BIM-based electromechanical installation position analysis method and device

Abstract

The invention provides a BIM-based electromechanical installation position analysis method, and relates to the field of equipment position detection, and the method comprises the steps: obtaining a BIM-based space structure model of a to-be-analyzed building through point cloud data; performing enhanced display on the obtained fuzzy region in the BIM-based building space model by fusing IFC semantic information and geometric similarity; and analyzing and judging the actual mechanical and electrical installation position by combining the processed BIM building space model for the building and the set mechanical and electrical installation information. The invention further discloses an electromechanical installation position analysis device based on the BIM. The electromechanical installation position analysis device comprises an electromechanical equipment position acquisition module, a central processing module, a position coordinate extraction module and a comparative analysis module. Spatial comparison is carried out on the basis of the BIM building space model and the set installation position, whether the installation error between the electromechanical installation position and the set position exceeds the error allowable value or not is judged, error accumulation in the data transmission process can be reduced through cleaning use of the space model, and the accuracy of the electromechanical installation position is improved.

Description

Technical Field

[0001] The present invention relates to the field of equipment position detection, and in particular to a method and device for analyzing electromechanical installation position based on BIM. Background Art

[0002] When the electromechanical equipment is under construction, the installation position confirmation operation is performed to confirm whether the installation position data of multiple electromechanical equipment stored in the centralized controller is consistent with the actual installation position of the electromechanical equipment. However, the existing electromechanical equipment confirmation methods mostly rely on manpower to check in turn whether the installation position of the equipment in the building meets the use requirements. As the built-in structure of the building becomes more and more complex, it is impossible to achieve effective inspection by relying solely on human power, and it requires a lot of human resources, so a new electromechanical equipment is urgently needed.

[0003] Chinese patent document CN201510088665.7 "Position detection device and position detection method". It adopts a determination unit, which determines the type of indicator; a detection unit, which detects the indicated position of the indicator; a storage unit, which stores the reference change amount related to the change of the indicated position detected by the detection unit for each indicator; and a correction unit, which obtains the reference change amount corresponding to the type of the indicator determined by the determination unit from the storage unit, and corrects the indicated position detected by the detection unit according to the obtained reference change amount. According to this structure, the indicated position can be properly corrected regardless of the indicator used. However, this patent cannot quickly and effectively analyze whether the installation positions of electromechanical equipment at different positions in the building are correct. Summary of the invention

[0004] The present invention mainly solves the original technical problem that it is impossible to efficiently check whether the installation position of electromechanical equipment in a building is accurate, and provides a BIM-based electromechanical installation position analysis method and device. The accuracy of obtaining the internal information of building B is improved through the cooperation of point cloud data and BIM model, and the remaining fuzzy areas in the model are further enhanced and displayed in combination with IFC semantic information and geometric similarity, thereby improving the accuracy of electromechanical position information in the building, and completing the analysis of the electromechanical installation position in the building with a clear BIM building space model to determine whether the electromechanical equipment is installed in place.

[0005] The above technical problem of the present invention is mainly solved by the following technical solution: The present invention comprises S1 obtaining a BIM-based spatial structure model of the building to be analyzed by means of point cloud data; S2 uses the fusion of IFC semantic information and geometric similarity to enhance the display of fuzzy areas in the BIM-based building space model; S3 combines the processed BIM building space model with the building and the set mechanical and electrical installation information to complete the analysis and judgment of the actual mechanical and electrical installation location.

[0006] Preferably, when the building space model is obtained with the help of point cloud data, alignment is required, and n planes are selected for plane alignment, wherein n-1 planes are not parallel in pairs, and the remaining planes are not coplanar with the remaining n-1 planes at the same time, but are parallel to any of the planes, and then n plane sets are defined from the BIM model and point cloud data sets respectively. If the overall number of planes of the point cloud model is less than that of the BIM model, it is determined to be point cloud driven; if the overall number of planes of the point cloud model is more than that of the BIM model, it is determined to be BIM model driven.

[0007] Preferably, after the registration, on the one hand, the point cloud plane is extracted, and the eigenvalues ​​and eigenvectors based on the covariance matrix are used to extract the point cloud plane patches. Once the plane patch descriptors of all points in the point cloud are calculated, the plane patches are extracted using an iterative process; on the other hand, the BIM model plane is extracted, the model is rendered, and the plane patches are extracted from the model using a region growing algorithm. In several iterations, triangles that have not yet been defined are randomly selected and set as new plane bases.

