Glass curtain wall structure adhesive damage integrated detection wall-climbing robot and system

Abstract

The application discloses a glass curtain wall structure glue damage integrated detection wall-climbing robot and system, the wall-climbing robot includes robot main body, adsorption assembly, holder assembly and structure glue detection assembly;The adsorption assembly includes the negative pressure sealing cavity and the duct fan arranged on the chassis, the holder assembly includes holder support, first steering wheel and second steering wheel;The structure glue detection assembly includes mounting bracket, knocking executor, acceleration sensor and distance sensor, the system includes real-time monitoring module, control interaction module, waveform presentation module, identification feedback module, through the collection acceleration sensor vibration response signal, and the characteristic parameter is extracted input pre-training machine learning model, completes the automatic identification of glass curtain wall structure glue damage, realizes the duct fan negative pressure adsorption, acceleration response collection and machine learning identification structure glue damage integrated integration, can realize efficient, reliable structure glue damage detection on the high-rise glass curtain wall outer facade.

Description

Technical Field

[0001] This invention relates to the field of civil engineering structural testing technology, specifically to a wall-climbing robot and system for integrated detection of adhesive damage in glass curtain wall structures. Background Technology

[0002] Glass curtain walls, due to their advantages of transparency and aesthetics, are widely used in the facade engineering of high-rise buildings. Typical concealed or semi-concealed frame glass curtain wall structures rely on structural adhesive to connect and transfer force between the glass panels and the aluminum alloy frames, vertical frames, and horizontal beams. This is a key structural element ensuring the overall safety and durability of the curtain wall. During long-term service, glass curtain walls are affected by factors such as temperature and humidity cycles, ultraviolet aging, repeated wind loads, differential deformation of components, rainwater erosion, and construction defects. The structural adhesive may experience aging, degradation of bonding performance, localized delamination, and edge cracks. Once structural adhesive damage develops to a certain extent, it weakens the effective connection between the glass panels and the frame, potentially leading to loosening, displacement, or even detachment under strong winds, impacts, or accidental loads, posing a serious safety risk.

[0003] Currently, the detection of structural adhesive damage in glass curtain walls in engineering projects still mainly relies on manual inspection from heights, tapping with handheld probes, visual inspection, and simple prying tests. These methods have significant shortcomings: First, they depend on scaffolding, suspended platforms, and other high-altitude work platforms, resulting in low efficiency, high labor intensity, and high risks associated with working at heights. Second, manual methods are highly subjective, relying on experience for judgment and making quantitative and standardized assessments difficult. Third, close-up inspection of each glass panel requires substantial preparation time, leading to high overall inspection costs and long cycles. Fourth, when dealing with tall, densely distributed glass panels or complex facade structures, the comprehensiveness and reliability of manual methods are difficult to guarantee. In the periodic maintenance of large-scale curtain wall projects, the aforementioned inspection methods are no longer sufficient to meet the requirements of refined, long-term, and structured management.

[0004] In recent years, some studies have attempted to use drones equipped with visible light cameras, infrared thermal imagers, or 3D laser scanning equipment for remote inspection of glass curtain wall facades. However, drones have significant limitations in detecting structural adhesive damage: First, the high reflectivity of glass curtain wall surfaces easily generates strong reflections, glare, and mirror effects, making it difficult for visible light images to directly reflect the internal bonding state of the structural adhesive. Second, structural adhesive damage is a hidden defect of interfacial adhesion degradation, and infrared thermal imagers show extremely weak thermal anomalies and are easily affected by external factors such as sunlight, wind speed, curtain wall ventilation structure, and interlayer convection, making it difficult to obtain stable and reliable thermal fingerprints. Third, the hovering accuracy and wind resistance of drones are limited, making it impossible to achieve repeatable and stable close-range observations on a single glass panel. Fourth, the limited payload capacity of drones prevents them from carrying contact detection equipment such as tapping mechanisms and accelerometers, making it difficult to capture the glass vibration characteristics corresponding to changes in the state of the structural adhesive. Therefore, drone inspections are more often used for macroscopic appearance inspections and are not suitable for the quantitative identification of structural adhesive damage. Summary of the Invention

[0005] This invention aims to provide a wall-climbing robot and system for integrated detection of structural adhesive damage in glass curtain walls. It addresses the problems of poor safety of manual high-altitude operations and the limited ability of drones to identify hidden damage in structural adhesives in existing glass curtain wall structural adhesive damage detection methods. This invention enables contact excitation detection, vibration signal acquisition, intelligent feature extraction, and machine learning recognition of structural adhesive defects such as debonding, degradation, and loosening to be completed on the same platform. This significantly improves the efficiency, quantification level, and reliability of structural adhesive damage detection, while reducing the risks of high-altitude operations.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0007] A wall-climbing robot for integrated detection of structural adhesive damage in glass curtain walls includes a robot body, an adsorption component, a gimbal component, and a structural adhesive detection component.

