Building information model quality detection method and device and storage medium

Abstract

The application discloses a kind of building information model quality detection method, device and storage medium, it is related to building information detection technical field, mainly in can improve the quality detection precision and detection efficiency of building information model. Including: in response to the quality detection signal of target building information model, the detection specification information corresponding to the target building information model is obtained;The detection specification information is classified, and the detection specification information under different classification categories is obtained;The model data corresponding to the target building information model is determined, the model data is matched with the detection specification information under each classification category, and based on the matching result, the mapping relationship between the model data and the detection specification information under each classification category is constructed;Based on the mapping relationship, the model data is detected using the detection specification information, and the quality detection result of the target building information model is obtained.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of building information detection, and in particular to a building information model quality detection method and device and storage medium. BACKGROUND

[0002] All information of a building project is integrated into a digital model to obtain a building information model. The quality of the building information model directly affects the accuracy and value of the building information. Therefore, quality detection of the building information model is an important means to ensure the accuracy of the building information.

[0003] At present, the building information model is usually detected by manual detection. However, due to the negligence of the detection personnel and the uneven technical level, the manual detection method may cause detection errors or missed detection, thereby reducing the quality detection accuracy of the building information model. At the same time, the manual detection method also reduces the quality detection efficiency of the building information model. SUMMARY

[0004] The present application provides a building information model quality detection method and device and storage medium, which can improve the quality detection accuracy and efficiency of the building information model.

[0005] According to a first aspect of the present application, a building information model quality detection method is provided, comprising:

[0006] In response to a quality detection signal of a target building information model, detection specification information corresponding to the target building information model is obtained;

[0007] The detection specification information is classified to obtain detection specification information under different classification categories;

[0008] Model data corresponding to the target building information model is determined, the model data is matched with the detection specification information under each classification category, and based on the matching result, a mapping relationship between the model data and the detection specification information under each classification category is constructed;

[0009] Based on the mapping relationship, the model data is quality detected by using the detection specification information to obtain a quality detection result of the target building information model.

[0010] Optionally, the method further comprises:

[0011] Obtain component attribute data from model data of the target building information model, and classify the component attribute data according to component types to obtain to-be-detected attribute data corresponding to each component in the target building information model, wherein the to-be-detected attribute data includes at least one of attribute data item quantity, attribute value corresponding to an attribute data item, and attribute value type;

[0012] Obtain standard attribute data of each component in the detection specification information, and match the standard attribute data with the to-be-detected attribute data, and determine a component attribute detection result of the target building information model based on a matching result.

[0013] Optionally, the detection specification information is classified to obtain detection specification information under different classification categories, including:

[0014] Initialize a centroid vector of a centroid corresponding to a different classification category, and determine an information feature vector corresponding to the detection specification information;

[0015] Based on the information feature vector and the centroid vector, calculate a distance between each detection specification information and a centroid corresponding to each classification category, and based on the distance, divide each detection specification information into each classification category;

[0016] Based on an information feature vector corresponding to detection specification information in each classification category, determine an updated centroid vector corresponding to each classification category;

[0017] Based on the updated centroid vector, re-divide each detection specification information into each classification category until the updated centroid vector does not change, and determine detection specification information finally divided into each classification category as detection specification information under different classification categories.

[0018] Optionally, the different classification categories include an attribute type category, a space type category, a geometry type category, and a calculation type category.

[0019] Based on the mapping relationship, perform quality detection on the model data by using the detection specification information to obtain a quality detection result of the target building information model, including:

[0020] Identify each component in the target building information model, and determine to-be-detected attribute type data, to-be-detected space type data, to-be-detected geometry type data, and to-be-detected calculation type data of each component in the model data;

[0021] Based on the mapping relationship, determine standard attribute type data, standard space type data, standard geometry type data, and standard calculation type data of each component in the detection specification information;

[0022] perform attribute specification detection on the to-be-detected attribute type data by using the standard attribute type data corresponding to the same component, to obtain an attribute type quality detection result;

[0023] perform spatial position detection on the to-be-detected spatial type data by using the standard spatial type data corresponding to the same component, to obtain a spatial type quality detection result;

[0024] perform geometric dimension detection on the to-be-detected geometric type data by using the standard geometric type data corresponding to the same component, to obtain a geometric type quality detection result;

[0025] perform topological relation detection on the to-be-detected calculation type data by using the standard calculation type data corresponding to the same component, to obtain a calculation type quality detection result.

[0026] Optionally, the determining the model data corresponding to the target building information model comprises:

[0027] determining IFC model data corresponding to the target building information model, performing component entity identification in the IFC model data to obtain each component contained in the target building information model, and determining component data corresponding to each component based on the IFC model data, and determining the component data corresponding to each component as the model data, wherein the method of performing component entity identification in the IFC model data comprises:

[0028] determining each word segment contained in the IFC model data, and determining semantic information vectors corresponding to the each word segment;

[0029] inputting each semantic information vector into a preset position prediction model to perform position prediction, to obtain predicted position information of the each word segment in a to-be-predicted component entity, the preset position prediction model being obtained by being trained in advance based on a sample data set with position labels;

[0030] inputting the semantic information vector and the predicted position information into a preset entity identification model to perform entity identification, to obtain each component entity contained in the IFC model data, the preset entity identification model being obtained by being trained in advance based on a sample data set with entity category labels.

[0031] Optionally, after the quality detection of the model data by using the detection specification information based on the mapping relationship, to obtain the quality detection result of the target building information model, the method further comprises:

[0032] determining deviation information between the target building information model and the detection specification information based on the quality detection result;

[0033] Generate a quality detection report based on the deviation type corresponding to the deviation information, the component identifier and the corresponding component position, and the detection specification information, and send the quality detection report to an auditing terminal for auditing;

[0034] Receive the auditing result of the auditing terminal, and in the case of passing the auditing, construct a model correction statement based on the detection specification information, and automatically correct the deviation information using the model correction statement to obtain the corrected target building information model.

[0035] Optionally, after obtaining the detection specification information corresponding to the target building information model, the method further comprises:

[0036] Obtain detection requirement information for the target building information model, and determine the quality detection type for the target building information model based on the detection requirement information;

[0037] Determine a preset quality detection model corresponding to the quality detection type;

[0038] Determine a specification feature vector corresponding to the detection specification information, and determine a model feature vector corresponding to the model data;

[0039] Cross-process the specification feature vector and the model feature vector to obtain a cross-detection vector;

[0040] Input the cross-detection vector into the preset quality detection model for quality detection to obtain the quality detection result of the target building information model.

