On-site inspection system and method based on BIM + AR

Abstract

The invention relates to the technical field of building inspection, in particular to an on-site inspection system and method based on BIM + AR, and the system comprises a server and a user AR inspection end. The server comprises a model construction module, a historical inspection result acquisition module, an inspection task generation module, a transmission module, a monitoring data acquisition module, an inspection result uploading module, a processing module, an intelligent identification module and a comparison module; wherein the processing module is used for calling inspection skilled points corresponding to the identity information of the current inspection personnel according to the identity information corresponding to the current inspection personnel, and selecting inspection points corresponding to non-inspection skilled points; the intelligent identification module is used for identifying a predicted inspection result corresponding to each inspection point in non-inspection skilled points based on a preset AI inspection identification model; and the comparison module is used for identifying an inspection point with a difference between the predicted inspection result and the current inspection result according to the predicted inspection result and the corresponding current inspection result.

Description

Technical Field

[0001] The present invention relates to the technical field of building inspection, and in particular to a BIM+AR-based on-site inspection system and method. Background Art

[0002] Safety management at construction sites is crucial. Any potential safety hazard could lead to a serious accident, endangering workers' lives and causing significant economic losses. Inspections are a crucial tool for ensuring project quality. By monitoring the construction process, quality issues can be identified and corrected promptly, ensuring that construction projects are completed to standard.

[0003] Currently, most construction sites still rely on manual inspections, with dedicated inspectors conducting regular or irregular inspections of the construction site. Inspection results are usually saved in paper records, including photos and text descriptions. However, the existing inspection method still has the following problems:

[0004] The results of manual inspections are easily affected by factors such as the inspectors' experience and sense of responsibility. The standards of different inspectors may be inconsistent, resulting in low reliability of the inspection results.

[0005] Due to the lack of an effective supervision mechanism during the inspection process, inspectors may miss inspections or be perfunctory, which reduces the effectiveness of the inspection.

[0006] Inspection results usually take some time to be summarized and reported, making real-time monitoring impossible and problems difficult to detect and address in a timely manner. Summary of the Invention

[0007] One of the purposes of the present invention is to provide a BIM+AR-based on-site inspection system and method, which can improve the inspection accuracy of the corresponding inspection personnel during the on-site inspection process, while realizing real-time detection of the inspection results and improving the reliability of the inspection.

[0008] In order to achieve the above objectives, a BIM+AR-based on-site inspection system is provided, which includes a server and a user AR inspection terminal;

[0009] The server includes:

[0010] The model building module is used to build BIM models of each building corresponding to the sub-buildings to be inspected in each inspection time period;

[0011] A historical inspection result acquisition module is used to obtain historical inspection results corresponding to buildings similar to the building to be inspected from the database;

[0012] The inspection task generation module is used to determine the inspection points corresponding to each inspection time period and the location information of the inspection points in the corresponding building BIM model based on the acquired historical inspection results and the building BIM model corresponding to each inspection time period, and formulate the inspection tasks corresponding to each inspection point and store them in the corresponding database;

[0013] The transmission module is used to retrieve the inspection points corresponding to the corresponding inspection time period, the location information of the inspection points in the corresponding building BIM model, and the inspection tasks of each inspection point from the database and send them to the user's AR inspection terminal when the corresponding inspection time period arrives;

[0014] The monitoring data acquisition module is used to monitor the sub-buildings to be inspected corresponding to the current inspection time period in real time through the user's AR inspection terminal, and obtain the corresponding monitoring video data of the current inspection personnel during the inspection process in real time;

[0015] The inspection result upload module is used to upload the current inspection results corresponding to each inspection point during the inspection process;

[0016] A processing module is used to retrieve the inspection points corresponding to the identity information of the current inspection personnel according to the identity information corresponding to the current inspection personnel, and based on the inspection points, select the inspection points corresponding to the non-inspection points to form a corresponding non-inspection point list, and associate each inspection point in the non-inspection point list with the corresponding monitoring video data;

[0017] The intelligent recognition module is used to identify the predicted inspection results corresponding to each inspection point in the non-inspection strong points based on the list of non-inspection strong points and the corresponding monitoring video data, based on the preset AI inspection recognition model;

[0018] The comparison module is used to identify the inspection points where the predicted inspection results are different from the current inspection results based on the predicted inspection results corresponding to each inspection point in the non-inspection strong points and the corresponding current inspection results, and feed back the identified inspection points to the user AR inspection terminal corresponding to the current inspection personnel.

