An AI-assisted rapid repair process for damaged real stone paint curtain wall surfaces

Abstract

The present invention discloses an AI-intervened rapid repair process for surface damage of real stone paint curtain walls, which belongs to the field of curtain wall repair. This process utilizes infrared scanning of the concave and convex balance points on the curtain wall surface to quickly and preliminarily locate possible damage to the curtain wall, and then uses ultrasonic flaw detection to further confirm the points where possible damage occurs. After the confirmation is completed, repairs are carried out according to the situation. During the entire repair process, workers only need to lift and position the lifting equipment at the lifting equipment, and do not need to enter the cage. The equipment can be controlled to perform the actions of the predetermined program through a predetermined program, which greatly increases safety. At the same time, due to the clear division of labor and rapid positioning of the equipment, efficiency is improved and the process can be carried out 24 hours a day, eliminating the embarrassing situation that manual labor is not suitable for high-altitude operations at night.

Description

Technical Field

[0001] The invention discloses an AI-intervened rapid repair process for surface damage of a real stone paint curtain wall, belonging to the field of curtain wall repair. Background Art

[0002] Curtain wall repair, a technology used to repair curtain wall surfaces.

[0003] The existing curtain wall needs to be repaired manually, and when repairing the curtain wall manually, it is necessary to build a large lifting equipment on the top floor, and then lower the cage for workers to be inside the cage and perform manual inspection and repair along the cage. In addition, since different tools are used at different times during the repair, if a large number of tools are carried at one time, the overall weight of the cage will increase. If not all the tools are brought, the repair needs to be repeated many times, which is very dangerous and cumbersome for the workers in the cage.

[0004] Now a new repairing process is provided to solve the above problems. Summary of the Invention

[0005] The purpose of the present invention is to solve the above problems and provide an AI intervention real stone paint curtain wall surface damage rapid repair process.

[0006] The present invention achieves the above-mentioned purpose through the following technical solutions, which include an AI-assisted rapid repair process for damaged surfaces of real stone paint curtain walls, comprising the following steps:

[0007] The S1 coordinate point is established with the vertical direction as the Y axis, the horizontal direction of the curtain wall as the X axis, and one corner of the curtain wall as the origin;

[0008] S2 scanning uses an infrared rangefinder to connect the two ends of the curtain wall to form a baseline, then scans the curtain wall vertically and transmits the scanned information to the AI ​​cloud computing center to find the concave and convex points on the curtain wall surface. The coordinate positions of the concave and convex points are then uploaded to the central controller.

[0009] S3 problem confirmation: input the coordinates into the hoisting equipment, then use the hoisting equipment to transport the ultrasonic testing machine to the coordinates, perform ultrasonic scanning on the coordinates, and classify the scanning results into three states: hollow, missing, and normal. The normal state will send a signal to delete the corresponding coordinates to the central controller, and the hollow and missing states will be summarized into two lists respectively;

[0010] S4 hollowing processing: import the hollow coordinates into the hoisting equipment, then use the hoisting equipment to send the chiseling equipment to the corresponding coordinates, use the chiseling equipment to break and remove the hollow part, and after removal, put the corresponding coordinates into the missing list;

[0011] S5 missing processing, import the missing coordinates into the hoisting equipment, and then use the hoisting equipment to send the wire drawing equipment and spray cleaning equipment to the corresponding coordinates, and perform wire drawing and surface enhancement treatment on the metal curtain wall surface. After completion, use the water gun of the spray cleaning equipment to spray the treated area, and finally blow to remove impurities and dry at the same time. Finally, use the spray gun in the spray cleaning equipment to spray the interface agent to the treated surface;

[0012] S6 painting is finished. The missing coordinates are imported into the hoisting equipment, and then the painting equipment is sent to the corresponding coordinates through the hoisting equipment. After arriving at the coordinate point, the metal curtain wall surface is first heated by a drying gun. After the surface temperature of the metal curtain wall is heated to 60℃, painting can be carried out. After the painting is completed, the repair work is completed.

[0013] Preferably, the problem of S3 can also be confirmed through visual recognition, echo analysis and thermal imaging inspection.

[0014] Preferably, the hollowing treatment in S4 can clean and crush the chiseled real stone paint and then add it to the paint spraying equipment in S6, so as to reduce the generation of color difference by recycling the raw materials and re-spraying.

[0015] Preferably, the S5 wire drawing and water gun flushing can be performed simultaneously.

[0016] Preferably, the wire drawing direction is carried out in a horizontal direction.

[0017] Preferably, when finishing the painting of S6, a heat-insulating film can be pasted on the outside of the curtain wall to maintain the temperature during the initial heating and speed up the drying speed.