[0041] According to a second aspect of the present application, a building information model quality detection device is provided, comprising:

[0042] An acquisition unit is configured to acquire detection specification information corresponding to a target building information model in response to a quality detection signal of the target building information model;

[0043] A classification unit is configured to classify the detection specification information to obtain detection specification information under different classification categories;

[0044] A construction unit is configured to determine model data corresponding to the target building information model, match the model data with the detection specification information under each classification category, and construct a mapping relationship between the model data and the detection specification information under each classification category based on the matching result;

[0045] A detection unit is configured to perform quality detection on the model data using the detection specification information based on the mapping relationship to obtain a quality detection result of the target building information model.

[0046] According to a third aspect of the present application, a computer readable storage medium is provided, having stored thereon a computer program which, when executed by a processor, implements the above method for quality detection of building information model.

[0047] According to a fourth aspect of the present application, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above method for quality detection of building information model when executing the program.

[0048] The method, device and storage medium for quality detection of building information model provided by the present application can improve the quality detection efficiency and precision of building information model. Meanwhile, the present application can help to systematize the complex detection specification information, so that the quality detection process is clearer and more orderly, and the classified detection specification information is easier to manage and update, facilitating the subsequent quality detection work. BRIEF DESCRIPTION OF DRAWINGS

[0049] The accompanying drawings, which are included to provide a further understanding of the present application and are incorporated in and constitute a part of this application, illustrate embodiments of the present application and serve to explain the principles of the present application. In the drawings:

[0050] Figure 1 Fig. 1 shows a flow chart of a method for quality detection of building information model according to an embodiment of the present application;

[0051] Figure 2Another building information model quality detection method provided by the embodiment of the present application is shown in the flow chart;

[0052] Figure 3 The structure diagram of the building information model quality detection device provided by the embodiment of the present application is shown in the structure diagram;

[0053] Figure 4 The structure diagram of another building information model quality detection device provided by the embodiment of the present application is shown in the structure diagram;

[0054] Figure 5 The entity structure diagram of the computer device provided by the embodiment of the present application is shown in the entity structure diagram. DETAILED DESCRIPTION

[0055] The present application will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.

[0056] At present, the quality of the building information model is detected by manual detection, which may result in detection errors or missed detection. In addition, the manual detection method also reduces the quality detection efficiency of the building information model.

[0057] In order to solve the above problems, the embodiment of the present application provides a building information model quality detection method, as shown in the figure, the method comprises the steps of: Figure 1

[0058] 101、In response to the quality detection signal of the target building information model, the detection specification information corresponding to the target building information model is acquired.

[0059] The detection specification information is prepared in advance according to the actual demand and various standards. The detection specification information contains standard specification information of the integrity, specification, standard, consistency and specification of the target building information model. For example, the standard specification of the geometric size of each component in the target building information model, the standard specification of the spatial position, the standard specification of the topological relationship, the standard specification of the attribute information, etc.

[0060] 102、The detection specification information is classified to obtain the detection specification information under different classification categories.

[0061] The different classification categories include: attribute type category, spatial type category, geometric type category and calculation type category.

[0062] ​For the embodiment of the application, in order to improve the quality detection efficiency of the target building information model, first, the detection specification information needs to be classified, based on which, step 102 specifically includes: initializing the centroid vector of the centroid corresponding to different classification categories, and determining the information feature vector corresponding to the detection specification information; based on the information feature vector and the centroid vector, calculating the distance between each detection specification information and the centroid corresponding to each classification category, and based on the distance, dividing each detection specification information into each classification category; based on the information feature vector corresponding to the detection specification information in each classification category, determining the updated centroid vector corresponding to each classification category; based on the updated centroid vector, re-dividing each detection specification information into each classification category until the updated centroid vector does not change, and determining the detection specification information finally divided into each classification category as the detection specification information under different classification categories.

[0063] Specifically, the information feature vector corresponding to the detection specification information is determined by using word embedding and the like, and the centroid vector of the centroid corresponding to different classification categories is initialized, the distance of each information feature vector to different centroid vectors is calculated for the information feature vector respectively corresponding to the detection specification information, and each detection specification information is allocated to the classification category corresponding to the nearest centroid vector, then for each category, the centroid of each category and the centroid vector corresponding thereto are recalculated, and the detection specification information is re-divided into different classification categories, so as to continuously divide the detection specification information until the position of the centroid does not change, i.e. the centroid vector does not change, and finally the detection specification information divided into different classification categories is determined as the detection specification information under different classification categories. Finally, the building information model is detected by using the detection specification information under different classification categories, when the building information model is detected by using the detection specification information, the classified detection specification information can be more orderly, which can quickly match the specification related to the current detection task, and improve the quality detection efficiency of the building information model. At the same time, when new specifications are released or existing specifications are updated, they can be easily added to the corresponding classification, simplifying the updating process.

[0064] 103、determine the model data corresponding to the target building information model, match the model data with the detection specification information under each classification category, and based on the matching result, construct the mapping relationship between the model data and the detection specification information under each classification category.

[0065] For the embodiment of the present application, in order to perform quality detection on the building information model, the model data in the building information model needs to be determined first, based on which, step 103 specifically includes: determining the IFC model data corresponding to the target building information model, and performing component entity recognition in the IFC model data to obtain each component contained in the target building information model, and based on the IFC model data, determining the component data corresponding to each component, and determining the component data corresponding to each component as the model data, wherein the method of performing component entity recognition in the IFC model data includes: determining each word in the IFC model data, and determining the semantic information vector corresponding to each word; inputting each semantic information vector into a preset position prediction model to perform position prediction to obtain the predicted position information of each word in the to-be-predicted component entity, the preset position prediction model being obtained by pre-training based on a sample data set with position labels; inputting the semantic information vector and the predicted position information into a preset entity recognition model to perform entity recognition to obtain each component entity contained in the IFC model data, the preset entity recognition model being obtained by pre-training based on a sample data set with entity category labels.

