Sand and gravel aggregate conveying abnormal supervision and management system based on industrial vision analysis

Abstract

The application discloses a sand and gravel aggregate conveying abnormality supervision and management system based on industrial visual analysis and relates to the field of industrial conveying belt abnormality monitoring.The system comprises the following modules: a sensing and automatic control module that collects sensor information to sense the starting state of the conveying belt and automatically triggers image collection;an image recognition module that performs real-time processing based on the surface image of the conveying belt, extracts characteristic quantities and performs tear area screening; a tear size judgment module that estimates the length, width and depth of the tear area, triggers sound-light alarm and system interlock control according to a preset threshold; a positioning and automatic stop belt module that controls the accurate stop of the conveying belt according to the position coordinates of the tear area; and a data management module that automatically stores tear area data and generates a record set for report and visual display.Through high-robustness texture feature extraction, topological structure discrimination and image coordinate mapping, the tear recognition accuracy is improved, false positives and false negatives are reduced, and accurate positioning and automatic stop control of the tear area are realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of industrial conveyor belt anomaly monitoring, in particular to a sand and gravel aggregate conveying anomaly supervision management system based on industrial visual analysis. BACKGROUND

[0002] In the industrial processes of sand and gravel aggregate production and transportation, the conveyor belt as a key material conveying equipment, its stable operation is of great significance to ensure the overall efficiency and safety of the system. However, due to factors such as material impact, equipment aging, and foreign matter inclusion, the conveyor belt is prone to tearing damage during long-term operation. Once such defects occur, it may affect the conveying efficiency, cause the conveyor belt to break, the equipment to stop, and even cause safety accidents, bringing significant risks and economic losses to production.

[0003] Currently, the detection of conveyor belt tearing mainly relies on manual inspection and partial physical sensors for anomaly monitoring. Manual inspection is limited by high labor costs, high work intensity, detection delay and other problems, and it is difficult to achieve real-time and effective supervision of long-distance conveyor belts. Physical sensing methods such as steel wire rope sensing and electromagnetic detection can achieve a certain degree of automatic monitoring, but they have poor adaptability to non-metallic conveyor belts or surface tearing areas, and are easily affected by dust, vibration and other complex industrial site environments, resulting in high false alarm rate and difficulty in accurately assessing the actual size and severity of the tear.

[0004] In recent years, computer vision technology has been widely concerned and applied in the field of industrial monitoring. Through image acquisition devices to shoot the surface of the conveyor belt, and combined with image processing algorithms to identify and evaluate cracks, damage and other features, it has become an important means to improve the intelligent level of detection. Some existing systems attempt to use image recognition algorithms to detect conveyor belt defects, but still have the following problems: first, there is a lack of robust processing of image noise and complex background, resulting in insufficient tearing detection accuracy; second, the quantification ability of tearing size, especially depth and other three-dimensional deformation features, is limited, making it difficult to support accurate fault classification and response decision; third, most systems cannot be linked with the conveyor belt control system, lack precise positioning and automatic shutdown capabilities for tearing areas, and cannot effectively intercept and intervene before accidents.

[0005] In summary, the existing technology still has significant shortcomings in intelligent perception, image recognition accuracy, size evaluation accuracy, and control linkage response of conveyor belt tearing anomalies, and there is an urgent need to develop a more efficient, accurate, and intelligent anomaly supervision management system to meet the high-reliability operation requirements of conveying equipment in complex industrial environments. SUMMARY

[0006] Based on the shortcomings of the existing technology described above, the purpose of the present application is to provide a sand and gravel aggregate conveying anomaly supervision management system based on industrial visual analysis to solve the above technical problems.

[0007] To achieve the above object, the present application provides the following technical solutions: a sand and gravel aggregate conveying abnormal supervision and management system based on industrial visual analysis, comprising: a perception and automatic control module, an image recognition module, a tear size judgment module, a positioning and automatic stop belt module, and a data management module.

[0008] The perception and automatic control module: through the sensor information of the conveying belt, the real-time perception of the starting state of the conveying belt is realized, and the image acquisition is automatically carried out according to the starting state.

[0009] The image recognition module: based on the collected conveying belt surface image, the real-time processing is carried out, the feature quantity is extracted, and the tear area is screened.

[0010] The tear size judgment module: the length, width and depth size of the identified tear area are estimated, and the sound-light alarm and system interlock control are triggered according to the preset threshold.

[0011] The positioning and automatic stop belt module: the position coordinates of the tear area are linked to control the accurate stop of the conveying belt.

[0012] The data management module: the tear area data is automatically stored, a trigger record set is generated, and a report or a visual interface is generated.

[0013] The present application further provides that the perception and automatic control module comprises:

[0014] The real-time data of the sensor linked to the motor end of the conveying belt is collected to judge the running state of the conveying belt.

[0015] Through the linkage control of PLC and frequency converter, when it is detected that the running state of the conveying belt is running, a trigger signal is transmitted to the image acquisition module, the integrated dust removal equipment cleans the lens, and after cleaning, the lens starts to collect the original image data of the conveying belt surface.

[0016] The present application further provides that the image recognition module comprises: an image preprocessing unit, a feature extraction unit, a defect discrimination unit and an abnormal classification storage unit.