[0019] The technical principle and effect of this solution: In this solution, first, the model construction module is used to realize the construction of each building BIM model corresponding to each sub-building to be inspected in each inspection time period. By dividing the buildings to be inspected in each inspection time period, the buildings with a large inspection volume are divided into periodic or regular groups, which greatly improves the process and feasibility of the inspection.

[0020] Afterwards, by obtaining the historical inspection results corresponding to the buildings similar to the building to be inspected from the database, the inspection points corresponding to each inspection time period corresponding to the building to be inspected and the location information of the inspection points in the building BIM model are identified. At the same time, the inspection tasks corresponding to each inspection point are also formulated, and this information is sent to the user's AR inspection terminal. After that, when arriving at the corresponding inspection time period, the inspection personnel can wear the user's AR inspection terminal to conduct the inspection. During this process, the user's AR inspection terminal is used to monitor the sub-building to be inspected corresponding to the current inspection time period in real time, and obtain the corresponding monitoring video data of the current inspection personnel during the inspection process in real time, so as to realize real-time monitoring of the entire inspection process. After the inspection is completed, the current inspection results corresponding to each inspection point of the current inspection personnel during the inspection process are uploaded, so that the inspection results of the current inspection personnel during the inspection process can be collected.

[0021] Of course, in this process, the inspectors may miss or fail to discover problems during the inspection, that is, they may inspect things they are not good at. In order to avoid this situation, the inspection points corresponding to the current inspector's identity information are retrieved according to the identity information corresponding to the current inspector. Based on the inspection points, the inspection points corresponding to the non-inspection points are selected to form a corresponding list of non-inspection points of expertise. The various inspection points in the non-inspection points of expertise list are associated with their corresponding monitoring video data. Through this analysis of the professional skills of the inspectors, the inspection points that the current inspectors are good at and the inspection points that they are not good at are identified. , in order to achieve differentiation, that is, to verify these inspection points that are not good at again, and identify the predicted inspection results corresponding to each inspection point in the non-inspection strong points through the preset AI inspection recognition model. By comparing the predicted inspection results corresponding to each inspection point in the non-inspection strong points and whether there are differences between the current inspection results, the inspection results corresponding to the non-inspection strong points of the current inspector are verified, so as to judge the inspection accuracy of the current inspector at each non-inspection strong point, so as to inform the current inspector of the corresponding erroneous inspection at the first time, greatly improving the accuracy of subsequent inspections by the inspector.

[0022] Furthermore, the server also includes:

[0023] The cloud inspection module is used to identify inspection points where the predicted inspection results differ from the current inspection results, select the optimal inspection personnel corresponding to each inspection point based on the various pending inspection personnel stored in the database, and form a corresponding optimal inspection personnel plan. Based on the optimal inspection personnel plan, the surveillance video data corresponding to each identified inspection point is sent to the corresponding pending inspection personnel;

[0024] The analysis module is used to obtain the cloud inspection results corresponding to the inspection points uploaded by the pending inspection personnel, and based on the current inspection results corresponding to the current inspection personnel and the corresponding cloud inspection results, analyze the inspection difference results corresponding to the current inspection personnel and the pending inspection personnel and output the corresponding inspection difference report, and send the inspection difference report to the user AR inspection terminal corresponding to the current inspection personnel for visual display.

[0025] Beneficial effect: By introducing other patrol personnel to conduct secondary inspections, specifically, according to the patrol points where the predicted patrol results are different from the current patrol results, based on the various pending patrol personnel stored in the database, the optimal patrol personnel corresponding to each patrol point are selected, and a corresponding optimal patrol personnel plan is formed, and the monitoring video data corresponding to each identified patrol point is sent to the corresponding pending patrol personnel based on the optimal patrol personnel plan, that is, the optimal set of pending patrol personnel is matched from the database, thereby realizing secondary inspections of patrol points that the current patrol personnel are not good at, greatly improving the reliability and efficiency of the inspections of each patrol point in the list of non-inspection strong points, and being able to more accurately and reasonably realize the inspections of patrol points where the predicted patrol results are different from the current patrol results.

