2 is specifically a logistic regression algorithm, which recommends the rule set in the rule base based on the geographical location and building use factors of the project. The specific calculation formula is:

[0083]

[0084]

[0085]

[0086] Among them, z is the result of linear combination, β0 is the bias term, β i is the weight of the i-th feature, x i is the i input feature value, n is the total number of features, and finally through x1,x2,…,x n In the case of , the probability p that the feature combination meets a certain rule set is calculated, so as to automatically match the most suitable rule set.

[0087] Specifically, the system will use the logistic regression algorithm to automatically select the applicable rule set from the various compliance rule bases provided according to the specific requirements of the design project. The rule base includes but is not limited to compliance requirements in different fields such as building structure specifications, fire safety specifications, energy-saving and environmental protection standards. The logistic regression algorithm is the core algorithm of this step. By analyzing key factors such as the project's geographical location and building use, the system can intelligently recommend the most suitable rule set. Specifically, the logistic regression algorithm first assigns a weight to each input feature (such as geographical location, building use, etc.) and calculates the linear combination of these features. The formula is:

[0088]

[0089] Next, the system converts the linear combination result into a probability value through the Sigmoid function:

[0090]

[0091] This probability value p represents the applicability probability of a certain rule set to the current project. Based on this probability, the system can automatically identify and match the most suitable rule set, thereby ensuring that the project design process strictly follows the corresponding compliance requirements. This process not only improves the accuracy and efficiency of rule selection, but also dynamically adapts to the specific needs of different projects, ensuring that the system can provide the best rule set recommendation in different scenarios.

[0092] S3. Start the system's automated compliance check function. The system checks each design content in the BIM model one by one according to the configured rule base, and automatically identifies potential non-compliant items in the model through the set algorithm, and highlights them at the same time;

[0093] The S3 step uses a graph neural model to automatically identify model non-compliance items. Its calculation steps are:

[0094] S31, initializing a feature vector for each node (i.e., a component in the BIM model) to represent the initial state of the node;

[0095] S32. In each layer of propagation, a node receives “messages” from its neighboring nodes and updates its own feature representation. The node feature update formula is:

[0096]

[0097] in, is the feature representation of node v at layer k, is the set of neighbor nodes of node v, f message is the message passing function used to combine the features of neighbor nodes;

[0098] S33, after several layers of propagation, the final feature representation of the node It is used to determine whether there are any non-compliant items at the node, and the output is displayed through the classifier:

[0099]

[0100] Among them, W class is the weight matrix of the classifier, y v is the probability distribution of node v belonging to different categories (compliant / non-compliant).

[0101] Specifically, the system starts the automated compliance check function, relying on the configured rule base to check each design content in the BIM model one by one to ensure that each part of the model complies with relevant specifications and standards. This process uses the graph neural network (GNN) model to automatically identify non-compliant items. When processing the BIM model, the graph neural network treats each component as a node, and initially assigns a feature vector to each node to represent its initial state.

[0102] Next, the system uses a multi-layer propagation mechanism to allow each node to receive information from its neighboring nodes and continuously update its own feature representation. This process effectively captures the complex topological relationships and interdependencies between components, allowing the system to not only analyze the compliance of a single component, but also identify non-compliant patterns based on the global structure.

[0103] After several layers of message passing and feature updates, the final feature representation of the node can fully reflect its role and status in the entire model. Subsequently, the system analyzes these final features through a classifier to determine whether there are any non-compliant items in the node.

[0104] Identified non-compliances are automatically marked and highlighted in the model, allowing users to quickly locate and correct problems. A significant advantage of using graph neural networks for non-compliance identification is that it can handle complex building structures and accurately capture potential problems between components, thereby greatly improving the accuracy and efficiency of compliance checks. In addition, this method has good scalability and adaptability, and can handle building models of different sizes and complexities, allowing the system to maintain efficient inspection performance in various scenarios.

