Method for automatically repairing equipment in BIM model

Abstract

This invention provides an automatic equipment repair method in a BIM model, by establishing an electromechanical equipment image library EP = {(equipment type et)} j {device image ep} jk})}, train the intelligent recognition AI algorithm A1 for electromechanical equipment; use panoramic cameras to collect panoramic images q of each room ri. i This forms a panoramic image set QJ = {q} i}; Establish each room r i The virtual fusion model of BIM and panoramic images M i Algorithm A1 is used to identify each panoramic image q. i All mechanical and electrical equipment e ik Add to the device set E = {E i}; Analyze each room r i Consistency between the model and the equipment in the panoramic view; adding equipment to appropriate locations in the model; deleting redundant equipment from the model; calculating the difference Cn between the model and the real-world view for each room. ij Based on the allowable difference error (CN), rooms and equipment with errors exceeding a preset threshold are identified, and the model is modified. This invention aims to intelligently identify equipment information inconsistent between the as-built BIM model and the actual building structure, and automatically repair the as-built BIM model, thereby improving the consistency between the model and the actual building.

Description

Technical Field

[0001] This invention relates to an automatic equipment repair method in a BIM model. Background Technology

[0002] Mechanical and electrical (MEP) as-built BIM models are crucial data support for building operation and maintenance management. However, current MEP as-built BIM models created during the construction phase often lack models of many important equipment due to factors such as late equipment design finalization. This results in inconsistencies between the MEP as-built BIM model and the actual building, making it difficult to support building operation and maintenance management.

[0003] While panoramic cameras and similar methods can quickly capture panoramic images of the building's various rooms, effective methods for rapidly and intelligently identifying inconsistencies between the model and the actual building, and for developing strategies to repair the model, remain lacking. This is because inconsistencies between the model and the mechanical and electrical equipment within the building can manifest in several ways:

[0004] 1) The model is missing equipment that exists in the actual entity;

[0005] 2) The model includes additional equipment that does not exist in the physical model;

[0006] 3) There is a discrepancy between the location of the same device in the model and the actual object;

[0007] 4) Problems such as significant differences between the appearance of the model and the appearance of the actual object.

[0008] Therefore, it is difficult to simply use image recognition or similar methods to analyze and handle the inconsistency between the model and the entity. However, actual research revealed that problems 3) and 4) do not require modification; as long as maintenance personnel can see the equipment from a room, it will not affect maintenance; or the problems can be modified manually by moving the equipment. Therefore, problems 1) and 2) need to be fixed, namely the issues of equipment redundancy and missing components. Summary of the Invention

[0009] The purpose of this invention is to achieve automatic repair of equipment in the BIM model based on the information of the building entity.

[0010] To address the above problems, the present invention provides an automatic equipment repair method in a BIM model, comprising:

[0011] Step 1: Create an image library for electromechanical equipment EP = {(equipment type et)} j {device image ep} jk})}, Training the AI ​​algorithm A1 for intelligent recognition of electromechanical equipment;

[0012] Step 2: Use a panoramic camera to capture panoramic images of each room. i This forms a panoramic image set QJ = {q i};

[0013] Step 3: Establish each room r i The virtual fusion model of BIM and panoramic images M i ;

[0014] Step 4: Use AI algorithm A1 to identify each panoramic image q i All mechanical and electrical equipment e ik Add to the device set E = {E i};

[0015] Step 5: Analyze r for each room i Consistency between the model and the equipment in the panoramic view;

[0016] Step 6: Add devices to appropriate locations in the model so that the number of each type of element in the model, mj, equals the number of the same type of element in the entity, ej.

[0017] Step 7: Delete redundant devices in the model so that mj = ej;

[0018] Step 8: Calculate the difference Cn between the model and the actual scene for each room. ij ;

[0019] Step 9: Based on the tolerance error CN, identify rooms and equipment with errors greater than the preset threshold, and modify the model.

[0020] Furthermore, in the above method, step 1 includes:

[0021] Step 1.1: Collect images of building electromechanical equipment. jk and named et according to device type j ;

[0022] Step 1.2: Group images of devices of the same type into a set (device type et). j {device image ep} jk}), ultimately forming an image library of various electromechanical equipment EP = {(equipment type et)} j {device image ep} jk})};

[0023] Step 1.3: Select an open-source image classification algorithm as the AI ​​algorithm A1 for electromechanical equipment recognition;

[0024] Step 1.4: Extract each device type from the device image library EP. j The image set is randomly divided into two non-overlapping sets: the training set EPT and the test set EPS.