[0017] The feature extraction unit: based on the gradient intensity, local tension response factor and edge closure enhancement function, the tear response map of the conveying belt image obtained through the image preprocessing unit is generated, the difference analysis of multi-direction Gabor filter response and the comparison calculation of symmetric direction reflection image are carried out, and the microscopic texture disturbance map is generated combined with the direction confidence weight factor.

[0018] The defect discrimination unit: a structure / topology double threshold strategy is adopted, the structure consistency score is calculated combined with the Euler characteristic number and the morphological compactness index, the region screening is carried out to generate the high confidence tear area.

[0019] The application is further provided, characterized in that the feature extraction unit comprises: a tearing response map generation and a micro-texture disturbance map generation;

[0020] The tearing response map generation constructs a local tension response factor through the variability of the gradient direction in the unit field;

[0021] A closure factor is obtained by using the local edge direction consistency and the local convenient closure score, and an edge closure enhancement function is obtained by normalizing the closure factor;

[0022] The tearing response map is constructed by combining the local tension response factor and the edge closure enhancement function.

[0023] The application is further provided, and the micro-texture disturbance map generation obtains a direction confidence weight factor by calculating the deviation of the local gradient amplitude in each direction and the main direction of the local direction field;

[0024] The micro-texture disturbance map is calculated based on the difference between the multi-direction Gabor filter response and the symmetric direction reflection image Gabor filter response, and the direction confidence weight factor.

[0025] The application is further provided, and the defect discrimination unit comprises: candidate region screening, topological consistency calculation, tearing region discrimination;

[0026] The candidate region screening determines that the conveyor belt image region belongs to the candidate tearing region by judging that the tearing response intensity and the micro-texture disturbance intensity of the conveyor belt image region exceed the set threshold value;

[0027] The topological consistency calculation generates a shape compactness index by calculating the perimeter-to-area ratio of the candidate tearing region, and generates a structure consistency score by combining the Euler characteristic number of the local region and the shape compactness index;

[0028] The tearing region discrimination determines whether the shape consistency factor is less than the set threshold value, and if not, it is determined as a high-confidence tearing region.

[0029] The application is further provided, and the tearing size judgment module comprises: a tearing size calculation unit, a tearing depth calculation unit, a tearing severity evaluation and response triggering unit;

[0030] The tearing size calculation unit: converts the conveyor belt image pixel coordinates into actual tearing length and width size based on the main direction projection integral of the tearing response map and in combination with the camera calibration parameters;

[0031] The tearing depth calculation unit: quantifies the three-dimensional deformation characteristics of the tearing region by fusing the texture residual, gradient consistency and edge sharpness to obtain the tearing depth;

[0032] The tear severity evaluation and response triggering unit generates a severity score based on a nonlinear weighted scoring model of the length-width-depth multidimensional parameters, and compares the score with a set threshold to obtain a tear grade.

[0033] The application further provides that the positioning and automatic stopping module comprises a tear region center positioning unit, a coordinate mapping unit and a stopping control unit.

[0034] The tear region center positioning unit calculates the response intensity weighted centroid coordinates of the tear region based on the tear region mask and the tear response map.

[0035] The coordinate mapping unit converts the response intensity weighted centroid coordinates into a longitudinal absolute position in the physical coordinate system of the conveyor belt through a pulse encoder signal synchronized with the conveyor belt.

[0036] The stopping control unit generates a stopping instruction for the tear region greater than the set threshold according to the tear severity score and the corresponding rules, so that the conveyor belt can be accurately stopped when the defect region reaches the set position.

[0037] The application further provides that the data management module records each triggering event and adds the data generated in the triggering process to a triggering record set, which comprises the triggering time, the severity score, the tear length, the tear width, the tear depth, the response intensity weighted centroid coordinates and the tear image.

[0038] The application provides a sand and gravel aggregate conveying abnormality supervision and management system based on industrial vision analysis. The system comprises a perception and automatic control module, an image recognition module, a tear size judgment module, a positioning and automatic stopping module and a data management module. The perception and automatic control module realizes real-time perception of the starting state of the conveyor belt through the sensor information of the conveyor belt and automatically performs image acquisition according to the starting state. The image recognition module performs real-time processing based on the collected conveyor belt surface image, extracts characteristic quantities and performs tear region screening. The tear size judgment module estimates the length, width and depth of the identified tear region, triggers an audible and light alarm and system interlocking control according to the preset threshold. The positioning and automatic stopping module controls the accurate stopping of the conveyor belt according to the position coordinates of the tear region. The data management module automatically stores the tear region data and generates a triggering record set to generate a report or a visual interface.

[0039] The image recognition accuracy and anti-interference ability are improved by using a tear response map construction method that fuses local tension response factors and edge closing enhancement functions, and introducing multi-direction Gabor filtering and direction confidence weight analysis to enhance the extraction robustness of the texture disturbance features of the tear region and improve the tear detection accuracy in complex backgrounds.

[0040] A joint discrimination mechanism of topological structure and morphological index is introduced: the structural consistency score is constructed by combining the Euler characteristic number and the morphological compactness degree, and the structure / response double threshold decision logic is formed by cooperating with the tearing response map intensity, so as to realize the accurate screening of the high confidence tearing area and effectively reduce the false positive and false negative probability.