[0008] The robot's main body includes a chassis and drive wheels located on both sides of the chassis;

[0009] The adsorption assembly includes a negative pressure sealing chamber and a ducted fan mounted on the chassis. The ducted fan is installed inside the negative pressure sealing chamber and is used for adsorption with the glass curtain wall.

[0010] The gimbal assembly includes a gimbal bracket, a first servo motor, and a second servo motor; the gimbal bracket is equipped with a visible light camera for acquiring image information of the glass curtain wall;

[0011] One end of the gimbal bracket is connected to the chassis, and the other end is connected to the structural adhesive detection component, which is used to adjust the aerial attitude of the structural adhesive detection component; the first servo and the second servo are mounted on the gimbal bracket and are used to control the yaw and pitch movements of the structural adhesive detection component, respectively.

[0012] The structural adhesive testing assembly includes a mounting bracket, a tapping actuator, an accelerometer, and a distance sensor;

[0013] The mounting bracket is connected to the gimbal bracket, and the end of the mounting bracket is equipped with a striking actuator, an acceleration sensor, and a distance sensor.

[0014] Preferably, the negative pressure sealing cavity includes a cavity shell, a top plate connector, and a bottom plate connector;

[0015] The outer shell of the cavity is embedded inside the chassis.

[0016] The upper end of the cavity shell is connected to the top plate of the chassis via a top plate connector, and the lower end of the cavity shell is connected to the bottom plate of the chassis via a bottom plate connector; the ducted fan is installed inside the cavity shell.

[0017] The base around the negative pressure sealing chamber is equipped with sealing gaskets, which are made of silicone, rubber or polyurethane materials.

[0018] Preferably, the gimbal bracket includes a vertical support arm and a horizontal cantilever.

[0019] The bottom of the vertical support arm is connected to the chassis, the first servo is installed inside the vertical support arm, the second servo is installed at the top of the vertical support arm, and the second servo is connected to the horizontal cantilever.

[0020] Preferably, the striking actuator is a linear electromagnet with a metal cylindrical or hemispherical striking head installed at the output end, used to generate a perceptible vibration response in the structural adhesive area of ​​the glass curtain wall.

[0021] The accelerometer is a piezoelectric single-axis or triaxial vibration sensor, which is located next to the impact actuator;

[0022] The distance sensor uses a capacitive proximity switch to detect the contact distance between the bottom surface of the accelerometer and the surface of the glass curtain wall.

[0023] Preferably, the robot body is also equipped with control and communication components, including an onboard computing unit, a motor drive board, a power supply module and a wireless communication module, which are used to control the movement of the wall-climbing robot, process data and communicate with the ground. The control and communication components are electrically connected to the ducted fan, the first servo motor, the second servo motor, the striking actuator, the acceleration sensor and the distance sensor, respectively.

[0024] An integrated detection method for structural adhesive damage in glass curtain walls includes the following steps:

[0025] S1. Attachment and Positioning: Activate the adsorption component to allow the wall-climbing robot to adhere to the glass curtain wall and move to the test area by manual tracking control.

[0026] S2. Impact and Acceleration Signal Acquisition: The attitude of the structural adhesive detection component is adjusted by the gimbal assembly so that the bottom surface of the accelerometer is parallel to the surface of the glass curtain wall and maintains a preset gap. Then, pulse impact is performed at the preset impact point of the glass curtain wall to synchronously acquire the acceleration signal.

[0027] S3. Acceleration signal preprocessing and feature extraction: Preprocess the acceleration signal and extract time-domain, frequency-domain, and time-frequency-domain features;

[0028] S4. Structural Adhesive Damage Identification: Construct a structural adhesive damage identification model, input time domain, frequency domain, and time-frequency domain features, and output the boundary category and confidence level of structural adhesive damage. When the confidence level is lower than the preset threshold, it is marked as suspected damage and prompts manual review.