[0025] Step 1.5: Train A1 using images labeled with device types in EPT;

[0026] Step 1.6: Use the test in EPS that marks the device type to measure the accuracy of the completed A1 training; if the accuracy meets the requirements, end Step 1; otherwise, return to Step 1.4 and continue training.

[0027] Furthermore, in the above method, step 2 includes:

[0028] Step 2.1: Derive the room set R = {rooms r} from the BIM model i};

[0029] Step 2.2: For each room r i A panoramic camera was used to capture a panoramic view of the entire room. i If the room is too large, it is difficult to see the whole room in one panoramic view. Divide the room into multiple sub-rooms and take panoramic views of each sub-room.

[0030] Step 2.3: Record the initial coordinates (qcs) of the panoramic view for each room. i Initial direction qcv i and focal length f i Information, i.e., q i =(qcs i ,qcv i f i ), and add it to the set QT = {q i}

[0031] Furthermore, in the above method, step 3 includes:

[0032] Step 3.1 Traverse each frame q in the panoramic frame set QJ. i =(qcs i ,qcv i f i );

[0033] Step 3.2: In room r i Set the camera c of the rendering view in the model. i c i Let the origin coordinates be qcs i , direction set to qcv i The focal length is set to f. i ;

[0034] Step 3.3: Render the camera c in the model i The view is overlaid on the panoramic view. i In the process, a virtual-real fusion model M is established, integrating the BIM model with the panoramic view. i .

[0035] Furthermore, in the above method, step 4 includes:

[0036] Step 4.1: Traverse each frame q in the panoramic frame set QJ. i ;

[0037] Step 4.2: Algorithm A1 recognizes image q i All devices present in e ijk and its types et j and the pixel height in the image ijk And grouped by type and added to the device set E i ={(type et)} j ET ij ={device e ijk})}; and record the rotation angle ae of the panoramic image when the device is at the center of the panoramic image. ijk That is, e ijk =(r i et j ae ijk ,,pl ijk );

[0038] Step 4.3: For each device type et j The corresponding set ET ij Chinese equipment according to AE ijk Arrange in ascending order; place E i Add to device set E.

[0039] Furthermore, in the above method, step 5 includes:

[0040] Step 5.1: Iterate through each element E in E. i ;

[0041] Step 5.2: Traverse device E i each device type et j ET device set ij ;

[0042] Step 5.3: Extract room r from the BIM model i The type is et j Add the device information to the ME set. ij ={device me ijk}; and according to me ijk center point coordinates cme ijk Calculate me ijk With camera c i Rotation angle of direction (ame) ijk The specific formula is as follows: Ang(cme) ijk -qcs i ,qcv i),Ang is a function for calculating the included angle between two vectors; and calculate the distance lme from the device to the center point of the camera ijk = Len(cme ijk - qcs i ), Len is a function for calculating the vector length; record it in the device information, that is, me ijk = (r i , et j , ame ijk , lme ijk ); and calculate ME ij The average height hme of the devices in ij ;

[0043] Step 5.4: Arrange the devices in ME ij in ascending order according to the rotation angle ame of the devices ijk ; Add to E i = {(type et j , ET ij = {device e ijk}, ME ij = {device me ijk}, hme ij )};

[0044] Step 5.5: Analyze the relationship between the number of elements mj in the device set ME ij in the model and the number of elements ej in the device set ET ij in the panoramic view. If mj < ej, go to step 6; if mj > ej, go to step 7; if mj = ej, return to step 5.2;

[0045] Step 5.6: After traversing each device type et i in E j , return to step 5.1; until after traversing each element in E, end step 5 and enter step 8.

[0046] Furthermore, in the above method, step 6 includes:

[0047] Step 6.1: Set the deviation value set JC ij = {};

[0048] Step 6.2: Traverse each element e ij in ET ijk = (r i , et j , ae ijk );

[0049] Step 6.3: If ae ijk is smaller than the angle ame ij of the first element in ME ij1If they are all small, then JC ijk =ame ij1 -ae ijk Add to JC ij ={(e ijk jc ijk If the condition is not met, proceed to step 6.2; otherwise, proceed to step 6.4.

