An engineering inspection system and method based on artificial intelligence
Through the artificial intelligence system, the inspection routes and regional commands are updated in real time, the problems of low efficiency of manual or drone inspections and poor adaptability to scene changes in construction engineering inspections are solved, and efficient and accurate inspection tasks are achieved, reducing the implementation of inspection tasks. Human resources consumption.
Patent Information
- Application Number
- CN202111560033.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-12-20
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2041-12-20
AI Technical Summary
In the prior art, construction engineering inspection methods rely on manual or drone inspections, which have problems such as low efficiency, large subjective errors and heavy task command planning, making it difficult to adapt to the rapid changes in the construction site in real time.
Adopting an engineering inspection system based on artificial intelligence, image information is collected through inspection devices and actual models are generated with the BIM server. The intelligent analysis unit updates inspection routes and acquisition area instructions in real time, and uses incremental autonomous visual learning driven by visual strangeness. Algorithm, automatically adjusts inspection tasks and reduces manual intervention.
It improves the accuracy and efficiency of inspections, reduces labor costs and misjudgments, realizes real-time tracking of changes in construction sites, saves a lot of human resources, and improves construction quality control.
Smart Images

Figure CN114140899B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of engineering inspection, and in particular, to an engineering inspection system and method based on artificial intelligence. Background Art
[0002] The purpose of construction project inspection is to fully understand the on-site construction situation, timely discover and solve relevant problems in construction through rectification and review of project inspection quality results, and improve the company's ability to control project quality and other goals.
[0003] In the existing technology, there are two ways of engineering inspection: manual inspection and drone inspection. Compared with the inefficiency and subjective errors of manual inspection, drone inspection has the advantages of fast speed and timely response.
[0004] However, the on-site environment of construction projects is complex, and there are many engineering inspection and detection items (such as scaffolding, personnel safety helmets, fire hazards, etc.). In addition, as the project progresses, the various scene elements on the project site change rapidly. The task instructions for drone inspections are manually determined, and the planning of its instructions requires the collection of a large amount of data. In addition, as the scene on the project site changes, the inspection task instructions need to be updated in real time. If only manual planning is relied upon, the workload will be very heavy, making it difficult to carry out inspection work well. Summary of the invention
[0005] In view of this, the purpose of the present invention is to provide an engineering inspection system and method based on artificial intelligence, which are mainly used for intelligently updating inspection task data according to scene changes during engineering inspection.
[0006] In order to solve the above technical problems, the technical solution of the present invention is:
[0007] An engineering inspection system based on artificial intelligence comprises an inspection device and a control center, wherein the inspection device is connected to the control center via a wireless network; the inspection device comprises a collection mechanism for collecting image information at a construction project site, a driving mechanism for driving the collection mechanism to move, and a second transceiver unit; the control center comprises a first transceiver unit and a control instruction unit, wherein the control instruction unit stores inspection route instructions and collection area instructions for controlling the inspection device, wherein the collection area instructions include the location of the inspection device and the shooting angle of the collection mechanism; the control center comprises a BIM server and a problem analysis unit, wherein a reference model is pre-stored in the BIM server, the BIM server models the image information collected by the collection mechanism and generates an actual model, and the problem analysis unit compares the actual model with the reference model to generate a problem point; the control center also comprises an intelligent analysis unit, wherein the intelligent analysis unit judges and updates the inspection route instruction and the collection area instruction by real-time analysis of the image information collected by the collection mechanism.
[0008] Compared with the prior art, the advantages of the present invention are:
[0009] By collecting image information from construction sites through inspection devices, instead of manually collecting data, problems such as high labor costs, low collection efficiency, and easy subjective omissions can be avoided.
[0010] By analyzing problems based on the collected image information, the control center can present problem points in an intelligent and specific manner with high accuracy and efficiency, thus avoiding the high cost, low efficiency, and subjective misjudgment and omission of manual analysis.
[0011] By analyzing image information in real time through the intelligent analysis unit, the inspection route can be updated in real time according to the changes in scene elements, and new area ranges can be actively explored to update the image acquisition area. There is no need for manual data collection and planning, and updates are performed during each inspection. It can adapt to the scene changes at the construction project site in a timely manner, follow up the project progress, and save a lot of manpower.