[0105] S4, automatically generate a detailed compliance check report; S4 uses Pandoc and LaTeX to generate a compliance check report, which includes but is not limited to the non-compliance items found, the corresponding regulatory clauses, severity, and modification suggestions. S4 also includes the use of risk scoring formulas to quantitatively rate the non-compliance items found, including three severity scores and three occurrence probability recommendations, including:

[0106] Severity Rating:

[0107] 1: Minor impact, no emergency treatment required;

[0108] 3: Moderate impact, needs to be dealt with in the short term;

[0109] 5: Severe impact, must be dealt with immediately;

[0110] Probability of occurrence:

[0111] 1: Rarely occurs;

[0112] 3: Occasionally;

[0113] 5: Occurs frequently;

[0114] The specific risk scoring formula is: R = S × P, where S is the severity score, P is the probability of occurrence, and R is the risk score.

[0115] Specifically, in step S4, the system is responsible for automatically generating a detailed compliance check report, which is generated using Pandoc and LaTeX technology. This combination ensures the high quality of the report in terms of format, layout, and content presentation, and can meet professional and standardized requirements. The compliance check report not only includes the non-compliance items found, but also lists in detail the specific regulatory clauses violated by each non-compliance item, the severity, and the corresponding modification suggestions. In order to make the report more practical and instructive, the system also quantitatively rates the non-compliance items found by using a risk scoring formula.

[0116] In the specific risk assessment process, the system first scores the severity of each non-compliance item, and the score is divided into three levels: 1 point indicates a minor impact, no emergency treatment is required; 3 points indicate a moderate impact, which needs to be handled in the short term; 5 points indicate a serious impact, which must be handled immediately. Then, the system evaluates the probability of occurrence of each non-compliance item based on historical data or expert evaluation, which is also divided into three levels: 1 point indicates very rare occurrence; 3 points indicates occasional occurrence; 5 points indicates frequent occurrence. Then, the system combines the severity score with the probability of occurrence to calculate the risk score of each non-compliance item, which is the product of the two, that is, risk score = severity score × probability of occurrence.

[0117] Through this quantitative scoring, the compliance inspection report can provide decision makers with clearer risk priority information, helping them to allocate resources more effectively and prioritize high-risk issues. This data-driven quantitative analysis method makes the report not only highly objective and accurate, but also easy to understand and operate, significantly improving the efficiency of problem handling and the overall quality of project management. The report generation process is highly automated, ensuring that the results of each inspection can be presented in a timely and accurate manner, further strengthening the application value of the system in compliance inspections.

[0118] S5. After correcting the BIM model according to the report published in step S3, re-enter steps S1-S4 for compliance check until all non-compliance items have been resolved;

[0119] Specifically, in step S5, the system makes necessary corrections to the BIM model based on the compliance check report generated in step S3 for the non-compliances identified in the report. The correction process is performed by the design team or automated tools to ensure that the model complies with the modification suggestions and specification requirements in the report. The corrected BIM model will enter the compliance check process again and be fully reviewed again from steps S1 to S4. Specifically, the corrected model is first re-imported into the system by the BIM data management module, and undergoes a consistency check of the data format and coordinate system to ensure the integrity and standardization of the model data. Subsequently, the system automatically selects the most suitable design specifications based on the project requirements and the latest set of rules, and restarts the compliance check function to check the BIM model item by item. The system uses advanced algorithms to automatically identify and mark any non-compliances that still exist.

[0120] In this process, the system gives full play to its self-learning mechanism, dynamically adjusts the rule base and detection parameters based on historical inspection results and model correction feedback to further reduce the possibility of false positives and false negatives. This cycle will continue until all non-compliance items are resolved, ensuring that the BIM model fully complies with all relevant design specifications and regulatory requirements.

