Tower crane load size detection method and system based on multi-sensor fusion and tower crane

Abstract

The present application relates to the technical field of tower crane intelligent control, and specifically provides a tower crane object size detection method and system based on multi-sensor fusion and a tower crane, comprising the following steps: a laser radar obtains object point cloud data below a hook, and through and in combination with an advanced point cloud clustering algorithm and a point cloud filtering ranging technology, the planar size of the object and the distance of the highest point of the object from the radar are obtained; a weighing sensor is used to detect whether the object has left the ground, the planar size obtained by the laser radar and the data of the weighing sensor are fused and analyzed, and in combination with a hook height encoder, the height of the object is calculated; the collected data is fused in a central processor. The present application uses multi-sensor fusion technology to perform backend fusion on laser point cloud data, object weight data and tower crane motion state data, so as to improve the accuracy and comprehensiveness of object size detection.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology for tower cranes, and in particular to a method, system, and tower crane for detecting the size of loads lifted by a tower crane based on multi-sensor fusion. Background Technology

[0002] Tower cranes are critical pieces of equipment in the construction and engineering industries, used for the vertical transport and positioning of various heavy objects. In tower crane operation, accurate measurement of the dimensions of the load is essential for ensuring safe lifting operations. Accurate measurement of load dimensions is crucial for avoiding collisions, maximizing workspace utilization, reducing lifting accidents, and improving work efficiency.

[0003] Traditional methods typically rely on operator skills and experience, which can lead to subjective errors and operational mistakes, further impacting tower crane safety. An existing Chinese patent application (CN116958530A) for a method of detecting the dimensions of loads lifted by a tower crane uses a single camera with the YOLOv5 algorithm to identify the hook and a corresponding load identification method to calculate the load's dimensions. However, this method only detects the length and width of the load, ignoring its vertical height, which is a limitation and cannot fully meet the requirements for safe operation of automatic tower cranes.

[0004] Therefore, it is necessary to propose a more comprehensive and accurate method for detecting the dimensions of suspended objects to overcome the aforementioned limitations. This new method should be able to reduce the error of a single sensor and obtain complete dimensional information of the suspended object through multi-sensor fusion. By integrating data from different sensors, the size and shape of the suspended object can be better understood. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention develops a method, system, and tower crane for detecting the dimensions of objects lifted by a tower crane based on multi-sensor fusion. By utilizing multi-sensor fusion technology, laser point cloud data, weight data of the object being lifted, and motion status data of the tower crane are fused in the backend to improve the accuracy and comprehensiveness of the detection of the dimensions of the object being lifted.

[0006] The technical solution to the technical problem solved by the present invention is as follows: On the one hand, the present invention provides a 1. A method for detecting the dimensions of loads lifted by a tower crane based on multi-sensor fusion, characterized by comprising the following steps: S1. Install the lidar at an angle below the hook to acquire point cloud data of the suspended object below the hook. By combining point cloud clustering algorithm and point cloud filtering ranging technology, the planar dimensions of the suspended object and the distance from the highest point of the suspended object to the lidar can be obtained. S2. Use a load cell to detect whether the load has left the ground. Combine the planar dimensions obtained by the lidar with the data from the load cell for analysis. At the same time, combine the data with the hook height encoder to calculate the height of the load. S3. The collected data is fused in the central processing unit, including the spatial position coordinate transformation of the lidar, the lidar point cloud, the suspension status of the load from the weighing sensor, and the height information of the tower crane lifting encoder. The data is integrated through a data fusion algorithm to eliminate errors and output an accurate three-dimensional load size data model.