[0026] Finally, by comparing the inspection differences between the current inspection results corresponding to the current inspection personnel and the corresponding cloud inspection results, an inspection difference report is produced, thereby realizing inspection reminders for the current inspection personnel, greatly improving the inspection capabilities of on-site inspection personnel, and through timely notification of inspection difference reports and integration with the site, greatly improving the inspection impression of inspection personnel, which can better help inspection personnel improve the quality of inspections.

[0027] Furthermore, the cloud inspection module includes:

[0028] The initial screening module is used to randomly generate N initial populations based on the inspection points where the predicted inspection results differ from the current inspection results, based on the pending inspection personnel stored in the database. The individuals in the initial population are the set of pending inspection personnel corresponding to each inspection point in the non-inspection strong point list; based on the preliminary screening conditions, the individuals in the initial population are screened and judged. If the conditions are met, it is a feasible solution; otherwise, it is an infeasible solution, and the corresponding initial population is eliminated;

[0029] A fitness calculation module is used to calculate the fitness of the selected population according to a preset first fitness calculation formula and a second fitness calculation formula;

[0030] The first fitness calculation formula is:

[0031]

[0032] In the formula, F1 is the first fitness, D1 is the inspection income value of the pending inspection personnel corresponding to the population, S i is the inspection accuracy of the i-th pending inspector in the population when inspecting the corresponding inspection point, M i is the inspection task satisfaction of the i-th pending inspector in the population when inspecting the corresponding inspection point, α, β are the corresponding weighting coefficients, suiji is a randomly generated random number between 0 and 1, P is the random threshold, x ij is the accurate value corresponding to the inspection point of the jth inspection corresponding to the i-th pending inspection personnel in the population. If it is accurate, then x ij =1, otherwise 0; m is the number of pending inspectors in the population, n is the number of inspections of a certain inspection point by the pending inspectors; y is a dynamic value, the size of which is related to the number of times c that the random number is less than P and the preset value B;

[0033] The second fitness calculation formula is:

[0034]

[0035] In the formula, F2 is the second fitness, D2 is the total inspection efficiency corresponding to the population, L k is the inspection efficiency of the kth pending inspector in the population when inspecting the corresponding inspection point;

[0036] A selection module is configured to select, within a first preset number of iterations, populations whose first fitness is less than or equal to a first fitness threshold according to the first fitness corresponding to the populations, and select the top three populations with the smallest second fitness from among the populations eliminated at this time, and save them into a backup library;

[0037] When the first preset number of iterations is exceeded, the current population is combined with the population in the backup library to form a new population, and the population corresponding to the current population whose second fitness is less than the preset second fitness threshold is selected;

[0038] The crossover and mutation module is used to hybridize and mutate the selected population to obtain the offspring population;

[0039] A loop module is used to continue executing the fitness calculation module after obtaining the offspring population until a preset number of iterations is met;

[0040] The output module is used to output the offspring population as the optimal solution set for multi-objective optimization.

[0041] Beneficial effects: In this scheme, the optimal selection of pending inspectors is achieved by optimizing the setting of the algorithm, and the fitness function is used to calculate the population to achieve screening. Specifically, by introducing inspection accuracy and inspection task satisfaction, it is ensured that the selected inspectors have high accuracy and satisfaction in specific tasks, thereby improving the overall quality of inspection.

[0042] Secondly, by introducing random numbers and thresholds, the fitness calculation has a certain degree of randomness, which prevents the algorithm from falling into the local optimal solution. When suiji is less than P, the fitness is adjusted by the dynamic value y, which increases the flexibility of the algorithm.

[0043] Furthermore, the server also includes:

[0044] The abnormal point identification module is used to identify the inspection points where the predicted inspection results are different from the current inspection results, retrieve the location information corresponding to the identified inspection points, and obtain the current location information corresponding to the user AR inspection terminal corresponding to the current inspection personnel in real time, calculate the position distance difference between the identified inspection point and the user AR inspection terminal corresponding to the current inspection personnel, judge the size relationship between the position distance difference and the preset distance threshold, and if the position distance difference is less than or equal to the preset distance threshold, send an alarm signal to the user AR inspection terminal corresponding to the current inspection personnel.