[0012] Furthermore, the intelligent analysis unit includes a route analysis unit and an area analysis unit. The route analysis unit analyzes the image information of the acquisition mechanism in real time, confirms the changes of scene elements on the inspection route, re-plans the route, and generates new inspection route instructions; the area analysis unit analyzes the image information of the acquisition mechanism in real time, confirms the changes of scene elements on the inspection route, and controls the acquisition mechanism to acquire image information of areas outside the original acquisition area instructions. The area analysis unit analyzes the acquired image information to increase / decrease the scope of the acquisition area, generates new acquisition area instructions, analyzes scene changes in real time, and automatically updates inspection instructions, with a high degree of intelligence.
[0013] Furthermore, the operation of the intelligent analysis unit is based on an incremental autonomous visual learning algorithm driven by visual unfamiliarity.
[0014] Furthermore, the engineering inspection system also includes a manual feedback end, the control center is connected to the manual feedback end through a wireless network, the manual feedback end includes a third transceiver unit and a display unit, the control center sends the problem point to the manual feedback end, and the manual feedback end makes feedback instructions to the control center, and the feedback instructions include positive feedback and negative feedback. When positive feedback is made, the intelligent analysis unit stores the new inspection route instruction and the new collection area instruction in the control instruction unit and assigns the highest priority. When negative feedback is made, the intelligent analysis unit deletes the new inspection route instruction and the new collection area instruction. Manual feedback avoids system errors.
[0015] Furthermore, the problem analysis unit captures several frames of problem images from the image information of the acquisition mechanism according to the problem point, and the control center sends the problem images to the manual feedback terminal to promptly notify the staff to handle them.
[0016] Furthermore, the acquisition mechanism includes a panoramic camera for acquiring panoramic images, and a laser scanner for acquiring three-dimensional stereo data.
[0017] Furthermore, the driving mechanism includes a drone or a self-propelled vehicle.
[0018] A method for an engineering inspection system based on artificial intelligence, comprising the following steps:
[0019] S1. Collecting image information that complies with the specifications at the construction project site through the inspection device or manually and uploading it to the control center, and the BIM server generates a reference model based on the image information;
[0020] S2, the control center sends an inspection route instruction and a collection area instruction to the inspection device, and the inspection device starts the inspection work;
[0021] S3, the acquisition mechanism acquires image information and sends it to the BIM server, the BIM server models the image information to generate an actual model, and the problem analysis unit compares the actual model with the reference model to generate problem points;
[0022] S4, the problem analysis unit captures several frames of problem images according to the problem point, and sends them to the manual feedback end;
[0023] S5, the intelligent analysis unit analyzes the image information in real time, determines the changes of the inspection route and the collection area, generates new inspection route instructions and collection area instructions, and sends them to the manual feedback terminal;
[0024] S6. The manual feedback end provides manual feedback on the new inspection route instruction and the new collection area instruction. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] Figure 1 It is a schematic diagram of the system structure of the present invention;
[0026] Figure 2 Flowchart of the algorithm for incremental autonomous visual learning driven by visual unfamiliarity. DETAILED DESCRIPTION
[0027] The specific implementation modes of the present invention are further described below in conjunction with the accompanying drawings to make the technical solutions of the present invention easier to understand and grasp.
[0028] Embodiment 1:
[0029] This embodiment provides an engineering inspection system based on artificial intelligence. Figure 1 As shown, the inspection device includes a patrol device, which includes a collection mechanism, which includes a panoramic camera for collecting panoramic images and a laser scanner for collecting three-dimensional data. The inspection device also includes a driving mechanism for driving the collection mechanism to move, and the driving mechanism includes but is not limited to a drone or a self-propelled vehicle. The inspection device moves on the construction project site through the driving mechanism, and collects image information of the project site through the collection mechanism, thereby replacing manual inspection work.
[0030] The engineering inspection system also includes a control center, the inspection device is connected to the control center via a wireless network, the control center includes a first transceiver unit, the inspection device includes a second transceiver unit, and the control center and the inspection device exchange data via the first transceiver unit and the second transceiver unit.
[0031] The control center includes a control instruction unit, which stores inspection route instructions and collection area instructions for controlling the operation of the inspection device, wherein the collection area instruction includes the location of the inspection device and the shooting angle of the collection mechanism. The control instruction unit sends the inspection route instruction and the collection area instruction to the inspection device, and the inspection device moves according to the inspection route instruction and collects images at the location indicated by the collection area instruction, wherein the initial inspection route instruction and the collection area instruction are planned by the staff according to the actual scene conditions at the construction project site and entered into the control instruction unit.
[0032] The control center also includes a BIM server and a problem analysis unit. Before the construction project is put into operation on site, the corresponding initial standard image information is collected manually or by the inspection device. The BIM server generates a reference model based on the image information. The reference model includes geometric information, professional attributes and status information of building components, as well as status information of non-component objects (such as space, movement behavior), specifically scaffolding component information, large equipment information, and construction personnel's standard wear information.