[0029] S5. Result Recording and Facade Distribution Generation: Store the inspection number, inspection location, and damage identification results of the glass curtain wall to form a structural adhesive damage detection dataset; automatically map the damage identification results to the building facade schematic diagram according to the arrangement order of the glass curtain wall for visualization.

[0030] Preferably, in S2, the preset striking points include the center point of each glass curtain wall, the 1 / 4 point of the upper left diagonal, the 1 / 4 point of the upper right diagonal, the 3 / 4 point of the lower left diagonal, and the 3 / 4 point of the lower right diagonal.

[0031] As a preferred embodiment, in S3, the time-domain, frequency-domain, and time-frequency-domain features include the time-domain peak amplitude, vibration energy index, decay rate, frequency-domain dominant frequency, spectral centroid, spectral energy distribution, and time-frequency energy map features based on continuous wavelet transform, forming a structural adhesive state feature vector.

[0032] Preferably, in S4, the structural adhesive damage identification model adopts at least one of random forest, support vector machine, and lightweight neural network.

[0033] An integrated detection system for structural adhesive damage in glass curtain walls, comprising:

[0034] Real-time monitoring module: used to display the curtain wall image information transmitted back by the visible light camera in real time;

[0035] Control and interaction module: used to remotely issue motion commands to control the pose of the wall-climbing robot, the angle of the gimbal component, and the tapping action of the structural adhesive detection component, so as to realize remote controlled triggering of the acceleration acquisition process;

[0036] Waveform presentation module: used to receive and dynamically plot the acceleration response curves acquired by the accelerometer in real time;

[0037] Recognition Feedback Module: Used to display the structural adhesive damage recognition results and damage probability output by the machine learning model.

[0038] The present invention has the following beneficial effects:

[0039] 1. A negative pressure sealing cavity, composed of a ducted fan and a flexible sealing structure, is used to achieve stable adhesion to vertical glass curtain walls. The flexible sealing structure can adapt to gaps between glass panels and minor geometric differences in the facade, ensuring adhesion while taking into account robot crawling resistance and mobility. It is suitable for various glass curtain wall units and their different installation methods.

[0040] 2. The structural adhesive detection component, combined with a gimbal-controlled accelerometer and a striking electromagnet, applies multi-point pulse excitation to the center and 1 / 4 and 3 / 4 positions of the diagonal of the glass plate, while simultaneously acquiring acceleration response signals. This multi-point detection method can achieve comprehensive sampling of the structural adhesive status of the four boundaries of a single glass piece, improving the sensitivity and recognition rate of damage such as local debonding, degradation, or loosening.

[0041] 3. The airborne computing unit deploys a feature extraction module and a machine learning model (such as random forest, SVM, or lightweight neural network) for structural adhesive damage identification, performing end-to-end analysis on the preprocessed acceleration signal. Compared to manual judgment or traditional thresholding methods, this invention can fully exploit the subtle differences in the structural adhesive state in terms of vibration amplitude, decay rate, and spectral distribution, achieving automatic and quantitative identification of the health status of glass curtain wall adhesive joints, thus improving identification accuracy and robustness.

[0042] 4. This invention integrates an adsorption component, a gimbal component, a structural adhesive detection component, and a control and communication component on a wall-climbing robot platform. It also provides an integrated detection system for structural adhesive damage in glass curtain walls, including a real-time monitoring module, a control and interaction module, a waveform presentation module, and a recognition and feedback module. Through the collaborative interaction between the onboard computing unit and the detection system, it realizes a complete process from manual tracking and positioning, acceleration excitation acquisition, signal preprocessing and feature extraction, to machine learning structural adhesive recognition and result facade mapping. This avoids the high-risk and low-efficiency operation mode of traditional manual high-altitude knocking detection methods. It can adapt to various glass curtain wall unit sizes, installation forms, and facade complexities, and has both high-density detection capabilities and modular scalability, significantly improving the efficiency and operational safety of glass curtain wall structural adhesive detection. Attached Figure Description

[0043] Figure 1 This is a schematic diagram of the overall structure of a wall-climbing robot for integrated detection of structural adhesive damage in glass curtain walls, as described in this invention.

[0044] Figure 2 This is a schematic diagram of the internal structure of the robot body and adsorption component of the present invention.

[0045] Figure 3 This is a schematic diagram of the bottom structure of the adsorption component of the present invention.

[0046] Figure 4 This is a schematic diagram of the gimbal assembly of the present invention.

[0047] Figure 5 This is a schematic diagram of the structure of the structural adhesive detection component of the present invention.

