Laser radar belt tearing detection method based on time sequence analysis and regression analysis

Abstract

The invention discloses a laser radar belt tearing detection method based on time sequence analysis and regression analysis, and belongs to the field of industrial detection and point cloud identification. Belt point cloud data are processed and analyzed based on the time sequence analysis and the regression analysis, and belt tearing detection is achieved. As the field environment is variable and is greatly interfered by light factors, the two-dimensional processing of the point cloud data is realized by utilizing a projection or rasterization method and the like aiming at the read point cloud frame; aiming at winged insects, floating ash and the like in the air, point cloud frame data caching is carried out, a sequential analysis algorithm is utilized to carry out front and back point cloud data comparison, and abnormal interference items are eliminated; and belt tearing detection is carried out by using a double regression analysis algorithm, so that the detection effect of hundreds of frames per second can be achieved, and real-time belt tearing monitoring and alarming are realized.

Description

Technical Field

[0001] The present invention belongs to the field of industrial detection and point cloud recognition, and in particular relates to a laser radar belt tear detection method based on time series analysis and regression analysis. Background Art

[0002] In the field of research and development of belt longitudinal tear monitoring devices, various solutions have been proposed both domestically and internationally. Existing belt tear detection equipment can be divided into two main categories: one uses mechanical devices, including linear detectors and leak detectors; the other uses computer vision technology, primarily including deep learning-based computer vision and visual inspection methods using line lasers and CCD cameras.

[0003] 1. Linear detector: This detector is installed below the trough belt. A metal wire or nylon line is stretched along the belt's contour, with a spring-loaded limit switch installed at one end. When material piercing the belt catches the line, it breaks or increases its tension, triggering the corresponding limit switch and stopping the conveyor.

[0004] 2. Leakage Detector: This detector consists of a tray, fulcrum, counterweight, and switch. When a longitudinal tear occurs in the belt, material on the belt leaks through the tear into the tray. The weight of the material overcomes the weight of the counterweight, causing the entire assembly to rotate about the fulcrum, forcing the limit switch to operate and shutting down the conveyor. This detector has a simple structure and is easy to detect. However, when a belt tear occurs, the detector will only detect material on the belt and the tear is large enough to leak material. Furthermore, excessive dust accumulation on the rollers can cause false operation.

[0005] 3. Computer vision-based deep learning detection: Deep learning is a machine learning technique that trains a model to identify tear signatures in belt images. This approach requires a large amount of labeled data—that is, images of belts known to contain tears. Once the model is trained, it can be used to detect tears in new belt images.

[0006] 4. Visual inspection based on line lasers and CCD cameras: This method is a common 3D visual inspection method. It projects a line laser onto the surface of an object, which is then captured by a CCD camera to obtain 3D surface information. For longitudinal tear detection in belts, this method provides more detailed and accurate surface information, thereby improving inspection accuracy.

[0007] In view of this, the present invention proposes a laser radar belt tear detection method based on time series analysis and regression analysis, which performs belt tear detection through laser radar point cloud analysis to solve the problems of changeable on-site environment of belt conveyor corridors and insufficient data sets. Summary of the Invention

[0008] The technical problem to be solved by the present invention is to provide a laser radar belt tear detection method based on time series analysis and regression analysis to address the shortcomings of the background technology. The belt point cloud data is processed and analyzed based on time series analysis and regression analysis to realize belt tear detection and realize real-time monitoring and alarm of belt tear.

[0009] The present invention adopts the following technical solutions to solve the above technical problems: A laser radar belt tear detection method based on time series analysis and regression analysis is used to process and analyze belt point cloud data based on time series analysis and regression analysis to achieve belt tear detection. The method specifically includes the following steps: Step 1: Obtain radar point cloud frame; Step 2: 2D point cloud data conversion: For the read radar point cloud frame, the point cloud data is converted into 2D using the projection method; Step 3: caching point cloud data; caching point cloud frame data for flying insects and floating dust in the air; Step 4: Time series analysis to remove interference frames: Use the time series analysis algorithm to compare the previous and next point cloud data and remove abnormal interference items; Step 5: Detect belt tearing by regression analysis: Utilize the dual regression analysis algorithm to detect belt tearing and realize real-time monitoring and alarm of belt tearing.

[0010] As a further preferred embodiment of the laser radar belt tear detection method based on time series analysis and regression analysis of the present invention, in step 2, the point cloud data is two-dimensionalized, and the three-dimensional point cloud is mapped to a two-dimensional plane using a projection method to achieve dimensionality reduction; the point cloud is selected to be projected onto a specific plane, including orthogonal projection, perspective projection, and spherical projection.

