Building component dismantling equipment and method

Abstract

The invention relates to building component dismantling equipment and a building component dismantling method, belongs to the technical field of engineering machinery, and mainly solves the technical problems of high collision risk, path adjustment lagging and insufficient safety protection of traditional dismantling equipment in a complex operation environment. The equipment comprises a base platform and adjustable hydraulic supporting legs of the base platform, a multi-axis mechanical arm and a tail end hydraulic shear are arranged above the platform, and an image acquisition module is arranged at the front end of the platform and connected with a visual identification module with a built-in feature database. The control module receives the three-dimensional coordinate data through the path planning unit to generate a motion track, and the motion control unit drives the mechanical arm to execute operation. And when the distance enters a braking critical zone of 10 mm, the track is frozen immediately, a reverse compensation path with the 1.5-time real-time distance value is generated, the path generation period does not exceed 5 milliseconds, and the mechanical arm is controlled to move along the compensation path. The device can effectively avoid operation collision, realizes dynamic path correction with millimeter-level precision, and is suitable for efficient and safe dismantling operation of building steel structures.

Description

Technical Field

[0001] The present invention relates to the technical field of construction machinery, and in particular to a building component demolition device and method. Background Art

[0002] In the field of building demolition engineering, the demolition operation of steel structure components has long faced the contradiction between operation accuracy and safety control. Traditional demolition equipment mostly relies on the operator's experience to judge the movement trajectory of the robotic arm, and is prone to collision accidents in complex spatial environments. In the prior art, the path planning of the robotic arm usually adopts an offline programming method, which cannot respond to the dynamically changing operation environment in real time, resulting in a significant lag in controlling the safety distance between the end effector of the robotic arm and the surrounding structure during actual operation. This lag is due to the too long data processing cycle of the environmental perception system. The image information collected by the vision sensor and the laser ranging data often require more than 100 milliseconds of processing time to generate updated path instructions, which is difficult to match the actual movement speed of the robotic arm (usually 0.5 - 1 m / s). When the robotic arm approaches an obstacle, the remaining braking distance is not enough to avoid a collision.

[0003] Most of the existing collision detection algorithms adopt a single threshold trigger braking mechanism. For example, a fixed distance value (such as 50 mm) is set as the safety boundary, and once the robotic arm enters this area, an emergency stop is triggered. This method has two defects: First, the inertial impact generated by the emergency braking is likely to cause secondary damage to the contact surface between the hydraulic shear and the component. Especially when the shear stress is relatively high (exceeding 20 MPa), it may cause irregular expansion of the fracture surface of the component. Second, the fixed threshold cannot adapt to the dynamic requirements of different operation stages. For example, a higher movement efficiency is required in the initial approach stage, while a higher safety margin is required in the precision cutting stage. In addition, the safety protection mechanism of traditional equipment lacks the ability of hierarchical response. In the early warning stage, only an audible and visual alarm is used to prompt the operator to manually intervene, and the manual reaction time (usually more than 500 milliseconds) is much longer than the time window required for the robotic arm to brake, resulting in limited actual protection effect.

[0004] In terms of path correction, the prior art mostly adopts a simple compensation strategy of straight-line retraction, without considering the coupling effects of component deformation, tool posture and environmental obstacles. For example, when the hydraulic shear contacts the component and generates a lateral load, if the retraction path of the robotic arm has an angular deviation from the load direction, it may exacerbate the interference between the tool and the component. At the same time, the calculation efficiency of the path regeneration algorithm is insufficient. The existing path planning method based on geometric analysis usually takes 10 - 15 milliseconds to complete a trajectory correction, which cannot meet the real-time response requirement within 5 milliseconds in a high-dynamic operation scenario. This is mainly because the algorithm does not effectively integrate multi-source sensor data, and subsystems such as visual recognition, mechanical detection and position feedback often run independently, and the data synchronization error accumulates, resulting in a decrease in the accuracy of trajectory generation.

[0005] Another prominent problem is the insufficient adaptability of existing systems, which are unable to dynamically adjust control parameters according to the job load. For example, when cutting high-strength steel members, the load change rate of the hydraulic shear can reach 10 MPa / second, but the traditional control strategy still uses a PID controller with a fixed gain, resulting in overshoot or oscillation when the load suddenly changes. In addition, the setting of the safety boundary depends on a static model preset manually and is not associated with the deformation data of the members collected in real time. When the member undergoes an accidental displacement during the operation (such as an instantaneous offset of 2 - 5 mm due to local fracture), the deviation between the preset safety boundary and the actual dangerous area may exceed the allowable tolerance.

[0006] The root cause of the above problems lies in the bottleneck of the information processing architecture of traditional systems: First, the sensor data fusion level is single, and a spatio-temporal correlation model of visual, mechanical, and kinematic data has not been established; second, the control decision logic lacks a hierarchical design, and the coordination mechanism for emergency braking and path correction is imperfect; third, the update frequency of the dynamic environment modeling does not match the control period, resulting in system response lag. These technical difficulties make it difficult for existing equipment to balance operation efficiency and operation safety in complex demolition scenarios, restricting its application in precision operations such as the demolition of high-rise building steel structures. Summary of the Invention

[0007] According to an object of the present invention, it is to solve the problems of high collision risk, lag in path adjustment, and insufficient safety protection of traditional demolition equipment in complex working environments. Solve the problems of secondary damage caused by inertial impact during emergency braking and deviation in the path compensation direction. Solve the problems of control of high-temperature sparks, dust pollution, and overload risk during demolition operations. Solve the problems of insufficient support stability of the equipment and weak collision protection ability of the support legs. Solve the problems of low accuracy of environmental three-dimensional modeling and lag in safety boundary update. Solve the problem of poor adaptability of the path compensation strategy under dynamic loads. Solve the problems of insufficient sectional control accuracy of the compensation path and low cooling efficiency. Solve the problem of single deceleration strategy under different load states. Solve the problem of poor real-time performance of multi-source data fusion path planning. Solve the problem of insufficient robustness of path correction under the coupling action of multiple parameters.

[0008] The present invention provides a building component demolition device, including the following structures: The base platform is provided with adjustable hydraulic support legs; A multi-axis robotic arm is fixedly installed above the base platform, and a hydraulic shear is connected to the end of the multi-axis robotic arm; An image acquisition module is installed at the front end of the base platform, the output end of the image acquisition module is connected to a visual recognition module, and the visual recognition module has a building component feature database built therein; The control module includes a path planning unit and a motion control unit. The path planning unit receives the three-dimensional coordinate data output by the visual recognition module. The motion control unit generates a multi-axis robotic arm motion trajectory based on the three-dimensional coordinate data and the building component feature database. The three-dimensional coordinate data includes preset safety boundary data; The control module controls the multi-axis robotic arm according to the multi-axis robotic arm motion trajectory, drives the hydraulic shear to the location of the preset steel component to be demolished, and the control module controls the hydraulic shear to perform the demolition operation; Among them, the control module is built with a collision detection algorithm. When the real-time distance value between the hydraulic shear at the end of the multi-axis robotic arm and the preset safety boundary enters the 30mm warning range, the control module linearly reduces the motion speed of the robotic arm to 30%-50% of the current speed value, and the deceleration slope is set to reduce the original speed by 20% per second; When the real-time distance value enters the 10mm braking critical area, the motion control unit immediately freezes the current multi-axis robotic arm motion trajectory, and generates a reverse compensation path based on the real-time spatial coordinates of the hydraulic shear. The length of the reverse compensation path is 1.5 times the real-time distance value, and the path generation period does not exceed 5ms; The control module controls the robotic arm to move along the reverse compensation path.

[0009] Preferably, for the building component demolition equipment of the present invention, when the real-time distance value enters the 10mm braking critical area, the control module triggers an emergency brake; And, the method for generating the reverse compensation path is: the deviation angle between the compensation path and the current motion trajectory is constrained within the range of 10°-15°, and the trajectory smoothing process adopts the cubic B-spline interpolation method; During the execution of the reverse compensation path, the acceleration upper limit of each joint of the multi-axis robotic arm is set to 0.3m / s², the sampling frequency of the speed control loop is set to 1kHz, and the position feedback data is subjected to noise suppression using Kalman filtering.

[0010] Preferably, for the building component demolition equipment of the present invention, a water storage tank and a high-pressure water pump are arranged on the side of the base platform. The outlet of the high-pressure water pump is connected to an annular spray pipe. The annular spray pipe is arranged around the hydraulic shear blade opening. The annular spray pipe is provided with a solenoid valve, and the solenoid valve is connected to the control module; A water level monitoring device is arranged in the water storage tank, and the water level monitoring device is connected to the control module. When the water storage volume is lower than 20% of the set volume, the control module issues a shutdown instruction; The nozzle spacing of the annular spray pipe is 15mm, the nozzle aperture is 0.5mm, the spray water pressure is maintained at 2.5MPa, and the diameter of the atomized water particles formed by the spray is 5-10mm; A strain gauge array is embedded on the inner side of the blade of the hydraulic shear, and the strain gauge array is connected to the control module through a signal wire; The control module configures a strain threshold determination program. When the strain gauge array detects that the shear stress reaches 85% of the preset yield strength value of the steel member, the solenoid valve is triggered to open.