[0007] Based on the above-mentioned method for detecting the size of loads lifted by tower cranes using multi-sensor fusion, the specific processing procedure for the load point cloud data is as follows: (1) Radar coordinate transformation: The positional offset of the radar relative to the center point of the hook is obtained by manual measurement. In three-dimensional space, The translation matrix is ​​represented as ; Rotation matrix around the X-axis ; Rotation matrix around the Y-axis ; Rotation matrix around the Z-axis ; in, , , These represent the radar's translation in the X, Y, and Z axes, respectively, and must be obtained through manual measurement. , , The real-time output from the IMU mounted on the lidar reflects the current attitude angle of the lidar. The angle is output by the IMU on the lidar. A calibration board is placed in the acquisition area, and the position and orientation of the calibration board and lidar are obtained using a laser rangefinder. The error of the sensor is determined by comparing it with the output value of the IMU and applied to the calculation of the rotation matrix. The position and orientation of each point cloud point are converted into coordinates in the center coordinate system of the hook through translation and rotation matrices to correct the deflection of the point cloud. (2) Point cloud filtering: A point cloud filtering module is configured to obtain the relative position information of the suspension rope in the LiDAR point cloud, including angle and distance, based on the relative angle and distance between the LiDAR installation location and the suspension rope. Angle and distance thresholds are set respectively, and filtering conditions are defined to perform distance and angle filtering, thus removing the suspension rope's own point cloud data. (3) Inspection of the dimensions of the suspended object: By segmenting the height information of the laser point cloud, the suspended object is separated from other surfaces, and a height threshold is set. Point clouds below this threshold are classified as hanging object point cloud data. Point clouds exceeding this threshold are classified as other surface point cloud data. , , Represents the first point in the point cloud There are 1 point, where z represents the height value in the point cloud; (4) Point cloud data of suspended objects Use the calculation method to calculate the area of ​​the convex hull. The points on the convex hull are calculated from a subset of points selected from the point cloud. ,in These are the coordinates of a point on the convex hull. It is the number of points on the convex hull, used to determine whether the area of ​​the convex hull exceeds a threshold. If the threshold is exceeded, a hole exists on the suspended object, and a binary hole mask is generated. , in, This represents the discrete coordinate position of the mask matrix in the vertical direction. This represents the discrete coordinate position of the mask matrix in the horizontal direction; (5) Use a shape fitting algorithm to fit the above-mentioned cavity region with the point cloud of the suspended object to obtain the actual planar dimensions of the suspended object. And the plane height information Zh.

[0008] Based on the above-mentioned method for detecting the size of loads lifted by tower cranes using multi-sensor fusion, a median filtering algorithm is used to process the weight data from the weighing sensors. The specific implementation steps are as follows: (1) Define window size: Determine the window size N of the median filter. Select the window size according to the noise level of the collected weighing data. When the load leaves the ground or is suspended from the platform, it is a rapidly changing signal with less noise. Select a smaller window when the load is in the air and moves with the hook, it is a slowly changing signal that is affected by noise. Adjust the window size according to actual needs. (2) Sort window data: Traverse the weight data sequence, apply mean filtering to each data point, and select a data subsequence with a window size of N at each position i; (3) Calculate the median: Sort the data in the window and select the median value as the filtered value Wi; (4) Boundary handling: At the sequence boundary, the window is filled by repeating data points at the sequence edge; (5) Output median: This will yield the filtered sequence of the weight data of the suspended object; The weight change rate is calculated from the obtained filtered load weight data. Calculation: That is, the current weight Weight at the previous time point Difference divided by time interval By setting a threshold This indicates the allowable range of the rate of change when the load is off the ground. The threshold is adjusted according to the actual situation. When the rate of change of weight is lower than the set threshold, the load is considered to be off the ground, and a load-off flag is issued. .

[0009] Based on the above-mentioned method for detecting the size of loads lifted by tower cranes using multi-sensor fusion, Data fusion employs a backend fusion approach, which specifically includes the following process: defining a data structure. As part of the fusion output, use off-ground markers. Perform conditional judgment. When the ground clearance flag is true, the planar dimensions of the suspended object detected by the radar are used as the actual dimensions to calculate the maximum diameter of the suspended object. When the ground clearance flag is false, the system's default value is used to calculate the maximum diameter of the suspended object, where R is the maximum diameter of the suspended object. and Xm and Ym represent the actual planar dimensions of the suspended object detected by the lidar, respectively, while Xm and Ym are the default dimensions of the suspended object. When the ground clearance marker is true, the radar point cloud excluding the suspended object is used. The maximum value is taken as D. When the ground clearance flag is false, the default value of 1 is used, where D is the maximum distance between the lidar and the ground when the suspended object is off the ground. Finally, the fusion algorithm outputs a defined set of data. It is sent to the autonomous driving controller via serial port.