[0121] Through this iterative inspection and correction process, the system not only improves the compliance of the model, but also significantly enhances the accuracy and reliability of the overall design. Each round of inspection and correction provides a more solid foundation for subsequent evaluations, so that the final BIM model can be delivered with the highest quality standards. This process effectively reduces potential risks in the project and avoids compliance issues that may arise in the later construction process, thereby ensuring the smooth progress and successful delivery of the project.

[0122] S6. During the S1-S4 cycle, the system records the results of each compliance check and the corresponding corrective measures as the basic data for the self-learning mechanism. Then, the algorithm automatically optimizes the rule base and adjusts the detection parameters to reduce false positives and false negatives in future checks. When the rules are updated or the design specifications change, the system automatically identifies and introduces new rules to ensure that the compliance check is consistent with the latest standards.

[0123] The self-learning mechanism in step S6 includes the following calculation steps:

[0124] S61. Calculation of false alarm and missed alarm rates:

[0125] False Positive Rate:

[0126] Among them, False Positives are items that the system mistakenly marks as non-compliant, and True Negatives are items that the system correctly marks as compliant;

[0127] False negative rate:

[0128] Among them, False Negatives are non-compliant items that the system failed to detect, and True Positives are items that the system correctly marked as non-compliant;

[0129] S62, parameter adjustment; including gradient descent method and Bayesian update formula:

[0130] Gradient Descent:

[0131] Among them, θ t is the current detection parameter, η is the learning rate, is the gradient of the loss function with respect to the parameters;

[0132] Bayesian update formula:

[0133] Among them, P(H|D) is the confidence of the updated rule, P(D|H) is the probability of new data under the current rule, P(H) is the prior probability of the rule, and P(D) is the marginal probability of the data;

[0134] S63, Rule Importance Score: Evaluate the importance of rules to decide whether to retain or introduce new rules. The calculation formula is:

[0135] Among them, I j It is rule R j Importance score, Gain(R j ,D i ) is the rule R j Right D i The information gain of n is the total amount of data;

[0136] S64, Identification and introduction of new rules: The introduction of new rules is determined by threshold, the formula is: IfI j >Threshold, introduce R j into the rule set,

[0137] Here, Threshold is the preset importance threshold of the introduced rule.

[0138] After obtaining the rule base, detection parameters, and false positive and false negative rate data, step S6 uses deep reinforcement learning for optimization. The calculation formula is:

[0139]

[0140] Here, θ is the policy parameter and J(θ) is the expected cumulative reward.

[0141] Specifically, in step S6, the system continuously optimizes the accuracy and effectiveness of compliance checks through a self-learning mechanism to ensure that the system can dynamically adapt to changes in design specifications and improve the accuracy of checks. The core of this step is that the system automatically records the inspection results and corresponding corrective measures in each compliance check cycle, and uses these data as the basis for self-learning to provide support for subsequent optimization. First, the system calculates the false alarm rate and missed alarm rate in each inspection. The false alarm rate is used to measure the proportion of items that the system mistakenly marks as non-compliant, while the missed alarm rate is used to evaluate the proportion of actual non-compliant items that the system fails to detect. These indicators provide a clear direction for the optimization of the detection algorithm. In order to reduce false alarms and missed alarms, the system uses the gradient descent method to adjust the detection parameters. By minimizing the loss function, the system can gradually optimize the parameters to make the inspection results more accurate.

[0142] In addition, the system also applies the Bayesian update formula to adjust the confidence of the rules in the rule base according to the new data. This process ensures that the rule base can dynamically reflect the latest design requirements and actual conditions, reducing detection errors caused by rule lags. When evaluating the importance of rules, the system determines whether to retain or introduce new rules by calculating the contribution of the rules to the information gain. Rules with higher information gain will be retained and used first in future inspections. To further enhance the adaptability of the system, the system will use thresholds to determine the introduction of new rules after each rule base update to ensure that compliance checks are consistent with the latest design specifications.