[0011] Preferably, for the building component demolition equipment of the present invention, a pressure sensor is installed at the end of the hydraulic support leg, and the pressure sensor is electrically connected to the control module of the base platform; The hydraulic support leg includes a four-stage telescopic oil cylinder, and an anti-collision buffer layer is provided on the outer wall of each stage of the oil cylinder. The anti-collision buffer layer adopts a polyurethane-steel mesh composite structure; Among them, the polyurethane-steel mesh composite structure is: the inner layer is a 3-mm-thick polyurethane elastomer, the middle layer is a hexagonal mesh layer woven with 304 stainless steel wires with a diameter of 1 mm, and the outer layer is a 5-mm-thick polyurethane-glass fiber mixed coating.

[0012] Preferably, for the building component demolition equipment of the present invention, the image acquisition module includes a binocular camera and a laser rangefinder; The visual recognition module performs frame synchronization fusion on the RGB image collected by the binocular camera and the point cloud data of the laser rangefinder to obtain three-dimensional coordinate data. The fused three-dimensional coordinate data updates the preset safety boundary data at a period of 20 ms; The spatial coordinates of the preset safety boundary data are obtained by parsing the BIM model in the building component feature database. During the spatial coordinate matching process, the iterative closest point algorithm is used for error compensation. The number of iterations is set to 50 times, and the single-iteration calculation time does not exceed 0.1 ms; The sampling frequency of the image acquisition module is set to 50 Hz, the measurement accuracy of the laser rangefinder is ±1 mm, and the baseline distance of the binocular camera is 300 mm.

[0013] Preferably, for the building component demolition equipment of the present invention, the method for the control module to control the manipulator to move along the reverse compensation path further includes: The motion control unit continuously obtains the real-time shear stress data detected by the strain gauge array at the hydraulic shear opening. When the shear stress change rate exceeds 5 MPa / s, the reverse compensation path is triggered, and the reverse compensation amount is dynamically adjusted to 1.5 times the real-time distance value; The visual recognition module synchronously analyzes the morphological characteristics of the component fracture surface captured by the binocular camera. When it detects that there is a transverse crack extension on the fracture surface, it automatically adjusts the reverse compensation direction to form an angle of 15°-30° with the crack extension direction; The trajectory control of the reverse compensation movement adopts an adaptive PID algorithm, and the control parameters are dynamically adjusted according to the load distribution data of the pressure sensor of the hydraulic support leg. When the pressure of a single support leg exceeds 20% of the average value, the proportional coefficient Kp is increased from 0.8 to 1.2, and the integral time Ti is shortened from 0.5 s to 0.3 s; The motion control unit synchronously processes the joint encoder data of the multi-axis robotic arm and the distance feedback data of the laser rangefinder at a cycle of 50 ms. When the deviation between the two data exceeds 0.5 mm, the trajectory is corrected using the data from the laser rangefinder.

[0014] Preferably, for the building component demolition equipment of the present invention, during the reverse compensation execution process, the opening frequency of the solenoid valve of the annular spray pipe is linked with the compensation speed. For every 0.1 m / s increase in the compensation speed, the spray frequency increases by 10 Hz, and the maximum spray flow rate is limited to 20 L / min. The path planning unit decomposes the reverse compensation path into three segments of motion trajectories according to the real-time updated three-dimensional coordinate data. The first 5 mm stroke uses an S-shaped acceleration curve with a maximum acceleration limited to 0.2 m / s², the second 3 mm stroke maintains a constant speed, and the third 2 mm stroke uses an exponential decay deceleration curve. When it is detected that the displacement of the component does not reach 80% of the expected value during the reverse compensation motion duration, the control module automatically switches to the high-frequency vibration mode. The vibration frequency is set to 50 Hz, the amplitude is 0.1 mm, and the upper limit of the number of vibrations is set to 30 times.

[0015] Preferably, for the building component demolition equipment of the present invention, when the real-time distance value between the hydraulic shear at the end of the multi-axis robotic arm and the preset safety boundary enters the first-level warning area of 20 - 30 mm, the motion control unit dynamically calculates the deceleration parameters according to the current load state of the hydraulic shear. The load state is divided into three levels by the real-time shear stress value detected by the strain gauge array: When the shear stress is less than 15 MPa, the deceleration target value is 50% of the current speed, and the deceleration slope is 15% per second decrease. When the shear stress is in the range of 15 MPa to 30 MPa, the deceleration target value is 40% of the current speed, and the deceleration slope is 20% per second decrease. When the shear stress reaches or exceeds 30 MPa, the deceleration target value is 30% of the current speed, and the deceleration slope is 25% per second decrease. The visual recognition module synchronously evaluates the environmental complexity. It identifies the number of obstacles within 5 meters through binocular cameras, and combines the distance value of the nearest obstacle detected by the laser rangefinder to calculate the environmental complexity value as the ratio of the number of identified obstacles to one-thousandth of the distance of the nearest obstacle. When this ratio is greater than 2, the deceleration slope is additionally increased by an increment of 5% per second. The path planning unit generates a speed constraint trajectory in real time, and sets the upper limit of the angular acceleration of each joint of the robotic arm as follows: for every additional 10 mm of the real-time distance value exceeding 20 mm, the upper limit of the angular acceleration is increased by 0.8 rad / s². When the real-time distance value continuously decreases during the deceleration process and the decreasing rate exceeds 0.5 mm / ms, it automatically switches to the non-linear deceleration mode, and the deceleration curve is adjusted to the initial speed decaying according to the natural exponential function plus a compensation amount of 30% of the initial speed. The decay coefficient of the natural exponential function is 0.2 per second, and the compensation amount increases with the duration of deceleration. The opening strategy of the solenoid valve of the annular spray pipe is linked with the deceleration process. When the deceleration target value is less than or equal to 40% of the current speed, the spray frequency is increased to 15 Hz, and the on-time duty ratio within a single spray cycle is dynamically adjusted according to the real-time shear stress value, and the adjustment range is 30% to 50%.

[0016] Preferably, for the construction member demolition equipment of the present invention, the method for generating the reverse compensation path specifically includes: When the real-time distance value enters the 8 - 10 mm braking preparation area, the motion control unit starts a three-level response mechanism: The first-level response triggers path pre-calculation. Based on the component surface texture features collected by the binocular camera to match the BIM model data, three alternative reverse paths are generated, and the path lengths are 1.2 times, 1.5 times, and 1.8 times of the real-time distance value respectively; The second-level response activates dynamic parameter adjustment. According to the load-bearing distribution data of the hydraulic support leg pressure sensor, when the pressure difference between the maximum load-bearing leg and the minimum load-bearing leg exceeds 15%, the reverse compensation amount automatically increases by a compensation coefficient of 0.3 times the real-time distance value; The third-level response executes the final path selection. The optimal path is selected based on the component deformation rate measured in real time by the laser rangefinder. When the deformation rate > 0.2 mm / ms, the 1.8 times path is adopted; when the deformation rate ≤ 0.2 mm / ms, the 1.5 times path is adopted; When the path planning unit generates the reverse compensation path, the path is decomposed into three motion trajectories: The first 3 mm stroke adopts an S-shaped acceleration curve, and the maximum acceleration is limited to 0.15 m / s²; The middle 7 mm stroke maintains a constant speed, and the speed value is set to 25% - 40% of the current motion speed; The last 5 mm stroke adopts an exponential decay deceleration curve, and the deceleration gradient is set to decrease the speed value by 35% per second; The visual recognition module synchronously analyzes the characteristics of the component fracture surface. When longitudinal cracks are detected on the fracture surface, the reverse compensation direction automatically deflects by an angle of 15° - 25° in the crack extension direction, and the deflection angle accuracy is controlled within the range of ±1°; The motion control unit synchronously collects the torque sensor data of each joint of the multi-axis robotic arm at a sampling frequency of 1 kHz. When it is detected that the torque of any joint exceeds 80% of the rated value, the reverse compensation amount is immediately reduced to 1.2 times the real-time distance value; During the execution of the reverse compensation path, the annular spray system starts the pulse cooling mode. The spray cycle is synchronized with the movement cycle of the robotic arm. Each compensation movement cycle includes 3 intermittent sprays. The duration of a single spray is set to 50 ms, and the interval time is set to 20 ms; After the path planning unit completes the reverse compensation, it automatically performs trajectory verification detection. By comparing the deviation value between the actual displacement data of the laser rangefinder and the theoretical path, when the deviation exceeds 0.3 mm, the secondary compensation program is triggered, and the secondary compensation amount is calculated as 1.1 times the deviation value.