[0008] Preferably, the point cloud plane extraction or BIM model plane extraction result is repaired, a given model plane patch list and a pre-calculated geometric relationship are selected, i plane bases are randomly selected from the point cloud plane alignment or BIM model plane alignment result, and a test is performed to see whether each plane constitutes a valid i plane base. If the test conditions are met, it is determined that no repair is required; if the test conditions are not met, it is determined that repair is required.

[0009] Preferably, in step S2, under the premise of considering the IF geometric reference semantic information when enhancing the display, the ICP algorithm and Hausdorff distance are used to calculate the geometric similarity, so as to achieve accurate extraction of BIM component instance information, and quickly extract instance information of special stretching bodies for common stretching body components in BIM scenes.

[0010] Preferably, step S3 includes comparing the distance between the BIM model and the actual point cloud data through a collision detection method based on the geometric information of the BIM model to determine whether an actual collision occurs between the electromechanical elements in the building.

[0011] Preferably, the pipeline and equipment positioning error detection method is based on the pipeline and equipment positioning error detection of BIM model and point cloud data. By comparing the spatial positions of pipeline equipment in the BIM model and point cloud scene, the method of judging its actual deviation situation takes the center point of the pipeline axis extracted from the point cloud as the actual position, establishes a corresponding relationship between the two, and then calculates the deviation distance between the two in the three-dimensional scene. After recording the actual deviation of all pipeline equipment, the position information is statistically analyzed to judge the factors affecting the positioning accuracy; and analyzes the error distribution law.

[0012] Preferably, in step S1, the BIM model plane and the point cloud plane are converted according to the actual plane requirements, and the conversion steps are as follows: S11 only considers the i centroids of two groups of i plane bases of congruence for centroid transformation; S12 considers all point cloud plane patches for plane transformation; S13 clusters similar transformations; S14 performs point conversion considering actual point cloud data.

[0013] A BIM-based electromechanical installation position analysis device, comprising: The electromechanical equipment position acquisition module includes a plurality of electromechanical equipment data acquisition and transmission devices connected to the central processing module; the central processing module performs centralized processing on the electromechanical equipment position information to generate processing results; The position coordinate extraction module performs position extraction processing and obtains the theoretical installation values ​​of different electromechanical equipment relative to the building in which they are located. The comparative analysis module determines whether there are problems with the installation positions of electromechanical equipment in different areas of the building by analyzing the actual positions and theoretical installation values ​​of the electromechanical equipment and combining the allowable installation error values.

[0014] Preferably, the comparative analysis module includes: a comparison unit, which compares the actual position of the electromechanical equipment with the standard installation position and determines whether the difference exceeds the allowable position error; a prediction unit, which analyzes and predicts the value curve of the electromechanical equipment installation position offset and the time of occurrence of exceeding the allowable position offset error based on the existing building status and the installation position information of the electromechanical equipment.

[0015] The beneficial effects of the present invention are as follows: under the preparation conditions, point cloud data and BIM models are used to synchronously assist in obtaining the building space structure information and the installation location information of the electromechanical equipment in the building, and the accuracy of the spatial information obtained in the two forms is improved as much as possible through plane registration and patching. In addition, the method of integrating IFC semantic information and geometric similarity is used to strengthen the clear and fuzzy areas for possible spatial fuzzy areas, so that relevant users can confirm the electromechanical installation information more clearly. After clarifying the building space structure and the electromechanical installation location, a spatial comparison is performed based on the BIM building space model and the set installation location to determine whether the installation error between the electromechanical installation location and the set location exceeds the allowable error value. The line division judgment process reduces the dependence on manpower, and the cleaning and use of the spatial model can reduce the error accumulation during the data transmission process and improve the accuracy of the electromechanical installation location. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flow chart based on BIM electromechanical installation location analysis.