[0050] Step 6.4: If the angle between aeijk and the last element in MEij is ame ijK Both are large, then jc ijk =ae ijk -ame ijK Add to JC ij ={(e ijk jc ijk If the condition is not met, proceed to step 6.2; otherwise, proceed to step 6.5.

[0051] Step 6.5: Based on its angle ae ijk , in ME ij Find the element me in ijk’ , satisfy me ijk’ Angle

[0052] a me ijk’ than ae ijk Small, but ame ij(k’+1) Not less than ae ijk Then jc ijk =Min(ae) ijk -ame ijk’ ,ame ij(k’+1) -ae ijk Add to JC ij ={(e ijk jc ijk Return to step 6.2;

[0053] Step 6.6: ET ij After the traversal is complete, for JC ij Chinese elements according to JC ijk Perform reverse sorting and extract JC. ij The first mj-ej elements sorted in the middle are added to JD. ij ;

[0054] Step 6.7: Traverse JD ij Each element e in ijk According to e ijk angle ae ijk Using the virtual-real fusion model M i Determine the location of the equipment in the BIM model. ijk = (distance from camera origin) ijkThe angle between the camera direction and the camera direction (ae) ijk ), le ijk It is in camera c i The polar coordinates in the polar coordinate system, where, In le ijk Add device et j Model.

[0055] Furthermore, in the above method, step 7 includes:

[0056] Step 7.1: Set the deviation value set JC ij ;

[0057] Step 7.2: Traverse ME ij Each element me ijk (r i et j ,ame ijk ,lme ijk );

[0058] Step 7.3: If ame ijk Compared to ET ij The angle of the first element in the middle ae ij1 If they are all small, then JC ijk =ae ij1 -ame ijk Add to JC ij ={(me ijk jc ijk If the condition is not met, proceed to step 7.2; otherwise, proceed to step 7.4.

[0059] Step 7.4: If ame ijk Compared to ET ij The angle of the last element in the middle (ae) ijK Both are large, then jc ijk =ame ijk -aeijK, add to JC ij ={(me ijk jc ijk If the condition is not met, proceed to step 7.2; otherwise, proceed to step 7.5.

[0060] Step 7.5: Based on its angle (ame) ijk In ET ij Search for element e in ijk’ , satisfy e ijk’ angle ae ijk’ than ame ijk Small, but ae ij(k’+1) Not less than ame ijk Then jc ijk =Min(ae) ijk -ameijk’ ,ame ij(k’+1) -ae ijk ), add to JCij = {(e ijk jc ijk Return to step 7.2;

[0061] Step 7.6: ME ij After the traversal is complete, for JC ij Chinese elements according to JC ijk Perform reverse sorting and extract JC. ij The first ej-mj elements in the sorted list are added to JD. ij ;

[0062] Step 7.7: Iterate through each element in JDij and store the device me in the model. ijk delete.

[0063] Furthermore, in the above method, step 8 includes:

[0064] Step 8.1: Iterate through each element E in E. i ;

[0065] Step 8.2: Traverse device E i each device type et j Set Cn ij =0;

[0066] Step 8:3: Traverse ME ij and ET ij For each element, calculate the corresponding two devices e. ijk and me ijk The absolute value of the angle difference: jc = Abs(ame) ijk -ae ijk );Cn ij =Cn ij +jc;

[0067] Step 8.4: Put Cn ij Add to E i In, that is, E i ={(type et)} j ET ij ={device e ijk}, Cn ij )}.

[0068] Step 8.5: Traverse all devices E i each device type et j Then, return to step 8.1.

[0069] Furthermore, in the above method, step 9 includes:

[0070] Step 9.1 Input the tolerance error CN for the degree of difference;

[0071] Step 9.2 Traverse the set of all devices E i ;

[0072] Step 9.3 Traverse E i each element in et ij If its difference Cn ij If >CN is displayed, a manual repair is requested.