[0143] Finally, after acquiring all relevant data, the system inputs it into the deep reinforcement learning model, and through continuous learning and optimization, further improves the efficiency and accuracy of compliance checks. Deep reinforcement learning optimizes the system's detection strategy through the policy gradient method to maximize the cumulative reward. This process can not only dynamically adjust the detection parameters and rule base, but also automatically optimize the system's inspection performance in different scenarios based on historical inspection data. Through these sophisticated self-learning steps, the system can continuously improve its compliance inspection capabilities in complex building design scenarios, ensuring that each inspection result can accurately reflect the current design specifications and actual project requirements, thereby greatly reducing compliance risks in the design phase and improving the overall quality and compliance management level of the project.

[0144] S7. Through multiple rounds of compliance checks and model corrections, the non-compliant items in the model are gradually reduced until all compliance requirements are met. After completing the final compliance check, a comprehensive compliance assessment report is automatically generated to meet all applicable design specifications and regulatory requirements;

[0145] Specifically, in this step, the system gradually reduces the non-compliant items in the BIM model through multiple rounds of compliance checks and model corrections, ensuring that the design model gradually approaches and eventually fully complies with all applicable design specifications and regulatory requirements. After each round of compliance checks, the system will generate a detailed inspection report, indicating the non-compliant items that still exist and their corresponding specification clauses, and put forward specific modification suggestions. After the design team corrects the BIM model according to the report, the system will conduct a compliance check again, troubleshooting and correcting non-compliant items round by round. Through this iterative process, the non-compliant items in the model will gradually decrease, and the compliance of the design model will continue to improve. In this process, the system's self-learning mechanism also continuously optimizes the rule base and detection parameters to improve the accuracy and efficiency of the inspection.

[0146] When the system detects that all non-compliances in the model have been resolved and the model fully complies with all currently applicable design specifications and regulatory requirements, the system will automatically generate a comprehensive compliance assessment report. This report not only summarizes all inspection results, but also records in detail the process of each round of inspection and correction to ensure the transparency and traceability of the entire design process. The final compliance assessment report covers all key design elements and confirms that the model has met all prescribed compliance requirements, providing a solid guarantee for the subsequent implementation of the project. This process ensures the high standard of execution of architectural design and greatly reduces the risk of compliance issues in the later construction and operation stages, ultimately laying the foundation for the successful delivery of the project.

[0147] S8. Archive the final compliance assessment report and the compliant BIM model, and feed back new findings, special circumstances and correction experiences in this project into the system's knowledge base through data comparison.

[0148] Specifically, in this step, after completing the final compliance check and confirming that the BIM model complies with all applicable design specifications and regulatory requirements, the system will archive the generated compliance assessment report and the final compliant BIM model. The purpose of archiving is to ensure that all compliance check results, correction records, and the final compliant model of the project can be properly preserved, providing reliable reference materials for future audits, project reviews, or subsequent projects. This archiving process includes storing the report and model data in the system's secure repository and performing version control to ensure that the design and compliance status of each stage are fully recorded and preserved.

[0149] In addition, in order to improve the self-learning ability of the system, the system will also conduct data comparison and analysis on new situations discovered in this project, special problems encountered, and corrective measures taken. These data will be sorted and fed back to the system's knowledge base to form valuable experience data. Through this feedback mechanism, the system's knowledge base will be continuously updated and expanded, accumulating more actual cases and processing experience on compliance checks. As the knowledge base continues to enrich, the system will conduct compliance checks more intelligently and efficiently in future projects, and will be able to better foresee and solve potential problems. Ultimately, this process not only provides a complete compliance record for the current project, but also provides solid data support for the continuous optimization and improvement of the system, allowing the system to meet compliance requirements with higher accuracy and efficiency when handling new projects.

[0150] Embodiment 2:

[0151] Please see attached Figure 2 The embodiment of the present invention provides a design compliance evaluation system based on BIM, including:

[0152] BIM data management module: This module is responsible for importing and preprocessing BIM model data to ensure the integrity and consistency of model data. Preprocessing includes data format conversion, coordinate system consistency check, and confirmation of connections and dependencies between components, laying the foundation for subsequent compliance checks.