[0048] Figure 6 This is a display image of an integrated detection system for structural adhesive damage in glass curtain walls according to the present invention.

[0049] Figure 7 This is a physical image of the wall-climbing robot of the present invention.

[0050] Figure 8 This is a real-time acceleration waveform diagram of the present invention.

[0051] Figure 9 This is a real-time spectrum diagram of the present invention.

[0052] The components include: 1. Robot body; 2. Adsorption assembly; 3. Gimbal assembly; 4. Structural adhesive detection assembly.

[0053] 11. Chassis; 12. Drive motor; 13. Drive wheels;

[0054] 21. Ducted fan; 22. Negative pressure sealing chamber; 23. Sealing gasket;

[0055] 221. Cavity shell; 222. Top plate connector; 223. Bottom plate connector;

[0056] 31. Vertical support arm; 32. Visible light camera; 33. Second servo motor; 34. First servo motor; 35. Lateral cantilever;

[0057] 41. Mounting bracket; 42. Striking actuator; 43. Accelerometer; 44. Distance sensor. Detailed Implementation

[0058] The present invention will now be described in further detail with reference to the accompanying drawings and specific preferred embodiments.

[0059] In the description of this invention, it should be understood that the terms "left side," "right side," "upper part," "lower part," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. "First," "second," etc., do not indicate the importance of the components, and therefore should not be construed as a limitation of this invention. The specific dimensions used in this embodiment are only for illustrating the technical solution and do not limit the scope of protection of this invention.

[0060] like Figures 1-9 As shown, a wall-climbing robot for integrated detection of structural adhesive damage in glass curtain walls includes a robot body 1, an adsorption component 2, a gimbal component 3, a structural adhesive detection component 4, and a control and communication component. The adsorption component 2, the gimbal component 3, the structural adhesive detection component 4, and the control and communication component are all integrated on the chassis 11 of the robot body 1 and are uniformly coordinated and controlled by the control and communication component.

[0061] The robot body 1 includes a chassis 11 and drive wheels 13 disposed on both sides of the chassis 11. The drive wheels 13 are preferably covered with a high-friction silicone layer to improve grip on the glass surface and ensure stable movement of the robot on vertical or inclined curtain walls. Drive motors 12 control the rotational speed and direction of the drive wheels 13 on both sides, enabling the robot to move straight, turn, and turn in place on vertical walls. The drive motors 12 are either DC geared motors or brushless DC motors.

[0062] The adsorption component 2 is located at the bottom center of the chassis 11 and includes a ducted fan 21 and a negative pressure sealing cavity 22 composed of a flexible sealing structure. The ducted fan 21 is installed inside the negative pressure sealing cavity 22 to form a stable negative pressure adsorption force on the glass surface, so that the wall-climbing robot can be reliably attached to the vertical glass curtain wall. The ducted fan 21 adjusts the adsorption force by adjusting its rotation speed.

[0063] The negative pressure sealing cavity 22 specifically includes a cavity shell 221, a top plate connector 222, and a bottom plate connector 223. The cavity shell 221 is embedded inside the chassis 11. The upper end of the cavity shell 221 is connected to the top plate of the chassis 11 through the top plate connector 222, and the lower end of the cavity shell 221 is connected to the bottom plate of the chassis 11 through the bottom plate connector 223. The ducted fan 21 is installed inside the cavity shell 221. During operation, the ducted fan 21 draws air from inside the sealing cavity, making the air pressure inside the cavity lower than the external atmospheric pressure, thereby generating a strong negative pressure adsorption force, enabling the robot to reliably adhere to the tile wall surface, providing sufficient adhesion stiffness for subsequent contact excitation and precise measurement.

[0064] A sealing gasket 23 is provided on the base plate 11 around the negative pressure sealing cavity 22. The sealing gasket 23 is made of silicone, rubber or polyurethane material and is used to compensate for the gaps and unevenness of the glass curtain wall surface to ensure the sealing effect.

[0065] The gimbal assembly 3 includes a gimbal bracket, a first servo motor 34, and a second servo motor 33. The gimbal bracket is specifically an inverted L-shaped cantilever structure, including a vertical support arm 31 and a horizontal cantilever arm 35. The bottom of the vertical support arm 31 is connected to the chassis 11, the first servo motor 34 is installed inside the vertical support arm 31, and the second servo motor 33 is installed at the top of the vertical support arm 31. The second servo motor 33 is connected to the horizontal cantilever arm 35, which is connected to the structural adhesive detection assembly 4. The gimbal bracket is used to adjust the aerial attitude of the structural adhesive detection assembly 4 to adapt to different installation angles and geometric changes in the facade of different glass curtain wall units. The precise two-degree-of-freedom rotation of the gimbal ensures that the detection module is ultimately aligned with the target detection area, improving the stability and accuracy of structural adhesive damage identification.