[0011] As a further preferred embodiment of the present invention's laser radar belt tear detection method based on time series analysis and regression analysis, orthogonal projection: projects a three-dimensional point perpendicularly onto a plane along a certain coordinate axis, ignoring the coordinate values perpendicular to the plane and retaining the other two coordinates as two-dimensional coordinates; wherein ground plane projection maps the points in the point cloud onto a two-dimensional coordinate system with the ground as the XY plane according to their coordinates on the ground, i.e., discarding the Z-axis coordinate; Perspective projection: The points in the point cloud are projected onto the camera's image plane by simulating the camera imaging principle, considering the effect of distance on the projection, defining the viewpoint position and projection plane; Spherical projection: project the point cloud onto a spherical surface and then expand it into a two-dimensional image; According to the on-site environment of the coal conveyor belt, the orthogonal projection method is selected to two-dimensionalize the point cloud data. The point cloud data is projected onto the XZ plane, retaining the cross-sectional direction and height direction information of the belt.

[0012] As a further preferred embodiment of the present invention's method for detecting belt tearing using a laser radar based on time series analysis and regression analysis, in step 4, a Mad algorithm is used to perform time series analysis, specifically comprising the following steps: Step 4.1, calculate the median of the data set : Sort the data by size and find the value in the middle, which is an odd number of data points, or the average of the two middle numbers, which is an even number of data points; Step 4.2, calculate the absolute deviation of each data point from the median: for each data point ,calculate ; Step 4.3, calculate the median of absolute deviation, that is, MAD value, for the data set , the calculation formula is: .

[0013] As a further preferred embodiment of the present invention's method for detecting belt tearing using a laser radar based on time series analysis and regression analysis, in step 5, polynomial regression is used for abnormality detection, specifically comprising the following steps: Step 5.1: The application of polynomial regression in curve anomaly detection is to identify abnormal data points that deviate from the expected pattern by establishing a mathematical model between variables. This method combines statistical modeling with machine learning and can effectively identify abnormal patterns in curve relationships while maintaining interpretability. Step 5.2, use polynomial regression to establish the functional relationship between the dependent variable and the independent variable. For the point cloud coordinate (X, Y) dataset and , the fitted regression model formula is: ;in, It is the coefficient of the regression model equation fitted according to the coordinate data (x, y); regression analysis is performed on the original point cloud data and the interpolated data respectively, and the residuals between the predicted value and the actual value are calculated. The abnormal points are identified by analyzing the residual distribution; the abnormal points are further analyzed, and single independent abnormal points are excluded, and continuous abnormal points are retained; and the results of the two types of data analysis are comprehensively judged. It is possible to avoid the interference of jump points while retaining the characteristics of the original data and realize accurate detection of belt tears.

[0014] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects: The present invention processes and analyzes belt point cloud data based on time series analysis and regression analysis to achieve belt tear detection. Due to the changeable field environment and the significant interference from light factors, the point cloud data is two-dimensionalized using projection or rasterization methods for the read point cloud frames. Point cloud frame data is cached to account for flying insects and floating dust in the air, and a time series analysis algorithm is used to compare the previous and next point cloud data to eliminate abnormal interference items. A double regression analysis algorithm is then used to detect belt tears, achieving a detection effect of 100 frames per second, realizing real-time monitoring and alarming of belt tears. Belt tear detection is performed based on high-frame rate lidar, which has high detection accuracy and sensitivity. It can accurately detect surface changes and irregularities of the belt and identify tiny distance changes. Even in the initial stage of belt tearing, detection and early warning can be performed. The detection effect of lidar is not affected by ambient light, dust, moisture, etc. Therefore, reliable detection can be achieved in various complex environments. LiDAR is a non-contact detection equipment that will not cause additional physical damage to the detected object, which helps to extend the service life of the belt. Through the automated data collection and analysis of lidar, the detection efficiency can be greatly improved, the investment in human resources can be reduced, and the user's usage cost can be reduced, thereby achieving the purpose of reducing costs, increasing efficiency and improving safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a flow chart of a method for detecting belt tearing using a laser radar based on time series analysis and regression analysis. DETAILED DESCRIPTION

[0016] The technical solution of the present invention is further described in detail below with reference to the accompanying drawings: The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. The present invention is described in detail below based on the drawings and preferred embodiments. The purpose and effect of the present invention will become more clear. It should be understood that the specific embodiments described here are only used to explain the present invention and are not used to limit the present invention.

[0017] Currently, most common belt tear detection methods rely on additional contact devices or sensors. Although existing tear detection technology has made some progress, it still has some technical defects, including: large errors, high missed detection rate, complex installation and debugging, high maintenance costs, and slow technology update and iteration.