[0066] Wherein, the IFC (Industry Foundation Classes, a data model standard) model data is information constructed based on the IFC model data. The IFC model data describes a standardized digital model of a building and its construction process, including the fields of housing and infrastructure. The IFC model data can contain all building information, such as structural components, mechanical and electrical systems, materials, colors, positional relationships, abstract concepts (such as performance, cost), and construction process information. The component data includes geometric information, spatial information, attribute information, topological relationships, and other information of the component. The to-be-predicted component entity can be an entity that exists in reality, or an entity that we are going to identify.

[0067] Specifically, in order to improve the position prediction accuracy of the preset position prediction model, firstly, the preset position prediction model needs to be constructed, based on which, the method comprises: constructing an initial position prediction model and obtaining a sample data set, wherein each sample word segmentation with a position label in the sample text is contained in the sample data set; the sample data set is divided into training data and test data, the initial position prediction model is trained by using the training data, and the initial position prediction model after training is tested by using the test data, and finally the initial position prediction model meeting the test condition is determined as the preset position prediction model. Similarly, in order to improve the entity recognition accuracy of the preset entity recognition model, firstly, the preset entity recognition model needs to be constructed, based on which, the method comprises: constructing an initial entity recognition model and obtaining a sample data set, wherein each sample word segmentation with an entity category label and its corresponding sample position information is contained in the sample data set; the sample data set is divided into training data and test data, the initial entity recognition model is trained by using the training data, and the entity recognition model after training is tested by using the test data, and finally the initial entity recognition model meeting the test condition is determined as the preset entity recognition model.

[0068] Further, the target building information model is converted to obtain IFC model data, each standard token in the preset token configuration table is used to split the IFC model data to obtain each token contained in the IFC model data, and a query vector, a key vector and a value vector corresponding to each token in the IFC model data are obtained; the query vector corresponding to a first target token in each token is multiplied by the key vector corresponding to each token to obtain an attention score of each token for the first target token; and the attention score corresponding to each token is multiplied by the value vector and summed to obtain a semantic information vector corresponding to the first target token. Then, the semantic information vector corresponding to each token is input into a preset position prediction model to perform first position prediction, middle position prediction and last position prediction, that is, whether each token is a start token of a to-be-predicted component entity is predicted, whether each token is an end token of the to-be-predicted component entity is predicted, and whether each token is a token in an intermediate part of the to-be-predicted component entity is also predicted. At the same time, according to the first position prediction information, the middle position prediction information and the last position prediction information, the number of entities corresponding to the target token at the first position of the to-be-predicted component entity can be determined. Finally, according to the semantic information vector of each token, the first position prediction information, the middle position prediction information and the last position prediction information of each token in the to-be-predicted component entity, each entity contained in the IFC model data is determined, and finally the component data corresponding to each component is determined based on the IFC model data. Thus, by first determining the position information of each token in the to-be-predicted component entity in the IFC model data, the position of each token in the entity can be predicted, and finally each entity contained in the IFC model data is recognized according to the position information. Not only the component entity of continuous words in the IFC model data can be recognized, but also the component entity of discontinuous words in the IFC model data can be recognized, thereby improving the recognition accuracy of the entity and avoiding missing detection.

[0069] Further, after determining the model data corresponding to the target building information model, the model data can be matched with the detection specification information under each classification category based on the component identifier, that is, the model data and the detection specification information corresponding to each component are matched, and a mapping relationship between the model data and the detection specification information corresponding to each component is established. When quality detection of each component in the target building information model is needed, the detection specification information corresponding to the model data of the component is linked based on the component identifier and the mapping relationship, and finally the model data of the component is quality detected by using the detection specification information. Thus, by establishing the mapping relationship between the model data and the detection specification information under each classification category, the correspondence between the model data and the detection specification can be automatically recognized through the mapping relationship, thereby quickly locating potential problems, realizing automatic detection, reducing manual intervention, and improving detection efficiency and detection accuracy.

[0070] 104. Based on the mapping relationship, the model data is detected by using the detection specification information to obtain a quality detection result of the target building information model.

[0071] For the embodiment of the present application, for example, if the quality of the beam in the building information model is to be detected, first, based on the mapping relationship between the model data corresponding to the beam (the model data can be the geometric data, spatial position data, topological relationship data with other structures, etc. of the beam) and the detection specification information, the model data corresponding to the beam and the detection specification information can be automatically and quickly linked in the database, and the model data of the beam is detected by using the detection specification information. The embodiment of the present application can help to systematize and organize the complex detection specification information, so that the quality detection process is more rapid, clear and orderly. At the same time, the classified detection specification information is easier to manage and update, which is convenient for subsequent quality detection work. The embodiment of the present application can realize the rapid and accurate positioning of the detection specification information and the automatic detection of the model data by establishing the mapping relationship between the model data and the detection specification information, so as to improve the quality detection efficiency and the quality detection precision of the building information model.

[0072] In another embodiment of the present application, the quality detection includes standard and specification detection. For the standard detection, the detection method includes: obtaining component attribute data from the model data of the target building information model, and classifying the component attribute data according to the component type to obtain the to-be-detected attribute data corresponding to each component in the target building information model, wherein the to-be-detected attribute data includes at least one of the attribute data item number, the attribute value corresponding to the attribute data item and the attribute value type; obtaining the standard attribute data of each component in the detection specification information, and matching the standard attribute data with the to-be-detected attribute data, and determining the component attribute detection result of the target building information model based on the matching result.

[0073] The component attribute data includes: attribute information naming, description, attribute value, unit.