[0017] Preferably, a method for controlling the demolition of building components according to the present invention includes the following steps: Establish the equipment support reference through the adjustable hydraulic support legs of the base platform; Perform component demolition operations through the hydraulic shear at the end of the multi-axis robotic arm. The movement trajectory of the multi-axis robotic arm is generated by the path planning unit of the control module; Collect the operation environment data in real time through the image acquisition module at the front end of the base platform and transmit the collected data to the visual recognition module; Call the building component feature database through the visual recognition module for three-dimensional coordinate matching to generate a three-dimensional space model containing preset safety boundary data; Receive the three-dimensional coordinate data output by the visual recognition module through the path planning unit of the control module and generate the movement trajectory of the multi-axis robotic arm in combination with the building component feature database; When it is detected that the real-time distance between the hydraulic shear and the preset safety boundary enters the 30-mm warning range, the control module linearly reduces the movement speed of the robotic arm to 30%-50% of the current speed value, and the deceleration slope is set to reduce the original speed by 20% per second; When the real-time distance enters the 10-mm braking critical area, the motion control unit immediately freezes the current movement trajectory of the robotic arm and generates a reverse compensation path based on the real-time spatial coordinates of the hydraulic shear. The length of the reverse compensation path is 1.5 times the real-time distance value and the generation period does not exceed 5 ms; Drive the multi-axis robotic arm to move along the reverse compensation path through the motion control unit, and at the same time control the hydraulic shear to perform the shearing demolition operation of the preset steel component; The following steps are synchronously executed during the generation of the reverse compensation path: Real-time monitor the component deformation rate through the laser rangefinder. When the deformation rate exceeds 0.2 mm / ms, automatically extend the compensation path to 1.8 times the real-time distance value; Detect the shear stress change rate of the hydraulic shear through the strain gauge array. When the change rate exceeds 5 MPa / s, dynamically adjust the direction of the compensation path to form an angle of 15°-25° with the stress gradient direction; The fracture surface feature data captured by the binocular camera are used to correct and compensate the path curvature, ensuring that the deviation between the path and the normal direction of the component fracture surface does not exceed ±2°.

[0018] The present invention at least includes the following beneficial effects: Through the hierarchical early warning braking mechanism and the dynamic path compensation algorithm, real-time path correction with millimeter-level accuracy is achieved, reducing the collision probability by more than 90% and improving the operation efficiency by 40%.

[0019] By adopting the cubic B-spline interpolation and Kalman filtering technologies, the path smoothness is increased by 75%, and the joint impact load is reduced by 60%, ensuring the safety of precision operations.

[0020] The intelligent spraying system starts 0.3 seconds before overload, can reduce the operation temperature by 150°C, control the dust concentration below 5 mg / m³, and extend the equipment life by 30%.

[0021] The four-stage telescopic oil cylinder is combined with the composite buffer layer, increasing the support stability by 55% and enhancing the anti-lateral impact ability by 80%, adapting to complex terrains with an inclination angle ≤ 15°.

[0022] The multi-modal data fusion technology controls the three-dimensional modeling error within ±0.8 mm, the safety boundary update frequency reaches 50 Hz, and the environmental perception response speed is increased by 5 times.

[0023] The adaptive PID control algorithm shortens the response time to 8 ms during load mutation, the path tracking error is less than 0.5 mm, and the adaptability to complex working conditions is improved by 65%.

[0024] The segmented trajectory control reduces the compensation movement time by 40%, and the high-frequency vibration mode can improve the cutting efficiency by 25%, meeting the requirements of high-intensity operations.

[0025] The dynamic speed reduction strategy reduces the emergency braking frequency by 60%, and the non-linear compensation algorithm increases the safety margin to more than 95% under different load conditions.

[0026] The three-stage response mechanism compresses the path generation time to 3.2 ms, and the secondary compensation program enables the trajectory deviation correction rate to reach 98.7%, ensuring ultra-precision operations.

[0027] The multi-parameter collaborative control technology enables the path correction accuracy to reach ±0.3 mm, the fracture surface normal deviation is controlled within ±1.5°, and the overall operation safety is improved by 70%. Description of the Drawings

[0028] Figure 1 It is a schematic diagram of an embodiment of the present application; Figure 2 It is a schematic diagram of another embodiment of the present application; Among them, 1: Hydraulic support leg; 2: Base platform; 3: Multi-axis robotic arm; 4: Hydraulic shear; 4a: Strain gauge; 5: Ring-shaped spray pipe; 5a: Solenoid valve; 5b: High-pressure water pump; 6: Water storage tank. Detailed implementation mode

[0029] The following further elaborates on the present invention in conjunction with the attached drawings so that those skilled in the art can implement it with reference to the text of the specification.

[0030] According to an embodiment of the present invention, the equipment basic structure consists of a base platform 2 and four groups of adjustable hydraulic support legs 1. The hydraulic support legs 1 can select the HSZ series four-stage telescopic cylinders of Rexroth, with a single-stage stroke of 300 mm and a maximum load-bearing capacity of 8 tons. A HBM U10M type pressure sensor is installed at the end of the support leg, with an accuracy of ±0.1% FS, for real-time monitoring of the ground reaction force. The base platform 2 uses a welded frame of Q345B high-strength steel, with a 5-mm thick anti-slip rubber pad laid on the surface. The hydraulic support legs 1 are rigidly connected to the four corners of the bottom surface of the base platform 2 through flange plates, and the installation inclination angle does not exceed ±3°.

[0031] The multi-axis robotic arm 3 can select the KUKA KR60 HA model, with a 6-degree-of-freedom design and a repeat positioning accuracy of ±0.05 mm. The end effector is a Hydrauvision HV-200 type hydraulic shear 4, with a maximum shear force of 200 tons, and the knife edge material is 9CrSi tool steel, with a blade thickness of 8 mm. The robotic arm 3 is rigidly connected to the hydraulic shear 4 through an ISO 9409-1 standard flange, and a Hydac pressure transmitter is configured at the connection to monitor the hydraulic system pressure in real time. The image acquisition module uses a Point Grey Bumblebee XB3 binocular camera, in cooperation with a SICK TIM561 laser rangefinder. The baseline distance of the camera is 300 mm, and the laser ranging accuracy is ±1 mm. It is installed on the front protection bracket of the base platform 2, maintaining a horizontal distance of 1.2 meters from the robotic arm 3.

[0032] The control module uses an Advantech UNOC-3362 industrial computer, equipped with an Intel Core i7-9700 processor, and integrates a National Instruments PCIe-7344 motion control card. The vision recognition module has an Autodesk Revit parsing engine built-in, which can read IFC format BIM model data. The collision detection algorithm sets a 30mm warning range and a 10mm braking critical zone. When the real-time distance value enters the braking critical zone, the motion control unit freezes the current trajectory, solves the inverse kinematic equation of the robotic arm 3 through the D-H parameter method, and generates a reverse compensation path that is 1.5 times the real-time distance. Cubic spline interpolation algorithm is used for path generation, and the calculation period is controlled within 5ms. Measured data shows that at a motion speed of 1.2m / s, the braking distance of this system can be controlled within 15mm, and the collision accident rate is reduced by 87% compared with traditional equipment.

[0033] Description of the working process: After the equipment is started, the hydraulic support legs 1 are extended to the preset height, and the pressure sensors monitor the support stability in real time. The binocular camera and the laser rangefinder collect environmental data at a frequency of 50Hz, and the vision recognition module fuses the point cloud data to generate a three-dimensional space model. When the hydraulic shear 4 approaches the preset safety boundary, the system responds in levels according to the real-time distance value: the 30mm warning range triggers linear deceleration, and the 10mm critical zone freezes the trajectory and generates a reverse compensation path. The compensation path is smoothed by cubic spline interpolation, and the robotic arm 3 is controlled to retract with an acceleration of 0.3m / s². At the same time, the position deviation is corrected through Kalman filtering. During the whole process, the control module updates the control parameters at a frequency of 1kHz to ensure the trajectory accuracy.

[0034] This implementation method realizes real-time path correction with millimeter-level accuracy through a hierarchical early warning braking mechanism and a dynamic path compensation algorithm. Measured data shows that at a motion speed of 1.2m / s, the braking distance is shortened to 15mm, which is 67% higher than that of traditional equipment; the collision accident rate is reduced by 87%, and the operation efficiency is increased by 40%. The multi-modal sensor fusion technology controls the three-dimensional modeling error within ±0.8mm, and the safety boundary update frequency reaches 50Hz, effectively adapting to complex working environments. The robotic arm joints use Kalman filtering to reduce noise, and the position feedback accuracy is improved to ±0.02mm, ensuring the requirements of ultra-precision demolition operations.

[0035] According to another embodiment of the present invention, when the device is running and the real-time distance value enters the 10 mm braking critical zone, the control module will trigger an emergency brake. The control module can select the Siemens S7-1500 series PLC, which has powerful logic operation and control capabilities. During braking, the control module will quickly cut off the power source of the multi-axis robotic arm. For example, a hydraulic valve group from Rexroth is used to quickly close the hydraulic oil circuit to achieve emergency braking. This braking method can ensure that the device stops quickly when approaching the dangerous area and avoid collision accidents.

[0036] When generating the reverse compensation path, the deviation angle between the compensation path and the current motion trajectory can be controlled within the range of 10° - 15°. In order to achieve smooth trajectory processing, the cubic B-spline interpolation method can be used. The cubic B-spline interpolation method is a mathematical algorithm widely used in industrial control software. For example, the MATLAB software can be used to implement the calculation of this algorithm. The path processed by this method can make the multi-axis robotic arm move more smoothly during reverse movement and reduce the impact on the device.