[0045] Beneficial effects: In this solution, by automatically identifying abnormal points, the time for manual inspection is reduced, making the inspection work more efficient. When an abnormal situation is detected, an alarm is promptly issued to the inspection personnel, which can effectively prevent the occurrence of potential accidents and ensure the safety of personnel. Real-time acquisition of the location information of the inspection personnel and comparison with the location of the abnormal point can help to quickly locate the problem and shorten the time to deal with the problem. Providing accurate data support to help managers make timely and effective decisions, such as dispatching the nearest maintenance team to the abnormal point, etc., can prevent small faults from turning into big problems through early warning, thereby reducing unnecessary maintenance costs and equipment downtime. For inspection personnel using the AR inspection terminal, this function simplifies the operating process, improves the convenience of work, and enhances the user experience.

[0046] The present invention also provides a BIM+AR-based on-site inspection method, which uses the above-mentioned BIM+AR-based on-site inspection system and includes the following steps:

[0047] S1. Construct BIM models of each building corresponding to the sub-buildings to be inspected in each inspection time period of the building to be inspected;

[0048] S2. Obtain historical inspection results corresponding to buildings similar to the building to be inspected from the database;

[0049] S3. Based on the historical inspection results obtained and the building BIM model corresponding to each inspection time period, determine the inspection points corresponding to each inspection time period and the location information of the inspection points in the corresponding building BIM model, and formulate the inspection tasks corresponding to each inspection point and store them in the corresponding database;

[0050] S4. When the corresponding inspection time period is reached, the inspection points corresponding to the corresponding inspection time period, the location information of the inspection points in the corresponding building BIM model, and the inspection tasks of each inspection point are retrieved from the database and sent to the user AR inspection terminal;

[0051] S5. Use the user's AR inspection terminal to monitor the sub-building to be inspected corresponding to the current inspection time period in real time, and obtain the corresponding monitoring video data of the current inspection personnel during the inspection process in real time;

[0052] S6. Upload the current inspection results corresponding to each inspection point during the inspection process by the current inspection personnel;

[0053] S7. Retrieving the inspection points corresponding to the identity information of the current inspection personnel based on the identity information of the current inspection personnel, selecting inspection points corresponding to the non-inspection points based on the inspection points, and forming a corresponding non-inspection point list. Each inspection point in the non-inspection point list is associated with its corresponding surveillance video data;

[0054] S8. Based on the list of non-inspection strong points and the corresponding surveillance video data, the AI ​​inspection recognition model with preset settings is used to identify the predicted inspection results corresponding to each inspection point in the non-inspection strong points;

[0055] S9. Based on the predicted inspection results corresponding to each inspection point in the non-inspection expertise points and the corresponding current inspection results, identify the inspection points where the predicted inspection results are different from the current inspection results, and feed back the identified inspection points to the user AR inspection terminal corresponding to the current inspection personnel. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 This is a logic block diagram of the BIM+AR-based on-site inspection system in Example 1 of the present invention;

[0057] Figure 2 This is a flow chart of the on-site inspection method based on BIM+AR in Example 1 of the present invention.

[0058] Figure 3 This is a map of a first location on-site inspection performed by a user carrying the user AR inspection terminal in the first embodiment of the present invention;

[0059] Figure 4This is a map of an on-site inspection at a second location performed by a user carrying the user AR inspection terminal in the first embodiment of the present invention;

[0060] Figure 5 This is a diagram of a user carrying an AR inspection terminal to conduct an on-site inspection at a third location in the first embodiment of the present invention. DETAILED DESCRIPTION

[0061] The following is further described in detail through specific implementation methods:

[0062] Example 1

[0063] A BIM+AR based on-site inspection system, basically Figure 1 As shown, it includes the server and the user AR inspection terminal;

[0064] The server includes:

[0065] The model building module is used to build BIM models of each building corresponding to the sub-buildings to be inspected in each inspection time period;