[0153] Rule Library Management Module: This module is used to manage and select applicable design specification rule sets. According to the specific needs of the project, the system automatically selects the most suitable rule set from the provided rule library for application. This module also supports dynamic update of the rule library to ensure that compliance checks are always based on the latest design specifications and regulatory requirements.

[0154] Compliance Check Module: In the Compliance Check Module, the system automatically performs compliance check tasks, identifying and marking potential non-compliance items in the BIM model one by one. This module uses advanced algorithms, such as the neural network model shown in the figure, to identify problems in complex building structures and highlight these problems for the design team to correct.

[0155] Report generation module: This module is responsible for automatically generating compliance check reports. The report not only includes detailed information on non-compliance items, but also quantitatively rates non-compliance items through risk scoring. The system uses Pandoc and LaTeX to generate PDF reports in a standardized format to ensure that the report content is clear and well-structured, which is convenient for review by the project management team and relevant stakeholders.

[0156] Self-learning optimization module: To continuously improve the system's compliance checking capabilities, the self-learning optimization module records the results of each check and automatically optimizes the rule base and detection parameters based on this data. By reducing false positives and false negatives, the system becomes smarter and more efficient in each compliance check. In addition, over time, the system is able to extract patterns from historical data to improve the accuracy of future checks.

[0157] External data acquisition module: This module is used to connect to external data sources to obtain the latest compliance requirements and design specification changes in real time. In order to ensure the timely update of the rule base content, the external data acquisition module supports connecting to external databases through RESTful or GraphQL API, or crawling real-time compliance information from public websites through Web Scraping technology. According to project requirements, the system can set a daily, weekly or monthly scheduled update frequency to ensure that the content in the rule base always reflects the current industry standards and regulatory requirements.

[0158] The BIM model data imported and preprocessed by the BIM data management module is passed to the compliance check module; the rule base management module provides the applicable design specification rule set to the compliance check module; the non-compliant items and related information identified and marked by the compliance check module are passed to the report generation module; the inspection results recorded by the compliance check module are passed to the self-learning optimization module; the compliance check report generated by the report generation module, including the risk score and modification suggestions for the non-compliant items, is provided as feedback to the self-learning optimization module; the self-learning optimization module updates the optimized rule base and detection parameters to the rule base management module; the external data acquisition module obtains the latest compliance requirements and design specification changes from the external data source and passes them to the rule base management module.

[0159] Specifically, BIM data management module → compliance check module: The BIM model data imported and preprocessed by the BIM data management module will be passed to the compliance check module. The preprocessed model data provides a basis for compliance checking, ensuring the accuracy and effectiveness of the check.

[0160] Rule Base Management Module → Compliance Check Module: The Rule Base Management Module provides the applicable design specification rule sets to the Compliance Check Module based on project requirements. The Compliance Check Module uses these rule sets to check the BIM model item by item to identify potential non-compliance items.

[0161] Compliance Check Module → Report Generation Module: After the compliance check module identifies and marks the non-compliant items in the BIM model, it passes these non-compliant items and related information to the report generation module. After receiving this data, the report generation module generates a detailed compliance check report.

[0162] Compliance Check Module → Self-Learning Optimization Module: Each inspection result recorded by the Compliance Check Module, including information on detected non-compliance items, false positives, and missed negatives, will be passed to the Self-Learning Optimization Module. The Self-Learning Optimization Module uses this data for analysis and optimization to improve the accuracy and efficiency of future inspections.

[0163] Report generation module → self-learning optimization module: The compliance check report generated by the report generation module not only records the non-compliance items in detail, but also includes the risk score and modification suggestions of the non-compliance items. This information will be fed back to the self-learning optimization module to help it better optimize the rule base and detection parameters.