[0037] During the execution of the reverse compensation path, the acceleration upper limit of each joint of the multi-axis robotic arm is set to 0.3 m / s², and this parameter can be accurately set through the program of the control module. The sampling frequency of the speed control loop is set to 1 kHz, which means that the speed is sampled and adjusted 1000 times per second. An Omron high-speed counter can be selected to implement this function. The position feedback data is processed by Kalman filtering to suppress noise. Kalman filtering is an optimal estimation algorithm and can be implemented using an open-source Kalman filter library such as the Eigen library. Through these measures, the accuracy and stability of the robotic arm movement can be improved.

[0038] Working process: When the device is running, the distance between the hydraulic shear at the end of the multi-axis robotic arm and the preset safety boundary is monitored in real time. Once the distance enters the 10 mm braking critical zone, the control module immediately triggers an emergency brake and cuts off the power source. Then, the control module uses the built-in algorithm to calculate the reverse compensation path with a deviation angle of 10° - 15° from the current motion trajectory and uses the cubic B-spline interpolation method to smooth the path. During the execution of the reverse compensation path, the control module strictly controls the acceleration of each joint not to exceed 0.3 m / s², and at the same time adjusts the speed at a sampling frequency of 1 kHz, and uses Kalman filtering to process the position feedback data to suppress noise interference, ensuring that the robotic arm moves accurately and smoothly along the reverse compensation path.

[0039] Through the above technical solutions, when the equipment enters the braking critical area, it can brake quickly, effectively avoiding the occurrence of collision accidents. The precise calculation and processing of the reverse compensation path make the robotic arm move more smoothly during reverse movement, reducing the impact on the equipment and extending the service life of the equipment. At the same time, the precise control of the acceleration of each joint, the high-speed speed sampling frequency, and the effective noise suppression measures improve the accuracy and stability of the movement of the robotic arm, ensuring the quality and efficiency of the demolition operation. After actual testing, after adopting this technical solution, the incidence rate of collision accidents of the equipment has been significantly reduced, and the repeat positioning accuracy of the movement of the robotic arm has been increased by about 30%.

[0040] According to another embodiment of the present invention, a water storage tank 6 (with a volume of 200L) made of 304 stainless steel can be installed on the side of the base platform. A Vickers PVQ series high-pressure water pump (with a head of 100m and a flow rate of 30L / min) is configured on the top of the tank body. The outlet of the high-pressure water pump 5b is connected to a brass annular spray pipe 5 with an inner diameter of 12mm. The spray pipe is fixed around the hydraulic shear opening through a hoop, and is kept at a distance of 15mm from the cutting edge. An SMC VQZ series solenoid valve (with a response time of 15ms) is installed on the pipeline, and the valve body is fixed to the front end of the base platform through a bracket. An E+H FMP50 type ultrasonic level gauge (with an accuracy of ±2mm) is installed at the bottom of the water storage tank. When the water level is lower than 40L, the control module sends a shutdown instruction to the high-pressure water pump through the MODBUS protocol.

[0041] The annular spray pipe is processed by a laser drilling process. The nozzle spacing is 15mm, the aperture is 0.5mm, and the hole positions are arranged staggeredly at 120°. When the spray system works, the water pressure is maintained at 2.5MPa by the high-pressure water pump, and atomized water curtains with a particle size of 5 - 10μm can be formed through calculation. The measured data shows that the water mist coverage rate under this parameter can reach more than 90%, effectively suppressing the splash of cutting sparks. An HBM K - C4A - 200N type strain gauge (with a sensitivity of 2.1±0.1%) is embedded inside the hydraulic shear opening and is pasted to the stress concentration area at the root of the cutting edge through epoxy resin glue (Loctite 326). The signal wire is connected to the control module by a shielded twisted pair wire (AWG24).

[0042] The control module is built-in with a strain threshold determination program. When the detection value of the strain gauge 4a reaches 85% of the preset yield strength of the steel member (such as 345MPa for Q345 steel), the solenoid valve 5a is triggered to open. The spray cycle is set to a 500ms on-off cycle, and the single opening time is 300ms. Experiments show that this strategy can keep the temperature in the cutting area stable below 120℃, reducing 150℃ compared with traditional dry cutting. At the same time, the response time of the water level monitoring system is less than 200ms, ensuring that the equipment stops in time when the water volume is low, avoiding idling and damaging the water pump.

[0043] Working process: When the equipment is running, the strain gauge monitors the stress at the hydraulic shear edge in real time. When the stress value reaches the threshold (such as 293 MPa), the control module immediately sends an electrical signal to the solenoid valve to activate the spraying system. The high-pressure water pump pressurizes the water in the water storage tank to 2.5 MPa, and forms an atomized water curtain through the annular spray pipe to cover the cutting area. At the same time, the ultrasonic level gauge continuously monitors the water level. When the water volume is lower than the set value, the control module cuts off the power supply of the high-pressure water pump and alarms. During the whole process, the spraying system starts and stops synchronously with the shearing operation, realizing the interlocking control of spark suppression and equipment protection.

[0044] This embodiment realizes the active cooling and dust control during the cutting process through real-time stress monitoring and the intelligent spraying system. The measured data shows that the splashing distance of the cutting sparks is shortened from 1.2 m to 0.3 m, and the dust concentration in the working environment is reduced from 15 mg / m³ to below 5 mg / m³. The strain threshold triggering mechanism enables the spraying system to start 0.3 seconds before overload, effectively avoiding overheating and wear of the edge, and extending the service life of the hydraulic shear by 30%. The design of the water level monitoring system eliminates the risk of the water pump running idly, and reduces the equipment failure rate by 45%. In addition, the atomized water curtain can absorb part of the cutting noise, reducing the noise value in the working area from 95 dB to 82 dB, and improving the construction environment.

[0045] According to another embodiment of the present invention, an HBM U10M type pressure sensor (range 0 - 500 kN, accuracy ±0.1% FS) can be installed at the end of the hydraulic support leg, and is rigidly connected to the support leg through an M12×1.5 threaded interface. The signal cable of the sensor uses a shielded cable (wire diameter 1.5 mm²) and is laid along the inner wall of the oil cylinder through a pipe to the control module on the base platform. The measured data shows that the response time of this sensor is less than 5 ms, and it can monitor the change of the supporting force in real time. When the deviation of the unilateral supporting force exceeds 20% of the average value, the control module triggers an alarm.

[0046] The hydraulic support leg adopts a Rexroth HSZ series four-stage telescopic oil cylinder (stroke 1.2 m, cylinder diameter 120 mm), and the outer wall of each stage of the oil cylinder is wrapped with an anti-collision buffer layer. The inner layer of the buffer layer is a 3 mm thick polyurethane elastomer (Shore hardness 90A), the middle layer is a 1 mm diameter 304 stainless steel wire braided mesh (mesh side length 5 mm), and the outer layer is a 5 mm thick polyurethane-glass fiber hybrid coating (glass fiber content 25%). The three-layer materials are compounded through a heat vulcanization process to form a 9 mm thick buffer structure. After the drop hammer impact test (10 kg heavy object, 1 m height), the energy absorption rate can reach 78%.

[0047] The four-stage telescopic cylinder is connected to the bottom surface of the base platform through a flange, and the installation angle deviation is controlled within ±2°. A magnetostrictive displacement sensor (MTS Temposonics R series) with a resolution of 0.01 mm is configured inside the cylinder to monitor the telescopic amount in real time. When the support leg is subjected to a lateral impact, the buffer layer absorbs energy through the deformation of the polyurethane elastomer, the steel mesh layer provides tear resistance, and the fiberglass coating enhances surface wear resistance. Experimental data shows that this structure can withstand a lateral load of 30 kN, with a deformation of less than 5 mm and a recovery time of less than 200 ms.

[0048] Operation process: During equipment deployment, the hydraulic support legs are adjusted to a horizontal state through four-stage cylinders, and the pressure sensors monitor the support force of each leg in real time. When the pressure of a certain support leg is abnormal (such as exceeding the average value by 20%), the control module automatically adjusts the telescopic amount of the corresponding cylinder for compensation. If a collision occurs during operation, the buffer layer absorbs the impact energy through the viscoelastic deformation of the polyurethane elastomer, the steel mesh layer restricts excessive deformation, and the fiberglass coating prevents surface damage. Throughout the process, the displacement sensor continuously feeds back the cylinder state to ensure the stability of the support system.

[0049] This embodiment, through the collaborative design of the pressure sensor and the multi-stage cylinder, improves the support stability of the equipment by 55% and can adapt to complex terrains with an inclination ≤15°. The composite structure of the anti-collision buffer layer increases the lateral impact energy absorption rate to 78%, and the anti-impact ability is enhanced by 80% compared with the traditional rubber buffer layer. Measured data shows that in a 10 Hz vibration environment, the amplitude of the support leg decreases from 12 mm to 4 mm, and the risk of equipment overturning is reduced by 62%. At the same time, the fiberglass coating extends the life of the buffer layer to 3000 hours, and the maintenance cycle is increased by 2 times compared with ordinary materials.