[0073] Compared with existing technologies, this invention establishes an image library of electromechanical equipment EP = {(equipment type et)} j {device image ep} jk})}, train the intelligent recognition AI algorithm A1 for electromechanical equipment; use panoramic cameras to collect panoramic images q of each room ri. i This forms a panoramic image set QJ = {q i}; Establish each room r i The virtual fusion model of BIM and panoramic images M i Algorithm A1 is used to identify each panoramic image q. i All mechanical and electrical equipment e ik Add to the device set E = {E i}; Analyze each room r i Consistency between the model and the equipment in the panoramic view; add equipment to appropriate locations in the model such that mj = ej; remove redundant equipment from the model such that mj = ej; calculate the difference Cn between the model and the real-world view for each room. ij Based on the allowable error (CN) for discrepancies, this invention identifies rooms and equipment with errors exceeding a preset threshold, modifies the model accordingly, and aims to intelligently identify inconsistencies between the as-built BIM model and the actual building structure, automatically repairing the as-built BIM model to improve model-to-building consistency and support building operation and maintenance management. This invention can automatically identify deviations between the model and the actual building structure based on real-world building images, automatically adding missing equipment and deleting redundant equipment from the model; it also marks the types of equipment with significant deviations in each room, guiding modelers to quickly and manually repair the model, thus saving considerable time on model review and modification; ultimately ensuring high integrity of the mechanical and electrical equipment in the model, making it suitable for operation and maintenance management. Attached Figure Description

[0074] Figure 1 This is a flowchart of an automatic equipment repair method in a BIM model according to an embodiment of the present invention;

[0075] Figure 2 This is a schematic diagram of a virtual-real fusion model according to an embodiment of the present invention;

[0076] Figure 3 This is a schematic diagram of a model in an embodiment of the present invention that lacks a device that exists in reality. Detailed Implementation

[0077] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0078] like Figure 1 As shown, the present invention provides an automatic equipment repair method in a BIM model, comprising:

[0079] Step 1: Create an image library for electromechanical equipment EP = {(equipment type et)} j {device image ep} jk The training of the intelligent recognition AI algorithm A1 for electromechanical equipment includes the following steps:

[0080] Step 1.1: Collect images of building electromechanical equipment. jk and named et according to device type j ;

[0081] For example, et j = "Distribution cabinet";

[0082] Step 1.2: Group images of devices of the same type into a set (device type et). j {device image ep} jk}), ultimately forming an image library of various electromechanical equipment EP = {(equipment type et)} j {device image ep} jk})};

[0083] Step 1.3: Select an open-source image classification algorithm as the AI ​​algorithm A1 for electromechanical equipment recognition, with PanddleClas algorithm being the preferred choice;

[0084] Step 1.4: Extract each device type from the device image library EP. j The image set is randomly divided into two non-overlapping sets: the training set EPT and the test set EPS.

[0085] Step 1.5: Train A1 using images labeled with device types in EPT;

[0086] Step 1.6: Use the test in EPS that marks the device type to measure the accuracy of the completed A1 training; if the accuracy meets the requirements, end Step 1; otherwise, return to Step 1.4 and continue training.

[0087] Step 2: Use a panoramic camera to capture panoramic images of each room. i This forms a panoramic image set QJ = {qi};

[0088] Step 2.1: Derive the room set R = {rooms r} from the BIM model i};

[0089] For example, R = {room 1, room 2};

[0090] Step 2.2: For each room r i A panoramic camera was used to capture a panoramic view of the entire room. i If the room is too large to be seen clearly in a single panoramic view, the room can be divided into multiple sub-rooms, and panoramic views can be taken for each sub-room.

[0091] Step 2.3: Record the initial coordinates (qcs) of the panoramic view for each room. i Initial direction qcv i and focal length f i Information, i.e., q i =(qcs i ,qcv i f i ), and add it to the set QT = {q i};

[0092] For example, for room 1, the panoramic view's q1 = (qcs) i =(2,2,1),qcv i =(06,0.8,0), f i =1.2);

[0093] Step 3: Establish each room r i The virtual fusion model of BIM and panoramic images M i Specifically, it includes the following steps:

[0094] Step 3.1 Traverse each frame q in the panoramic frame set QJ. i =(qcs i ,qcv i f i ),

[0095] Step 3.2: In room r i Set the camera c of the rendering view in the model. i c i Let the origin coordinates be qcs i , direction set to qcv i The focal length is set to f. i ;

[0096] Step 3.3: Render the camera c in the model i The view is overlaid on the panoramic view.i In the process, a virtual-real fusion model M is established, integrating the BIM model with the panoramic view. i ,like Figure 2 As shown;

[0097] Step 4: Use Algorithm A1 to identify each panoramic image q i All mechanical and electrical equipment e ik Add to the device set E = {E i};

[0098] Step 4.1: Traverse each frame q in the panoramic frame set QJ. i ;

[0099] Step 4.2: Algorithm A1 recognizes image q i All devices present in e ijk and its types et j and the pixel height in the image ijk And grouped by type and added to the device set E i ={(type et)} j ET ij ={device e ijk})}; and record the rotation angle ae of the panoramic image when the device is at the center of the panoramic image. ijk That is, e ijk =(r i et j ae ijk ,,pl ijk );

[0100] Step 4.3: For each device type et j The corresponding set ET ij Chinese equipment according to AE ijk Arrange in ascending order; place E i Add to device set E.