[0041] The spatial positioning and accurate stop control of the tearing area are realized: through the mapping of the tearing area image coordinates and the conveyor belt encoder signal, the actual position of the defect in the conveyor belt physical coordinate system is accurately positioned, and the stop control logic is combined to automatically stop the belt when the tearing area reaches the preset position, which significantly improves the real-time and accuracy of fault interception.

[0042] The above description is only a summary of the technical solutions of the present application. In order to more clearly understand the technical means of the present application, the specific embodiments of the present application can be implemented according to the content of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the following specific embodiments of the present application are described. BRIEF DESCRIPTION OF DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creating laborious work.

[0044] In the drawings:

[0045] Figure 1 The structural schematic diagram of the sand and gravel aggregate conveying abnormal supervision and management system based on industrial visual analysis is shown for an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0046] The embodiments of the present application will be described below with reference to the drawings and preferred embodiments, and those skilled in the art can easily understand other advantages and effects of the present application from the content disclosed in the specification. The present application can also be implemented or applied by different specific embodiments, and the details in the specification can be modified or changed based on different views and applications without departing from the spirit of the present application. It should be understood that the preferred embodiments are only for illustration of the present application, and are not intended to limit the protection scope of the present application.

[0047] It should be noted that the diagrams provided in the following embodiments only illustrate the basic concept of the present application in a schematic manner, and only the components related to the present application are shown in the diagrams, not the number, shape and size of the components during actual implementation. The actual implementation of each component may be a random change, and the component layout pattern may be more complex.

[0048] In the following description, numerous specific details are discussed in order to provide a thorough understanding of embodiments of the application. However, it will be apparent to one of ordinary skill in the art that embodiments of the application can be practiced without these specific details. In other instances, well-known structures and devices are not described in detail in order to avoid obscuring embodiments of the application.

[0049] Embodiments

[0050] The sand and gravel aggregate conveying abnormal supervision and management system based on industrial vision analysis, as shown in the figure, comprises: Figure 1

[0051] The perception and automatic control module: the real-time perception of the starting state of the conveying belt is realized by collecting the sensor information of the conveying belt, and the image collection is automatically carried out according to the starting state;

[0052] The image recognition module: based on the collected conveying belt surface image, the real-time processing is carried out, the characteristic quantity is extracted, and the tearing area is screened;

[0053] The tearing size judgment module: the length, width and depth size of the identified tearing area are estimated, and the sound-light alarm and system interlock control are triggered according to the preset threshold value;

[0054] The positioning and automatic stop belt module: the accurate stop belt of the conveying belt is linkage controlled according to the position coordinates of the tearing area;

[0055] The data management module: the tearing area data is automatically stored, the trigger record set is generated, and the report or visual interface is generated.

[0056] The application is further provided with the perception and automatic control module comprising:

[0057] The running state of the conveying belt is judged by collecting the real-time data of the sensor linkage with the motor end of the conveying belt;

[0058] ​Through linkage control of the PLC and the frequency conversion controller, when it is detected that the running state of the conveying belt is in running, a transmission trigger signal is transmitted to the graphic acquisition module, the integrated dust removal equipment carries out lens cleaning, and after the cleaning is completed, the lens starts to collect original image data of the conveying belt surface. Specifically, the motor speed information is collected through the sensor installed at the shaft end of the conveying belt driving motor, if the motor speed exceeds the set threshold and the time exceeds one second, the running state of the conveying belt is changed to running state 1, if the motor speed is less than the set threshold and the time exceeds three seconds, the running state of the conveying belt is changed to stop state 0; based on the ModbusRTU / TCP protocol, a real-time communication link is established with the frequency conversion controller, when the PLC receives the running signal and detects that the running state is 1, the pulse type air curtain dust removal device is started for 50 ms, then delayed for 100 ms to wait for dust settlement, and finally a trigger signal is sent to the industrial camera to start automatic collection of the conveying belt surface image.

[0059] The application is further provided that the image recognition module comprises an image preprocessing unit, a feature extraction unit, a defect discrimination unit and an abnormality classification storage unit.

[0060] The feature extraction unit: generates a tearing response map based on gradient intensity, local tension response factor and edge closure enhancement function for the conveying belt image obtained through the image preprocessing unit, generates a microscopic texture disturbance map through comparison calculation of multi-direction Gabor filter response difference analysis and symmetric direction reflection image, and combines a direction confidence weight factor.

[0061] The defect discrimination unit adopts a structure / topology double threshold strategy, combines Euler characteristic number and morphological compactness index to calculate structure consistency score, and performs regional screening to generate high confidence tearing region.

[0062] The feature extraction unit includes: tearing response map generation and micro-texture disturbance map generation.

[0063] The tearing response map generation constructs a local tension response factor through the variability of the gradient direction in the unit field.

[0064] A closure factor is obtained by using local edge direction consistency and local convenient closure score, and the closure factor is normalized to obtain an edge closure enhancement function.