[0010] On the other hand, embodiments of the present invention provide a tower crane load size detection device based on multi-sensor fusion, which includes: The data acquisition module is used to acquire point cloud data of the load below the hook by scanning with lidar. Clustering and filtering module, point cloud clustering algorithm and point cloud filtering ranging technology, to obtain the planar dimensions of the suspended object and the distance of the highest point of the suspended object from the radar; The segmentation module segments the height information of the suspended object point cloud data, separating the suspended object from other surfaces. The cavity detection algorithm module calculates the cavity region and performs shape fitting with the point cloud of the suspended object to obtain the actual planar dimensions and planar height information of the suspended object. The median filtering algorithm module processes the weight data from the weighing sensor. The data fusion module determines the maximum diameter of the suspended object and the distance between the lidar and the ground when the object is off the ground, and outputs the data to the autonomous driving controller.

[0011] This invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement a method for detecting the size of objects lifted by a tower crane based on multi-sensor fusion.

[0012] This invention also provides a storage medium storing a computer program that, when executed by a processor, implements a method for detecting the size of objects lifted by a tower crane based on multi-sensor fusion.

[0013] This invention also provides a tower crane load size detection system based on multi-sensor fusion, comprising: The central processing unit is used to collect real-time data from various sensors, including spatial coordinate transformation based on lidar, laser point cloud processing, detection of load leaving the ground based on load cells, and height information from the tower crane lifting encoder. The lidar is installed directly below the hook and is connected to the central processor via Ethernet, enabling it to acquire point cloud data below the hook in real time. The load cell is installed in front of the tower arm and connected to the hoisting rope. The weight data from the load cell is processed using a median filtering algorithm. The lifting encoder, installed on the shaft of the lifting motor, automatically controls the tower crane's lifting motor via commands from the tower crane's automatic driving controller to lift objects and sends the current hook height in real time. To the central processing unit.

[0014] The effects described in the invention are merely those of the embodiments, and not all the effects of the invention. The above technical solutions have the following advantages or beneficial effects: By utilizing multi-sensor fusion technology, laser point cloud data, load weight data, and tower crane motion status data are fused in the backend to improve the accuracy and comprehensiveness of load size detection. Attached Figure Description

[0015] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0016] Figure 1 This is a structural diagram of the tower crane lifting object size detection system based on multi-sensor fusion according to the present invention.

[0017] Figure 2 This is a flowchart illustrating the specific processing of point cloud data for the present invention.

[0018] Figure 3 This is a flowchart of the multi-sensor data fusion process of the present invention.

[0019] Figure 4 This is a schematic diagram illustrating the radar identification of the dimensions of a suspended object according to the present invention.

[0020] Figure 5 This is a schematic diagram illustrating the radar identification of the dimensions of two suspended objects according to the present invention.

[0021] Figure 6 This is a schematic diagram illustrating the dimensions of three suspended objects identified by radar according to the present invention.

[0022] Figure 7 This is a schematic diagram illustrating the dimensions of four suspended objects identified by the radar in this invention. Detailed Implementation

[0023] To clearly illustrate the technical features of this solution, the invention will be described in detail below through specific embodiments and in conjunction with the accompanying drawings. The following disclosure provides many different embodiments or examples for implementing different structures of the invention. To simplify the disclosure of the invention, the components and arrangements of specific examples are described below.