[0050] According to another embodiment of the present invention, the image acquisition module can be composed of a Point Grey Bumblebee XB3 binocular camera and a SICK TIM561 laser rangefinder. The baseline distance of the binocular camera is 300 mm, and it can acquire RGB images. The measurement accuracy of the laser rangefinder is ±1 mm, and it can obtain point cloud data. The binocular camera and the laser rangefinder can be installed at the front end of the base platform through brackets. Such an installation position can ensure that its field of view covers the operation area and is not easily interfered by other components of the equipment. The sampling frequency of the image acquisition module is set to 50 Hz, which means that 50 sets of image and distance data can be acquired per second, providing sufficient information for subsequent analysis.

[0051] The visual recognition module can adopt a system built based on the NVIDIA Jetson Xavier NX development board, which has powerful computing capabilities and can efficiently process the collected data. The visual recognition module will perform frame synchronization fusion on the RGB images collected by the binocular camera and the point cloud data of the laser rangefinder, and update the preset safety boundary data with a period of 20 ms. This process is like integrating information from different perspectives and types to form a more comprehensive and accurate three-dimensional scene description. Through this fusion method, more accurate three-dimensional coordinate data can be obtained, providing a reliable basis for subsequent path planning.

[0052] The spatial coordinates of the preset safety boundary data are obtained by parsing the BIM model in the building component feature database. During the spatial coordinate matching process, the iterative closest point algorithm is used for error compensation, with the number of iterations set to 50 times and the single-iteration calculation time not exceeding 0.1 ms. This can ensure that the coordinate matching is completed in a short time and the error is controlled within a small range. In practical applications, to verify the accuracy and stability of the system, a simulated building demolition scenario can be selected as the experimental object, and experiments can be carried out multiple times by changing different environmental parameters and component positions. Statistical analysis is performed on the experimental data to calculate indicators such as the error range and the response time of the system to evaluate the performance of the system.

[0053] Working process: After the device is started, the image acquisition module starts to work. The binocular camera collects RGB images at a sampling frequency of 50 Hz, and the laser rangefinder synchronously obtains the point cloud data. These data are transmitted to the visual recognition module in real time. The visual recognition module performs frame synchronization fusion on the RGB images and the point cloud data at a period of 20 ms to obtain three-dimensional coordinate data. At the same time, the BIM model data is extracted from the building component feature database, the spatial coordinates of the preset safety boundary data are parsed, and error compensation is performed 50 times through the iterative closest point algorithm. The updated preset safety boundary data will be provided to the path planning unit of the control module to generate the motion trajectory of the multi-axis robotic arm. The whole process loops to ensure that the device can make accurate path planning according to the real-time environmental changes.

[0054] By adopting an image acquisition module composed of a binocular camera and a laser rangefinder, combined with the efficient data processing ability of the visual recognition module, high-precision three-dimensional environment perception can be achieved. The multi-modal data fusion technology controls the three-dimensional modeling error within ±0.8 mm, greatly improving the accurate description of the operating environment. The preset safety boundary data is updated at a period of 20 ms, enabling the system to respond to environmental changes in a timely manner. The safety boundary update frequency reaches 50 Hz, and the environmental perception response speed is increased by 5 times. The application of the iterative closest point algorithm further reduces the coordinate matching error, ensuring the accuracy of the preset safety boundary data. These technical effects enable the device to operate more safely and efficiently in a complex building demolition environment, reducing the occurrence of collision accidents and improving the operation efficiency.

[0055] According to another embodiment of the present invention, the control module can adopt Siemens S7-1500 PLC, and its analog input module supports 16-bit precision. An HBM K-C4A-200N strain gauge (sensitivity coefficient 2.1) is pasted on the inner side of the hydraulic shear mouth and connected to the AI module of the PLC through a half-bridge circuit. When the strain gauge detects that the shear stress change rate exceeds 5 MPa / s, the PLC triggers a compensation path adjustment instruction and sets the reverse compensation amount to 1.5 times the real-time distance value. This threshold can be calibrated through the WinCC configuration software. Experiments show that the response time of this trigger mechanism is less than 12 ms during the shearing of Q345 steel.

[0056] The fracture surface image collected by the binocular camera (such as Point Grey Bumblebee XB3) is transmitted to the visual recognition module through the GigE interface. The OpenCV library is used for edge detection and morphological analysis. When a transverse crack expansion is detected, the algorithm calculates the angle between the crack direction and the compensation path, and adjusts the mechanical arm joint angle through a servo drive (such as the Beckhoff AX5000 series). The measured data shows that when the crack expansion rate ≤ 0.3 mm / ms, the direction adjustment error can be controlled within the range of ±2°.

[0057] The adaptive PID algorithm can be tuned through Matlab / Simulink. The initial value of the proportional coefficient Kp is set to 0.8, and the integral time Ti is 0.5 s. When the hydraulic support leg pressure sensor (HBM U10M) detects that the unilateral pressure exceeds 20% of the average value, the PLC sends a parameter update instruction to the servo drive through the Modbus protocol. In addition, the data of the mechanical arm joint encoder (Heidenhain ERN 1387) and the laser rangefinder (SICK TIM561) are fused by Kalman filter at a period of 50 ms. When the deviation between the two exceeds 0.5 mm, the system automatically switches to the laser data as the position reference.

[0058] Working process: When the equipment is running, the strain gauge monitors the change of shear stress in real time. When the change rate breaks through 5 MPa / s, the adjustment of the compensation amount is triggered. The binocular camera synchronously analyzes the morphology of the fracture surface. If a transverse crack is detected, the compensation direction automatically deflects by 15° - 30°. At the same time, the pressure sensors of the support legs continuously monitor the load distribution. When the unilateral pressure deviation exceeds 20%, the PID parameters are dynamically adjusted. During the movement of the robotic arm, the data of the joint encoder and the laser rangefinder are fused every 50 ms. If the deviation exceeds the limit, the trajectory is corrected with the laser data to ensure the accurate execution of the compensation path.

[0059] Through multi-parameter coupling control, the adjustment accuracy of the compensation path in this embodiment is improved to ±0.3 mm. Experimental data shows that when the shear stress change rate reaches 10 MPa / s, the system response time can still be controlled within 8 ms, and the path tracking error is less than 0.5 mm. The recognition accuracy of the transverse crack direction reaches 92%, and the adjustment error of the compensation direction ≤ ±2°. The adaptive PID algorithm reduces the overshoot of the robotic arm during load mutation by 60%, and shortens the pressure balance response time of the support legs to 150 ms. The multi-source data fusion technology improves the trajectory deviation correction rate to 98.2%, and the operation stability under complex working conditions is improved by 65% compared with the traditional scheme.

[0060] According to another embodiment of the present invention, the annular spray pipe can be configured with SMC VQZ series solenoid valves (response time 15 ms), which are linked with Rexroth A10VSO high-pressure pumps (flow rate 20 L / min). The compensation speed is monitored in real time by the robotic arm encoder (Heidenhain ERN 1387). When the speed exceeds 0.1 m / s, the control module (Siemens S7-1500) increases the spray frequency proportionally. The measured data shows that when the compensation speed is 0.3 m / s, the spray frequency reaches 30 Hz, and the water mist coverage rate reaches 85%. The nozzle is made of brass with a pore diameter of 0.5 mm, and is fixed to the periphery of the hydraulic shear opening by a clamp, keeping a distance of 15 mm from the blade edge.

[0061] The path planning unit can select the Beckhoff CX5140 controller with TwinCAT 3 software built in. The reverse compensation path is decomposed into three segments: the first segment of 5 mm uses an S-shaped acceleration curve (maximum 0.2 m / s²), which is executed by the robotic arm joint servo driver (Yaskawa Σ-7 series); the middle segment of 3 mm maintains a constant speed (the speed value is 30% of the current speed); the last segment of 2 mm uses exponential decay deceleration (time constant 0.1 s). Through MATLAB simulation verification, this segmentation strategy reduces the jitter amplitude at the end of the robotic arm to 0.08 mm.

[0062] The high-frequency vibration mode is achieved by superimposing a Rexroth M-SR series vibrator on the hydraulic system. The vibration frequency of 50 Hz is precisely controlled by a frequency converter (Siemens G120). The amplitude of 0.1 mm is adjusted by closed-loop feedback of a displacement sensor (MTS Temposonics R series). When the laser rangefinder (SICK TIM561) detects that the displacement of the component is less than 80% of the expected value, the control module triggers the vibration mode, and the duration does not exceed 30 cycles. Experimental data shows that this mode can increase the cutting efficiency by 25% and is applicable to high-strength materials such as Q345 steel.

[0063] Working process: When the reverse compensation path is started, the control module monitors the compensation speed in real time and increases the spraying frequency in a gradient of 0.1 m / s, with a maximum not exceeding 20 L / min. The path planning unit decomposes the trajectory into three segments. The first segment realizes smooth start through an S-shaped acceleration curve, the middle segment maintains a stable speed, and the last segment decelerates exponentially to ensure precise stop. If the laser rangefinder detects insufficient displacement of the component, the system automatically switches to the high-frequency vibration mode to assist cutting with a hydraulic vibrator. During the whole process, the data of the displacement sensor and the encoder are fed back at a frequency of 1 kHz to ensure the trajectory accuracy.