[0065] The tearing response map is constructed by combining the local tension response factor and the edge closure enhancement function. Specifically, the local tension factor is used to reflect the change degree of the image texture direction in the local region, simulates the "material tension distortion" caused by tearing, and the tearing region is often accompanied by direction field disturbance, so the local tension factor can be calculated according to the "variability" of the gradient direction. wherein, is a local tension factor; is a gradient direction, used to reflect the local texture orientation, which is obtained by calculating the first-order gradient of the image through Sobel, Prewitt, Scharr operator, and then calculating the gradient direction of the first-order gradient of the image through the four-quadrant inverse tangent function; is a standard deviation of the gradient direction in the local region, indicating the degree of direction dispersion, and the calculation method is a prior art which will not be described here; is a coefficient for adjusting the influence of direction variation on the response factor, the greater the value, the more sensitive the influence of direction disturbance, and the value range is , and the closure factor is used to strengthen the response of the structure of the tear edge which is a "shape close to closed" structure, which can effectively suppress the interference of background edges or stray textures; the specific calculation logic is: wherein, is a closure factor; is a boundary closure score, used to reflect whether the tear edge constitutes a closed structure, which is obtained by estimating through edge connectivity or boundary tracking method, and the value range is between , 1 indicates complete closure, and 0 indicates that the edge is broken and does not have closure, which is a prior art and will not be described here; is an edge direction consistency, indicating the degree of concentration of the edge direction in the local region, used to determine whether the tear structure has a coherent edge orientation, and the value range is between , 1 indicates that the edge direction is highly consistent, and 0 indicates that the edge direction is chaotic and disordered, and the specific calculation formula is: , is the number of effective edge pixels in the field, which is calculated by the image edge detection algorithm, is a gradient direction, is a unit vector in complex form, indicating the direction, is a central field window. The edge closure enhancement function can strengthen the structural response characteristics of the "shape tends to be closed" in the tear edge, effectively suppress the interference of background edges or stray textures while highlighting the tear shape, and the value range is ; the edge closure enhancement function calculation logic is constructed by normalizing the closure factor : wherein, is an edge closure enhancement function, is a weight coefficient of the closure degree amplification effect on the tear response, the higher the closure degree, the more obvious the enhancement effect, and the non-tear edge is suppressed, and the value range is between 0 and 1, which can be modified according to the actual scene. The purpose of generating the tearing response map is to extract the possible tearing position through the gradient change, texture disturbance and edge feature of the local area of the image. This process combines the physical texture disturbance mechanism, such as local tension distortion and geometric edge morphology, such as crack edge closure, by constructing the tearing response map , highlighting the significance of the tearing area in the visual feature map, realizing the preliminary screening of the tearing candidate area; the specific calculation logic is: , wherein, is the tearing response map; is the overall scaling coefficient, used to adjust the numerical range of the response map intensity, the value range is between , and is usually set to ; is the current pixel direction confidence factor, which is obtained by the maximum direction of the Gabor filter response, which is the prior art and will not be described here; is the image global average direction, which is obtained by calculating the sum of the direction confidence factors of all valid pixel points in the image and taking the average value; is the gradient amplitude, which represents the gradient intensity of the current pixel point and reflects the edge strength of the image, which is calculated by Sobel, Scharr or Prewitt operator, which is the prior art and will not be described here; is the local tension factor; is the edge closure enhancement function.

[0066] The application further provides that the micro texture disturbance map is generated by calculating the deviation of the local gradient amplitude in each direction from the main direction of the local direction field to obtain the direction confidence weight factor;

[0067] Based on the difference between the multi-direction Gabor filter response and the symmetric direction reflected image Gabor filter response, the micro texture disturbance map is calculated combined with the direction confidence weight factor. Specifically, the direction confidence weight factor is a value for measuring the direction consistency of a pixel in a specific direction in the image, which is obtained by calculating the difference between the local gradient change of the pixel in the direction and the global main direction; the specific calculation formula is: , wherein, is the direction confidence weight factor, is the direction, indicating the Gabor filter direction; is the pixel in the i-th The local gradient direction value is obtained by calculating the gradient of each pixel in the conveyor belt image using classic edge detection algorithms such as the Sobel operator, Prewitt operator, or Roberts operator. Then, the response of a Gabor filter in different directions provides local feature information in that direction. Based on this information, the gradient direction is obtained by calculating the angle of gradient change within the local neighborhood. After calculating the local gradient direction of the image in a certain direction... This refers to the local gradient direction in that direction. This method is existing technology and will not be elaborated on here. The average principal direction of the current pixel's neighborhood represents the central tendency of the local directional distribution. It is obtained through the structure tensor or principal direction histogram, which are existing techniques and will not be elaborated upon here. Microscopic texture perturbation map. This image is calculated based on the differences in Gabor filter responses between the original image and the reflected image in various directions. It generates a map reflecting texture perturbation by measuring the degree of texture symmetry disruption in different directions; the specific calculation formula is as follows: ,in, This is a micro-texture perturbation map. The number of directional filters is the number of Gabor filter directions set. The default value is 8. Setting it to 4 results in high efficiency but poor resolution. Setting it to 16 results in detailed peaks but high computational cost. The more directions, the more accurately the asymmetric texture disturbances caused by tearing can be captured. The specific number can be adjusted according to actual needs. For conveyor belt images in the first The Gabor response in each direction is obtained through the image. With the The result of convolving the filter kernels of Gabor filters in each direction is existing technology and will not be elaborated further here; The Gabor response in the k-th direction of the symmetrical image is obtained by mirroring the conveyor belt image and then processing it with a filter in the same direction. The structure is similar, but the difference lies in the fact that the conveyor belt image needs to be mirrored first. Mirroring is achieved by subtracting the image's length and width from its current position for each pixel, obtaining the mirror position of each pixel. These mirrored pixels are then combined to obtain a complete mirrored conveyor belt image. Finally, the position of the first pixel in the mirrored conveyor belt image is calculated. The Gabor response in each direction yields the Gabor response in the k-th direction of the final symmetrical image; The symmetry perturbation term is used to directly quantify the degree of symmetry disruption. It can measure the structural damage to the texture caused by tearing. The larger the difference, the greater the probability that it is a torn region. Finally, by summing the values ​​for each direction in combination with the directional confidence weight factor, a map reflecting the micro-texture perturbation is obtained. The higher the value obtained, the more severe the local symmetry is disrupted, and the more likely it is to be a torn region.