[0024] Example 1 A method for detecting the dimensions of loads lifted by a tower crane based on multi-sensor fusion includes the following steps: S1. Install the lidar at an angle below the hook to acquire point cloud data of the suspended object below the hook. By combining point cloud clustering algorithm and point cloud filtering ranging technology, the planar dimensions of the suspended object and the distance from the highest point of the suspended object to the lidar can be obtained. S2. Use a load cell to detect whether the load has left the ground. Combine the planar dimensions obtained by the lidar with the data from the load cell for analysis. At the same time, combine the data with the hook height encoder to calculate the height of the load. S3. The collected data is fused in the central processing unit, including the spatial position coordinate transformation of the lidar, the lidar point cloud, the suspension status of the load from the weighing sensor, and the height information of the tower crane lifting encoder. The data is integrated through a data fusion algorithm to eliminate errors and output an accurate three-dimensional load size data model.

[0025] In this embodiment, the lidar 2 uses an area array radar as its sensing hardware platform and is connected to the central processing unit 1 via Ethernet. The lidar 2 is installed directly below the hook and can acquire point cloud data below the hook in real time. The lidar 2 has a built-in inertial measurement unit, which can acquire the real-time attitude information of the hook. Figure 2 The specific processing procedure for the point cloud data of the suspended object shown is as follows: (1) Radar coordinate transformation: The positional offset of the radar relative to the center point of the hook is obtained by manual measurement. In three-dimensional space, The translation matrix is ​​represented as ; Rotation matrix around the X-axis ; Rotation matrix around the Y-axis ; Rotation matrix around the Z-axis ; in, , , These represent the radar's translation in the X, Y, and Z axes, respectively, and must be obtained through manual measurement. , , The real-time output from the IMU mounted on the lidar reflects the current attitude angle of the lidar. The angle is output by the IMU on the lidar. A calibration board is placed in the acquisition area, and the position and orientation of the calibration board and lidar are obtained using a laser rangefinder. The error of the sensor is determined by comparing it with the output value of the IMU and applied to the calculation of the rotation matrix. The position and orientation of each point cloud point are converted into coordinates in the center coordinate system of the hook through translation and rotation matrices to correct the deflection of the point cloud. (2) Point cloud filtering: Based on the installation location of the LiDAR, the radar scans the suspension rope below the hook. A point cloud filtering module is designed to filter the point cloud on the suspension rope. This module uses a PCL CropBox filter. According to the relative angle distance between the LiDAR installation location and the suspension rope, angle thresholds and distance thresholds are set respectively, filtering conditions are defined, and distance filtering and angle filtering are performed to remove the point cloud of the suspension rope itself. (3) Inspection of the dimensions of the suspended object: By segmenting the height information of the laser point cloud, the suspended object is separated from other surfaces, and a height threshold is set. Point clouds below this threshold are classified as hanging object point cloud data. Point clouds exceeding this threshold are classified as other surface point cloud data. , , Represents the first point in the point cloud There are 1 point, where z represents the height value in the point cloud; (4) Point cloud data of suspended objects Use the calculation method to calculate the area of ​​the convex hull. The points on the convex hull are calculated from a subset of points selected from the point cloud. ,in These are the coordinates of a point on the convex hull. It is the number of points on the convex hull, used to determine whether the area of ​​the convex hull exceeds a threshold. If the threshold is exceeded, a hole exists on the suspended object, and a binary hole mask is generated. , ,in, This represents the discrete coordinate position of the mask matrix in the vertical direction. This represents the discrete coordinate position of the mask matrix in the horizontal direction; (5) Use a shape fitting algorithm to fit the above-mentioned cavity region with the point cloud of the suspended object to obtain the actual planar dimensions of the suspended object. And the plane height information Zh.

[0026] In this embodiment, the load cell 3 adopts the INS570D model lifting weighing system. The load cell 3 is installed in front of the tower arm and connected to the lifting rope. The weight data information from the load cell is processed using a median filtering algorithm. The specific implementation steps are as follows: (1) Define the window size: Determine the window size N of the median filter. Select the window size according to the noise level of the collected weighing data. When the load leaves the ground or is suspended from the platform, it is a rapidly changing signal with less noise. Select a smaller window. When the load is in the air and moves with the hook, it is a slowly changing signal with noise interference. Select a larger window. Adjust the window size according to actual needs. A smaller window size can be selected as 3 to 10 data points, and a larger window is generally 30-50 data points. (2) Sort window data: Traverse the weight data sequence, apply mean filtering to each data point, and select a data subsequence with a window size of N at each position i; (3) Calculate the median: Sort the data in the window and select the median value as the filtered value Wi; (4) Boundary handling: At the sequence boundary, the window is filled by repeating data points at the sequence edge; (5) Output median: This will yield the filtered sequence of the weight data of the suspended object; The weight change rate is calculated from the obtained filtered load weight data. Calculation: That is, the current weight Weight at the previous time point Difference divided by time interval By setting a threshold This indicates the allowable range of the rate of change when the load is off the ground. The threshold is adjusted according to the actual situation. When the rate of change of weight is lower than the set threshold, the load is considered to be off the ground, and a load-off flag is issued. .