[0017] Figure 2 It is a process diagram of a planar registration judgment method. DETAILED DESCRIPTION

[0018] In this embodiment, a BIM-based building electromechanical installation error detection is used, and the steps include obtaining a BIM-based spatial structure model of the building to be analyzed with the help of point cloud data. In this embodiment, when performing BIM and point cloud registration, any four planes are used as "four plane bases", of which three planes are not parallel in pairs, and the fourth plane is not coplanar with the other three planes, but can be parallel to any of the three planes. Based on this, four plane sets are defined from the data sets of the BIM model and the 3D point cloud, respectively, provided that the two sets of four plane sets can be aligned using rigid transformation within the tolerance range. Planes are extracted from the BIM model and the point cloud, and for all planes, their paired geometric relationships will be stored in a search list by the computer for future use. In this embodiment, the plane sets extracted from the BIM model and the point cloud are respectively. If the overall number of planes of the point cloud model is less than that of the BIM model, the process is considered to be point cloud driven; if the overall number of planes of the point cloud model is more than that of the BIM model, the process is considered to be BIM model driven. Then, multiple sets of different four plane bases are randomly searched to define these multiple sets of four plane bases. If the four planes present the same geometric relationship as the four planes that make up the four planes, the four plane bases are considered to be congruent. Finally, given a pair of congruent four plane bases, the transformation matrix that transforms one of the four plane bases into another four plane base is calculated. Then the plane is extracted, including point cloud plane extraction, and the point cloud plane patch is extracted using the method based on the eigenvalue and eigenvector of the covariance matrix. The covariance matrix is ​​calculated using the principal component analysis method. An ellipsoid is defined to represent each area. This ellipsoid can be described by three geometric features: linear, planar, and scattered. Once the planar patch descriptors of all points in the point cloud are calculated, an iterative process is used to extract the planar patch. In each iteration, the point with the highest value is used as the basis, and the eigenvector with the smallest associated eigenvalue is used to define the plane direction. All point clouds within the plane of the highest value point are added to the median point of the patch. The plane is usually set according to the laser scanner. The highest value point that has not yet been added to any patch is used as the new basis. Repeat the process, and the iteration continues until all points are added to a patch. Each patch can eventually be split into separate planar patches using a region growing algorithm and a sufficient distance close to a threshold. Finally, the centroid, normal vector and area are calculated for each patch and stored in a lookup table for the area of ​​the planar patch.

[0019] It also includes BIM model plane extraction, in which it is assumed that the surfaces of all objects in the BIM model are defined as triangular meshes that are easy to obtain. In fact, it is often used for real-time rendering models using technical models such as OpenGL. Then a region growing algorithm is used to extract plane patches from the model. In each iteration, the algorithm randomly selects triangles that have not yet defined a plane and sets them as new plane bases, and considers adjacent triangles to expand the patch. If any of these triangles are coplanar with the plane, they are added to it. At this point, any remaining triangles are used as new bases, and the iteration continues until no new triangular meshes are added to the plane. Similar to the extraction of point cloud planes, the centroid, normal vector and area of ​​each patch are calculated and stored in a lookup table. After using the above two methods for plane extraction, it is also necessary to patch the extracted point cloud plane with the BIM plane to improve the accuracy of the subsequent model display of the building. By comparing the set of 4 points between the BIM model and the 3D point cloud, if these 4 points exist in both sets of data, there is a matching set. However, in the case of plane-based registration, the plane patches completed in the BIM model data are likely to exist only partially in the point cloud data due to the turbulence in the scanning process. It is important to match the 4 plane bases using geometric features that are not affected by such factors. Therefore, the following relationships between the sets of planes are calculated: between the angles of the normals between each pair of plane patches, and the normal angle is assumed to be the orthogonal projection distance of the centroid of one plane into the centroid of the other plane. These pairwise relationships are calculated for all BIM plane patches and point cloud plane patches and stored in a lookup table.

[0020] Then select the 4-plane basis of the BIM model. Given the list of model plane patches and the pre-calculated geometric relationship, randomly select 4 plane bases from the model and test to see if they form a valid 4-plane basis. If the following conditions occur, they are considered valid. In the first case, no plane is coplanar with any other three planes; in the second case, the three planes are not parallel in pairs, that is, there is a set of three planes in the 4-plane basis whose angle is not 0. The search is stopped until 50 valid sets of plane bases are found, and these 4-plane bases are called. These 50 sets of plane bases ensure a series of different transformation matrices and help limit the overall calculation time. If the algorithm fails to find the correct transformation from the set, another 50 sets of valid plane bases are selected for the second transformation until all valid sets are found to be correctly transformed. At the same time, the 4-plane basis of the point cloud needs to be selected. For each valid set of 4-plane basis, it can be searched in the point cloud. For this reason, the plane patches extracted from the point cloud satisfy the same angle relationship. The two groups of patches form a candidate 4-plane basis. If the angular relationship between each plane basis has been verified and the included angle is not 0, the 4 plane basis are considered to be congruent. The result is calculated using the formula of rigid transformation in each pair of congruent bases.