[0066] The first servo motor 34 is used to control the yaw motion of the structural adhesive detection component 4, and the second servo motor 33 is used to control the pitch motion of the structural adhesive detection component 4. A visible light camera 32 is installed on the gimbal bracket to acquire image information of the glass curtain wall. This allows operators to observe the wall condition from the ground workstation or remote control terminal, assist the robot in manual tracking and precise docking, and the acquired images can be used to mark the detection point positions and assist in adjusting the gimbal attitude and the contact angle with the accelerometer 43, thereby improving the stability and accuracy of structural adhesive detection.

[0067] The structural adhesive detection component 4 includes a mounting bracket 41, a striking actuator 42, an acceleration sensor 43, and a distance sensor 44. The mounting bracket 41 is connected to the gimbal bracket, and the striking actuator 42, acceleration sensor 43, and distance sensor 44 are located at the end of the mounting bracket 41. Both the gimbal bracket and the mounting bracket 41 are made of carbon fiber, which is lightweight and high-strength, reducing the overall load on the robot while maintaining stability.

[0068] The striking actuator 42 is a linear electromagnet with a metal cylindrical or hemispherical striking head installed at the output end. It is used to apply short pulse excitation to the glass panel, causing a perceptible vibration response in the structural adhesive area. The excitation position can be adjusted by gimbal posture or robot movement. The accelerometer 43 is a piezoelectric single-axis or three-axis vibration sensor, arranged close to the striking position, used to detect the vibration response signal of the structural adhesive under excitation. Debonding, loosening or degradation of the structural adhesive will cause significant changes in frequency response, attenuation characteristics, etc. The acceleration response can be used as the main criterion. The distance sensor 44 is preferably a capacitive proximity switch distance sensor 44, used to detect the contact distance between the bottom surface of the accelerometer 43 and the surface of the glass curtain wall, and can achieve stable support or light attachment through gimbal control, so as to improve the repeatability of excitation-response detection.

[0069] The control and communication components include an onboard computing unit, a motor drive board, a power supply module, and a wireless communication module, used to control the wall-climbing robot's movement, process data, and communicate with the ground. The onboard computing unit is the core of the entire system. It connects all actuators (drive motor 12, ducted fan 21, first servo motor 34, second servo motor 33, and striking actuator 42) and sensors (accelerometer 43, visible light camera 32, and distance sensor 44), and works with the motor drive board and power supply module to achieve motion control and actuator drive. The onboard computing unit establishes a communication link with the ground workstation through the wireless communication module to transmit detection results, working status, and control commands between the robot and the ground workstation.

[0070] This embodiment also includes an integrated detection system for structural adhesive damage in glass curtain walls. The system operates on a ground workstation and has a monitoring and interactive page for real-time display of visible light images and acceleration response curves. It also supports remote issuance of motion commands and acceleration acquisition commands to control the movement of the robot body, the gimbal pose, and the controlled acceleration acquisition process.

[0071] The system includes a real-time monitoring module, a control and interaction module, a waveform display module, and a recognition and feedback module. For example... Figure 6 As shown, the left half displays the monitoring module and the control interaction module, while the right half displays the waveform presentation module and the recognition feedback module.

[0072] Real-time monitoring module: used to display the curtain wall image information transmitted back by the visible light camera in real time;

[0073] Control and interaction module: used to remotely issue motion commands to control the pose of the wall-climbing robot, the angle of the gimbal component, and the tapping action of the structural adhesive detection component, so as to realize remote controlled triggering of the acceleration acquisition process;

[0074] Waveform presentation module: Used to receive and dynamically plot the acceleration response curves acquired by the accelerometer in real time; such as... Figure 8 , Figure 9 As shown;

[0075] Recognition Feedback Module: Used to display the structural adhesive damage recognition results and damage probability output by the machine learning model.