[0033] During the inspection work, the image information transmitted by the acquisition mechanism to the control center is modeled on the BIM server to generate an actual model. The problem analysis unit calls professional software to compare the actual model with the reference model, and determines the problem points in the actual model according to the differences in the models, thereby determining the problems and hidden dangers at the construction project site. Problems and hidden dangers include but are not limited to construction workers not wearing safety helmets or wearing safety belts, flames or high temperature spots in non-fire areas, scaffolding spacing exceeding the safety regulations, and scaffolding deformation monitoring.
[0034] Among them, the engineering inspection system also includes a manual feedback end, the manual feedback end is connected to the control center through a wireless network, the manual feedback end includes a third transceiver unit and a display unit, the control center and the manual feedback end realize data exchange through the first transceiver unit and the third transceiver unit, the problem analysis unit intercepts several frames of images in the corresponding image information according to the problem point, sends them to the manual feedback end and displays them on the display unit, and is manually confirmed by the staff, and the corresponding maintenance personnel are notified to request processing.
[0035] In actual construction projects, the scenes at the construction site are complex and changeable. As the project progresses, the buildings, equipment, personnel, etc. in the scene are changing. The inspection route instructions and the collection area instructions in the control center are specific and need to be updated manually. Additional image information and modeling are required, and new instructions need to be manually verified and determined, which can easily consume a lot of time and manpower. Sites with fast construction progress require more frequent updates, which can easily increase human resource costs.
[0036] Therefore, the control center of the present embodiment is further provided with an intelligent analysis unit, which includes a path analysis unit and a regional analysis unit. Both the path analysis unit and the regional analysis unit analyze the image information of the acquisition mechanism in real time. The path analysis unit confirms the changes of scene elements on the original inspection route according to the image information, and determines whether it is feasible for the inspection mechanism to move according to the original inspection route instruction. If it is feasible, the inspection mechanism moves according to the original inspection route instruction. If it is not feasible, a feasible new inspection route instruction is generated.
[0037] The regional analysis unit confirms the changes of scene elements on the original inspection route based on the image information. If there is no change, the acquisition is performed according to the original acquisition area instruction. If there is a change, the control center controls the acquisition mechanism to acquire image information of areas outside the original image acquisition area. The regional analysis unit analyzes the acquired image information and makes a judgment to increase / decrease the range of the acquisition area instruction and generate a new acquisition area instruction.
[0038] Therefore, when the inspection device is performing inspection work, it is possible to update the inspection route in real time according to changes in scene elements, and actively explore new area ranges to update the image acquisition area, without the need for manual data collection and planning. Moreover, updates are performed during each inspection work, which can timely adapt to scene changes at the construction project site, follow up on project progress, and save a lot of manpower.
[0039] The operation of the intelligent analysis unit is based on an incremental autonomous visual learning algorithm driven by visual unfamiliarity. The specific steps of the incremental autonomous visual learning algorithm driven by visual unfamiliarity are as follows: Figure 2 shown.
[0040] The new inspection route instruction and the new acquisition area instruction generated by the intelligent analysis unit are sent to the manual feedback end and displayed on the display unit. The staff manually confirms the new inspection route instruction and the new acquisition area instruction to make feedback instructions to the control center through the manual feedback end. The feedback instructions include positive feedback and negative feedback. When the new inspection route instruction and the new acquisition area instruction meet the manual confirmation requirements, positive feedback is made, and the intelligent analysis unit stores the new inspection route instruction and the new acquisition area instruction in the control instruction unit and assigns the highest priority; when the new inspection route and the new image acquisition area do not meet the manual confirmation requirements, negative feedback is made, and the intelligent analysis unit deletes them. The control instruction unit retains the old inspection route instruction and the old acquisition area instruction as the current highest priority. Therefore, this manual confirmation can avoid the accidental generation of erroneous instructions and the erroneous execution of inspection work.
[0041] Embodiment 2:
[0042] This embodiment provides a method for an engineering inspection system based on artificial intelligence, comprising the following steps:
[0043] S1. Collecting image information that complies with the specifications at the construction project site through the inspection device or manually and uploading it to the control center, and the BIM server generates a reference model based on the image information;
[0044] S2, the control center sends an inspection route instruction and a collection area instruction to the inspection device, and the inspection device starts the inspection work;
[0045] S3, the acquisition mechanism acquires image information and sends it to the BIM server, the BIM server models the image information to generate an actual model, and the problem analysis unit compares the actual model with the reference model to generate problem points;
[0046] S4, the problem analysis unit captures several frames of problem images according to the problem point, and sends them to the manual feedback end;
[0047] S5, the intelligent analysis unit analyzes the image information in real time, determines the changes of the inspection route and the collection area, generates new inspection route instructions and collection area instructions, and sends them to the manual feedback terminal;
[0048] S6. The manual feedback end provides manual feedback on the new inspection route instruction and the new collection area instruction.