[0018] A laser radar belt tear detection method based on time series analysis and regression analysis is used to process and analyze belt point cloud data to achieve belt tear detection. Figure 1 As shown, the specific steps include: Step 1: Obtain radar point cloud frame; Step 2: 2D point cloud data conversion: For the read radar point cloud frame, the point cloud data is converted into 2D using the projection method; Step 3: caching point cloud data; caching point cloud frame data for flying insects and floating dust in the air; Step 4: Time series analysis to remove interference frames: Use the time series analysis algorithm to compare the previous and next point cloud data and remove abnormal interference items; Step 5: Detect belt tearing by regression analysis: Utilize the dual regression analysis algorithm to detect belt tearing and realize real-time monitoring and alarm of belt tearing.

[0019] The key core algorithms of this invention are point cloud data 2D conversion, time series analysis, and regression analysis. The point cloud data 2D conversion is achieved using projection methods; time series analysis uses data analysis algorithms such as MAD to correct deviations, improving robustness against dust and rain; and regression analysis uses a polynomial regression model to identify tear anomalies.

[0020] 2D point cloud data: Projection methods map a 3D point cloud onto a 2D plane to achieve dimensionality reduction. You can choose to project the point cloud onto a specific plane, such as orthographic, perspective, or spherical projections.

[0021] Orthogonal projection: projects a 3D point perpendicularly onto a plane along a certain coordinate axis, ignoring the coordinate values perpendicular to the plane and retaining the other two coordinates as 2D coordinates. Ground plane projection maps the points in the point cloud onto a 2D coordinate system with the ground as the XY plane, i.e., discarding the Z-axis coordinate.

[0022] Perspective projection: Simulating the camera's imaging principle, taking into account the effect of distance on projection, defines the viewpoint position and projection plane, and projects the points in the point cloud onto the camera's image plane. This requires the camera's intrinsic and extrinsic matrix.

[0023] Spherical projection: Project the point cloud onto a spherical surface and then expand it into a two-dimensional image; commonly used methods include latitude and longitude projection, cube map projection, etc. The present invention selects to use the orthogonal projection method to two-dimensionalize the point cloud data according to the on-site environment of the coal conveyor belt, projects the point cloud data to the XZ plane, and retains the belt cross-sectional direction and height direction information.

[0024] Mad algorithm for timing analysis; MAD (Median Absolute Deviation) is a robust measure of dispersion, primarily used to assess the variability of a dataset. It is particularly useful for analyzing data with outliers. MAD measures the median of the absolute deviations of a data point from the median of the dataset. Its calculation steps are as follows: Calculate the median of a data set : Sort the data by size and find the value in the middle, which is an odd number of data points, or the average of the two middle numbers, which is an even number of data points; Calculate the absolute deviation of each data point from the median: ,calculate ; Calculate the median of absolute deviations, or MAD, for the data set , the calculation formula is: .

[0025] Polynomial regression for anomaly detection: The application of polynomial regression in curve anomaly detection is to identify abnormal data points that deviate from the expected pattern by establishing a mathematical model between variables. By combining statistical modeling with machine learning, it can effectively identify abnormal patterns in curve relationships while maintaining interpretability. It is particularly suitable for industrial detection scenarios with clear physical relationships.

[0026] Use polynomial regression to establish the functional relationship between the dependent variable and the independent variable. Use polynomial regression to establish the functional relationship between the dependent variable and the independent variable. For the point cloud coordinate (X, Y) dataset and , the fitted regression model formula is: ;in, It is the coefficient of the regression model equation fitted according to the coordinate data (x, y); regression analysis is performed on the original point cloud data and the interpolated data respectively, and the residuals between the predicted value and the actual value are calculated. The abnormal points are identified by analyzing the residual distribution; the abnormal points are further analyzed, and single independent abnormal points are excluded, and continuous abnormal points are retained; and the results of the two types of data analysis are comprehensively judged. It is possible to avoid the interference of jump points while retaining the characteristics of the original data and realize accurate detection of belt tears.

[0027] The present invention processes and analyzes belt point cloud data based on time series analysis and regression analysis to realize belt tear detection; due to the changeable on-site environment and the significant interference from light factors, the read point cloud frames are used to realize two-dimensionalization of point cloud data by projection or rasterization methods; for situations such as flying insects and floating dust in the air, the point cloud frame data is cached, and the time series analysis algorithm is used to compare the previous and next point cloud data to eliminate abnormal interference items; and then the double regression analysis algorithm is used to detect belt tears, which can achieve a detection effect of 100 frames per second and realize real-time monitoring and alarm of belt tears.