[0076] Operators can view the curtain wall images transmitted back by the visible light camera 32 in real time through the real-time monitoring module. They can remotely send commands to the onboard computing unit via the motion control panel to control the robot's movement and steering, as well as the pose fine-tuning of the structural adhesive detection component 4. This also controls the operation of the striking actuator 42 and the data collection by the accelerometer 43. During the detection process, the control interaction module can receive and process the raw data transmitted back by the accelerometer 43 in real time and dynamically display the acceleration response curve on the waveform display module.

[0077] The onboard computing unit deploys machine learning models. These models are used to identify structural adhesive defects such as debonding, degradation, or loosening based on the characteristics of acceleration response signals (such as frequency domain peak value, dominant frequency, damping ratio, etc.). Simultaneously, they can integrate signal filtering, noise suppression, and feature extraction algorithms to improve identification accuracy. The identification results and corresponding confidence levels are fed back to the identification feedback module in real time.

[0078] The onboard computing unit is connected to the gimbal assembly 3, enabling control of the servo motors, driving the camera, and the structural adhesive detection assembly 4 to achieve precise rotation in two degrees of freedom within the pitch direction (−45°~45°) and yaw direction (−90°~90°). This ensures that the attitude of the accelerometer 43 is aligned with the normal to the glass surface, guaranteeing stable and consistent excitation direction and improving vibration response measurement accuracy. It also ensures that the accelerometer 43 is in close contact with the glass curtain wall surface and transmits camera images to the ground workstation in real time, simultaneously recording the robot's visual reference information at each detection point for subsequent data analysis and result mapping.

[0079] The present invention also provides a method for detecting damage to structural adhesive in glass curtain walls, comprising the following steps:

[0080] Step 1: Wall adhesion and manual tracking positioning:

[0081] The ducted fan 21 in the adsorption assembly 2 is activated to create a stable negative pressure in the negative pressure sealing chamber 22, allowing the wall-climbing robot to reliably attach to the glass curtain wall facade. Operators can view the high-definition images transmitted from the visible light camera 32 in real time through the monitoring interface of the glass curtain wall structural adhesive damage detection system, and use the interface control commands to move the robot along the glass curtain wall to the detection area of ​​each unit glass panel. After stopping near each glass panel, the attitude of the two-degree-of-freedom gimbal is adjusted, and the attachment status is confirmed by the distance sensor 44 value fed back by the interface, ensuring that the accelerometer 43 is stably attached to the glass surface. This provides reliable contact conditions for subsequent structural adhesive tapping and vibration response acquisition, ensuring that subsequent tapping excitation can be effectively transmitted to the glass panel and obtain a high-quality vibration response signal. The visible light camera 32 works synchronously, transmitting real-time images to the ground workstation to assist operators in observing the wall condition and accurately stopping the robot.

[0082] Step 2: Acquisition of five-point tapping and acceleration signals:

[0083] Once the robot has come to a stable stop, a cyclic detection process begins based on the coordinates of five preset detection points (center point, upper left 1 / 4 point, upper right 1 / 4 point, lower left 3 / 4 point, and lower right 3 / 4 point). For each detection point, the operator can adjust the robot's body movement and the angle of the gimbal assembly to achieve precise alignment and pose adjustment. The operator observes the pose adjustment process in real time through the system's interactive page. After confirming the pose is ready, the operator issues a tapping command through the page, controlling the tapping actuator 42 to perform multiple pulse taps and using the accelerometer 43 to collect acceleration. Each tap maintains a consistent excitation intensity and duration to enhance the stability and comparability of the detection results. The time interval between adjacent taps is greater than the sound wave decay time to avoid interference between adjacent tapping signals. The accelerometer 43 module synchronously collects the vibration acceleration signal of the glass plate caused by the taps and dynamically plots the real-time acceleration response curve on the waveform display module of the monitoring interactive page, allowing the operator to instantly judge the quality and completeness of the signal acquisition.

[0084] Step 3: Acceleration signal preprocessing and feature extraction:

[0085] After data acquisition, the onboard computing unit systematically preprocesses the raw acceleration signal, including environmental noise suppression, bandpass filtering, and amplitude normalization, to improve the signal-to-noise ratio and highlight response characteristics related to structural adhesive damage. Based on this, the system extracts multi-dimensional feature parameters from the response sequence at each impact point, including time-domain, frequency-domain, and time-frequency-domain features. These parameters include: time-domain peak amplitude, vibration energy index, decay rate, frequency-domain dominant frequency, spectral centroid, spectral energy distribution, and time-frequency energy map features based on continuous wavelet transform. These features constitute a structural adhesive state feature vector reflecting the glass plate constraint stiffness and adhesive bonding state.