[0074] Specifically, all components in the building information model (such as columns, beams, plates, walls, windows, etc.) are obtained, and the stored component attribute data in each component is further obtained. By component classification and detection specification information classification, the corresponding component attribute data and detection specification information of each component are associated and mapped. The inspection configuration item of the additional attribute value in the detection specification information (such as: the numerical attribute inspection range can be set to 0-100, and the enumeration attribute inspection range can be set to the related type), and the attribute value inspection configuration item is dynamically configured and managed in the attribute inspection item module, which can extend multiple inspection rule items. The building information model analysis engine is used to open and load the building information model and manage the model components. The model quality review is performed, and through the detection specification information in the mapping relationship, the component naming, attribute information item, attribute value type, attribute value range and other elements can be checked, and the quality of the building information model is comprehensively guaranteed. The review results component list and related error information (attribute missing, attribute value incorrect, attribute value missing) are listed, the component error position is located and browsed, and the review report is exported. If the standard is found to be deviated, the correction scheme is provided, the user confirms, and the data elements of the original building information model are automatically corrected (batch hanging missing attributes, setting attribute values, etc.). The specific quality detection results are shown in the following table:

[0075]

[0076]

[0077] After quality detection, the detection report is issued, and one type of detection report is shown below:

[0078]

[0079] For the problematic building information, an automatic correction function is provided. For example, the attribute item is incomplete, and the attribute information name is filled in error. When the attribute item is incomplete, the corresponding detection specification information is directly added to the related component to increase the attribute item, ensuring the integrity of the component attribute information, which is mainly realized by adding IFC statements in the building information model. If the attribute information name is filled in error, the attribute information name in the corresponding detection specification information can be modified, which is mainly realized by modifying the related statements of the IFC model.

[0080] According to the quality detection method of the building information model provided by the application, compared with the current quality detection method of the building information model by artificial, the application obtains the detection specification information corresponding to the target building information model in response to the quality detection signal of the target building information model, classifies the detection specification information to obtain the detection specification information under different classification categories, determines the model data corresponding to the target building information model, matches the model data with the detection specification information under each classification category, constructs the mapping relationship between the model data and the detection specification information under each classification category based on the matching result, and finally detects the quality of the model data by using the detection specification information based on the mapping relationship to obtain the quality detection result of the target building information model. Therefore, by classifying the detection specification information and establishing the mapping relationship between the model data and the detection specification information of different classification categories, the quality of the target building information model is directly detected by using the mapping relationship. Since the detection specification information has been systematically classified and the accurate mapping relationship has been established with the model data, the quality detection result can comprehensively and accurately reflect the quality condition of the model data, and the time wasted by artificial detection and the detection error can be avoided, so that the quality detection efficiency and the quality detection precision of the building information model can be improved. At the same time, by classifying the detection specification information, the application can help to systematize and systematize the complex detection specification information, so that the quality detection process is more clear and orderly. Meanwhile, the classified detection specification information is easier to manage and update, which is convenient for subsequent quality detection work.

[0081] Further, in order to better illustrate the above process of detecting the quality of the building information model, as a refinement and expansion of the above embodiment, the embodiment of the application provides another quality detection method of the building information model, as shown in Figure 2 The method comprises the following steps:

[0082] 201, in response to the quality detection signal of the target building information model, the detection specification information corresponding to the target building information model is obtained.

[0083] Specifically, the building information model detection database provides the building information model classification and coding standard file (PDF / Word format), and the program opens the standard file and uses the PDF / Word analysis module to obtain the standard file document content (detection specification information). The standard related attribute rule data is parsed and stored in a structured manner, which is used as the basis for subsequent checking.

[0084] 202, the detection specification information is classified to obtain the detection specification information under different classification categories, wherein the different classification categories include: attribute type category, space type category, geometry type category and calculation type category.

[0085] Specifically, the detection specification information is classified and typed, and is divided into attribute type, space type, geometric type, calculation type and the like. The building information model data is matched with the digital structure of different types to realize automatic mapping of the specification provisions and the model data.

[0086] 203. Determine the model data corresponding to the target building information model, match the model data with the detection specification information under each classification category, and based on the matching result, construct the mapping relationship between the model data and the detection specification information under each classification category.

[0087] Specifically, each structural member contained in the target building information model is identified, and based on the model data, the member data (such as member attribute data, member geometric data, member spatial data, member topological relationship data, etc.) corresponding to each structural member is determined. According to the construction attribute data, the attribute detection specification information corresponding to the member attribute data is matched in the detection specification information under the attribute type, and then the mapping relationship between the member attribute data corresponding to each member and the attribute detection specification information corresponding thereto is established. According to the member geometric data, the geometric detection specification information corresponding to the member geometric data is matched in the detection specification information under the geometric type, and then the mapping relationship between the member geometric data corresponding to each member and the geometric detection specification information corresponding thereto is established. According to the member spatial position data, the spatial detection specification information corresponding to the member spatial position data is matched in the detection specification information under the space type, and then the mapping relationship between the member spatial position data corresponding to each member and the spatial detection specification information corresponding thereto is established. According to the member topological relationship data, the topological detection specification information corresponding to the member topological relationship data is matched in the detection specification information under the calculation type, and then the mapping relationship between the member topological relationship data corresponding to each member and the topological detection specification information corresponding thereto is established. Finally, according to the mapping relationship, the detection specification information corresponding to each member can be automatically linked, the automatic detection of the member data is realized, the manual participation is reduced, and thus the quality detection efficiency and the quality detection precision of the building information model are improved.

[0088] 204. Identify each member in the target building information model, and determine the attribute type data to be detected, the space type data to be detected, the geometric type data to be detected, and the calculation type data to be detected of each member in the model data.

[0089] For the embodiment of the application, each component in the target building information model is identified by using a preset entity recognition model. Since the model data records the component data (attribute type data, spatial type data, geometric type data, and calculation type data) corresponding to each component, the to-be-detected attribute type data, the to-be-detected spatial type data, the to-be-detected geometric type data, and the to-be-detected calculation type data corresponding to each component can be identified in the model data based on the component identifier of each component.

[0090] 205. Based on the mapping relationship, the standard attribute type data, the standard spatial type data, the standard geometric type data, and the standard calculation type data of each component are determined in the detection specification information.

[0091] Specifically, when the component data of each component needs to be detected, the detection specification information corresponding to each component (including the standard attribute type data, the standard spatial type data, the standard geometric type data, and the standard calculation type data) can be automatically linked according to the mapping relationship between the various types of data corresponding to each component and the detection specification information of the corresponding type.

[0092] 206. The attribute specification detection is performed on the to-be-detected attribute type data by using the standard attribute type data corresponding to the same component, and an attribute type quality detection result is obtained.

[0093] The attribute type data includes the name, material, weight, type, performance, supplier, shelf life, mandatory items, and attribute items of the component.

[0094] Specifically, first, the component to be checked is identified, and then all the attribute information items of the component are traversed. After indexing to a specific attribute item, the value of the attribute item is compared with the standard attribute type data based on the mapping relationship. If the value, quantity, and type of the attribute item are consistent with the standard attribute type specification data, it means that the attribute type quality detection result is passed. If they are not consistent, it means that the attribute type quality detection result is failed.