[0027] refer to Figure 3 The data fusion adopts a backend fusion method, which specifically includes the following process: defining a data structure. As part of the fusion output, use off-ground markers. Perform conditional judgment. When the ground clearance flag is true, the planar dimensions of the suspended object detected by the radar are used as the actual dimensions to calculate the maximum diameter of the suspended object. When the ground clearance flag is false, the system's default value is used to calculate the maximum diameter of the suspended object, where R is the maximum diameter of the suspended object. and Xm and Ym represent the actual planar dimensions of the suspended object detected by the lidar, respectively, while Xm and Ym are the default dimensions of the suspended object. When the ground clearance marker is true, the radar point cloud excluding the suspended object is used. The maximum value is taken as D. When the ground clearance flag is false, the default value of 1 is used, where D is the maximum distance between the lidar and the ground when the suspended object is off the ground. Finally, the fusion algorithm outputs a defined set of data. It is sent to the autonomous driving controller via serial port.

[0028] Example 2 A tower crane load size detection device based on multi-sensor fusion includes: The data acquisition module is used to acquire point cloud data of the load below the hook by scanning with lidar. Clustering and filtering module, point cloud clustering algorithm and point cloud filtering ranging technology, to obtain the planar dimensions of the suspended object and the distance of the highest point of the suspended object from the radar; The segmentation module segments the height information of the suspended object point cloud data, separating the suspended object from other surfaces. The cavity detection algorithm module calculates the cavity region and performs shape fitting with the point cloud of the suspended object to obtain the actual planar dimensions and planar height information of the suspended object. The median filtering algorithm module processes the weight data from the weighing sensor. The data fusion module determines the maximum diameter of the suspended object and the distance between the lidar and the ground when the object is off the ground, and outputs the data to the autonomous driving controller.

[0029] Example 3 An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, when the processor executes the computer program, it implements the method for detecting the size of a tower crane load based on multi-sensor fusion as described in any one of claims 1 to 7.

[0030] Example 4 A storage medium storing a computer program that, when executed by a processor, implements a method for detecting the dimensions of objects lifted by a tower crane based on multi-sensor fusion.

[0031] Example 5 A tower crane load size detection system based on multi-sensor fusion includes: The central processing unit is used to collect real-time data from various sensors, including spatial coordinate transformation based on lidar, laser point cloud processing, detection of load leaving the ground based on load cells, and height information from the tower crane lifting encoder. The lidar is installed directly below the hook and is connected to the central processor via Ethernet, enabling it to acquire point cloud data below the hook in real time. The load cell is installed in front of the tower arm and connected to the hoisting rope. The weight data from the load cell is processed using a median filtering algorithm. The lifting encoder, installed on the shaft of the lifting motor, automatically controls the tower crane's lifting motor via commands from the tower crane's automatic driving controller to lift objects and sends the current hook height in real time. To the central processing unit.

[0032] Example 6 A tower crane includes the tower crane load size detection system.

[0033] Although the specific embodiments of the invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the invention. Based on the technical solutions of the invention, various modifications or variations that can be made by those skilled in the art without creative effort are still within the scope of protection of the invention.