[0021] In addition, in order to facilitate the actual needs of users, the conversion between BIM planes and point cloud planes can also be performed to meet the actual usage conditions of users. For every two groups of 4-plane bases to be converted, the process is divided into 4 stages: 1. Centroid conversion. Only the 4 centroids of the two groups of 4-plane bases that are congruent are considered; 2. Plane conversion. Consider all point cloud plane patches; 3. Clustering. Cluster similar conversions; 4. Point conversion. Consider the actual point cloud data. When converting the centroid, first determine whether the centroids of the 4 plane patches that make up the point cloud congruence basis match the plane patches of the BIM. If any of the 4 centroids is not projected in the BIM model, the congruence basis is discarded, and another set of valid 4-plane bases is used. It should be noted that for this test, it is important to project the centroid of the point cloud plane patch onto the BIM plane patch. When a plane is transformed, the possibility of transformation is judged by calculating the number of point cloud plane patches of the transformation matrix. If the following conditions are met, the point cloud plane is considered to support the transformation: 1. The projection distance d from the point cloud plane patch to the BIM plane patch is less than the maximum value; 2. The point cloud and BIM plane patches have similar directions. Once the individual plane patches are matched, the overall plane transformation degree is given by a percentage. When classifying similar transformations, if there are multiple groups of 4-plane basis transformations with similar situations, these similar 4-plane basis transformations should be grouped. The homogeneous transformation matrix distance measure D is currently used to evaluate the similarity of two rigid transformations. Given two homogeneous transformation matrices in the form of 4×4, the homogeneous transformation matrix distance of the two homogeneous transformation matrices in the form of 4×4 is calculated. The closer the value is to 0, the more similar the matrices are. After calculating the distances between all similar transformations, hierarchical clustering is used to group them. The segmentation of groups during hierarchical clustering is done using the average distance between links (formed by the similarity distance) as a threshold. Finally, a representative transformation matrix is ​​selected from each cluster as the transformation matrix for plane transformation. After completing the above steps to perform point transformation, we extend the plane transformation to all point clouds. It is recommended to simply extend it to the points associated with the matching point cloud plane patch, not just the center of mass of the plane.

[0022] To this end, similar to the conversion of a plane, check whether each point cloud plane is close to the BIM plane and within its projection, and then use two indicators to measure the conversion percentage of the point: After completing the basic restoration of the BIM spatial structure model, the fuzzy area in the BIM-based architectural spatial model is enhanced by integrating IFC semantic information and geometric similarity. The integration of IFC semantic information and geometric similarity requires the definition of all BIM component entity object sets in the IFC file, the introduction of vector P to describe the BIM component entity object set and IFC semantic information, and the definition of the order in the attribute vector P describing the object in order to calculate and compare the IFC semantic information of different objects and compare the semantic information between different objects. The IFC geometric semantic information of the BIM object is further distinguished, and the explicit geometric expression is defined and the display of the fuzzy area is enhanced according to the following steps. First, based on the instance information extraction of IFC semantic information, a typical feature of the BIM model is the frequent reuse of the same building components. In order to reduce the design, production and maintenance costs, the use of multiple identical components, such as doors, windows, walls, etc., is common in any building design. However, when the modeling software exports these repeated components as IFC files, a large amount of redundant data will be generated. These duplications include: multiple entities have different IDs, but the content describing the entities is the same; entity IDs are different, but the content of the data container used to store reference information in the entity is the same; except for the ID, the content of the default attribute value of the entity representing its own information is different, but the content of the data container used to store reference information in the entity is the same. In order to remove the above redundant information, this embodiment is based on the idea of ​​instance information extraction of IFC semantic information, analyzes the object geometric expression mapping relationship, and checks whether the geometric expression content of each object is the same based on the semantic content, thereby removing redundant information.