[0028] The present invention performs belt tear detection based on a high-frame rate laser radar with high detection accuracy and high sensitivity. It can accurately detect surface changes and irregularities of the belt and identify small distance changes. Therefore, even in the initial stage of belt tearing, detection and early warning can be performed. The detection effect of the laser radar is not affected by ambient light, dust, moisture, etc., so reliable detection can be achieved in various complex environments. The laser radar is a non-contact detection device that does not cause additional physical damage to the detected object, which helps to extend the service life of the belt. Through the automated data collection and analysis of the laser radar, the detection efficiency can be greatly improved and the investment in human resources can be reduced. The user's usage cost is reduced, achieving the purpose of reducing costs, increasing efficiency and improving safety.

[0029] Those skilled in the art will understand that the above descriptions are merely preferred embodiments of the invention and are not intended to limit the invention. Although the invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art will still be able to modify the technical solutions described in the aforementioned embodiments or substitute equivalents for some of the technical features. Any modifications, equivalent substitutions, etc. made within the spirit and principles of the invention shall be included within the scope of protection of the invention. All technical features in this embodiment may be freely combined according to actual needs.

[0030] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A laser radar belt tear detection method based on time series analysis and regression analysis, characterized by: The belt point cloud data is processed and analyzed based on time series analysis and regression analysis to achieve belt tear detection, which specifically includes the following steps: Step 1: Obtain radar point cloud frame; Step 2: 2D point cloud data conversion: For the read radar point cloud frame, the point cloud data is converted into 2D using the projection method; Step 3: point cloud data cache; Cache point cloud frame data for flying insects and floating dust in the air; Step 4: Time series analysis to remove interference frames: Use the time series analysis algorithm to compare the previous and next point cloud data and remove abnormal interference items; Step 5: Detect belt tearing by regression analysis: Utilize the dual regression analysis algorithm to detect belt tearing and realize real-time monitoring and alarm of belt tearing.

2. The laser radar belt tear detection method based on time series analysis and regression analysis according to claim 1 is characterized in that: In step 2, the point cloud data is two-dimensionalized, and the three-dimensional point cloud is mapped to a two-dimensional plane using the projection method to achieve dimensionality reduction; the point cloud is projected onto a specific plane, including orthogonal projection, perspective projection, and spherical projection.

3. The laser radar belt tear detection method based on time series analysis and regression analysis according to claim 2 is characterized in that: Orthogonal projection: projects a 3D point perpendicularly onto a plane along a certain coordinate axis, ignoring the coordinate values perpendicular to the plane and retaining the other two coordinates as 2D coordinates. Ground plane projection maps the points in the point cloud onto a 2D coordinate system with the ground as the XY plane, i.e., discarding the Z-axis coordinate. Perspective projection: The points in the point cloud are projected onto the camera's image plane by simulating the camera imaging principle, considering the effect of distance on the projection, defining the viewpoint position and projection plane; Spherical projection: project the point cloud onto a spherical surface and then expand it into a two-dimensional image; According to the on-site environment of the coal conveyor belt, the orthogonal projection method is selected to two-dimensionalize the point cloud data. The point cloud data is projected onto the XZ plane, retaining the cross-sectional direction and height direction information of the belt.

4. The laser radar belt tear detection method based on time series analysis and regression analysis according to claim 1 is characterized in that: In step 4, the Mad algorithm is used to perform timing analysis, which specifically includes the following steps: Step 4.1, calculate the median of the data set : Sort the data by size and find the value in the middle, which is an odd number of data points, or the average of the two middle numbers, which is an even number of data points; Step 4.2, calculate the absolute deviation of each data point from the median: for each data point ,calculate ; Step 4.3, calculate the median of absolute deviation, that is, MAD value, for the data set , the calculation formula is: .

5. The laser radar belt tear detection method based on time series analysis and regression analysis according to claim 1 is characterized in that: In step 5, double regression is used for anomaly detection, which specifically includes the following steps: Step 5.1: Use linear interpolation to process the original point cloud data to obtain interpolated point cloud data, which can eliminate the interference of jump points in the original radar data; Step 5.2: The application of polynomial regression in curve anomaly detection is to identify abnormal data points that deviate from the expected pattern by establishing a mathematical model between variables. This method combines statistical modeling with machine learning and can effectively identify abnormal patterns in curve relationships while maintaining interpretability. Step 5.3, use polynomial regression to establish the functional relationship between the dependent variable and the independent variable. For the point cloud coordinate (X, Y) dataset and , the fitted regression model formula is: ;in, It is the coefficient of the regression model equation fitted according to the coordinate data (x, y); regression analysis is performed on the original point cloud data and the interpolated data respectively, and the residuals between the predicted value and the actual value are calculated. The abnormal points are identified by analyzing the residual distribution; the abnormal points are further analyzed, and single independent abnormal points are excluded, and continuous abnormal points are retained; and the results of the two types of data analysis are comprehensively judged. It is possible to avoid the interference of jump points while retaining the characteristics of the original data and realize accurate detection of belt tears.

Images