[0049] The above are only typical examples of the present invention. In addition, the present invention may have many other specific implementations. Any technical solutions formed by equivalent replacement or equivalent transformation shall fall within the scope of protection required by the present invention.
Claims
1. An engineering inspection system based on artificial intelligence, characterized in that: It includes an inspection device and a control center, and the inspection device is connected to the control center through a wireless network; the inspection device includes a collection mechanism for collecting image information on the construction project site, and a driving mechanism for driving the collection mechanism to move, and also includes a second transceiver unit; the control center includes a first transceiver unit and a control instruction unit, and the control instruction unit stores inspection route instructions and collection area instructions for controlling the inspection device, wherein the collection area instructions include the location of the inspection device and the shooting angle of the collection mechanism; the control center includes a BIM server and a problem analysis unit, and the BIM server has a reference model pre-stored therein, and the BIM server models the image information collected by the collection mechanism and generates an actual model, and the problem analysis unit compares the actual model with the reference model to generate a problem point; the control center also includes an intelligent analysis unit, and the intelligent analysis unit judges and updates the inspection route instruction and the collection area instruction by real-time analysis of the image information collected by the collection mechanism; The intelligent analysis unit includes a route analysis unit and a region analysis unit. The route analysis unit analyzes the image information of the acquisition mechanism in real time, confirms the changes of scene elements on the inspection route, re-plans the route, and generates new inspection route instructions; the region analysis unit analyzes the image information of the acquisition mechanism in real time, confirms the changes of scene elements on the inspection route, and controls the acquisition mechanism to acquire image information of areas outside the original acquisition region instructions. The region analysis unit analyzes the acquired image information to increase / decrease the scope of the acquisition region, and generates new acquisition region instructions; The engineering inspection system also includes a manual feedback end. The control center is connected to the manual feedback end through a wireless network. The manual feedback end includes a third transceiver unit and a display unit. The control center sends the problem point to the manual feedback end. The manual feedback end makes a feedback instruction to the control center. The feedback instruction includes positive feedback and negative feedback. When positive feedback is made, the intelligent analysis unit stores the new inspection route instruction and the new collection area instruction in the control instruction unit and assigns the highest priority. When negative feedback is made, the intelligent analysis unit deletes the new inspection route instruction and the new collection area instruction.
2. According to claim 1, an artificial intelligence-based engineering inspection system is characterized in that: The operation of the intelligent analysis unit is based on an incremental autonomous visual learning algorithm driven by visual unfamiliarity.
3. According to the artificial intelligence-based engineering inspection system of claim 1, it is characterized in that: The problem analysis unit intercepts several frames of problem images from the image information of the acquisition mechanism according to the problem points, and the control center sends the problem images to the manual feedback terminal.
4. According to the artificial intelligence-based engineering inspection system of claim 1, it is characterized in that: The acquisition mechanism includes a panoramic camera for acquiring panoramic images and a laser scanner for acquiring three-dimensional stereo data.
5. The engineering inspection system based on artificial intelligence according to claim 1 is characterized in that: The driving mechanism includes a drone or a self-propelled vehicle.
6. The method of an engineering inspection system based on artificial intelligence according to claim 3 is characterized in that: The following steps are involved: S1. Collecting image information that complies with the specifications at the construction project site through the inspection device or manually and uploading it to the control center, and the BIM server generates a reference model based on the image information; S2, the control center sends an inspection route instruction and a collection area instruction to the inspection device, and the inspection device starts the inspection work; S3, the acquisition mechanism acquires image information and sends it to the BIM server, the BIM server models the image information to generate an actual model, and the problem analysis unit compares the actual model with the reference model to generate problem points; S4, the problem analysis unit captures several frames of problem images according to the problem point, and sends them to the manual feedback end; S5, the intelligent analysis unit analyzes the image information in real time, determines the changes of the inspection route and the collection area, generates new inspection route instructions and collection area instructions, and sends them to the manual feedback terminal; S6. The manual feedback end provides manual feedback on the new inspection route instruction and the new collection area instruction.