[0064] This implementation method shortens the compensation movement time by 40% through segmented trajectory control, and the path tracking error is less than 0.3 mm. The linkage strategy of the spraying system stabilizes the temperature in the cutting area below 120 °C, a reduction of 150 °C compared with the traditional method. In the Q345 steel cutting experiment, the high-frequency vibration mode reduces the average cutting force from 180 kN to 140 kN and reduces the tool wear rate by 35%. The measured data shows that when the compensation speed reaches 0.3 m / s, the system can still operate stably, and the water mist dust suppression efficiency reaches 90%, meeting the requirements of high-intensity operations.

[0065] According to another embodiment of the present invention, an HBM K-C4A-200N strain gauge (sensitivity coefficient 2.1) can be pasted on the inner side of the hydraulic shear mouth and connected to the control module (Siemens S7-1500) through a half-bridge circuit. When the shear stress < 15 MPa, the deceleration target value is 50% of the current speed, with a slope of 15% / s; when it is 15 - 30 MPa, the target is 40%, with a slope of 20% / s; when it is ≥ 30 MPa, the target is 30%, with a slope of 25% / s. This threshold can be calibrated by the WinCC configuration software. Experiments show that when shearing Q345 steel, the stress measurement error ≤ ±1.5 MPa.

[0066] The binocular camera (Point Grey Bumblebee XB3) works in collaboration with the laser rangefinder (SICK TIM561) to identify the number of obstacles within 5 meters and measure the closest distance. The environmental complexity value = the number of obstacles / (the closest distance × 0.001). When the ratio > 2, the deceleration slope increases by an additional 5% / s. This algorithm can be implemented through the OpenCV library. When measured with 3 obstacles and the closest distance of 2m, the complexity value reaches 1.5, triggering the slope increment.

[0067] The upper limit of the joint angular acceleration of the robotic arm is determined by the part of the real-time distance exceeding 20mm. For every additional 10mm, it increases by 0.8rad / s², which is set through the parameters of the Beckhoff AX5000 servo drive. When the distance reduction rate > 0.5mm / ms during the deceleration process, it switches to the non-linear deceleration mode, and the curve is v(t)=v0×e^(-0.2t)+0.3v0, and the compensation amount increases linearly with time. At the same time, the SMC VQZ solenoid valve increases the frequency to 15Hz when the target speed ≤ 40%, and the duty cycle is dynamically adjusted between 30% - 50% according to the stress value.

[0068] Working process: When the real-time distance enters the warning area of 20 - 30mm, the control module divides the load level according to the strain gauge data and dynamically calculates the deceleration parameters. The binocular camera and the laser rangefinder evaluate the environmental complexity in real time and adjust the slope increment. The path planning unit calculates the upper limit of the angular acceleration according to the real-time distance and generates a speed constraint trajectory. If the deceleration rate exceeds the limit, it switches to the non-linear mode. At the same time, the spraying system adjusts the working parameters according to the speed target. During the whole process, the robotic arm encoder (Heidenhain ERN 1387) feeds back data at a frequency of 1kHz to ensure the control accuracy.

[0069] This implementation method reduces the emergency braking frequency by 60% and increases the safety margin to 95% through the multi-parameter coupling deceleration strategy. Experimental data shows that when the shear stress of Q345 steel is 30MPa, the deceleration response time is shortened to 12ms, and the path tracking error < 0.5mm. The environmental complexity algorithm improves the adaptability of the deceleration slope under complex working conditions by 40%, and the non-linear compensation curve reduces the impact load by 55%. The spraying system linkage strategy stabilizes the temperature in the cutting area below 120℃ and controls the dust concentration within 5mg / m³, which is 40% higher than the traditional scheme.

[0070] According to another embodiment of the present invention, the fault self-diagnosis module may use the Siemens S7-1500 series PLC as the core controller, which has powerful computing and data processing capabilities. The Bayesian network model can be constructed and trained using MATLAB software, and the fault code library can be stored in the data block of the PLC. This module can be installed in the electrical control cabinet of the device to facilitate connection with other electrical components. When the device is running, various data collected by the sensors are transmitted to the PLC, and the Bayesian network model performs fault diagnosis based on this data and the fault code library, outputting the fault probability and confidence level. The threshold of the fault probability can be set to 0.6. When the fault probability exceeds 0.6, it is considered that the device may have a fault. The range of the confidence level can be set from 0 to 1, and the closer the value is to 1, the more reliable the diagnostic result is.

[0071] The fault warning threshold can be dynamically adjusted according to the running time and working conditions of the equipment. For example, during the running-in period of the equipment (running time less than 100 hours), the temperature warning threshold can be set to 60°C; during the normal operation period (running time 100 - 500 hours), the temperature warning threshold can be set to 70°C; during the aging period (running time greater than 500 hours), the temperature warning threshold can be set to 80°C. For different working conditions, such as under heavy load conditions, the vibration warning threshold can be set to 5 mm / s², and under light load conditions, the vibration warning threshold can be set to 3 mm / s². These thresholds can be adjusted in real time through the PLC program to adapt to different operating conditions.

[0072] Multi-modal information fusion can adopt the D-S evidence theory and be implemented through a dedicated information fusion algorithm chip, such as the Intel Xeon series processor. The temperature sensor can select a PT100 platinum resistance temperature sensor and be installed at the key heat-generating parts of the equipment; the vibration sensor can select a piezoelectric vibration sensor and be installed near the vibration source of the equipment; the pressure sensor can select a diffused silicon pressure sensor and be installed on the pipeline of the hydraulic system. The data collected by these sensors are transmitted to the PLC through the analog input module, and then the D-S evidence theory is used for fusion processing to improve the accuracy of fault diagnosis. At the same time, the device can transmit the collected data and diagnostic results to the remote monitoring center through an industrial Ethernet module, such as the Siemens CP 1543-1 module. The remote monitoring center can use professional monitoring software, such as WinCC, for data display and analysis to achieve remote diagnosis and warning functions.

[0073] Working process: During the operation of the equipment, sensors such as temperature, vibration, and pressure collect data in real time and transmit the data to the PLC in the fault self-diagnosis module. The Bayesian network model in the PLC performs fault diagnosis based on this data and the fault code library, calculating the fault probability and confidence level. At the same time, the PLC dynamically adjusts the fault warning threshold according to the running time and working conditions of the equipment. The multi-modal information fusion module performs fusion processing on the multi-modal information collected by the sensors to improve the accuracy of diagnosis. If the fault probability exceeds the set threshold, the PLC issues a fault warning signal. In addition, the equipment transmits the collected data and diagnosis results to the remote monitoring center through the industrial Ethernet, and the staff in the remote monitoring center can understand the running state of the equipment in real time and conduct remote diagnosis and warning.

[0074] Through the fault self-diagnosis module and the dynamically adjusted fault warning threshold, potential faults of the equipment can be detected in time, advancing the fault discovery time by about 30% and reducing the sudden fault shutdown time of the equipment. The multi-modal information fusion technology improves the accuracy of fault diagnosis, and the diagnosis accuracy rate has increased from the original 70% to over 90%. The remote monitoring and diagnosis function enables technicians to perform real-time monitoring and diagnosis of the equipment remotely, improving the efficiency of fault handling and reducing the maintenance cost of the equipment. According to statistics, the maintenance cost of the equipment has been reduced by about 25%.

[0075] According to another embodiment of the present invention, Rexroth HSZ series hydraulic support legs can be installed at the four corners of the bottom surface of the base platform, and each leg is equipped with an HBM U10M pressure sensor (range 0 - 500 kN, accuracy ±0.1% FS). During the leveling process, the pressure sensor monitors the supporting force in real time, and the control module adjusts the telescopic amount of the oil cylinder through the PID algorithm to ensure that the deviation of the base levelness < ±0.5°. The measured data shows that under the working condition of a 15° slope, the system can complete leveling within 30 seconds, shortening the time by 40% compared with the traditional method.

[0076] The vision recognition module uses the NVIDIA Jetson Xavier NX development board and integrates the Autodesk Revit engine to parse the BIM model. The binocular camera and the laser rangefinder collect data at a frequency of 50 Hz, and perform point cloud fusion through the OpenCV library to generate a three-dimensional model including a preset safety boundary. The safety boundary update period is 20 ms, and error compensation is performed through the iterative closest point algorithm (ICP). Experiments show that after 50 iterations of this algorithm, the modeling error can converge to ±0.8 mm.

[0077] When the real-time distance enters the 10mm critical zone, the control module freezes the current trajectory and generates a reverse compensation path. Based on the real-time distance value, the path length is automatically extended by 1.8 times the original path length. If the component deformation rate exceeds 0.2mm / ms, as detected by the strain gauge, the path direction is deflected by 15°-25° when the strain gauge detects a stress change rate greater than 5MPa / s. Measured data shows that this strategy achieves a path correction accuracy of ±0.3mm and a fracture surface normal deviation of ±1.5°.

[0078] Working process: When the equipment is deployed, the hydraulic support legs unfold to the preset height, and pressure sensors monitor the support stability in real time. The binocular camera and laser rangefinder collect environmental data at a frequency of 50Hz, and the visual recognition module integrates the point cloud data to generate a three-dimensional spatial model. When the hydraulic shear approaches the preset safety boundary, the system responds in a graded manner based on the real-time distance value: the 30mm warning zone triggers linear deceleration, and the 10mm critical zone freezes the trajectory and generates a reverse compensation path. The compensation path is smoothed using cubic spline interpolation, and the robotic arm is controlled to retract at an acceleration of 0.3m / s². The position deviation is simultaneously corrected through Kalman filtering. Throughout the process, the control module updates the control parameters at a frequency of 1kHz to ensure trajectory accuracy.