[0066] A historical inspection result acquisition module is used to obtain historical inspection results corresponding to buildings similar to the building to be inspected from the database;

[0067] The inspection task generation module is used to determine the inspection points corresponding to each inspection time period and the location information of the inspection points in the corresponding building BIM model based on the acquired historical inspection results and the building BIM model corresponding to each inspection time period, and formulate the inspection tasks corresponding to each inspection point and store them in the corresponding database;

[0068] The transmission module is used to retrieve the inspection points corresponding to the corresponding inspection time period, the location information of the inspection points in the corresponding building BIM model, and the inspection tasks of each inspection point from the database and send them to the user's AR inspection terminal when the corresponding inspection time period arrives;

[0069] The monitoring data acquisition module is used to monitor the sub-buildings to be inspected corresponding to the current inspection time period in real time through the user's AR inspection terminal, and obtain the corresponding monitoring video data of the current inspection personnel during the inspection process in real time;

[0070] The inspection result upload module is used to upload the current inspection results corresponding to each inspection point during the inspection process;

[0071] A processing module is used to retrieve the inspection points corresponding to the identity information of the current inspection personnel according to the identity information corresponding to the current inspection personnel, and based on the inspection points, select the inspection points corresponding to the non-inspection points to form a corresponding non-inspection point list, and associate each inspection point in the non-inspection point list with the corresponding monitoring video data;

[0072] The intelligent recognition module is used to identify the predicted inspection results corresponding to each inspection point in the non-inspection strong points based on the list of non-inspection strong points and the corresponding monitoring video data, based on the preset AI inspection recognition model;

[0073] The comparison module is used to identify the inspection points where the predicted inspection results are different from the current inspection results according to the predicted inspection results corresponding to each inspection point in the non-inspection points and the corresponding current inspection results, and feed back the identified inspection points to the user AR inspection terminal corresponding to the current inspection personnel. Figure 3 、 Figure 4 as well as Figure 5 As shown, the user carries the corresponding user AR inspection terminal to conduct on-site inspections of multiple locations of the building to be inspected.

[0074] The server also includes:

[0075] The cloud inspection module is used to identify inspection points where the predicted inspection results differ from the current inspection results, select the optimal inspection personnel corresponding to each inspection point based on the various pending inspection personnel stored in the database, and form a corresponding optimal inspection personnel plan. Based on the optimal inspection personnel plan, the surveillance video data corresponding to each identified inspection point is sent to the corresponding pending inspection personnel;

[0076] The cloud inspection module includes:

[0077] The initial screening module is used to randomly generate N initial populations based on the inspection points where the predicted inspection results differ from the current inspection results, based on the pending inspection personnel stored in the database. The individuals in the initial population are the set of pending inspection personnel corresponding to each inspection point in the non-inspection strong point list; based on the preliminary screening conditions, the individuals in the initial population are screened and judged. If the conditions are met, it is a feasible solution; otherwise, it is an infeasible solution, and the corresponding initial population is eliminated;

[0078] A fitness calculation module is used to calculate the fitness of the selected population according to a preset first fitness calculation formula and a second fitness calculation formula;

[0079] The first fitness calculation formula is:

[0080]

[0081] In the formula, F1 is the first fitness, D1 is the inspection income value of the pending inspection personnel corresponding to the population, S i is the inspection accuracy of the i-th pending inspector in the population when inspecting the corresponding inspection point, M i is the inspection task satisfaction of the i-th pending inspector in the population when inspecting the corresponding inspection point, α, β are the corresponding weighting coefficients, suiji is a randomly generated random number between 0 and 1, P is the random threshold, x ij is the accurate value corresponding to the inspection point of the jth inspection corresponding to the i-th pending inspection personnel in the population. If it is accurate, then x ij =1, otherwise 0; m is the number of pending inspectors in the population, n is the number of inspections of a certain inspection point by the pending inspectors; y is a dynamic value, the size of which is related to the number of times c that the random number is less than P and the preset value B;

[0082] The second fitness calculation formula is:

[0083]

[0084] In the formula, F2 is the second fitness, D2 is the total inspection efficiency corresponding to the population, L k is the inspection efficiency of the kth pending inspector in the population when inspecting the corresponding inspection point;