[0068] The present invention is further configured such that the defect discrimination unit includes: candidate region screening, topology consistency calculation, and tear region discrimination;

[0069] Candidate region selection is based on the fact that both the tear response intensity and the micro-texture disturbance intensity of the conveyor belt image region exceed the set thresholds, thus determining that the conveyor belt image region belongs to the candidate tear region.

[0070] Topological consistency calculation generates a morphological compactness index by calculating the perimeter-to-area ratio of candidate tear regions, and combines the Euler feature number of local regions with the morphological compactness index to generate a structural consistency score.

[0071] Tear region identification is performed by determining whether the morphological consistency factor is less than a set threshold. If it is not less than the threshold, it is identified as a high-confidence tear region. Specifically, the purpose of region selection screening is to initially determine whether there are tear defects in image regions using set thresholds. Using a single threshold can easily lead to judgment errors or missed detections, so a dual threshold judgment is used to increase the accuracy of the initial screening. The initial screening results are obtained by comparing the intensity threshold of the tear response map and the intensity threshold of the micro-texture perturbation map respectively. The tear response map intensity threshold is set to 0.4 by default, with a range of values ​​within [missing value]. The specific value can be adjusted according to the actual situation; the default threshold for the intensity of the micro-texture perturbation map is 2.5, and the value range is within... The specific values ​​can be modified according to actual needs. A second screening is performed on suspected torn areas that pass the initial screening, and the morphological consistency factor is calculated by combining geometric and topological features; the calculation logic for the morphological consistency factor is as follows: ,in, For morphological consistency factor; These are suspected tear areas that passed the initial screening. Euler feature number for local regions is used to quantify the complexity of the topology. It is obtained by calculating the difference between the number of connected components and the number of holes. A value of 1 indicates simple connectivity without holes, while a value less than or equal to 0 indicates multiple holes or breaks. This method is existing technology and will not be elaborated on here. The morphological compactness index is defined as the normalized ratio of the total length of the suspected tear region boundary to the total number of pixels in the suspected tear region, with the output value ranging from [value missing]. Between these parameters, the closer the shape is to 1, the more regular and compact it is, and the lower the probability of it being a torn region. The total length of the torn region boundary is obtained by extracting the boundary contour of the region using OpenCV methods and calculating the sum of the Euclidean distances between the contour points. The total number of pixels in the suspected torn region is obtained using a statistical summation function. The torn region identification in the secondary screening is based on the morphological consistency factor. Is it less than the set morphological consistency threshold? The default morphological consistency threshold is 0.7, and its value range is within... The process can be modified according to the actual scenario. The torn areas that meet the conditions are marked as 1, and the rest as 0, generating an initial mask. This initial mask is then optimized based on morphology. Finally, the GrabCut algorithm, combined with the gradient of the original image, is used to optimize the mask edge fit to obtain the torn area mask. This method is existing technology and will not be elaborated on further here.

[0072] The present invention is further configured such that the tear size determination module includes: a tear size calculation unit, a tear depth calculation unit, and a tear severity assessment and response triggering unit;

[0073] Tear size calculation unit: Based on the physical dimensions of the main direction projection integral of the tear response map and combined with camera calibration parameters, the pixel coordinates of the conveyor belt image are converted into the actual tear length and width dimensions;

[0074] Tear depth calculation unit: The tear depth is obtained by fusing texture residuals, gradient consistency and edge clarity to measure the three-dimensional deformation characteristics of the tear region;