[0079] This implementation method achieves real-time path correction with millimeter-level accuracy through a graded warning braking mechanism and a dynamic path compensation algorithm. Measured data shows that at a speed of 1.2 m / s, the system's braking distance is reduced to 15 mm, a 67% improvement over traditional equipment. The collision accident rate is reduced by 87%, and operational efficiency is increased by 40%. Multimodal sensor fusion technology controls 3D modeling errors to ±0.8 mm, with a safety boundary update frequency of 50 Hz, effectively adapting to complex operating environments. Kalman filtering is used to reduce noise in the robotic arm joints, increasing position feedback accuracy to ±0.02 mm, ensuring ultra-precision demolition operations.

[0080] Experiments show that a 1.5-fold compensation distance can cover 95% of the error tolerance of the braking inertial displacement of the robotic arm, ensuring a safe withdrawal. By optimizing the cubic B-spline interpolation algorithm and adopting a parallel computing architecture (such as GPU acceleration), the path planning cycle is compressed from 10 ms to 5 ms. The feature database is parsed based on the BIM model, extracting component geometric features (such as cross-sectional shape, bolt hole positions) and storing them in JSON format. When the pressure deviation of the support leg exceeds 20%, the proportional coefficient Kp is increased from 0.8 to 1.2, and the integral time Ti is shortened from 0.5 s to 0.3 s to enhance the load balancing response. Through the weighted fusion of binocular camera (±0.5 mm accuracy) and laser rangefinder data, the final positioning error is reduced to ±0.3 mm. Under 50 Hz vibration, the amplitude at the end of the robotic arm ≤ 0.05 mm (measured), and the accuracy loss is within the allowable range. The experiment selected Q345B steel beams (section 200×200×8×12), and the comparison equipment was the DT-300 demolition machine of XX Company, with 200 operation times. The collision rate of the present invention is 1.5%, and that of the traditional equipment is 12%; the average cutting time is shortened from 4.2 minutes to 2.5 minutes. The environmental complexity = the number of obstacles / (the distance to the nearest obstacle × 0.001), and the obstacle is defined as a rigid object with a size ≥ 10 cm. The reverse compensation path is a smooth curve with an angle of 10° - 15° with the current trajectory, generated by cubic B-spline interpolation.

[0081] According to another embodiment of the present invention, through dynamic mechanical simulation and experimental verification, when the robotic arm moves at a speed of 1.2 m / s, the reverse compensation path of 1.5 times the real-time distance can cover 95% of the braking inertial displacement, ensuring the hydraulic shear is withdrawn to a safe area. For example, when the real-time distance is 10 mm, the compensation path is 15 mm, leaving a margin of 5 mm to offset system errors. Adopting a parallel computing architecture (such as GPU acceleration of NVIDIA Jetson Xavier NX), the calculation time of the control points of the cubic B-spline interpolation algorithm is compressed from 10 ms to 3 ms. Combining with the optimization of inverse kinematics, the total path generation cycle is controlled within 5 ms. The building component feature database is parsed based on the BIM model, extracting the geometric features (such as H-shaped steel section dimensions, bolt hole position coordinates) and material properties (such as yield strength) of the components, and storing them in a standardized JSON format. For example, the feature data of Q345B steel components include: section 200×200×8×12, yield strength 345 MPa, elastic modulus 206 GPa.

[0082] The data of the laser rangefinder (±1 mm) and the binocular camera (±0.5 mm) are weighted and fused, where the weight of the laser data is 0.6 and the weight of the visual data is 0.4, and the final positioning error is reduced to ±0.3 mm. The fusion formula is: D_fusion = 0.6×D_laser + 0.4×D_visual In the 50Hz high-frequency vibration mode, the amplitude at the end of the robotic arm is closed-loop controlled by a displacement sensor. The measured amplitude is ≤0.05mm (Heidenhain encoder data), and the vibration direction is perpendicular to the cutting load direction to avoid affecting the shearing accuracy.

[0083] Although the embodiments of the present invention have been disclosed above, it is not limited to the applications listed in the specification and embodiments. It can be fully applied to various fields suitable for the present invention. For those familiar with the field, additional modifications can be easily made. Therefore, without departing from the general concept defined by the claims and the equivalent scope, the present invention is not limited to the specific details and the illustrated examples here.

Claims

1. A building component demolition device, characterized in that, It includes the following structures: The base platform is provided with adjustable hydraulic support legs; Above the base platform, a multi-axis robotic arm is fixedly installed, and a hydraulic shear is connected to the end of the multi-axis robotic arm; An image acquisition module is installed at the front end of the base platform. The output end of the image acquisition module is connected to a visual recognition module, and the visual recognition module has a building component feature database built-in; The control module includes a path planning unit and a motion control unit. The path planning unit receives the three-dimensional coordinate data output by the visual recognition module. The motion control unit generates the motion trajectory of the multi-axis robotic arm according to the three-dimensional coordinate data and the building component feature database. The three-dimensional coordinate data includes preset safety boundary data; The control module controls the multi-axis robotic arm according to the motion trajectory of the multi-axis robotic arm, drives the hydraulic shear to the location of the preset steel component to be demolished, and the control module controls the hydraulic shear to perform the demolition operation; Among them, the control module has a collision detection algorithm built-in. When the real-time distance value between the hydraulic shear at the end of the multi-axis robotic arm and the preset safety boundary enters the 30mm warning range, the control module linearly reduces the motion speed of the robotic arm to 30%-50% of the current speed value, and the deceleration slope is set to reduce the original speed by 20% per second; When the real-time distance value enters the 10mm braking critical area, the motion control unit immediately freezes the current motion trajectory of the multi-axis robotic arm, and generates a reverse compensation path based on the real-time space coordinates of the hydraulic shear. The length of the reverse compensation path is 1.5 times the real-time distance value, and the path generation period does not exceed 5ms; The control module controls the robotic arm to move along the reverse compensation path.

2. The building component demolition device according to claim 1, characterized in that When the real-time distance value enters the 10mm braking critical area, the control module triggers an emergency brake; And, the method for generating the reverse compensation path is: the deviation angle between the compensation path and the current motion trajectory is constrained within the range of 10°-15°, and the trajectory smoothing process uses the cubic B-spline interpolation method; During the execution of the reverse compensation path, the acceleration upper limit of each joint of the multi-axis robotic arm is set to 0.3m / s², the sampling frequency of the speed control loop is set to 1kHz, and the position feedback data uses Kalman filtering for noise suppression.

3. The building component demolition device according to claim 1, characterized in that A water storage tank and a high-pressure water pump are arranged on the side of the base platform. The outlet of the high-pressure water pump is connected to an annular spray pipe, and the annular spray pipe is arranged around the hydraulic shear opening. The annular spray pipe is provided with a solenoid valve, and the solenoid valve is connected to the control module; A water level monitoring device is arranged in the water storage tank, and the water level monitoring device is connected to the control module. When the water storage volume is lower than 20% of the set volume, the control module issues a shutdown instruction; The nozzle spacing of the annular spray pipe is 15mm, the nozzle aperture is 0.5mm, the spray water pressure is maintained at 2.5MPa, and the diameter of the atomized water particles formed by the spray is 5-10mm; A strain gauge array is embedded on the inner side of the cutting edge of the hydraulic shear, and the strain gauge array is connected to the control module through a signal line; The control module is configured with a strain threshold determination program. When the strain gauge array detects that the shear stress reaches 85% of the preset yield strength value of the steel component, the solenoid valve is triggered to open.

4. The building component demolition equipment according to claim 1, characterized in that a pressure sensor is installed at the end of the hydraulic support leg, and the pressure sensor is electrically connected to the control module of the base platform; the hydraulic support leg includes a four-stage telescopic oil cylinder, and an anti-collision buffer layer is provided on the outer wall of each stage of the oil cylinder. The anti-collision buffer layer adopts a polyurethane-steel mesh composite structure; wherein, the polyurethane-steel mesh composite structure is: the inner layer is a 3mm thick polyurethane elastomer, the middle layer is a hexagonal mesh layer woven with 1mm diameter 304 stainless steel wires, and the outer layer is a 5mm thick polyurethane-glass fiber mixed coating.

5. The building component demolition equipment according to claim 1, characterized in that the image acquisition module includes a binocular camera and a laser rangefinder; the visual recognition module performs frame synchronization fusion on the RGB image collected by the binocular camera and the point cloud data of the laser rangefinder to obtain three-dimensional coordinate data, and the fused three-dimensional coordinate data updates the preset safety boundary data at a period of 20ms; the spatial coordinates of the preset safety boundary data are obtained by parsing the BIM model in the building component feature database. The iterative closest point algorithm is used for error compensation during the spatial coordinate matching process. The number of iterations is set to 50 times, and the single iteration calculation time does not exceed 0.1ms; the sampling frequency of the image acquisition module is set to 50Hz, the measurement accuracy of the laser rangefinder is ±1mm, and the baseline distance of the binocular camera is 300mm.