[0023] Based on the instance information extraction of component geometry similarity, the traditional BIM instance information extraction method is relatively simple and rough in judging instance BIM components. It mainly checks whether entities have the same geometry information based on IFC semantic information, thereby performing instance compression. These methods ignore two common geometric redundancy situations: different entity IDs have different contents describing entity geometry information (such as vertex coordinates), but the topological information and shape information expressed by the two geometry are the same; different entity IDs have different contents describing entity geometry information, and the topological information and shape information expressed by the geometry information are also different, but there is an affine transformation matrix. Calculate the alignment matrix between the two point sets, select BIM components oj and ok with the same number of vertices N for alignment (N = pg(oj)l = Ilpg(ok)), and the process can be described as ICP point cloud matching algorithm can usually accurately calculate the translation component t and rotation matrix R of the two point sets. For the convenience of calculation, the uniform scale scaling factor can be eliminated. After eliminating the uniform scale scaling factor, the numerical calculation results in the formula can be calculated by the ICP algorithm. However, the traditional ICP is easily affected by the local minimum, its performance is heavily dependent on the quality of the initialized point cloud, and can only guarantee the local optimal solution. This embodiment selects the global optimal algorithm, based on the alignment process of ICP, and uses the Hausdorff distance (HD) to define the similarity given in the equation. If the similarity of two aligned representations is within a certain threshold, they are considered to be duplicates, and then the transformation matrix between them is recorded.

[0024] Then, based on the instance information extraction of the special parametric model, stretch modeling is one of the most common modeling methods in BIM modeling, and the number of cylindrical stretch bodies and rectangular stretch bodies (referred to as cylindrical component tubes and square components) in any building accounts for a large proportion

[11] , such as various complex pipe networks in large buildings, walls in building structures, etc. The shape of any circular or square component can be obtained by affine transformation of a standard cylinder or cube. Therefore, this embodiment proposes a fast instantiation method based on the semantic information of cylindrical and rectangular component stretch bodies, which can quickly identify a part of instantiated components based on IFC semantic information. Define the instantiation objects of unit cylinders and cubes. By calculating the scaling matrix between all square and circular stretch bodies and unit cylinders and cubes and calculating the new world coordinate system transformation matrix, the instantiation components of circular and square stretch body components can be quickly identified. Finally, the component information and the updated coordinate transformation are recorded separately.

[0025] After the BIM space model is processed, the actual electromechanical installation location is analyzed and judged by combining the processed BIM building space model with the building and the set electromechanical installation information. The built-in rule engine of the BIM software can be used to realize the automatic detection of the spatial gaps between different building electromechanical systems and equipment, and accurately identify whether there are foreign body conflicts such as collisions and overlaps in the model. This rule-based detection method pre-sets space restrictions and allowable error ranges on the BIM platform by referring to relevant standards and specifications, empirical parameters, etc., and then automatically calculates the actual distance or gap between each building electromechanical element to determine whether the rule requirements are met and quickly find all conflicting parts. Compared with traditional methods, rule-based BIM conflict detection does not require manual judgment and measurement, and can automatically realize the spatial rationality detection of building electromechanical systems in batches, greatly improving detection efficiency.

[0026] Among them, the data collection method based on laser scanning point cloud uses laser radar to perform high-density scanning of the interior of the building based on laser scanning technology, quickly obtains three-dimensional point cloud data of each spatial area, and records the position and morphological characteristics of all building mechanical and electrical elements in the scene. Unlike traditional methods, laser scanning can remotely collect a large amount of target information without contacting the object, and can obtain rich on-site data without affecting the normal operation of the building's mechanical and electrical systems, providing high-precision basic data support for later detection and analysis. After integrating and registering the BIM model with the laser point cloud data, the actual deviation of the building's mechanical and electrical elements can be clearly and intuitively identified from a macro perspective.

[0027] Among them, the collision detection method based on the geometric information of the BIM model is to judge whether the mechanical and electrical elements of the building actually collide by comparing the distance between the BIM model and the actual point cloud data. Its basic principle can be expressed as follows: Let the three-dimensional curve of the pipe i in the BIM model be Ci(u), and the three-dimensional curve of another pipe j in the point cloud data be Cj(v). Then the shortest distance between the two pipes is: Dmin=minu,v|Ci(u)-Cj(v)|where Dmin represents the closest distance or minimum distance between the two pipe surfaces. If Dmin is less than the minimum distance dmin allowed between the two pipe surfaces, a collision is confirmed. When performing this calculation, it is necessary to consider the geometric shape of the pipe, surface treatment, etc. to determine a reasonable dmin value. Unlike manual collision detection, this method can quickly and efficiently calculate the distance between a large number of pipes and automatically identify all potential collision problems. At the same time, the calculation results are more objective and accurate, and are not limited by personnel experience and subjective judgment. By setting the dmin threshold for different pipe types and locations, more flexible collision detection and evaluation can be achieved.