[0095] 207. The spatial position detection is performed on the to-be-detected spatial type data by using the standard spatial type data corresponding to the same component, and a spatial type quality detection result is obtained.

[0096] The spatial type data refers to the position, shape, size, and spatial position relationship with other components of the component in the three-dimensional space.

[0097] Specifically, first, one component or several components or spatial entities to be checked are identified, the positions of the components or spaces are obtained, the standard spatial type data corresponding to each component is determined based on the mapping relationship, the envelope relationship of the components in the three-dimensional space is detected using the standard spatial type data, it is judged whether the spatial type data of the components conforms to the standard spatial type specification data, if it conforms, it is determined that the spatial type quality detection result of the component passes, otherwise it is determined that the spatial type quality detection result of the component fails.

[0098] 208. Geometric dimension detection is performed on the to-be-detected geometric type data using the standard geometric type data corresponding to the same component, and a geometric type quality detection result is obtained.

[0099] The geometric type data refers to the shape, length, width, height, curvature, angle, etc. of the component.

[0100] Specifically, first, the component to be checked is identified, the specific geometric information of the component is obtained, the standard geometric data of the component is determined based on the mapping relationship, and then the to-be-detected geometric type data of the component is compared with the standard geometric type data, it is judged whether the to-be-detected geometric type data conforms to the requirements of the standard geometric type data, if it conforms, it is determined that the geometric type quality detection result passes, otherwise, it is determined that the geometric type quality detection result fails.

[0101] 209. Topological relationship detection is performed on the to-be-detected calculation type data using the standard calculation type data corresponding to the same component, and a calculation type quality detection result is obtained.

[0102] The calculation type data refers to the topological relationship of the component, such as the evacuation distance, i.e. the distance from the point of each component to the evacuation port.

[0103] Specifically, it is necessary to calculate and analyze the geometric and topological relationship of the relevant entity object in the IFC model, such as evacuation distance checking, which requires calculating the distance from each component point in the project to the evacuation port, and based on the mapping relationship, the standard calculation type data corresponding to the component is determined, it is judged whether the distance of the evacuation port conforms to the standard calculation type data, if it conforms, it is determined that the calculation type quality detection result passes, otherwise, it is determined that the calculation type quality detection result fails. Calculation type data detection requires comprehensive extraction of geometric dimension, spatial position and other information of the component. The specific value is calculated according to the index of the standard specification, and compared with the range specified by the standard specification.

[0104] Further, after quality detection is performed on the target building information model, a quality detection report can be generated based on the detection results to trace problems and automatically correct the detected deviation information. Based on this, the method comprises: determining deviation information between the target building information model and the detection specification information based on the quality detection results; generating a quality detection report based on the deviation type corresponding to the deviation information, the component identifier and the component position corresponding thereto, and the detection specification information, and sending the quality detection report to an audit terminal for auditing; receiving the audit result of the audit terminal, and in the case of passing the audit, constructing a model correction statement based on the detection specification information, and automatically correcting the deviation information using the model correction statement to obtain a corrected target building information model.

[0105] The deviation information refers to the deviation between the model data and the to-be-detected specification information. For example, the constructed length is 50 m, and the standard length corresponding to the construction in the detection specification information is 80 m, so the deviation information is 30 m shorter than the standard length. The construction has no name attribute, and the name corresponding to the construction in the detection specification information is beam, so the deviation information is the lack of the name beam. The deviation type is, for example, length shortening type, attribute missing type, position offset type, etc. The component position refers to the spatial position of the component with deviation. The detection specification information refers to the detection specification information of the component with deviation under the deviation.

[0106] Specifically, a quality detection report is generated based on the deviation type corresponding to the deviation information, the component identifier and the component position corresponding thereto, and the detection specification information, and the quality detection report is sent to an auditor for auditing. After the audit is correct, the system can automatically correct the deviation based on the detection specification information corresponding to the deviation information using a pre-constructed correction statement to obtain a corrected building information model. By automatically correcting the deviation information, the correction efficiency and accuracy can be improved. The following table shows an exemplary quality detection report:

[0107]

[0108] In another embodiment of the present application, the quality detection of the building information model can also be performed by the following method: obtaining detection requirement information for the target building information model, and determining the quality detection type for the target building information model based on the detection requirement information; determining a preset quality detection model corresponding to the quality detection type; determining a specification feature vector corresponding to the detection specification information, and determining a model feature vector corresponding to the model data; performing cross processing on the specification feature vector and the model feature vector to obtain a cross detection vector; inputting the cross detection vector into the preset quality detection model for quality detection to obtain the quality detection result of the target building information model.

[0109] The detection requirement information refers to the type of quality detection performed on the building information model, such as specification detection, standard detection, etc. Different detection types correspond to different preset quality detection models.

[0110] Specifically, first, the preset quality detection model corresponding to different detection types is constructed, and the specific construction method includes: constructing an initial model, and obtaining a sample data set, wherein the sample data set includes sample detection specification information with quality detection labels and model data of sample building information models; dividing the sample data set into training data and test data, training the initial model using the training data, and testing the trained initial model using the test data, and finally determining the initial model that meets the test conditions (the number of test times meets the requirements or the test accuracy meets the requirements) as the preset quality detection model.

[0111] Further, the specification feature vector corresponding to the detection specification data, the model feature vector corresponding to the model data are determined respectively by using word embedding, etc., and the specification feature vector, the model feature vector are processed at the feature level to obtain a feature cross vector; the specification feature vector, the model feature vector are processed at the element level to obtain an element cross vector; the specification feature vector, the model feature vector are processed at the low-order level to obtain a low-order cross vector; the feature cross vector, the element cross vector, the low-order cross vector are transformed by using a preset transformation function to obtain a cross detection vector. Finally, the cross detection vector is input into the preset quality detection model, and the quality detection model can output the quality detection result.