6. The building component demolition equipment according to claim 3, characterized in that, The method for the control module to control the mechanical arm to move along the reverse compensation path further includes: the motion control unit continuously obtains the real-time shear stress data detected by the strain gauge array at the hydraulic shear opening. When the shear stress change rate exceeds 5MPa / s, the reverse compensation path is triggered, and the reverse compensation amount is dynamically adjusted to 1.5 times the real-time distance value; the visual recognition module synchronously analyzes the morphological characteristics of the component fracture surface captured by the binocular camera. When it detects that there is a transverse crack extension on the fracture surface, it automatically adjusts the reverse compensation direction to form an angle of 15° - 30° with the crack extension direction; the trajectory control of the reverse compensation movement adopts an adaptive PID algorithm, and the control parameters are dynamically adjusted according to the load distribution data of the pressure sensor of the hydraulic support leg. When the pressure of a single support leg exceeds 20% of the average value, the proportional coefficient Kp is increased from 0.8 to 1.2, and the integral time Ti is shortened from 0.5s to 0.3s; the motion control unit synchronously processes the joint encoder data of the multi-axis mechanical arm and the distance feedback data of the laser rangefinder at a period of 50ms. When the deviation between the two data exceeds 0.5mm, the laser rangefinder data is used to correct the trajectory.

7. The building component demolition equipment according to claim 6, characterized in that during the execution of the reverse compensation, the opening frequency of the solenoid valve of the annular spray pipe is linked with the compensation speed. For every 0.1m / s increase in the compensation speed, the spray frequency is increased by 10Hz, and the maximum spray flow rate is limited to 20L / min. Based on the real-time updated three-dimensional coordinate data, the path planning unit decomposes the reverse compensation path into three segments of motion trajectories. For the first segment with a 5-mm travel, an S-shaped acceleration curve is adopted, with the maximum acceleration limited to 0.2 m / s². For the second segment with a 3-mm travel, uniform motion is maintained. For the third segment with a 2-mm travel, an exponential decay deceleration curve is adopted; When it is detected that the displacement of the component does not reach 80% of the expected value during the duration of the reverse compensation motion, the control module automatically switches to the high-frequency vibration mode, with the vibration frequency set to 50 Hz, the amplitude to 0.1 mm, and the upper limit of the vibration times set to 30 times.

8. The construction component demolition device according to claim 3, characterized in that When the real-time distance value between the hydraulic shear at the end of the multi-axis robotic arm and the preset safety boundary enters the first-level warning area of 20 - 30 mm, the motion control unit dynamically calculates the deceleration parameters according to the current load state of the hydraulic shear. The load state is divided into three levels by the real-time shear stress value detected by the strain gauge array: When the shear stress is less than 15 MPa, the deceleration target value is 50% of the current speed, and the deceleration slope is to decrease by 15% per second; When the shear stress is in the range of 15 MPa to 30 MPa, the deceleration target value is 40% of the current speed, and the deceleration slope is to decrease by 20% per second; When the shear stress reaches or exceeds 30 MPa, the deceleration target value is 30% of the current speed, and the deceleration slope is to decrease by 25% per second; The visual recognition module synchronously evaluates the environmental complexity. By identifying the number of obstacles within 5 meters through a binocular camera and combining the distance value of the nearest obstacle detected by the laser rangefinder, the environmental complexity value is calculated as the ratio of the number of identified obstacles to one-thousandth of the distance of the nearest obstacle. When this ratio is greater than 2, the deceleration slope is additionally increased by an increment of 5% per second; The path planning unit generates a speed constraint trajectory in real time, and sets the upper limit of the angular acceleration of each joint of the robotic arm as follows: for each additional 10-mm distance beyond the 20-mm real-time distance value, the upper limit of the angular acceleration is increased by 0.8 rad / s²; When the real-time distance value continuously shrinks during the deceleration process and the shrinking rate exceeds 0.5 mm / ms, it automatically switches to the non-linear deceleration mode, and the deceleration curve is adjusted to the initial speed decaying according to the natural exponential function plus a compensation amount of 30% of the initial speed. The decay coefficient of the natural exponential function is 0.2 per second, and the compensation amount increases with the duration of the deceleration; The opening strategy of the solenoid valve of the annular spray pipe is linked with the deceleration process. When the deceleration target value is less than or equal to 40% of the current speed, the spray frequency is increased to 15 Hz, and the on-time duty cycle within a single spray cycle is dynamically adjusted according to the real-time shear stress value, with the adjustment range being 30% to 50%.

9. The building component demolition equipment according to claim 1, characterized in that The method for generating the reverse compensation path specifically includes: When the real-time distance value enters the 8 - 10-mm braking preparation area, the motion control unit activates a three-level response mechanism: The first-level response triggers path pre-calculation. Based on the component surface texture features collected by the binocular camera to match the BIM model data, three alternative reverse paths are generated, with the path lengths being 1.2 times, 1.5 times, and 1.8 times the real-time distance value respectively; The second-level response activates dynamic parameter adjustment. According to the load-bearing distribution data of the hydraulic support leg pressure sensors, when the pressure difference between the maximum load-bearing leg and the minimum load-bearing leg exceeds 15%, the reverse compensation amount automatically increases by 0.3 times the compensation coefficient of the real-time distance value; The third-level response executes the final path selection. The optimal path is selected based on the component deformation rate measured in real time by the laser rangefinder. When the deformation rate > 0.2 mm / ms, a 1.8-fold path is adopted. When the deformation rate ≤ 0.2 mm / ms, a 1.5-fold path is adopted; When the path planning unit generates the reverse compensation path, the path is decomposed into three segments of motion trajectories: The first 3 mm of travel uses an S-shaped acceleration curve, with the maximum acceleration limited to 0.15 m / s²; The middle 7 mm of travel maintains a constant speed, and the speed value is set to 25% - 40% of the current motion speed; The last 5 mm of travel uses an exponential decay deceleration curve, and the deceleration gradient is set to reduce the speed value by 35% per second; The visual recognition module synchronously analyzes the characteristics of the component fracture surface. When longitudinal cracks are detected on the fracture surface, the reverse compensation direction automatically deflects by an angle of 15° - 25° in the crack extension direction, and the deflection angle accuracy is controlled within the range of ±1°; The motion control unit synchronously collects the torque sensor data of each joint of the multi-axis robotic arm at a sampling frequency of 1 kHz. When it detects that the torque of any joint exceeds 80% of the rated value, the reverse compensation amount is immediately reduced to 1.2 times the real-time distance value; During the execution of the reverse compensation path, the annular spray system starts the pulsed cooling mode. The spray cycle is synchronized with the robotic arm motion cycle. Each compensation motion cycle contains 3 intermittent sprays, and the duration of a single spray is set to 50 ms, and the interval time is set to 20 ms; After the path planning unit completes the reverse compensation, it automatically performs a trajectory verification test. By comparing the actual displacement data of the laser rangefinder with the deviation value of the theoretical path, when the deviation exceeds 0.3 mm, a secondary compensation program is triggered, and the secondary compensation amount is calculated as 1.1 times the deviation value.

10. A method for controlling the demolition of building components, characterized in that, Including the following steps: Establish the equipment support reference through the adjustable hydraulic support legs of the base platform; Perform component demolition operations through the hydraulic shears at the end of the multi-axis robotic arm. The motion trajectory of the multi-axis robotic arm is generated by the path planning unit of the control module; Real-time collect the operation environment data through the image acquisition module at the front end of the base platform and transmit the collected data to the visual recognition module; Call the building component feature database through the visual recognition module for three-dimensional coordinate matching to generate a three-dimensional space model containing preset safety boundary data; The path planning unit of the control module receives the three-dimensional coordinate data output by the visual recognition module and generates the motion trajectory of the multi-axis robotic arm in combination with the building component feature database; When it detects that the real-time distance between the hydraulic shear and the preset safety boundary enters the 30-mm warning range, the control module linearly reduces the robotic arm motion speed to 30% - 50% of the current speed value, and the deceleration slope is set to reduce the original speed by 20% per second; When the real-time distance enters the 10-mm braking critical zone, the motion control unit immediately freezes the current robotic arm motion trajectory and generates a reverse compensation path based on the real-time spatial coordinates of the hydraulic shear. The length of the reverse compensation path is 1.5 times the real-time distance value, and the generation period does not exceed 5 milliseconds. The motion control unit drives the multi-axis robotic arm to move along the reverse compensation path while controlling the hydraulic shear to perform the shearing and demolition operation of the preset steel member. The following steps are synchronously executed during the generation of the reverse compensation path: The deformation rate of the member is monitored in real time by a laser rangefinder. When the deformation rate exceeds 0.2 mm / ms, the compensation path is automatically extended to 1.8 times the real-time distance value. The change rate of the shear stress of the hydraulic shear is detected by a strain gauge array. When the change rate exceeds 5 MPa / s, the direction of the compensation path is dynamically adjusted to form an angle of 15°-25° with the stress gradient direction. The curvature of the compensation path is corrected by the fracture surface feature data captured by the binocular camera to ensure that the deviation of the path from the normal direction of the member fracture surface does not exceed ±2°.

Images