[0018] Preferably, after all curtain walls are repaired, S6 can send hot air into between the curtain wall and the film through a heat pump to accelerate the drying speed.

[0019] Preferably, in S6, the humidity between the film and the curtain wall is adjusted using a dehumidifier or a humidifier according to the regional humidity to ensure that the humidity is maintained at 40%-60%.

[0020] Preferably, the equipment used in S1-S6 is equipped with a negative pressure device for making the equipment fit against the curtain wall.

[0021] Preferably, in the scanning of S2, AI is used to further analyze the concave and convex points on the curtain wall surface using the following process:

[0022] S2-1 point cloud segmentation and denoising: using deep learning models to remove clutter and noise from point cloud data, eliminate environmental interference, and retain surface information;

[0023] S2-2 Concave and convex feature recognition: Through convolutional neural networks or Transformer models, the features of concave and convex points are learned to distinguish between normal surface roughness and abnormal areas of real stone paint;

[0024] S2-3 precision optimization, combined with transfer learning, migrates the labeled concave and convex samples to the previously scanned real stone paint scene, improving the distinction between the surface roughness and concave and convex points of the real stone paint, and improving the detection accuracy of complex surfaces.

[0025] Compared with the prior art, the beneficial effects of the present invention are: this process utilizes infrared scanning of the concave and convex balance points on the surface of the curtain wall to quickly and preliminarily locate possible damage to the curtain wall, and then uses ultrasonic testing to further confirm the points where damage may occur. After the confirmation is completed, repairs are carried out according to the situation. During the entire repair process, workers only need to lift and position the lifting equipment at the lifting equipment. There is no need for workers to enter the cage. The equipment can be controlled to perform predetermined program actions through a predetermined program, which greatly increases safety. Due to the clear division of labor and rapid positioning of the equipment, efficiency is improved and it can be carried out 24 hours a day, eliminating the embarrassing situation that manual work is not suitable for high-altitude operations at night. DETAILED DESCRIPTION

[0026] An AI-assisted rapid repair process for damaged stone paint curtain wall surfaces includes the following steps:

[0027] The S1 coordinate point is established with the vertical direction as the Y axis, the horizontal direction of the curtain wall as the X axis, and one corner of the curtain wall as the origin;

[0028] S2 scanning uses an infrared rangefinder to connect the two ends of the curtain wall to form a baseline, and then scans the curtain wall in the vertical direction of the curtain wall, and transmits the scanned information to the AI ​​cloud computing center to find the concave and convex points on the curtain wall surface, and uploads the coordinate positions of the concave and convex points to the central controller. The computing center uses a network cloud server base station, and the central controller uses a local server or computer. Here, the damage to the curtain wall of real stone paint is mostly due to the poor bonding between the paint surface and the aluminum plate, which causes the real stone paint to become hollow or fall off. The fall will cause concave points to form on the curtain wall surface, and hollowing will cause convex points to form on the current surface. Therefore, when the infrared rangefinder scans the curtain wall, it can mark the convex and concave points. At this time, the X-axis coordinate can be obtained by the horizontal projection distance of the convex point or concave point from the origin, and the distance of the convex point in the vertical direction is the displacement distance of the infrared rangefinder. At this time, the infrared rangefinder can be used to obtain the concave and convex data of each point on the same horizontal plane, and the concave and convex points can be marked by using the summarized data to achieve the effect of quickly locating suspicious points.

[0029] S3 problem confirmation, input the coordinates into the hoisting equipment, and then use the hoisting equipment to transport the ultrasonic detection machine to the coordinates, perform ultrasonic scanning on the coordinate parts, and divide the scanning results into three states: hollowing, missing and normal. The normal state will send a signal to delete the corresponding coordinates to the central controller, and summarize the hollowing and missing states into two lists respectively. Here, classification can be performed automatically through software programs or through manual observation. Using prior data, the detection equipment is sent to the coordinate point corresponding to the suspicious point, and then the suspicious point is further confirmed by the detection equipment. Ultrasonic waves can effectively present the connection between real stone paint and metal curtain wall. Through imaging, it can be known that: a gap between real stone paint and metal curtain wall is hollowing, only the metal curtain wall is missing, and the real stone paint and metal curtain wall are tightly combined without a gap, which is normal;

[0030] S4 hollowing processing: import the hollow coordinates into the hoisting equipment, then use the hoisting equipment to send the chiseling equipment to the corresponding coordinates, use the chiseling equipment to break and remove the hollow part, and after removal, put the corresponding coordinates into the missing list;