[0075] Tear Severity Assessment and Response Trigger Unit: A severity score is generated based on a nonlinear weighted scoring model using multi-dimensional parameters (length, width, and depth), and compared with a set threshold to obtain the tear level. Specifically, through projection transformation and dynamic compensation, the torn area in the image is converted into its actual physical length. and width To resolve measurement errors caused by camera angle, conveyor belt movement, or lens distortion, the core principle lies in combining gradient direction weights with spatial resolution calibration to achieve precise pixel-to-millimeter conversion; the formulas for calculating tear length and width are as follows: , ,in, This refers to the tear length; This refers to the tear width; This is the torn area; For part tension factor; The gradient direction; The main tear direction is extracted using PCA and determined by the eigenvectors of the regional covariance matrix, representing the tear extension direction. , The spatial resolution of an industrial camera refers to the actual physical size corresponding to each pixel. It is obtained through the initial calibration of the industrial camera during setup. The calibration process first fixes the distance between the industrial camera and the object, then photographs a calibration board of known size, and finally calculates the ratio of pixels to actual size to obtain the spatial resolution of the industrial camera. , Quantize the contribution of each pixel to the length and width separately, if it is the gradient direction. With the main direction Consistency If it is close to or approximately equal to 1, it is in the gradient direction. Perpendicular to the main direction but Approximately equal to or close to 1; This is the image scaling compensation factor. Because lens distortion or image compression can affect calculations and cause size deviations, this factor is used for adjustment. Its value ranges from [value range missing]. Ideally, the value should be set to 1 if the image is unscaled and distortion-free. If the image is compressed, the value should be greater than 1; if the image is stretched, the value should be less than 1. The specific value should be calculated simultaneously when calibrating the spatial resolution of the industrial camera. The calibration value is obtained by calculating the ratio of the product of the side length pixel value of the calibration board square and the spatial resolution to the actual size of the calibration board square. For example, if the actual size of the calibration board square is 10 and the side length pixel value is 80, the resulting spatial resolution is 0.125. The tear depth is calculated based on shadow differences and texture distortion to estimate the physical depth of the tear. The specific calculation formula is as follows: ,in, This refers to the tear depth; The edge sharpness index is used to quantify the sharpness of the boundary within the torn area. The sharper the torn edge, the more severe the physical structure of the material has been damaged, and the higher its positioning and discrimination value. The edge sharpness index is obtained by processing the torn area image with the Laplacian operator, extracting the second derivative information of the torn area image to reflect the degree of edge abruptness, and then normalizing the absolute value of the Laplacian response to finally obtain the edge sharpness score of each pixel. The illumination direction correction factor is calculated by quantifying the impact of shadow occlusion on depth estimation by calculating the cosine of the angle between the illumination direction and the local gradient direction. This is an existing technology and will not be elaborated on here. The texture energy of the torn region is extracted by using the gradient magnitude variance of the local image to extract the texture integrity of the torn region, reflecting the degree of structural damage. This is an existing technology and will not be elaborated on here. As a reference area texture energy, the average texture energy is extracted from intact pixels around the torn area and used as a lossless reference benchmark for texture anomaly comparison. The calculation method is the same as that for the texture energy of the torn area. This is a numerical stability constant used to avoid the denominator being zero; it is set to zero by default. . By comparing the texture energy difference between the damaged area and the reference area, and combining the light shielding effect, non-contact depth measurement is realized. The tear severity evaluation and response triggering unit quantifies the severity of the detected tear defects based on a multi-dimensional nonlinear scoring model, and triggers a graded response strategy such as shutdown and alarm according to the score. The core is to capture high-risk defect features more sensitively through an asymmetric gain function, while avoiding overreaction to minor tears; the severity score calculation logic is: wherein, is the severity score; , , is the tear dimension weight coefficient, used to distinguish the impact of different tear dimensions on safety, wherein the depth weight coefficient is the highest, because a penetrating tear can cause a conveyor belt to break and other chain failures, the default value is , the length weight coefficient is second, the default value is , and the width weight coefficient is slightly lower than the first two, the default value is . , , is the nonlinear saturation threshold, used to prevent large-size tear damage from causing an explosive increase in the score, wherein the default value is: , , , the specific value can be modified according to the toughness of different materials, the default value is the commonly used rubber conveyor belt, if custom high-elasticity rubber is used, experts can modify it according to experiments. The tear grade is obtained by the severity score according to the graded response strategy, different grade thresholds are set to judge the severity of the tear area, if the severity score is greater than or equal to the grade threshold 1, it is determined as a serious tear, in order to avoid defect expansion leading to conveyor belt rupture or material leakage, immediate shutdown is required to notify the staff to repair, if the severity score is less than the grade threshold 1 but greater than or equal to the grade threshold 2, it is determined as a slight tear, the tear area needs to be recorded, all data is transmitted to the data management module for recording and triggering sound and light alarm, reminding the staff to pay attention to processing next time maintenance, if the severity score is less than the grade threshold 2, it is determined as a suspected tear, the suspected tear area needs to be recorded, all data is transmitted to the data management module for recording, to provide data support for staff verification; the grade threshold 1 is 2.5 by default, the grade threshold 2 is 1.2 by default, the specific value is adjusted according to the material quality or environmental temperature, if the material is high-elasticity rubber, the grade threshold 1 can be appropriately increased to 3.0, if the environment is in a low-temperature state, such as winter temperature below -20℃, or in the northern low-temperature area, the brittleness of rubber material increases, the grade threshold 2 can be appropriately reduced to 1.0.

[0076] The application is further configured to, the positioning and automatic stopping module comprises: a tear region center positioning unit, a coordinate mapping unit and a stopping control unit;

[0077] The tear region center positioning unit: based on the tear region mask and the tear response map, the response intensity weighted centroid coordinates of the tear region are calculated. Specifically, the tear region center positioning unit is used to detect the center point of the tear region, so as to accurately position the damage region and realize accurate stopping to ensure that the error is within 1 meter; the coordinate mapping unit is used to convert the image coordinates into the actual position of the conveyor belt, which is used to control the tear region to stop at a specified position or notify the maintenance personnel to urgently go to the specified position for maintenance, and the stopping control unit is used to execute stopping or alarm according to the position and severity score. The response intensity weighted centroid coordinates include the horizontal coordinate and the vertical coordinate, and the horizontal coordinate calculation logic is: , and the vertical coordinate calculation logic is: , wherein, is the horizontal coordinate center; is the vertical coordinate center; is the tear response map, the value range , which is used to quantify the probability that the pixel point belongs to the tear defect, and here it is used as a weight function, and the high response region contributes more to the centroid coordinates; is the pixel horizontal coordinate; is the pixel vertical coordinate; represents the weighted value of the pixel horizontal coordinate, which is used to calculate the weighted average in the horizontal coordinate direction; represents the total response value of the tear region, which is a normalization denominator, and is used to ensure that the centroid coordinates are within a reasonable range.