[0101] Step 5: Analyze the r of each room. i The consistency between the model and the device in the panoramic image includes the following steps:

[0102] Step 5.1: Iterate through each element E in E. i ;

[0103] Step 5.2: Traverse device E i each device type et j ET device set ij ;

[0104] Step 5.3: Extract room r from the BIM model i The type is et j Add the device information to the ME set.ij = {device me ijk}; and calculate the rotation angle ame ijk between me ijk and the camera c ijk according to the center point coordinate cme i of me, the specific formula is as follows: Ang(cme ijk - qcs ijk , qcv i ), where Ang is a function to calculate the included angle between two vectors; and calculate the distance lme i from the device to the center point of the camera = Len(cme ijk - qcs ijk ), where Len is a function to calculate the vector length; record it in the device information, that is, me i = (r ijk , et i , ame j , lme ijk ); and calculate the average height hme ijk of the devices in ME ij ;

[0105] Step 5.4: Arrange the devices in ME ij in ascending order according to the rotation angle ame ij of the devices; add them to E ijk = {(type et i , ET j = {device e ij}, ME ijk = {device me ij}, hme ijk )};

[0106]

[0107] Step 5.5: Analyze the relationship between the number of elements mj in the device set ME ij in the model and the number of elements ej in the device set ET ij in the panoramic view. If mj < ej, go to Step 6; if mj > ej, go to Step 7; if mj = ej, return to Step 5.2;

[0108] Step 5.6: After traversing each device type et ij in E i , return to Step 5.1; until each element in E is traversed, end Step 5 and enter Step 8.

[0109] Step 6: Add devices at appropriate positions in the model to make mj = ej, which specifically includes the following steps: Figure 3 As

[0110] shown, Step 6.1: Set the deviation value set JCij ={};

[0110] Step 6.2: Traverse ET ij Each element e in ijk =(r i et j ae ijk );

[0111] Step 6.3: If ae ijk Than ME ij The angle of the first element in the middle is ame ij1 If they are all small, then JC ijk =ame ij1 -ae ijk Add to JC ij ={(e ijk jc ijk If the condition is not met, proceed to step 6.2; otherwise, proceed to step 6.4.

[0112] Step 6.4: If the angle between aeijk and the last element in MEij is ame ijK Both are large, then jc ijk =ae ijk -ame ijK Add to JC ij ={(e ijk jc ijk If the condition is not met, proceed to step 6.2; otherwise, proceed to step 6.5.

[0113] Step 6.5: Based on its angle ae ijk , in ME ij Find the element me in ijk’ , satisfy me ijk’ Angle a me ijk’ than ae ijk Small, but ame ij(k’+1) Not less than ae ijk Then jc ijk =Min(ae) ijk -ame ijk’ ,ame ij(k’+1) -ae ijk Add to JC ij ={(e ijk jc ijk Return to step 6.2;

[0114] Step 6.6: ET ij After the traversal is complete, for JC ij Chinese elements according to JC ijk Perform reverse sorting and extract JC. ijThe first mj-ej elements sorted in the middle are added to JD. ij ;

[0115] Step 6.7: Traverse JD ij Each element e in ijk According to e ijk angle ae ijk Using the virtual-real fusion model M i Determine the location of the equipment in the BIM model. ijk = (distance from camera origin) ijk The angle between the camera direction and the camera direction (ae) ijk ), le ijk It is in camera c i The polar coordinates in the polar coordinate system, where, In le ijk Add device et j Model;

[0116] Step 6.8: Traverse ET ij For each element, step 6 ends.

[0117] Step 7: Remove redundant devices from the model so that mj = ej. This includes the following steps:

[0118] Step 7.1: Set the deviation value set JC ij ;

[0119] Step 7.2: Traverse ME ij Each element me ijk (r i et j ,ame ijk ,lme ijk );

[0120] Step 7.3: If ame ijk Compared to ET ij The angle of the first element in the middle ae ij1 If they are all small, then JC ijk =ae ij1 -ame ijk Add to JC ij ={(me ijk jc ijk If the condition is not met, proceed to step 7.2; otherwise, proceed to step 7.4.