[0031] S5 missing processing, import the missing coordinates into the hoisting equipment, and then use the hoisting equipment to send the wire drawing equipment and spray cleaning equipment to the corresponding coordinates, and perform wire drawing and surface enhancement treatment on the metal curtain wall surface. After completion, use the water gun of the spray cleaning equipment to spray the treated area, and finally blow to remove impurities and dry at the same time. Finally, use the spray gun in the spray cleaning equipment to spray the interface agent to the treated surface for wire drawing to increase the surface roughness and effectively increase the bonding area, so that the paint surface can better bond with the metal curtain wall;

[0032] S6 painting is finished. The missing coordinates are imported into the hoisting equipment, and then the painting equipment is sent to the corresponding coordinates through the hoisting equipment. After arriving at the coordinate point, the metal curtain wall surface is first heated by a drying gun. After the surface temperature of the metal curtain wall is heated to 60℃, painting can be carried out. After the painting is completed, the repair work is completed.

[0033] Through the above process, work can be carried out in windy conditions without having to select specific days due to manual risks, which greatly increases work efficiency. In addition, this product can straighten the cables of the lifting equipment by fixing them up and down, allowing the equipment to carry out repair work in an orderly manner even in strong winds.

[0034] Problem confirmation for S3 can also be achieved through visual recognition, echo analysis, and thermal imaging inspections. A variety of inspection equipment is available, and operators can select specific equipment based on the environment.

[0035] Among them, the hollowing treatment of S4 can wash and crush the chiseled real stone paint and then add it to the S6 paint spraying equipment. The color difference can be reduced by re-spraying the raw materials. The real stone paint can be reused after adding chemicals for washing. This process can recycle the chiseled real stone paint through a vacuum cleaner and then crush and wash it for secondary use.

[0036] Among them, S5 brushing and water gun flushing can be carried out at the same time. Flushing and brushing are carried out synchronously. Water can not only cool the equipment, but also avoid the deposition of debris and cause adhesion and tightening. The brushing direction is horizontal. The horizontal brushing can produce horizontal grooves in the microscopic direction, and the interface agent is used to penetrate into the groove to form a hanging form, so that the real stone paint is not easily affected by gravity and falls off.

[0037] Among them, S6 can finish painting by sticking insulation film on the outside of the curtain wall. The film is pasted by placing a whole roll of insulation film on the top of the painting equipment, and installing a pressure wheel on the line to abut the two sides of the insulation film through the pressure wheel, so as to achieve the adhesion of the insulation film edge to the curtain wall surface. After all the curtain walls are repaired, S6 can use the heat pump to send hot air into the space between the curtain wall and the film to speed up the drying speed. In S6, according to the regional humidity, a dehumidifier or humidifier is used to adjust the humidity between the film and the curtain wall to ensure that the humidity is maintained at 40%-60%, so as to maintain the temperature during the early heating and speed up the drying speed. The coating can effectively create the most suitable drying environment for real stone paint, so that the adhesion effect of the real stone paint after drying is stronger.

[0038] Here, the thermal insulation film can also be directly adhered to the cables of the lifting equipment. The characteristic that the cables need to be attached to the surface of the curtain wall during lifting can separate the thermal insulation film from most of the outside world, thereby reducing the air flow between the curtain wall and the thermal insulation film and achieving the effect of heat preservation.

[0039] The equipment used in S1-S6 are all equipped with a negative pressure device for making the equipment fit against the curtain wall. The negative pressure device is suitable for not straightening the cables in the lifting equipment, and through negative pressure adsorption, the equipment can be better fitted to the curtain wall surface.

[0040] For S2 scanning, AI was used to further analyze the concave and convex points on the curtain wall surface using the following process:

[0041] S2-1 point cloud segmentation and denoising: using deep learning models to remove clutter and noise from point cloud data, eliminate environmental interference, and retain surface information;

[0042] S2-2 Convex and concave point feature recognition: Through convolutional neural networks or Transformer models, the features of concave and convex points are learned to distinguish between normal surface roughness and abnormal areas of real stone paint;

[0043] S2-3 precision optimization, combined with transfer learning, migrates the labeled concave and convex samples to the previously scanned real stone paint scene, improving the distinction between the surface roughness and concave and convex points of the real stone paint, and improving the detection accuracy of complex surfaces.

[0044] Deep learning models use PointNET or PointCNN to learn features, effectively removing image noise and thus increasing the accuracy of collected information. Convolutional neural networks (CNNs) or Transformer models learn features of concave and convex points, enabling better distinction between surface roughness and abnormal areas of real stone paint, leading to more precise positioning and effective understanding of the overall scene, reducing misjudgments. The precision-optimized door system, combined with transfer learning, transfers samples of concave and convex points to a previously scanned real stone paint scene, leveraging these differences to enhance learning and achieve precision optimization. Transfer learning also allows for better annotation of sample data in other scenes similar to real stone paint, allowing the model to quickly adapt to different scenarios.