[0086] Step 4: Structural Adhesive Damage Identification

[0087] For each glass panel, the onboard computing unit calls a pre-trained machine learning structural adhesive damage identification model (such as at least one of random forest, support vector machine, or lightweight neural network) to comprehensively determine the bonding state of the four glass boundaries. The identification model outputs the damage categories and corresponding confidence levels for the four adhesive seams (top, bottom, left, and right), enabling boundary-level localization of potential risks such as localized degradation, debonding, or deformation of the adhesive seams. The identification results are immediately transmitted back to the identification feedback module on the monitoring interaction page. When the model confidence level is lower than a preset threshold or there are conflicting results, the system marks the glass panel as "suspected damage state," prompting that it needs to be manually reviewed or confirmed by increasing the number of taps.

[0088] Step 5: Result Recording and Facade Distribution Generation:

[0089] The onboard computing unit stores the inspection number, location information, and boundary damage results determined by the model for each glass panel in a structured manner, forming a glass curtain wall structural adhesive damage detection dataset. Subsequently, the robot transmits the detection data back to the ground workstation via a wireless communication module. Based on a glass curtain wall layout database, the glass curtain wall structural adhesive damage detection system automatically synchronizes the identification results to the building facade schematic diagram in the facade mapping area of ​​the interactive page, marking the identified damage areas with color, symbols, or highlighted borders. Operators can directly click on abnormal points in the facade diagram on the page to retrieve historical monitoring images and original acceleration curves for in-depth source analysis, providing maintenance personnel with an intuitive structural adhesive degradation distribution map to support maintenance decisions.

[0090] This invention provides a wall-climbing robot and system for integrated detection of structural adhesive damage in glass curtain walls. The robot can closely adhere to the surface of the glass curtain wall and move stably along the facade using negative pressure adsorption. This allows the robot to deploy a tapping mechanism and acceleration sensor 43 for contact detection at close range, making it more suitable for the large-area, refined, and quantitative structural adhesive damage detection needs of high-rise glass curtain walls. The accompanying glass curtain wall structural adhesive damage detection system achieves deep visualization and active control of the detection process through an interactive interface, ensuring high-standard data acquisition even in complex environments. This invention organically integrates negative pressure adsorption of the ducted fan 21, contact and tapping, acceleration response acquisition, feature extraction and machine learning recognition, visual positioning, and result mapping, achieving efficient, safe, and reliable detection of the structural adhesive condition of high-rise curtain walls, providing key technical support for the safe maintenance of building facades.

[0091] The preferred embodiments of the present invention have been described in detail above. However, the present invention is not limited to the specific details in the above embodiments. Within the scope of the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all fall within the protection scope of the present invention.

Claims

1. A wall-climbing robot for integrated detection of structural adhesive damage in glass curtain walls, characterized in that, This includes the robot body, adsorption components, gimbal components, and structural adhesive detection components; The robot's main body includes a chassis and drive wheels located on both sides of the chassis; The adsorption assembly includes a negative pressure sealing chamber and a ducted fan mounted on the chassis. The ducted fan is installed inside the negative pressure sealing chamber and is used for adsorption with the glass curtain wall. The gimbal assembly includes a gimbal bracket, a first servo motor, and a second servo motor; the gimbal bracket is equipped with a visible light camera for acquiring image information of the glass curtain wall; One end of the gimbal bracket is connected to the chassis, and the other end is connected to the structural adhesive detection component, which is used to adjust the aerial attitude of the structural adhesive detection component; the first servo and the second servo are mounted on the gimbal bracket and are used to control the yaw and pitch movements of the structural adhesive detection component, respectively. The structural adhesive testing assembly includes a mounting bracket, a tapping actuator, an accelerometer, and a distance sensor; The mounting bracket is connected to the gimbal bracket, and the end of the mounting bracket is equipped with a striking actuator, an acceleration sensor, and a distance sensor.

2. The wall-climbing robot for integrated detection of structural adhesive damage in glass curtain walls according to claim 1, characterized in that, The negative pressure sealing cavity includes a cavity shell, a top plate connector, and a bottom plate connector; The outer shell of the cavity is embedded inside the chassis. The upper end of the cavity shell is connected to the top plate of the chassis via a top plate connector, and the lower end of the cavity shell is connected to the bottom plate of the chassis via a bottom plate connector; the duct fan is installed inside the cavity shell. The base around the negative pressure sealing chamber is equipped with sealing gaskets, which are made of silicone, rubber or polyurethane materials.