[0121] Step 7.4: If ame ijk Compared to ET ij The angle of the last element in the middle (ae) ijK Both are large, then jc ijk =ame ijk -aeijK, add to JCij ={(me ijk jc ijk If the condition is not met, proceed to step 7.2; otherwise, proceed to step 7.5.

[0122] Step 7.5: Based on its angle (ame) ijk In ET ij Search for element e in ijk’ , satisfy e ijk’ angle ae ijk’ than ame ijk Small, but ae ij(k’+1) Not less than ame ijk Then jc ijk =Min(ae) ijk -ame ijk’ ,ame ij(k’+1) -ae ijk ), add to JCij = {(e ijk jc ijk Return to step 7.2;

[0123] Step 7.6: ME ij After the traversal is complete, for JC ij Chinese elements according to JC ijk Perform reverse sorting and extract JC. ij The first ej-mj elements in the sorted list are added to JD. ij ;

[0124] Step 7.7: Iterate through each element in JDij and store the device me in the model. ijk delete;

[0125] Step 7.8: Traverse all ME ij For each element, step 7 is complete.

[0126] Step 8: Calculate the difference Cn between the model and the actual scene for each room. ij Specifically, it includes the following steps:

[0127] Step 8.1: Iterate through each element E in E. i ;

[0128] Step 8.2: Traverse device E i each device type et j Set Cn ij =0;

[0129] Step 8:3: Traverse ME ij and ET ij For each element, calculate the corresponding two devices e. ijk and me ijkThe absolute value of the angle difference: jc = Abs(ame) ijk -ae ijk );Cn ij =Cn ij +jc;

[0130] Step 8.4: Put Cn ij Add to E i In, that is, E i ={(type et)} j ET ij ={device e ijk}, Cn ij )}.

[0131] Step 8.5: Traverse all devices E i each device type et j Then, return to step 8.1;

[0132] Step 8.6 Iterate through each element E in E. i Then, step 8 is complete.

[0133] Step 9: Based on the tolerance error CN, identify rooms and equipment with errors exceeding the preset threshold, and modify the model. This includes the following steps:

[0134] Step 9.1 Input the tolerance error CN for the degree of difference;

[0135] Step 9.2 Traverse the set of all devices E i ;

[0136] Step 9.3 Traverse E i each element in et ij If its difference Cn ij If >CN is displayed, a manual repair is requested.

[0137] Step 9.4: After traversing all elements in E, end the process.

[0138] This invention aims to intelligently identify inconsistencies between the as-built BIM model of mechanical and electrical systems (MEP) and the actual building structure, and automatically repair the MEP as-built BIM model, improving model-to-building consistency and supporting building operation and maintenance management. Based on real-world building images, this invention can automatically identify deviations between the model and the MEP equipment in the building structure, automatically add missing equipment to the model, and delete redundant equipment. It also marks the equipment types with significant deviations in each room, guiding modelers to quickly and manually repair the model, thus saving considerable time on model review and modification. Ultimately, it ensures a high degree of integrity of the MEP equipment in the model, making it suitable for operation and maintenance management.

[0139] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0140] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0141] Obviously, those skilled in the art can make various modifications and variations to the invention without departing from the spirit and scope of the invention. Therefore, if these modifications and variations fall within the scope of the claims of the invention and their equivalents, the invention is also intended to include these modifications and variations.