[0078] The application is further configured to, the coordinate mapping unit: the response intensity weighted centroid coordinates are converted into the longitudinal absolute position in the physical coordinate system of the conveyor belt through the pulse encoder signal synchronized with the conveyor belt;

[0079] The stopping control unit: according to the tear severity score and the corresponding rules, a stopping instruction is generated for the tear region greater than the set threshold, so that the conveyor belt can be accurately stopped when the defect region reaches the set position. Specifically, the coordinate mapping unit is used to map the weighted centroid coordinates of the tear region in the image to the physical coordinate system of the conveyor belt, to generate the longitudinal absolute position, which provides a spatial reference for the stopping control; first, the spatial resolution of the industrial camera is multiplied by the weighted response intensity weighted centroid coordinates, to calculate the coordinate position of the tear region in the conveyor belt image refracted into the actual space; the conveyor belt motor end encoder records the pulse number in real time, and reads the encoder value at the camera exposure time, to obtain the distance moved by the conveyor belt; the longitudinal absolute position of the tear region is obtained by adding the distance moved by the conveyor belt and the longitudinal coordinate position of the tear region According to the tear level and the physical position , a hierarchical stop belt instruction is triggered, ensuring that the defective area is accurately stopped at the set position, if the tear level is in a serious tear, a stop instruction is immediately generated and sent to the worker, arranging the maintenance personnel to the designated position to wait, and then calculating the distance between the current physical position and the designated position, gradually reducing the running speed of the conveyor belt until the distance between the current physical position and the designated position is 0.

[0080] The application is further provided that the data management module adds the data generated in the triggering process to the trigger record set by recording each triggering event, the trigger record set including: trigger time, severity score, tear length, tear width, tear depth, response intensity weighted centroid coordinates, tear image. Specifically, the data management module is an important part of the whole system, responsible for recording and storing each triggering tear event, and archiving it in detail for subsequent analysis, statistics, monitoring and optimization. The function of the data management module is to provide detailed event logs and data, support abnormal event tracking, historical data analysis, and provide data support for system improvement and early warning strategies. The tear image in the trigger record set includes the conveyor belt image collected by the industrial camera and the processed tear response map, microscopic texture disturbance map. The trigger record set records each tear event in detail, and the system can sort the events by time and priority, quickly retrieve historical records when problems occur, analyze the development trajectory of the event, and locate the problem source; the module can also statistically analyze all triggering events, such as the number of tear events per month, the distribution of different severity, and the frequent occurrence of different parts. This provides data support for equipment maintenance and production optimization; the severity score of the triggering event and other data can be linked with the alarm system to alarm in real time according to the set threshold. High severity tear can trigger production shutdown or emergency repair; recorded trigger event data provides a basis for later algorithm optimization, and developers can adjust the algorithm according to historical data to optimize system performance; all trigger records are stored for a long time to prevent data loss and can be recovered to trace and analyze historical events.

[0081] The above-described embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented by software, the above-described embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are wholly or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another, for example, the computer instructions can be transferred from one website, computer, server, or data center to another via wired (for example, infrared, wireless, microwave, etc.) or wireless means. The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server, data center, or the like, which includes one or more available medium collections. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state disk.

[0082] It should be understood that the term "and / or" herein merely describes an association relationship of associated objects, which means that there can be three relationships, for example, A and / or B can represent the following three cases: A exists alone, A and B exist together, and B exists alone, where A and B can be singular or plural. In addition, the character " / " herein generally represents an "or" relationship between the associated objects before and after it, but it can also represent an "and / or" relationship. The specific meaning can be understood according to the context before and after it.

[0083] In this application, "at least one" means one or more, and "multiple" means two or more. "At least one of the following" or the like means any combination of the items, including any combination of single or multiple items. For example, at least one of a, b, or c can represent a, b, c, a-b, a-c, b-c, or a-b-c, where a, b, and c can be single or multiple.

[0084] It should be understood that in various embodiments of the present application, the size of the sequence number of the above-described processes does not mean the order of execution, and the execution order of the processes should be determined according to their functions and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0085] Those skilled in the art can clearly understand that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0086] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working processes of the above-described system, device and unit can refer to the corresponding processes in the foregoing method embodiments, which will not be repeated here.

[0087] In several embodiments provided in the present application, it should be understood that the disclosed system can be implemented in other ways. For example, the above-described device embodiments are only schematic, for example, the division of the units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interface, device or unit, and can be electrical, mechanical or other forms.

[0088] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on a plurality of network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0089] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically independently, or two or more units can be integrated into one unit.