[0112] Specifically, in order to make full use of the relationship between the data, extract more implicit features, and at the same time take into account the high-order and low-order processing, make the data more fully utilized, and the prediction result obtained later more accurate, meet the needs of actual application scenarios, the specification feature vector and the model feature vector need to be cross-processed. The specific cross-processing method is as follows: if the specification feature vector is (a1, a2) and the model feature vector is (b1, b2), the specific cross-processing method includes: cross-processing between different feature vectors at the feature level, that is, after Hadamard product is performed on all elements between the vectors, convolution transformation is performed again under a certain weight, and the feature cross vector is f(w*(a1*b1, a2*b2)); at the same time, cross-processing between all feature vector data at the element level, that is, after Hadamard product is performed on each element between the vectors, different weight values are assigned to each product result, and linear transformation is performed again, and the element cross vector is f(w1*a1*b1, w2*a2*b2); in addition, low-order cross-processing is performed on all feature vectors, a weight coefficient is assigned to the cross-processed result, and linear transformation is performed again, and the low-order cross vector is f(w2(a1, a2, b1, b2)); finally, the above feature cross vector, element cross vector, and low-order cross vector are transformed by using a preset transformation function (the preset transformation function can be set according to actual conditions, and the embodiment does not limit this). It should be noted that the above examples are only illustrative and do not limit the embodiments of the present application. Therefore, by cross-processing the specification feature vector and the model feature vector, different features can be automatically or explicitly combined to generate new feature combinations. These combined features can contain complex nonlinear relationships between original features, so that the model can capture more detailed and rich information in the data, that is, various relationships between data can be fully utilized, more implicit features can be extracted, and high-order and low-order processing can be taken into account, so that data utilization is more sufficient, and the quality detection result obtained later is more accurate, meeting the needs of actual application scenarios.

[0113] According to the quality detection method of the building information model provided by the application, compared with the current quality detection method of the building information model by artificial, the application obtains the detection specification information corresponding to the target building information model in response to the quality detection signal of the target building information model, classifies the detection specification information to obtain the detection specification information under different classification categories, determines the model data corresponding to the target building information model, matches the model data with the detection specification information under each classification category, constructs the mapping relationship between the model data and the detection specification information under each classification category based on the matching result, and finally detects the quality of the model data by using the detection specification information based on the mapping relationship to obtain the quality detection result of the target building information model. Therefore, by classifying the detection specification information and establishing the mapping relationship between the model data and the detection specification information of different classification categories, the quality of the target building information model is directly detected by using the mapping relationship. Since the detection specification information has been systematically classified and accurately mapped with the model data, the quality detection result can comprehensively and accurately reflect the quality condition of the model data, and can avoid the time wasted and detection errors caused by artificial detection, so that the quality detection efficiency and accuracy of the building information model can be improved. At the same time, by classifying the detection specification information, the application can help to systematize and systematize the complex detection specification information, so that the quality detection process is more clear and orderly. Meanwhile, the classified detection specification information is easier to manage and update, which is convenient for subsequent quality detection work.

[0114] Further, as a specific implementation of Figure 1 , the embodiment of the application provides a quality detection device of a building information model, as shown in Figure 3 , the device comprises an acquisition unit 31, a classification unit 32, a construction unit 33 and a detection unit 34.

[0115] The acquisition unit 31 can be used to acquire the detection specification information corresponding to the target building information model in response to the quality detection signal of the target building information model.

[0116] The classification unit 32 can be used to classify the detection specification information to obtain the detection specification information under different classification categories.

[0117] The construction unit 33 can be used to determine the model data corresponding to the target building information model, match the model data with the detection specification information under each classification category, and construct the mapping relationship between the model data and the detection specification information under each classification category based on the matching result.

[0118] The detection unit 34 can be configured to perform quality detection on the model data based on the mapping relationship and the detection specification information, to obtain a quality detection result of the target building information model.

[0119] In a specific application scenario, in order to detect the construction attribute data in the target building information model, the classification unit 32 can be further configured to obtain construction attribute data from the model data of the target building information model, and classify the construction attribute data according to the construction type, to obtain the to-be-detected attribute data corresponding to each construction in the target building information model, wherein the to-be-detected attribute data includes at least one of the attribute data item quantity, the attribute value corresponding to the attribute data item, and the attribute value type.

[0120] The detection unit 34 can be further configured to obtain the standard attribute data of each construction in the detection specification information, and match the standard attribute data with the to-be-detected attribute data, and determine the construction attribute detection result of the target building information model based on the matching result.

[0121] In a specific application scenario, in order to classify the detection specification information, as shown in Figure 4 The classification unit 32 includes an initialization module 321, a division module 322, and a first determination module 323.

[0122] The initialization module 321 can be configured to initialize the centroid vector of the centroid corresponding to each classification category, and determine the information feature vector corresponding to the detection specification information.

[0123] The division module 322 can be configured to calculate the distance between each detection specification information and the centroid corresponding to each classification category based on the information feature vector and the centroid vector, and divide each detection specification information into each classification category based on the distance.

[0124] The first determination module 323 can be configured to determine the updated centroid vector corresponding to each classification category based on the information feature vector corresponding to the detection specification information in each classification category.

[0125] The division module 322 can be specifically configured to re-divide each detection specification information into each classification category based on the updated centroid vector, until the updated centroid vector does not change, and determine the detection specification information finally divided into each classification category as the detection specification information under different classification categories.

[0126] In a specific application scenario, the different classification categories include: an attribute type category, a space type category, a geometry type category, and a calculation type category; in order to perform quality detection on the model data, the detection unit 34 includes an identification module 341, a second determination module 342, and a detection module 343.

[0127] The identification module 341 can be configured to identify each component in the target building information model, and determine, in the model data, attribute type data to be detected, space type data to be detected, geometry type data to be detected, and calculation type data to be detected of each component.

[0128] The second determination module 342 can be configured to determine, in the detection specification information, standard attribute type data, standard space type data, standard geometry type data, and standard calculation type data of each component based on the mapping relationship.

[0129] The detection module 343 can be configured to perform attribute specification detection on the attribute type data to be detected by using the standard attribute type data corresponding to the same component, to obtain an attribute type quality detection result.

[0130] The detection module 343 can also be configured to perform spatial position detection on the space type data to be detected by using the standard space type data corresponding to the same component, to obtain a space type quality detection result.

[0131] The detection module 343 can also be configured to perform geometry size detection on the geometry type data to be detected by using the standard geometry type data corresponding to the same component, to obtain a geometry type quality detection result.

[0132] The detection module 343 can also be configured to perform topology relationship detection on the calculation type data to be detected by using the standard calculation type data corresponding to the same component, to obtain a calculation type quality detection result.