[0164] Self-learning optimization module → rule base management module: After analysis and optimization, the self-learning optimization module updates the adjusted and optimized rule base and detection parameters to the rule base management module. In this way, the rule base management module can provide a more accurate set of rules for future compliance checks.

[0165] External data acquisition module → rule base management module: The external data acquisition module obtains the latest compliance requirements and design specification changes from external data sources through API or WebScraping technology, and passes these updated information to the rule base management module. The rule base management module dynamically updates the rule set accordingly to ensure that compliance checks are always based on the latest industry standards and regulatory requirements.

[0166] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A design compliance evaluation method based on BIM, characterized in that: The following steps are involved: S1. Import and prepare the BIM model of the design project, ensure that it contains the design data of building structure, facility layout, and material selection, and pre-process the design data of data format, coordinate system consistency, and connection and dependency between components; S2. According to the specific requirements of the design project, the algorithm automatically selects the applicable rule set from the various compliance rule libraries provided; S3. Start the system's automated compliance check function. The system checks each design content in the BIM model one by one according to the configured rule base, and automatically identifies potential non-compliant items in the model through the set algorithm, and highlights them at the same time; S4. Automatically generate detailed compliance inspection reports; S5. After correcting the BIM model according to the report published in step S3, re-enter steps S1-S4 for compliance check until all non-compliance items have been resolved; S6. During the S1-S4 cycle, the system records the results of each compliance check and the corresponding corrective measures as the basic data for the self-learning mechanism. Then, the algorithm automatically optimizes the rule base and adjusts the detection parameters to reduce false positives and false negatives in future checks. When the rules are updated or the design specifications change, the system automatically identifies and introduces new rules to ensure that the compliance check is consistent with the latest standards. S7. Through multiple rounds of compliance checks and model corrections, the non-compliant items in the model are gradually reduced until all compliance requirements are met. After the final compliance check is completed, a comprehensive compliance assessment report is automatically generated to meet all applicable design specifications and regulatory requirements; S8. Archive the final compliance assessment report and the compliant BIM model, and feed back new findings, special circumstances and correction experiences in this project into the system's knowledge base through data comparison.

2. According to the BIM-based design compliance evaluation method of claim 1, it is characterized in that: In the step S2, the rule base includes but is not limited to building structure specifications, fire safety specifications, energy-saving and environmental protection standards, and the algorithm in the step S2 is specifically a logistic regression algorithm, which recommends the rule set in the rule base according to the geographical location and building use factors of the project. The specific calculation formula is: Among them, z is the result of linear combination, β0 is the bias term, β i is the weight of the i-th feature, x i is the i input feature value, n is the total number of features, and finally through x1,x2,…,x n In the case of , the probability p that the feature combination meets a certain rule set is calculated, so as to automatically match the most suitable rule set.

3. The design compliance evaluation method based on BIM according to claim 1 is characterized in that: The S3 step uses a graph neural model to automatically identify model non-compliant items, and its calculation steps are: S31, initializing a feature vector for each node (i.e., a component in the BIM model) to represent the initial state of the node; S32. In each layer of propagation, a node receives "messages" from its neighboring nodes and updates its own feature representation. The node feature update formula is: in, is the feature representation of node v at layer k, is the set of neighbor nodes of node v, f message is the message passing function used to combine the features of neighbor nodes; S33, after several layers of propagation, the final feature representation of the node It is used to determine whether there are any non-compliant items at the node, and the output is displayed through the classifier: Among them, W class is the weight matrix of the classifier, y v is the probability distribution of node v belonging to different categories (compliant / non-compliant).

4. The design compliance evaluation method based on BIM according to claim 1 is characterized in that: The S4 step uses Pandoc and LaTeX to generate a compliance check report, which includes but is not limited to the non-compliance items found, the corresponding regulatory clauses, the severity and modification suggestions.