[0085] A selection module is configured to select, within a first preset number of iterations, populations whose first fitness is less than or equal to a first fitness threshold according to the first fitness corresponding to the populations, and select the top three populations with the smallest second fitness from among the populations eliminated at this time, and save them into a backup library;

[0086] When the first preset number of iterations is exceeded, the current population is combined with the population in the backup library to form a new population, and the population corresponding to the current population whose second fitness is less than the preset second fitness threshold is selected;

[0087] The crossover and mutation module is used to hybridize and mutate the selected population to obtain the offspring population;

[0088] A loop module is used to continue executing the fitness calculation module after obtaining the offspring population until a preset number of iterations is met;

[0089] The output module is used to output the offspring population as the optimal solution set for multi-objective optimization.

[0090] The analysis module is used to obtain the cloud inspection results corresponding to the inspection points uploaded by the pending inspection personnel, and based on the current inspection results corresponding to the current inspection personnel and the corresponding cloud inspection results, analyze the inspection difference results corresponding to the current inspection personnel and the pending inspection personnel and output the corresponding inspection difference report, and send the inspection difference report to the user AR inspection terminal corresponding to the current inspection personnel for visual display.

[0091] The abnormal point identification module is used to identify the inspection points where the predicted inspection results are different from the current inspection results, retrieve the location information corresponding to the identified inspection points, and obtain the current location information corresponding to the user AR inspection terminal corresponding to the current inspection personnel in real time, calculate the position distance difference between the identified inspection point and the user AR inspection terminal corresponding to the current inspection personnel, judge the size relationship between the position distance difference and the preset distance threshold, and if the position distance difference is less than or equal to the preset distance threshold, send an alarm signal to the user AR inspection terminal corresponding to the current inspection personnel.

[0092] like Figure 2 As shown, this embodiment also discloses a BIM+AR-based on-site inspection method, which uses the above-mentioned BIM+AR-based on-site inspection system and includes the following steps:

[0093] S1. Construct BIM models of each building corresponding to the sub-buildings to be inspected in each inspection time period of the building to be inspected;

[0094] S2. Obtain historical inspection results corresponding to buildings similar to the building to be inspected from the database;

[0095] S3. Based on the historical inspection results obtained and the building BIM model corresponding to each inspection time period, determine the inspection points corresponding to each inspection time period and the location information of the inspection points in the corresponding building BIM model, and formulate the inspection tasks corresponding to each inspection point and store them in the corresponding database;

[0096] S4. When the corresponding inspection time period is reached, the inspection points corresponding to the corresponding inspection time period, the location information of the inspection points in the corresponding building BIM model, and the inspection tasks of each inspection point are retrieved from the database and sent to the user AR inspection terminal;

[0097] S5. Use the user's AR inspection terminal to monitor the sub-building to be inspected corresponding to the current inspection time period in real time, and obtain the corresponding monitoring video data of the current inspection personnel during the inspection process in real time;

[0098] S6. Upload the current inspection results corresponding to each inspection point during the inspection process by the current inspection personnel;

[0099] S7. Retrieving the inspection points corresponding to the identity information of the current inspection personnel based on the identity information of the current inspection personnel, selecting inspection points corresponding to the non-inspection points based on the inspection points, and forming a corresponding non-inspection point list. Each inspection point in the non-inspection point list is associated with its corresponding surveillance video data;

[0100] S8. Based on the list of non-inspection strong points and the corresponding surveillance video data, the AI ​​inspection recognition model with preset settings is used to identify the predicted inspection results corresponding to each inspection point in the non-inspection strong points;

[0101] S9. Based on the predicted inspection results corresponding to each inspection point in the non-inspection expertise points and the corresponding current inspection results, identify the inspection points where the predicted inspection results are different from the current inspection results, and feed back the identified inspection points to the user AR inspection terminal corresponding to the current inspection personnel.