[0034] In summary, using the lidar provided by this invention to detect the planar dimensions of suspended objects, the detection error is less than 20cm; using the multi-sensor fusion calculation method provided by this invention to calculate the height of suspended objects, the height error is less than 30cm; using the lidar point cloud processing method to remove the ground and suspension rope, the accuracy rate for identifying suspended objects below the hook is 90%. Figure 4 Figures 5, 6, and 7 show one, two, three, and four objects directly below the hook, respectively, detected by the radar. The detection method of this invention can meet the characteristics of tower cranes lifting a wide variety of materials with irregular sizes and shapes.

Claims

1. A method for detecting the dimensions of loads lifted by a tower crane based on multi-sensor fusion, characterized in that, Includes the following steps: S1. Install the lidar at an angle below the hook to acquire point cloud data of the suspended object below the hook. By combining point cloud clustering algorithm and point cloud filtering ranging technology, obtain the planar dimensions of the suspended object and the distance from the highest point of the suspended object to the lidar. S2. Use a load cell to detect whether the load has left the ground, fuse and analyze the planar dimensions obtained by the lidar and the data from the load cell, and calculate the height of the load by combining the data from the hook height encoder. S3. The collected data is fused in the central processing unit. The collected data includes the spatial position coordinate transformation of the lidar, the lidar point cloud, the suspension status of the load from the weighing sensor, and the height information of the tower crane lifting encoder. The data is integrated and error is eliminated through the data fusion algorithm, and an accurate three-dimensional load size data model is output. The data fusion algorithm employs a backend fusion method, specifically... This includes the following process: defining a data structure As part of the fusion output, use off-ground markers. Perform conditional judgment. When the ground clearance flag is true, the planar dimensions of the suspended object detected by the radar are used as the actual dimensions to calculate the maximum diameter of the suspended object. When the ground clearance flag is false, the system's default value is used to calculate the maximum diameter of the suspended object, where R is the maximum diameter of the suspended object. and These are the actual planar dimensions of the suspended object detected by the lidar. and These are the default dimensions for the suspended object. When the ground clearance marker is true, the radar point cloud excluding the suspended object is used. The maximum value is taken as D. When the ground clearance flag is false, the default value of 1 is used, where D is the maximum distance between the lidar and the ground when the suspended object is off the ground. Finally, the fusion algorithm outputs a defined set of data. It is sent to the autonomous driving controller via serial port.

2. The method for detecting the size of a tower crane load based on multi-sensor fusion according to claim 1, characterized in that, The specific processing procedure for the point cloud data of the suspended object is as follows: (1) Radar coordinate transformation: The positional offset of the radar relative to the center point of the hook is obtained by manual measurement. In three-dimensional space, The translation matrix is ​​represented as ; Rotation matrix around the X-axis ; Rotation matrix around the Y-axis ; Rotation matrix around the Z-axis ; in, , , These represent the radar's translation in the X, Y, and Z axes, respectively, and must be obtained through manual measurement. , , The real-time output from the IMU mounted on the lidar reflects the current attitude angle of the lidar. The angle is output by the IMU on the lidar. A calibration board is placed in the acquisition area, and the position and orientation of the calibration board and lidar are obtained using a laser rangefinder. The error of the sensor is determined by comparing it with the output value of the IMU and applied to the calculation of the rotation matrix. The position and orientation of each point cloud point are converted into coordinates in the center coordinate system of the hook through translation and rotation matrices to correct the deflection of the point cloud. (2) Point cloud filtering: The point cloud filtering module is set up to obtain the relative position information of the suspension rope in the LiDAR point cloud based on the relative angle and distance between the LiDAR installation position and the suspension rope. This includes the angle and distance. Based on the angle and distance, angle thresholds and distance thresholds are set respectively, filtering conditions are defined, and distance filtering and angle filtering are performed to clear the point cloud of the suspension rope itself. (3) Inspection of the dimensions of the suspended object: By segmenting the height information of the laser point cloud, the suspended object is separated from other surfaces, and a height threshold is set. Point clouds below this threshold are classified as hanging object point cloud data. Point clouds exceeding this threshold are classified as other surface point cloud data. , , Represents the first point in the point cloud There are 1 point, where z represents the height value in the point cloud; (4) Point cloud data of suspended objects Use the calculation method to calculate the area of ​​the convex hull. The points on the convex hull are calculated from a subset of points selected from the point cloud. ,in These are the coordinates of a point on the convex hull. It is the number of points on the convex hull, used to determine whether the area of ​​the convex hull exceeds a threshold. If the threshold is exceeded, a hole exists on the suspended object, and a binary hole mask is generated. , ,in, This represents the discrete coordinate position of the mask matrix in the vertical direction. This represents the discrete coordinate position of the mask matrix in the horizontal direction; (5) Use a shape fitting algorithm to fit the above-mentioned cavity region with the point cloud of the suspended object to obtain the actual planar dimensions of the suspended object. And the plane height information Zh.