[0045] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations that come within the meaning and range of equivalents of the claims be embraced therein.

[0046] In addition, it should be understood that although this specification is described in terms of implementation methods, not every implementation method contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

Claims

1. An AI-assisted rapid repair process for damaged stone paint curtain wall surfaces, characterized in that: The following steps are involved: The S1 coordinate point is established with the vertical direction as the Y axis, the horizontal direction of the curtain wall as the X axis, and one corner of the curtain wall as the origin; S2 scanning uses an infrared rangefinder to connect the two ends of the curtain wall to form a baseline, then scans the curtain wall vertically and transmits the scanned information to the AI ​​cloud computing center to find the concave and convex points on the curtain wall surface. The coordinate positions of the concave and convex points are then uploaded to the central controller. S3 problem confirmation: input the coordinates into the hoisting equipment, then use the hoisting equipment to transport the ultrasonic testing machine to the coordinates, perform ultrasonic scanning on the coordinates, and classify the scanning results into three states: hollow, missing, and normal. The normal state will send a signal to delete the corresponding coordinates to the central controller, and the hollow and missing states will be summarized into two lists respectively; S4 hollowing processing: import the hollow coordinates into the hoisting equipment, then use the hoisting equipment to send the chiseling equipment to the corresponding coordinates, use the chiseling equipment to break and remove the hollow part, and after removal, put the corresponding coordinates into the missing list; S5 missing processing, import the missing coordinates into the hoisting equipment, and then use the hoisting equipment to send the wire drawing equipment and spray cleaning equipment to the corresponding coordinates, and perform wire drawing and surface enhancement treatment on the metal curtain wall surface. After completion, use the water gun of the spray cleaning equipment to spray the treated area, and finally blow to remove impurities and dry at the same time. Finally, use the spray gun in the spray cleaning equipment to spray the interface agent to the treated surface; S6 painting is finished. The missing coordinates are imported into the hoisting equipment, and then the painting equipment is sent to the corresponding coordinates by the hoisting equipment. After arriving at the coordinate point, the metal curtain wall surface is first heated by the drying gun. After the surface temperature of the metal curtain wall is heated to 60℃, the painting can be carried out. After the painting is completed, the repair work is completed; When finishing the painting of S6, a heat-insulating film can be pasted on the outside of the curtain wall to maintain the temperature during the initial heating and speed up the drying process. After all curtain walls are repaired, S6 can use a heat pump to send hot air between the curtain wall and the film to speed up the drying process.

2. The AI-assisted rapid repair process for damaged stone paint curtain wall surfaces according to claim 1, characterized in that: Problems with S3 can also be confirmed through visual recognition, echo analysis, and thermal imaging inspections.

3. The AI-assisted rapid repair process for damaged stone paint curtain wall surfaces according to claim 1 is characterized by: Among them, the hollowing treatment of S4 can wash and crush the chiseled real stone paint and then add it to the S6 painting equipment, and reduce the color difference by recycling the raw materials and re-spraying.

4. The AI-assisted rapid repair process for damaged stone paint curtain wall surfaces according to claim 1 is characterized by: Among them, S5 wire drawing and water gun washing can be carried out at the same time.

5. The AI-assisted rapid repair process for damaged stone paint curtain wall surfaces according to claim 1 is characterized by: The drawing direction is along the horizontal direction.

6. The AI-assisted rapid repair process for damaged stone paint curtain wall surfaces according to claim 1 is characterized by: In S6, a dehumidifier or humidifier is used to adjust the humidity between the film and the curtain wall according to the regional humidity to ensure that the humidity is maintained at 40%-60%.

7. The AI-assisted rapid repair process for damaged stone paint curtain wall surfaces according to claim 1 is characterized by: The equipment used in S1-S6 is equipped with a negative pressure device to make the equipment fit against the curtain wall.

8. The AI-assisted rapid repair process for damaged stone paint curtain wall surfaces according to claim 1 is characterized by: For S2 scanning, AI was used to further analyze the concave and convex points on the curtain wall surface using the following process: S2-1 point cloud segmentation and denoising: using deep learning models to remove clutter and noise from point cloud data, eliminate environmental interference, and retain surface information; S2-2 Concave and convex feature recognition: Through convolutional neural networks or Transformer models, the features of concave and convex points are learned to distinguish between normal surface roughness and abnormal areas of real stone paint; S2-3 precision optimization, combined with transfer learning, migrates the labeled concave and convex samples to the previously scanned real stone paint scene, improving the distinction between the surface roughness and concave and convex points of the real stone paint, and improving the detection accuracy of complex surfaces.