[0102] The above is only an embodiment of the present invention. Common knowledge such as the known specific structures and characteristics in the scheme is excessively described here. Ordinary technicians in the relevant field are aware of all common technical knowledge in the technical field of the invention before the application date or priority date, can obtain all existing technologies in the field, and have the ability to apply conventional experimental means before that date. Ordinary technicians in the relevant field can improve and implement this scheme in combination with their own abilities under the inspiration given by this application. Some typical known structures or known methods should not become obstacles for ordinary technicians in the relevant field to implement this application. It should be pointed out that for those skilled in the art, without departing from the structure of the present invention, several variations and improvements can be made, which should also be regarded as the scope of protection of the present invention. These will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. A BIM+AR-based on-site inspection system, characterized by: Including the server and user AR inspection terminal; The server includes: A model building module is used to build BIM models of each building corresponding to the sub-buildings to be inspected in each inspection time period of the building to be inspected; A historical inspection result acquisition module is used to obtain historical inspection results corresponding to buildings similar to the building to be inspected from the database; The inspection task generation module is used to determine the inspection points corresponding to each inspection time period and the location information of the inspection points in the corresponding building BIM model based on the acquired historical inspection results and the building BIM model corresponding to each inspection time period, and formulate the inspection tasks corresponding to each inspection point and store them in the corresponding database; The transmission module is used to retrieve the inspection points corresponding to the corresponding inspection time period, the location information of the inspection points in the corresponding building BIM model, and the inspection tasks of each inspection point from the database and send them to the user AR inspection terminal when the corresponding inspection time period is reached; The monitoring data acquisition module is used to monitor the sub-buildings to be inspected corresponding to the current inspection time period in real time through the user's AR inspection terminal, and obtain the corresponding monitoring video data of the current inspection personnel during the inspection process in real time; The inspection result upload module is used to upload the current inspection results corresponding to each inspection point during the inspection process of the current inspection personnel; A processing module is used to retrieve the inspection points corresponding to the identity information of the current inspection personnel according to the identity information corresponding to the current inspection personnel, and based on the inspection points, select the inspection points corresponding to the non-inspection points to form a corresponding non-inspection point list, and associate each inspection point in the non-inspection point list with the corresponding monitoring video data; The intelligent recognition module is used to identify the predicted inspection results corresponding to each inspection point in the non-inspection strong points according to the list of non-inspection strong points and the corresponding monitoring video data, based on the preset AI inspection recognition model; The comparison module is used to identify the inspection points where the predicted inspection results are different from the current inspection results based on the predicted inspection results corresponding to each inspection point in the non-inspection strong points and the corresponding current inspection results, and feed back the identified inspection points to the user AR inspection terminal corresponding to the current inspection personnel.

2. The BIM+AR-based on-site inspection system according to claim 1, characterized in that: The server also includes: The cloud inspection module is used to select the optimal inspection personnel corresponding to each inspection point based on the inspection points where the predicted inspection results are different from the current inspection results, and form a corresponding optimal inspection personnel plan based on the various pending inspection personnel stored in the database, and send the monitoring video data corresponding to each identified inspection point to the corresponding pending inspection personnel based on the optimal inspection personnel plan; The analysis module is used to obtain the cloud inspection results corresponding to the inspection points uploaded by the pending inspection personnel, and based on the current inspection results corresponding to the current inspection personnel and the corresponding cloud inspection results, analyze the inspection difference results corresponding to the current inspection personnel and the pending inspection personnel and output the corresponding inspection difference report, and send the inspection difference report to the user AR inspection terminal corresponding to the current inspection personnel for visual display.