[0028] It also includes a BIM-based pipeline and equipment positioning error detection method. The pipeline and equipment positioning error detection based on the BIM model and point cloud data is a method of comparing the spatial positions of pipeline equipment in the BIM model and point cloud scene to determine its actual deviation. Specifically, the BIM model is used as a reference for the design position of the pipeline equipment, and the center point of the pipeline axis extracted from the point cloud is used as the actual position. The corresponding relationship between the two is established, and then the deviation distance between the two in the three-dimensional scene is calculated. After recording the actual deviation of all pipeline equipment, it can also be summarized and analyzed. For example, the average positioning deviation of each pipeline type, different parts, each floor or partition is counted to determine the factors affecting the positioning accuracy; analyze the error distribution law to confirm the location of the systematic deviation; check whether the positioning deviation meets the relevant acceptance standards, etc.

[0029] When the connection error detection method of the building mechanical and electrical system is based on BIM, the connection error detection of the building mechanical and electrical system can be expressed as judging the deviation between the design connection and the actual connection of the BIM model. Assume the center point coordinates and normal vector of the interface between the two devices in the BIM model and the interface specifications. Then the connection error term can be expressed as: displacement deviation, direction deviation, specification deviation. The above method establishes an accurate correspondence between the connection points by analyzing the spatial position, normal direction, and size specification parameters of the same interface in the BIM model and the actual scene, and quantitatively analyzes whether the joints of the two match from the vector and specification level, overcoming the limitation that the traditional method cannot clearly establish a point-to-point connection correspondence between the design and the actual. At the same time, this technology can uniformly detect and evaluate the connection quality of the building mechanical and electrical system during the installation process in terms of displacement deviation, direction deviation, and specification deviation, and judge whether the interface has connection faults such as misalignment, direction flipping, and thread non-conformity, which is conducive to targeted rectification of existing connection problems in the later stage.

[0030] In the actual application process, the project in this embodiment uses laser scanning and BIM technology to detect the installation errors of its pipelines and equipment for the electromechanical system of the building project. The key steps are as follows: 1. For the water supply, fire protection, HVAC and other specialties of the project, a BIM model containing detailed information such as equipment, pipelines, and devices is established. 2. Use a laser scanner to scan the main indoor and outdoor areas of the building to obtain point cloud data. 3. Use relevant software to process the above two types of data, perform registration, meet the requirements of the accuracy control value, and complete the high-precision registration of virtual and real scenes. 4. Based on the registered BIM and point cloud integrated model, calculate the displacement distance, direction angle difference, etc. between the two, and detect the error between the actual installation of various pipeline equipment and the design requirements. 5. Generate a visual inspection report for key areas and typical problems. Mark pipelines with large dislocations, equipment joints with wrong directions, etc., output real-life pictures and BIM comparison pictures, and finally form a complete electromechanical inspection and evaluation report. Through the method in this embodiment, fine building electromechanical elements are added on the basis of the BIM model to form professional BIM models such as electromechanical facilities in the building. Secondly, 3D laser scanning and other technologies are used to obtain point cloud data generated by on-site construction, and it is registered with the BIM model to achieve virtual-real comparison. Finally, the system automatically or manually compares the difference between the registered model and the point cloud data to determine the actual installation error of the building's mechanical and electrical elements, providing a basis for later rectification. This principle based on the comparison of information-based models with actual construction overcomes the limitations of traditional manual measurement and realizes the automated detection and evaluation of the building's mechanical and electrical installation locations.

Claims

1. A BIM-based electromechanical installation location analysis method, characterized in that: The step S1 comprises obtaining a BIM-based spatial structure model of the building to be analyzed by means of point cloud data; S2 uses the fusion of IFC semantic information and geometric similarity to enhance the display of fuzzy areas in the BIM-based building space model; S3 combines the processed BIM building space model with the building and the set mechanical and electrical installation information to complete the analysis and judgment of the actual mechanical and electrical installation location.