[0133] In a specific application scenario, in order to determine the model data corresponding to the target building information model, the construction unit 33 can be specifically used to determine the IFC model data corresponding to the target building information model, perform component entity identification in the IFC model data to obtain each component contained in the target building information model, and determine the component data corresponding to each component based on the IFC model data, and determine the component data corresponding to each component as the model data. The method of performing component entity identification in the IFC model data includes: determining each word in the IFC model data, and determining the semantic information vector corresponding to each word; inputting each semantic information vector into a preset position prediction model to perform position prediction to obtain the predicted position information of each word in the to-be-predicted component entity, the preset position prediction model being obtained by pre-training based on a sample data set with position labels; inputting the semantic information vector and the predicted position information into a preset entity identification model to perform entity identification to obtain each component entity contained in the IFC model data, the preset entity identification model being obtained by pre-training based on a sample data set with entity category labels.

[0134] In a specific application scenario, in order to generate a detection report, the device further includes a generation unit 35 and a correction unit 36.

[0135] The generation unit 35 can be used to determine deviation information between the target building information model and the detection specification information based on the quality detection result, generate a quality detection report based on the deviation type, the component identifier and the component position corresponding thereto, and the detection specification information corresponding to the deviation information, and send the quality detection report to an audit terminal for auditing.

[0136] The correction unit 36 can be used to receive the audit result of the audit terminal, construct a model correction sentence based on the detection specification information in the case of passing the audit, and automatically correct the deviation information by using the model correction sentence to obtain the target building information model after correction.

[0137] In a specific application scenario, in order to perform quality detection on the target building information model, the device further includes a determination unit 37 and a cross processing unit 38.

[0138] The acquisition unit 31 can also be used to acquire detection requirement information for the target building information model, and determine the quality detection type for the target building information model based on the detection requirement information.

[0139] The determination unit 37 can be used to determine a preset quality detection model corresponding to the quality detection type.

[0140] The determining unit 37 can be specifically configured to determine a specification feature vector corresponding to the detection specification information, and determine a model feature vector corresponding to the model data.

[0141] The cross processing unit 38 can be configured to cross-process the specification feature vector and the model feature vector to obtain a cross detection vector.

[0142] The detection unit 34 can be further configured to input the cross detection vector into the preset quality detection model to perform quality detection, and obtain a quality detection result of the target building information model.

[0143] It should be noted that other corresponding descriptions of each functional module involved in the building information model quality detection device provided by the embodiments of the present application can be referred to the corresponding descriptions of the method shown in Figure 1 , and will not be repeated here.

[0144] Based on the above method shown in Figure 1 , correspondingly, the embodiments of the present application also provide a computer readable storage medium, which stores a computer program, and the program is executed by a processor to realize the following steps: in response to a quality detection signal of a target building information model, obtaining detection specification information corresponding to the target building information model; classifying the detection specification information to obtain detection specification information under different classification categories; determining model data corresponding to the target building information model, matching the model data with the detection specification information under each classification category, and based on the matching result, constructing a mapping relationship between the model data and the detection specification information under each classification category; based on the mapping relationship, performing quality detection on the model data by using the detection specification information, and obtaining a quality detection result of the target building information model.

[0145] Based on the above method shown in Figure 1 and the device shown in Figure 3 , the embodiments of the present application also provide a physical structure diagram of a computer device, as shown in Figure 5As shown, the computer device comprises a processor 41, a memory 42, and a computer program stored on the memory 42 and executable on the processor, wherein the memory 42 and the processor 41 are both arranged on a bus 43, and the processor 41 implements the following steps when executing the program: in response to a quality detection signal of a target building information model, obtaining detection specification information corresponding to the target building information model; classifying the detection specification information to obtain detection specification information under different classification categories; determining model data corresponding to the target building information model, matching the model data with the detection specification information under each classification category, and based on the matching result, constructing a mapping relationship between the model data and the detection specification information under each classification category; based on the mapping relationship, using the detection specification information to perform quality detection on the model data to obtain a quality detection result of the target building information model.

[0146] Through the technical scheme of the present application, the present application obtains detection specification information corresponding to a target building information model in response to a quality detection signal of the target building information model; and classifies the detection specification information to obtain detection specification information under different classification categories; then determines model data corresponding to the target building information model, matches the model data with the detection specification information under each classification category, and based on the matching result, constructs a mapping relationship between the model data and the detection specification information under each classification category; finally, based on the mapping relationship, uses the detection specification information to perform quality detection on the model data to obtain a quality detection result of the target building information model. Thus, by classifying the detection specification information and establishing a mapping relationship between the model data and the detection specification information under different classification categories, the quality of the target building information model is directly detected using the mapping relationship. Since the detection specification information has been systematically classified and a precise mapping relationship has been established with the model data, the quality detection result can comprehensively and accurately reflect the quality status of the model data, and can avoid the time wasted and errors caused by manual detection, thereby improving the quality detection efficiency and accuracy of the building information model. At the same time, by classifying the detection specification information, the present application can help to systematize and organize the complex detection specification information, making the quality detection process more clear and orderly, and at the same time, the classified detection specification information is easier to manage and update, facilitating subsequent quality detection work.

[0147] It should be apparent to those skilled in the art that the modules or steps of the application described above can be implemented with a general purpose computing device, which can be centralized on a single computing device or distributed over a network of multiple computing devices, and optionally implemented with program code executable by a computing device, which can be stored in a storage device and executed by a computing device, and in some cases, the steps shown or described can be performed in a different order than shown, or made into individual integrated circuit modules, or multiple modules or steps made into a single integrated circuit module. Thus, the application is not limited to any particular combination of hardware and software.

[0148] The preferred embodiments of the application described above are intended to be merely illustrative, and not limiting. Having now described some embodiments of the application, it will be apparent to those of ordinary skill in the art that many modifications, substitutions, improvements, and other changes can be made to the application without deviating from the spirit and scope of the application.