Claims

1. A method for automatic equipment repair in a BIM model, characterized in that, include: Step 1: Create an image library for electromechanical equipment EP={(equipment type et)} j {device image ep} jk })}, Training the AI ​​algorithm A1 for intelligent recognition of electromechanical equipment; Step 2: Use a panoramic camera to capture panoramic images of each room. i This forms a panoramic image set QJ={q i }; Step 3: Establish each room r i The virtual fusion model of BIM and panoramic images M i ; Step 4: Use AI algorithm A1 to identify each panoramic image q i All mechanical and electrical equipment e ik Add to the device set E={E i }; Step 5: Analyze r for each room i Consistency between the model and the equipment in the panoramic view; Step 6: Add devices to appropriate locations in the model so that the number of each type of element in the model, mj, equals the number of the same type of element, ej, in the entity. Step 7: Delete redundant devices in the model so that mj = ej; Step 8: Calculate the difference Cn between the model and the actual scene for each room. ij ; Step 9: Based on the tolerance error CN, identify rooms and equipment with errors greater than the preset threshold, and modify the model accordingly; Step 4 includes: Step 4.1: Traverse each frame q in the panoramic frame set QJ. i ; Step 4.2: Algorithm A1 recognizes image q i All devices present in e ijk and its types et j and the pixel height in the image ijk And grouped by type and added to the device set E i ={(type et) j ET ij ={device e ijk })};and record the rotation angle ae of the panoramic image when the device is in the center of the panoramic image. ijk That is, e ijk = (r i et j ae ijk ,pl ijk ); Step 4.3: For each device type et j The corresponding set ET ij Chinese equipment according to AE ijk Arrange in ascending order; place E i Add to device set E; Step 5 includes: Step 5.1: Iterate through each element E in E. i ; Step 5.2: Traverse device E i each device type et j ET device set ij ; Step 5.3: Extract room r from the BIM model i The type is et j Add the device information to the ME set. ij ={Device me ijk }; and according to me ijk center point coordinates cme ijk Calculate me ijk With camera c i Rotation angle of direction (ame) ijk The specific formula is as follows: Ang(cme) ijk - qcs i ,qcv i ), Ang is a function to calculate the angle between two vectors; and lme is the distance from the device to the center point of the camera. ijk = Len(cme ijk - qcs i Len is a function for calculating the length of a vector; it is recorded in the device information, i.e., me. ijk = (r i et j ,ame ijk ,lme ijk ); and calculate ME ij The average height of the equipment is hme ij ; Step 5.4: For ME ij The equipment is arranged according to the rotation angle of the equipment. ijk Sort in ascending order; add to E i ={(type et) j ET ij ={device e ijk }, ME ij ={Device me ijk }, hme ij )}; Step 5.5: Analyze the device set ME in the model ij The relationship between the number of elements mj in it and the device set ET in the panoramic picture ij The relationship between the number of elements ej. If mj < ej, go to Step 6; if mj > ej, go to Step 7; if mj = ej, return to Step 5.2; Step 5.6: Traverse E i each device type et j Then, return to step 5.1; until every element in E has been traversed, end step 5 and proceed to step 8.

2. The automatic equipment repair method in a BIM model as described in claim 1, characterized in that, Step 1 includes: Step 1.1: Collect images of building electromechanical equipment. jk and named et according to device type j ; Step 1.2: Group images of devices of the same type into a set (device type et). j {device image ep} jk }), ultimately forming an image library of various electromechanical equipment EP={(equipment type et)} j {device image ep} jk })}; Step 1.3: Select an open-source image classification algorithm as the AI ​​algorithm A1 for electromechanical equipment recognition; Step 1.4: Extract each device type from the device image library EP. j The image set is randomly divided into two non-overlapping sets: the training set EPT and the test set EPS. Step 1.5: Train A1 using images labeled with device types in EPT; Step 1.6: Use the test in EPS that marks the device type to measure the accuracy of the completed A1 training; if the accuracy meets the requirements, end Step 1; otherwise, return to Step 1.4 and continue training.

3. The automatic equipment repair method in a BIM model as described in claim 2, characterized in that, Step 2 includes: Step 2.1: Export the room set R = {rooms r} from the BIM model i }; Step 2.2: For each room r i A panoramic camera was used to capture a panoramic view of the entire room. i If the room is too large, it is difficult to see the whole room in one panoramic view. Divide the room into multiple sub-rooms and take panoramic views of each sub-room. Step 2.3: Record the initial coordinates (qcs) of the panoramic view for each room. i Initial direction qcv i and focal length f i Information, i.e., q i =(qcs i ,qcv i f i ), and add it to the set QT={q i } 4. The automatic equipment repair method in a BIM model as described in claim 3, characterized in that, Step 3 includes: Step 3.1 Traverse each frame q in the panoramic frame set QJ. i =(qcs i ,qcv i f i ); Step 3.2: In room r i Set the camera c of the rendering view in the model. i c i Let the origin coordinates be qcs i , direction set to qcv i The focal length is set to f. i ; Step 3.3: Render the camera c in the model i The view is overlaid on the panoramic view. i In the process, a virtual-real fusion model M is established, integrating the BIM model with the panoramic view. i .