3. The wall-climbing robot for integrated detection of structural adhesive damage in glass curtain walls according to claim 1, characterized in that, The gimbal support includes a vertical support arm and a horizontal cantilever. The bottom of the vertical support arm is connected to the chassis, the first servo is installed inside the vertical support arm, the second servo is installed at the top of the vertical support arm, and the second servo is connected to the horizontal cantilever.

4. The wall-climbing robot for integrated detection of structural adhesive damage in glass curtain walls according to claim 1, characterized in that, The striking actuator is a linear electromagnet with a metal cylindrical or hemispherical striking head installed at the output end, used to generate a perceptible vibration response in the structural adhesive area of ​​the glass curtain wall. The accelerometer is a piezoelectric single-axis or triaxial vibration sensor, which is located next to the impact actuator; The distance sensor uses a capacitive proximity switch to detect the contact distance between the bottom surface of the accelerometer and the surface of the glass curtain wall.

5. The wall-climbing robot for integrated detection of structural adhesive damage in glass curtain walls according to claim 1, characterized in that, The robot body is also equipped with control and communication components, which are electrically connected to the ducted fan, the first servo motor, the second servo motor, the striking actuator, the acceleration sensor, and the distance sensor, respectively.

6. A method for integrated detection of structural adhesive damage in glass curtain walls, using a wall-climbing robot for integrated detection of structural adhesive damage in glass curtain walls as described in any one of claims 1-5, characterized in that, Includes the following steps, S1. Attachment and Positioning: Activate the adsorption component to allow the wall-climbing robot to adhere to the glass curtain wall and move to the test area by manual tracking control. S2. Impact and Acceleration Signal Acquisition: The attitude of the structural adhesive detection component is adjusted by the gimbal assembly so that the bottom surface of the accelerometer is parallel to the surface of the glass curtain wall and maintains a preset gap. Then, pulse impact is performed at the preset impact point of the glass curtain wall to synchronously acquire the acceleration signal. S3. Acceleration signal preprocessing and feature extraction: Preprocess the acceleration signal and extract time-domain, frequency-domain, and time-frequency-domain features; S4. Structural Adhesive Damage Identification: Construct a structural adhesive damage identification model, input time domain, frequency domain, and time-frequency domain features, and output the boundary category and confidence level of structural adhesive damage. When the confidence level is lower than the preset threshold, it is marked as suspected damage and prompts manual review. S5. Result Recording and Facade Distribution Generation: Store the inspection number, inspection location, and damage identification results of the glass curtain wall to form a structural adhesive damage detection dataset; automatically map the damage identification results to the building facade schematic diagram according to the arrangement order of the glass curtain wall for visualization.

7. The integrated detection method for structural adhesive damage in glass curtain walls according to claim 6, characterized in that, In S2, the preset tapping points include the center point of each glass curtain wall, the 1 / 4 point of the upper left diagonal, the 1 / 4 point of the upper right diagonal, the 3 / 4 point of the lower left diagonal, and the 3 / 4 point of the lower right diagonal.

8. The integrated detection method for structural adhesive damage in glass curtain walls according to claim 6, characterized in that, In S3, the time domain, frequency domain, and time-frequency domain features include time domain peak amplitude, vibration energy index, decay rate, frequency domain dominant frequency, spectral centroid, spectral energy distribution, and time-frequency energy map features based on continuous wavelet transform, forming a structural adhesive state feature vector.

9. The integrated detection method for structural adhesive damage in glass curtain walls according to claim 6, characterized in that, In S4, the structural adhesive damage identification model employs at least one of random forest, support vector machine, and lightweight neural network.

10. An integrated detection system for structural adhesive damage in glass curtain walls, characterized in that, The method for integrated detection of structural adhesive damage in glass curtain walls according to any one of claims 6-9 is used. include, Real-time monitoring module: used to display the curtain wall image information transmitted back by the visible light camera in real time; Control and interaction module: used to remotely issue motion commands to control the pose of the wall-climbing robot, the angle of the gimbal component, and the tapping action of the structural adhesive detection component, so as to realize remote controlled triggering of the acceleration acquisition process; Waveform presentation module: used to receive and dynamically plot the acceleration response curves acquired by the accelerometer in real time; Recognition Feedback Module: Used to display the structural adhesive damage recognition results and damage probability output by the machine learning model.

Images