3. The method for detecting the size of a tower crane load based on multi-sensor fusion according to claim 1, characterized in that, The weight data from the weighing sensor is processed using a median filtering algorithm. The specific implementation steps are as follows: (1) Define window size: Determine the window size N of the median filter. Select the window size according to the noise level of the collected weighing data. When the load leaves the ground or is suspended from the platform, it is a rapidly changing signal with less noise. Select a smaller window when the load is in the air and moves with the hook, it is a slowly changing signal that is affected by noise. Adjust the window size according to actual needs. (2) Sort window data: Traverse the weight data sequence, apply mean filtering to each data point, and select a data subsequence with a window size of N at each position i; (3) Calculate the median: Sort the data in the window and select the median value as the filtered value Wi; (4) Boundary handling: At the sequence boundary, the window is filled by repeating data points at the sequence edge; (5) Output median: This will yield the filtered sequence of the weight data of the suspended object; The weight change rate is calculated from the obtained filtered load weight data. Calculation: Current weight Weight at the previous time point Difference divided by time interval By setting a threshold This indicates the allowable range of the rate of change when the load is off the ground. The threshold is adjusted according to the actual situation. When the rate of change of weight is lower than the set threshold, the load is considered to be off the ground, and a load-off flag is issued. .

4. A tower crane load size detection device for implementing the tower crane load size detection method based on multi-sensor fusion according to any one of claims 1-3, characterized in that, include: The data acquisition module is used to acquire point cloud data of the load below the hook by scanning with lidar. Clustering and filtering module, point cloud clustering algorithm and point cloud filtering ranging technology, to obtain the planar dimensions of the suspended object and the distance of the highest point of the suspended object from the radar; The segmentation module segments the height information of the suspended object point cloud data, separating the suspended object from other surfaces. The cavity detection algorithm module calculates the cavity region and performs shape fitting with the point cloud of the suspended object to obtain the actual planar dimensions and planar height information of the suspended object. The median filtering algorithm module processes the weight data from the weighing sensor. The data fusion module determines the maximum diameter of the suspended object and the distance between the lidar and the ground when the object is off the ground, and outputs the data to the autonomous driving controller.

5. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that... When the processor executes the computer program, it implements the method for detecting the size of a tower crane load based on multi-sensor fusion as described in any one of claims 1 to 3.

6. A storage medium having a computer program stored thereon, characterized in that... When the program is executed by the processor, it implements the method for detecting the size of a tower crane load based on multi-sensor fusion as described in any one of claims 1 to 3.

7. A tower crane load size detection system for implementing the multi-sensor fusion-based tower crane load size detection method according to any one of claims 1-3, characterized in that it comprises: The central processing unit is used to collect real-time data from various sensors, including spatial coordinate transformation based on lidar, laser point cloud processing, detection of load leaving the ground based on load cells, and height information from the tower crane lifting encoder. The lidar is installed directly below the hook and is connected to the central processor via Ethernet, enabling it to acquire point cloud data below the hook in real time. The load cell is installed in front of the tower arm and connected to the hoisting rope. The weight data from the load cell is processed using a median filtering algorithm. The lifting encoder, installed on the shaft of the lifting motor, automatically controls the tower crane's lifting motor via commands from the tower crane's automatic driving controller to lift objects and sends the current hook height in real time. To the central processing unit.

8. A tower crane, comprising the tower crane load size detection system of claim 5.

Images