[0090] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the parts that contribute to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0091] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A sand and gravel aggregate conveying abnormality supervisory management system based on industrial vision analysis, characterized by, include: Sensing and Automatic Control Module: Real-time sensing of the conveyor belt's start-up status is achieved by collecting sensor information from the conveyor belt, and image acquisition is automatically performed based on the start-up status; Image recognition module: Based on the acquired conveyor belt surface image, it performs real-time processing, extracts features, and filters for torn areas. The image recognition module includes: an image preprocessing unit, a feature extraction unit, a defect discrimination unit, and an anomaly classification and storage unit; Feature extraction unit: Based on the gradient intensity, local tension response factor, and edge closure enhancement function, it generates a tear response map from the conveyor belt image obtained by the image preprocessing unit. It also generates a micro-texture perturbation map by comparing the multi-directional Gabor filter response difference analysis with the symmetrical direction reflection image and combining the directional confidence weight factor; Defect discrimination unit: It adopts a structure / topology dual threshold strategy, combines Euler feature number and morphological compactness index to calculate the structural consistency score, and performs region filtering to generate high-confidence tear areas; Tear size determination module: Estimates the length, width, and depth of the identified tear area, and triggers an audible and visual alarm and system interlock control based on a preset threshold; Positioning and automatic stop module: It controls the conveyor belt to stop precisely based on the position coordinates of the torn area; Data Management Module: Automatically stores data from torn areas and generates a set of trigger records for use in generating reports or visualizations.

2. The sand aggregate conveying abnormal supervision and management system based on industrial vision analysis according to claim 1, characterized in that, The sensing and automatic control module includes: The operating status of the conveyor belt is determined by collecting real-time data from sensors linked to the motor end of the conveyor belt. The PLC and frequency converter work together to control the system. When the system detects that the conveyor belt is running, it sends a trigger signal to the image acquisition module. The integrated dust removal equipment then cleans the lens. After cleaning, the lens begins to acquire raw image data of the conveyor belt surface.

3. The sand aggregate conveying abnormal supervision and management system based on industrial vision analysis according to claim 1, characterized in that, The feature extraction unit includes: tear response map generation and micro-texture perturbation map generation; The tear response map is generated by constructing a local tension response factor through the variability of gradient direction within a unit neighborhood; The closure factor is obtained by using the local edge direction consistency and local convenience closure scores. The closure factor is then normalized to obtain the edge closure enhancement function. A tear response map is constructed by combining the local tension response factor and the edge closure enhancement function.

4. The sand aggregate conveying abnormal supervision and management system based on industrial vision analysis according to claim 3, characterized in that, The micro-texture perturbation map is generated by calculating the deviation between the local gradient magnitude in each direction and the main direction of the local directional field to obtain the directional confidence weight factor. Based on the difference between the multi-directional Gabor filter response and the Gabor filter response of the symmetrical directional reflection image, a micro-texture perturbation map is calculated by combining the directional confidence weight factor.

5. The sand aggregate conveying abnormal supervision and management system based on industrial vision analysis according to claim 1, characterized in that, The defect discrimination unit includes: candidate region screening, topology consistency calculation, and tear region discrimination; Candidate region selection is based on the fact that both the tear response intensity and the micro-texture disturbance intensity of the conveyor belt image region exceed the set thresholds, thus determining that the conveyor belt image region belongs to the candidate tear region. Topological consistency calculation generates a morphological compactness index by calculating the perimeter-to-area ratio of candidate tear regions, and combines the Euler feature number of local regions with the morphological compactness index to generate a structural consistency score. The tearing region discrimination discriminates the tearing region with high confidence by judging whether the shape consistency factor is less than a set threshold.

6. The sand aggregate conveying abnormal supervision and management system based on industrial vision analysis according to claim 1, characterized in that, The tearing size judging module comprises a tearing size calculation unit, a tearing depth calculation unit, and a tearing severity evaluation and response triggering unit. The tearing size calculation unit converts the pixel coordinates of the conveyor belt image into actual tearing length and width dimensions based on the physical size of the main direction projection integral of the tearing response map and in combination with the camera calibration parameters. The tearing depth calculation unit quantifies the three-dimensional deformation characteristics of the tearing region by fusing texture residuals, gradient consistency, and edge sharpness to obtain the tearing depth. The tearing severity evaluation and response triggering unit generates a severity score based on a nonlinear weighted scoring model of the length, width, and depth multidimensional parameters, and compares it with a set threshold to obtain the tearing grade.

7. The sand aggregate conveying abnormal supervision and management system based on industrial vision analysis according to claim 1, characterized in that, The positioning and automatic stop belt module comprises a tearing region center positioning unit, a coordinate mapping unit, and a stop belt control unit. The tearing region center positioning unit calculates the response intensity weighted centroid coordinates of the tearing region based on the tearing region mask and the tearing response map.

8. The sand aggregate conveying abnormal supervision and management system based on industrial vision analysis according to claim 7, characterized in that, The coordinate mapping unit converts the response intensity weighted centroid coordinates into the longitudinal absolute position in the conveyor belt physical coordinate system through the pulse encoder signal synchronized with the conveyor belt. The stop belt control unit generates a stop belt instruction for the tearing region greater than the set threshold according to the tearing severity score and the corresponding rules, so that the conveyor belt can be accurately stopped when the defect region reaches the set position.

9. The sand aggregate conveying abnormal supervision and management system based on industrial vision analysis according to claim 1, characterized in that, The data management module records each triggering event and adds the data generated during the triggering process to the triggering record set, which comprises the triggering time, severity score, tearing length, tearing width, tearing depth, response intensity weighted centroid coordinates, and tearing image.

Images