Claims

1. A method of quality detection of a building information model, characterized in that, The method comprises the following steps: obtaining detection specification information corresponding to the target building information model in response to a quality detection signal of the target building information model; classifying the detection specification information to obtain detection specification information under different classification categories; determining model data corresponding to the target building information model, matching the model data with the detection specification information under each classification category, and constructing a mapping relationship between the model data and the detection specification information under each classification category based on the matching result; based on the mapping relationship, using the detection specification information to perform quality detection on the model data to obtain a quality detection result of the target building information model; wherein classifying the detection specification information to obtain detection specification information under different classification categories comprises: initializing a centroid vector corresponding to the centroids of different classification categories, and determining an information feature vector corresponding to the detection specification information; based on the information feature vector and the centroid vector, calculating the distance between each detection specification information and the centroid corresponding to each classification category, and based on the distance, dividing each detection specification information into each classification category; based on the information feature vector corresponding to the detection specification information in each classification category, determining an updated centroid vector corresponding to each classification category; based on the updated centroid vector, re-divide each detection specification information into each classification category until the updated centroid vector does not change; the detection specification information finally divided into each classification category is determined as the detection specification information under different classification categories; after obtaining the detection specification information corresponding to the target building information model, the method further comprises: obtaining detection requirement information for the target building information model, and determining a quality detection type for the target building information model based on the detection requirement information; determining a preset quality detection model corresponding to the quality detection type; determining a specification feature vector corresponding to the detection specification information, and determining a model feature vector corresponding to the model data; cross-processing the specification feature vector and the model feature vector to obtain a cross-detection vector; inputting the cross-detection vector into the preset quality detection model for quality detection to obtain a quality detection result of the target building information model.

2. The method of claim 1, wherein, The method further comprises: obtaining component attribute data from the model data of the target building information model, and classifying the component attribute data according to component types to obtain to-be-detected attribute data corresponding to each component in the target building information model, wherein the to-be-detected attribute data includes at least one of attribute data item quantity, attribute value corresponding to attribute data item, and attribute value type; obtaining standard attribute data of each component in the detection specification information, and matching the standard attribute data with the to-be-detected attribute data, and determining a component attribute detection result of the target building information model based on the matching result.

3. The method of claim 1, wherein, The different classification categories include: attribute type category, space type category, geometry type category, and calculation type category. The quality detection result of the target building information model is obtained by performing quality detection on the model data based on the mapping relationship and the detection specification information. Each component is identified in the target building information model, and attribute type data, spatial type data, geometric type data, and calculation type data to be detected of each component are determined in the model data; Based on the mapping relationship, standard attribute type data, standard spatial type data, standard geometric type data, and standard calculation type data of each component are determined in the detection specification information; Attribute specification detection is performed on the attribute type data to be detected by using the standard attribute type data corresponding to the same component, and an attribute type quality detection result is obtained; Spatial position detection is performed on the spatial type data to be detected by using the standard spatial type data corresponding to the same component, and a spatial type quality detection result is obtained; Geometric size detection is performed on the geometric type data to be detected by using the standard geometric type data corresponding to the same component, and a geometric type quality detection result is obtained; Topology relationship detection is performed on the calculation type data to be detected by using the standard calculation type data corresponding to the same component, and a calculation type quality detection result is obtained.

4. The method of claim 1, wherein, The model data corresponding to the target building information model is determined, including: The IFC model data corresponding to the target building information model is determined, component entity identification is performed in the IFC model data, each component contained in the target building information model is obtained, and component data corresponding to each component is determined based on the IFC model data, and the component data corresponding to each component is determined as the model data, wherein the method of performing component entity identification in the IFC model data includes: Each word in the IFC model data is determined, and semantic information vectors corresponding to each word are determined; Each semantic information vector is input into a preset position prediction model to perform position prediction, and prediction position information of each word in a to-be-predicted component entity is obtained, the preset position prediction model is obtained by pre-training based on a sample data set with position labels; The semantic information vector and the prediction position information are input into a preset entity recognition model to perform entity recognition, and each component entity contained in the IFC model data is obtained, and the preset entity recognition model is obtained by pre-training based on a sample data set with entity category labels.

5. The method of claim 1, wherein, After the quality detection result of the target building information model is obtained by performing quality detection on the model data based on the mapping relationship and the detection specification information, the method further includes: Based on the quality detection result, deviation information between the target building information model and the detection specification information is determined; Based on the deviation type, the component identifier, the component position corresponding to the component identifier, and the detection specification information corresponding to the deviation information, a quality detection report is generated, and the quality detection report is sent to an auditing terminal for auditing. The audit terminal receives the audit result, and in the case of passing the audit, a model correction statement is constructed based on the detection specification information, and the bias information is automatically corrected using the model correction statement to obtain the corrected target building information model.

6. A quality detection device of a building information model, characterized by, Comprise: An acquisition unit is configured to acquire detection specification information corresponding to a target building information model in response to a quality detection signal of the target building information model; A classification unit is configured to classify the detection specification information to obtain detection specification information under different classification categories, wherein classifying the detection specification information to obtain detection specification information under different classification categories comprises: initializing a centroid vector of centroids corresponding to different classification categories, and determining an information feature vector corresponding to the detection specification information; based on the information feature vector and the centroid vector, calculating the distance between each detection specification information and the centroid corresponding to each classification category, and based on the distance, dividing each detection specification information into each classification category; based on the information feature vector corresponding to the detection specification information in each classification category, determining an updated centroid vector corresponding to each classification category; based on the updated centroid vector, re-divide each detection specification information into each classification category until the updated centroid vector does not change, and the detection specification information finally divided into each classification category is determined as the detection specification information under different classification categories; A construction unit is configured to determine model data corresponding to the target building information model, match the model data with the detection specification information under each classification category, and based on the matching result, construct a mapping relationship between the model data and the detection specification information under each classification category; A detection unit is configured to perform quality detection on the model data using the detection specification information based on the mapping relationship to obtain a quality detection result of the target building information model; The detection unit is configured to acquire detection requirement information for the target building information model, and determine a quality detection type for the target building information model based on the detection requirement information; determine a preset quality detection model corresponding to the quality detection type; determine a specification feature vector corresponding to the detection specification information, and determine a model feature vector corresponding to the model data; cross-process the specification feature vector and the model feature vector to obtain a cross-detection vector; input the cross-detection vector into the preset quality detection model for quality detection to obtain the quality detection result of the target building information model.

7. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method of any one of claims 1 to 5.

8. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The computer program is executed by the processor to implement the steps of the method of any one of claims 1 to 5.

Images