3. The BIM+AR-based on-site inspection system according to claim 2, characterized in that: The cloud inspection module includes: The initial screening module is used to randomly generate N initial populations based on the inspection points where the predicted inspection results are different from the current inspection results, based on the pending inspection personnel stored in the database. The individuals in the initial population are the pending inspection personnel set corresponding to each inspection point in the non-inspection strong point list; based on the preliminary screening conditions, the individuals in the initial population are screened and judged. If the conditions are met, it is a feasible solution, otherwise it is an infeasible solution, and the corresponding initial population is eliminated; A fitness calculation module, used to calculate the fitness of the selected population according to a preset first fitness calculation formula and a second fitness calculation formula; The first fitness calculation formula is: In the formula, F1 is the first fitness, D1 is the inspection benefit value of the pending inspection personnel corresponding to the population, S i is the inspection accuracy of the i-th pending inspector in the population when inspecting the corresponding inspection point, M i is the inspection task satisfaction of the i-th pending inspector in the population when inspecting the corresponding inspection point, α, β are the corresponding weighting coefficients, suiji is a randomly generated random number between 0 and 1, P is the random threshold, x ij is the accurate value corresponding to the inspection point corresponding to the jth inspection of the i-th pending inspector in the population. If it is accurate, then x ij =1, otherwise 0; m is the number of pending inspectors in the population, n is the number of inspections of a certain inspection point by the pending inspectors; y is a dynamic value, the size of y is related to the number of times c that the random number is less than P and the preset value B; The second fitness calculation formula is: In the formula, F2 is the second fitness, D2 is the total inspection efficiency corresponding to the population, L k is the inspection efficiency of the kth pending inspector in the population when inspecting the corresponding inspection point; A selection module is used to select a population whose first fitness is less than or equal to a first fitness threshold according to the first fitness corresponding to the population within a first preset number of iterations, and select the first three populations with the smallest second fitness from the populations eliminated at this time, and save them into a spare library; When the first preset number of iterations is exceeded, the current population is combined with the population in the standby library to form a new population, and the population corresponding to the current population whose second fitness is less than the preset second fitness threshold is selected; The crossover mutation module is used to cross and mutate the selected population to obtain the offspring population; A loop module is used to continue executing the fitness calculation module after obtaining the offspring population until a preset number of iterations is met; The output module is used to output the offspring population as the optimal solution set for multi-objective optimization.

4. The BIM+AR-based on-site inspection system according to claim 3 is characterized in that: The server also includes: The abnormal point identification module is used to retrieve the location information corresponding to the identified inspection points based on the inspection points where the predicted inspection results are different from the current inspection results, and obtain the current location information corresponding to the user AR inspection terminal corresponding to the current inspection personnel in real time, calculate the position distance difference between the identified inspection points and the user AR inspection terminal corresponding to the current inspection personnel, judge the relationship between the position distance difference and the preset distance threshold, and if the position distance difference is less than or equal to the preset distance threshold, send an alarm signal to the user AR inspection terminal corresponding to the current inspection personnel.

5. A BIM+AR-based on-site inspection method, using a BIM+AR-based on-site inspection system according to any one of claims 1 to 4, characterized in that: The following steps are involved: S1. Construct BIM models of each building corresponding to the sub-buildings to be inspected in each inspection time period of the building to be inspected; S2. Obtain historical inspection results corresponding to buildings similar to the building to be inspected from the database; S3. According to the historical inspection results obtained, based on the building BIM model corresponding to each inspection time period, determine the inspection points corresponding to each inspection time period and the location information of the inspection points in the corresponding building BIM model, and formulate the inspection tasks corresponding to each inspection point and store them in the corresponding database; S4. When the corresponding inspection time period is reached, the inspection points corresponding to the corresponding inspection time period, the location information of the inspection points in the corresponding building BIM model, and the inspection tasks of each inspection point are retrieved from the database and sent to the user AR inspection terminal; S5. Use the user's AR inspection terminal to monitor the sub-building to be inspected corresponding to the current inspection time period in real time, and obtain the corresponding monitoring video data of the current inspection personnel during the inspection process in real time; S6. Upload the current inspection results corresponding to each inspection point during the inspection process by the current inspection personnel; S7. According to the identity information corresponding to the current patrol inspector, retrieve the patrol points corresponding to the identity information of the current patrol inspector, select the patrol points corresponding to the non-patrol points based on the patrol points, and form a corresponding non-patrol point list. Each patrol point in the non-patrol point list is associated with the corresponding surveillance video data; S8. According to the list of non-inspection strong points and the corresponding monitoring video data, based on the preset AI inspection recognition model, identify the predicted inspection results corresponding to each inspection point in the non-inspection strong points; S9. According to the predicted inspection results corresponding to each inspection point among the non-inspection expertise points and the corresponding current inspection results, the inspection points where the predicted inspection results differ from the current inspection results are identified, and the identified inspection points are fed back to the user AR inspection terminal corresponding to the current inspection personnel.

Images