5. The automatic equipment repair method in a BIM model as described in claim 1, characterized in that, Step 6 includes: Step 6.1: Set the deviation value set JC ij ={}; Step 6.2: Traverse ET ij Each element e in ijk = (r i et j ae ijk ); Step 6.3: If ae ijk Than ME ij The angle of the first element in the middle is ame ij1 If they are all small, then JC ijk =ame ij1 -ae ijk Add to JC ij ={(e ijk jc ijk If the condition is not met, return to step 6.2; otherwise, proceed to step 6.

4. Step 6.4: If the angle between aeijk and the last element in MEij is ame ijK Both are large, then jc ijk = ae ijk -ame ijK Add to JC ij ={(e ijk jc ijk If the condition is not met, proceed to step 6.2; otherwise, proceed to step 6.

5. Step 6.5: Based on its angle ae ijk , in ME ij Find the element me in ijk’ , satisfy me ijk’ Angle a me ijk’ than ae ijk Small, but ame ij(k’+1) Not less than ae ijk Then jc ijk =Min(ae ijk - ame ijk’ ,ame ij(k’+1) -ae ijk Add to JC ij ={(e ijk jc ijk )}, then return to step 6.2; Step 6.6: ET ij After the traversal is complete, for JC ij Chinese elements according to JC ijk Perform reverse sorting and extract JC. ij The first mj-ej elements sorted in the middle are added to JD. ij ; Step 6.7: Traverse JD ij Each element e in ijk According to e ijk angle ae ijk Using the virtual-real fusion model M i Determine the location of the equipment in the BIM model. ijk = (distance from camera origin) ijk The angle between the camera direction and the camera direction (ae) ijk ), le ijk It is in camera c i polar coordinates in the polar coordinate system, where lene ijk = ; in le ijk Add device et j Model.

6. The automatic equipment repair method in a BIM model as described in claim 5, characterized in that, Step 7 includes: Step 7.1: Set the deviation value set JC ij ; Step 7.2: Traverse ME ij Each element me ijk (r) i et j ,ame ijk ,lme ijk ); Step 7.3: If ame ijk Compared to ET ij The angle of the first element in the middle ae ij1 If they are all small, then JC ijk =ae ij1 -ame ijk Add to JC ij ={(me ijk jc ijk If the condition is not met, return to step 7.2; otherwise, proceed to step 7.

4. Step 7.4: If ame ijk Compared to ET ij The angle of the last element in the middle (ae) ijK Both are large, then jc ijk = ame ijk -aeijK, add to JC ij ={(me ijk jc ijk If the condition is not met, proceed to step 7.2; otherwise, proceed to step 7.

5. Step 7.5: Based on its angle (ame) ijk In ET ij Search for element e in ijk’ , satisfy e ijk’ Angle ae ijk’ than ame ijk Small, but ae ij (k’+1) Not less than ame ijk Then jc ijk =Min(ae ijk - ame ijk’ ame ij(k’+1) -ae ijk ), add to JCij={(e ijk jc ijk )}, return to step 7.2; Step 7.6: ME ij After the traversal is complete, for JC ij Chinese elements according to JC ijk Perform reverse sorting and extract JC. ij The first ej-mj elements in the sorted list are added to JD. ij ; Step 7.7: Iterate through each element in JDij and store the device me in the model. ijk delete.

7. The automatic equipment repair method in a BIM model as described in claim 6, characterized in that, Step 8 includes: Step 8.1: Iterate through each element E in E. i ; Step 8.2: Traverse device E i each device type et j Set Cn ij =0; Step 8:3: Traverse ME ij and ET ij For each element, calculate the corresponding two devices e. ijk and me ijk The absolute value of the angle difference: jc = Abs(ame) ijk -ae ijk );Cn ij =Cn ij + jc; Step 8.4: Put Cn ij Add to E i In, that is, E i ={(type et) j ET ij ={device e ijk }, Cn ij )}; Step 8.5: Traverse all devices E i each device type et j Then, return to step 8.

1.

8. The automatic equipment repair method in a BIM model as described in claim 7, characterized in that, Step 9 includes: Step 9.1 Input the tolerance error CN for the degree of difference; Step 9.2 Traverse the set of all devices E i ; Step 9.3 Traverse E i each element in et ij If its difference Cn ij If >CN is displayed, a manual repair is requested.

Images