2. The BIM-based electromechanical installation location analysis method according to claim 1, characterized in that: When the building space model is obtained with the help of point cloud data, alignment is required, and n planes are selected for plane alignment, among which n-1 planes are not parallel in pairs, and the remaining planes are not coplanar with the remaining n-1 planes at the same time, but are parallel to any of them, and then n plane sets are defined from the BIM model and point cloud data sets respectively. If the overall number of planes of the point cloud model is less than that of the BIM model, it is determined to be point cloud driven; if the overall number of planes of the point cloud model is more than that of the BIM model, it is determined to be BIM model driven.

3. The BIM-based electromechanical installation location analysis method according to claim 2 is characterized in that: After the alignment, on the one hand, the point cloud plane is extracted, and the eigenvalue and eigenvector method based on the covariance matrix is ​​used to extract the point cloud plane patches. Once the plane patch descriptors of all points in the point cloud are calculated, the plane patches are extracted using an iterative process; on the other hand, the BIM model plane is extracted, the model is rendered, and the plane patches are extracted from the model using the region growing algorithm. In several iterations, triangles that have not yet been defined are randomly selected and set as new plane bases.

4. A BIM-based electromechanical installation location analysis method according to any one of claims 2 or 3, characterized in that: The point cloud plane extraction or BIM model plane extraction result is repaired, a given model plane patch list and a pre-calculated geometric relationship are selected, i plane bases are randomly selected from the point cloud plane alignment or BIM model plane alignment result, and a test is performed to see whether each plane constitutes a valid i plane base. If the test conditions are met, it is determined that no repair is required; if the test conditions are not met, it is determined that repair is required.

5. The BIM-based electromechanical installation location analysis method according to claim 1, characterized in that: In the step S2, the geometric similarity is calculated using the ICP algorithm and the Hausdorff distance under the premise of considering the IF geometric reference semantic information during the enhanced display, so as to achieve accurate extraction of BIM component instance information and quickly extract instance information of special stretching bodies for common stretching body components in the BIM scene.

6. The BIM-based electromechanical installation location analysis method according to claim 1, characterized in that: The step S3 includes comparing the distance between the BIM model and the actual point cloud data through a collision detection method based on the geometric information of the BIM model to determine whether an actual collision occurs between the electromechanical elements in the building.

7. The BIM-based electromechanical installation location analysis method according to claim 6, characterized in that: The pipeline and equipment positioning error detection method is based on the pipeline and equipment positioning error detection of BIM model and point cloud data. By comparing the spatial positions of pipeline equipment in the BIM model and point cloud scene, the method of judging its actual deviation situation takes the center point of the pipeline axis extracted from the point cloud as the actual position, establishes a corresponding relationship between the two, and then calculates the deviation distance between the two in the three-dimensional scene. After recording the actual deviation of all pipeline equipment, the position information is statistically analyzed to judge the factors affecting the positioning accuracy; and analyzes the error distribution law.

8. The BIM-based electromechanical installation location analysis method according to claim 4, characterized in that: In step S1, the BIM model plane and the point cloud plane are converted according to the actual plane requirements. The conversion steps are as follows: S11 only considers the i centroids of two groups of i plane bases of congruence for centroid transformation; S12 considers all point cloud plane patches for plane transformation; S13 clusters similar transformations; S14 performs point conversion considering actual point cloud data.

9. A BIM-based electromechanical installation location analysis device, adapted to the BIM-based electromechanical installation location analysis method according to any one of claims 1 to 8, characterized in that: include: The electromechanical equipment position acquisition module includes a number of electromechanical equipment data acquisition and transmission devices connected to the central processing module; A central processing module performs centralized processing on the location information of the electromechanical equipment to generate processing results; The position coordinate extraction module performs position extraction processing and obtains the theoretical installation values ​​of different electromechanical equipment relative to the building in which they are located; The comparative analysis module analyzes the actual position of the electromechanical equipment and the theoretical installation value, and combines the allowable installation error value to determine whether there are problems with the installation position of the electromechanical equipment in different areas of the building.

10. The BIM-based electromechanical installation position analysis device according to claim 9, characterized in that: The comparative analysis module includes: a comparison unit, which compares the actual position of the electromechanical equipment with the standard installation position and makes a difference to determine whether the difference exceeds the allowable position error; a prediction unit, which analyzes and predicts the value curve of the electromechanical equipment installation position offset and the time of occurrence of exceeding the allowable position offset error based on the existing building status and the installation position information of the electromechanical equipment.

Images