5. The design compliance evaluation method based on BIM according to claim 1 is characterized in that: The S4 step also includes using a risk scoring formula to quantify the non-compliance items found, including three severity scores and three occurrence probability recommendations, specifically including: Severity Rating: 1: Minor impact, no emergency treatment required; 3: Moderate impact, needs to be dealt with in the short term; 5: Serious impact, must be dealt with immediately; Probability of occurrence: 1: Rarely occurs; 3: Occasionally; 5: Occurs frequently; The specific risk scoring formula is: R = S × P, where S is the severity score, P is the probability of occurrence, and R is the risk score.

6. The design compliance evaluation method based on BIM according to claim 1 is characterized in that: The self-learning mechanism in step S6 includes the following calculation steps: S61. Calculation of false alarm and missed alarm rates: False Positive Rate: Among them, False Positives are items that the system mistakenly marks as non-compliant, and True Negatives are items that the system correctly marks as compliant; False negative rate: Among them, False Negatives are non-compliant items that the system failed to detect, and True Positives are items that the system correctly marked as non-compliant; S62, parameter adjustment; including gradient descent method and Bayesian update formula: Gradient Descent: Among them, θ t is the current detection parameter, η is the learning rate, is the gradient of the loss function with respect to the parameters; Bayesian update formula: Among them, P(H|D) is the confidence of the updated rule, P(D|H) is the probability of new data under the current rule, P(H) is the prior probability of the rule, and P(D) is the marginal probability of the data; S63, Rule Importance Score: Evaluate the importance of rules to decide whether to retain or introduce new rules. The calculation formula is: Among them, I j It is rule R j Importance score, Gain(R j ,D i ) is the rule R j Right D i The information gain of n is the total amount of data; S64, Identification and introduction of new rules: The introduction of new rules is determined by threshold, the formula is: IfI j >Threshold, introduce R j into the rule set, Here, Threshold is the preset importance threshold of the introduced rule.

7. The design compliance evaluation method based on BIM according to claim 1 is characterized in that: After obtaining the rule base, detection parameters, and false positive and false negative rate data, step S6 is optimized by deep reinforcement learning. The calculation formula is: Here, θ is the policy parameter and J(θ) is the expected cumulative reward.

8. A BIM-based design compliance evaluation system, used in a BIM-based design compliance evaluation method according to any one of claims 1 to 7, characterized in that: include: BIM data management module, used to import and pre-process BIM model data; A rule base management module is used to manage and select applicable design specification rule sets and dynamically update the rule base; A compliance checking module that automates compliance checking, identifies and marks potential non-compliance items in the model; Report generation module, which automatically generates compliance check reports, including details of non-compliance items and risk scores; Self-learning optimization module, which is used to record inspection results, automatically optimize the rule base and detection parameters, and reduce false positives and false negatives; External data acquisition module: used to connect to external data sources, obtain the latest compliance requirements and design specification changes in real time, and synchronize this updated information to the system's rule base management module.

9. The BIM-based design compliance evaluation system according to claim 8, characterized in that: The external data acquisition module includes but is not limited to crawling the real-time compliance rule library content through RESTful or GraphQL API and Web Scraping, and updating it on a daily, weekly or monthly basis during the process.

10. The BIM-based design compliance evaluation system according to claim 8, characterized in that: The BIM model data imported and preprocessed by the BIM data management module is transferred to the compliance checking module; the rule base management module provides the applicable design specification rule set to the compliance checking module; the non-compliant items identified and marked by the compliance checking module and related information are transferred to the report generation module; The inspection results recorded by the compliance inspection module are transmitted to the self-learning optimization module; The compliance check report generated by the report generation module includes risk scores and modification suggestions for non-compliant items, and provides feedback to the self-learning optimization module; The self-learning optimization module updates the optimized rule base and detection parameters to the rule base management module; The external data acquisition module acquires the latest compliance requirements and design specification changes from the external data source and transmits them to the rule base management module.

Images