Conveyor belt damage detection method

Abstract

The invention discloses a conveyor belt damage detection method. The method comprises the following steps: S1, constructing an initial feature library; s2, in the running process of the belt conveyor, the inspection machine collects image data of the belt at the frame rate R in real time; s3, the feature vector f of the belt image is obtained through the step S1, then the distance between the feature vector of the belt image and the normal belt feature clustering center is calculated, and whether the belt in the belt image is damaged or not is judged according to the result; and S4, dynamically updating the feature vector library. The method has the beneficial effects that through real-time image acquisition and unsupervised learning analysis, a large amount of labeled data is not needed, the damage condition of the belt can be quickly identified, the potential damage risk can be timely found, the burden of manual inspection is reduced, the monitoring efficiency is improved, and meanwhile, the feature vector library is dynamically updated, so that the detection efficiency is improved. Therefore, the influence caused by gradual change of the belt appearance in the long-term operation process is effectively overcome, and then it is guaranteed that the method has high robustness in practical application.

Description

Technical Field

[0001] The invention relates to the technical field of belt conveyors, in particular to a method for detecting damage of a conveyor belt. Background Art

[0002] Belt conveyors are widely used in mining, coal, steel, electricity, chemical and other industries. As a key material conveying equipment, they undertake continuous and stable transportation tasks. However, due to the influence of various factors, the belts may be damaged in various forms such as cracks, tears, and flanging during long-term operation. These damages will not only reduce transportation efficiency and increase maintenance costs, but may also cause equipment downtime, material loss, and even safety accidents. Therefore, real-time and accurate monitoring of the health status of the belt is the key to ensuring production safety and improving equipment reliability.

[0003] Traditional belt damage detection methods mostly rely on manual inspections and rule-based monitoring systems. Although manual inspections can detect obvious damage, the inspection process is time-consuming and it is easy to miss minor damage, which cannot meet the needs of efficient, all-weather, and uninterrupted monitoring. Rule-based monitoring systems usually require manual setting of fault modes and alarm thresholds, but this method has poor flexibility and is difficult to adapt to complex changes in different working environments or operating conditions. It cannot effectively identify unforeseen damage patterns. In addition, most existing computer vision-based detection methods rely on supervised learning and require a large amount of labeled data for training. However, labeled data for belt damage is difficult to obtain, especially in actual production, so it is difficult to be widely used in actual production environments. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a method for detecting damage to a conveyor belt.

[0005] The purpose of the present invention is achieved through the following technical solutions: A method for detecting damage to a conveyor belt comprises the following steps:

[0006] S1: build the initial feature library;

[0007] S2: During the operation of the belt conveyor, the inspection machine collects image data of the belt in real time at a frame rate R;

[0008] S3: Obtain the feature vector f of the belt image through step S1, then calculate the distance between the feature vector of the belt image and the center of the normal belt feature cluster, and determine whether the belt in the belt image is damaged based on the result;

[0009] S4: Dynamically update the feature vector library.

[0010] Preferably, step S1 further includes the following steps:

[0011] S11: Collect N pieces of normal belt image data through the camera of the inspection machine;

[0012] S12: filtering the image content of the non-belt area in the image;

[0013] S13: extracting features from the filtered image;

[0014] S14: construct a feature vector library of a normal belt image set, and calculate statistical features of the normal belt image feature set.

[0015] Preferably, step S12 further includes the following steps:

[0016] S12.1: The image is segmented into belt areas using the belt segmentation algorithm.

[0017] B(x,y)=S(I(x,y));

[0018] Where (x, y) is the image before filtering; S is the belt segmentation algorithm; B(x, y) is the mask of the belt area;

[0019] S12.2: Set the non-belt area in the image to zero to obtain an image containing only the belt part.

[0020] I filtered (x,y)=I(x,y)·B(x,y);

[0021] Among them, I filtered (x,y) is the filtered image.

[0022] Preferably, in step S13, features are extracted from the filtered image using an image feature extraction model.

[0023] Preferably, in step S14, calculating the statistical features of the normal belt image feature set further comprises the following steps:

[0024] S14.1: Calculate the cluster centers of normal categories,

[0025]

[0026] S14.2: Calculate the average distance between the sample and the cluster center.

[0027]

[0028] S14.3: Calculate the standard deviation of the sample distance from the cluster center.

[0029]

[0030] Preferably, in step S3, the formula for calculating the distance between the belt image feature vector and the normal belt feature cluster center is:

[0031] d=||ff c ||.

[0032] Preferably, in step S3, the criterion for judging whether the belt in the belt image is damaged is:

[0033]

[0034] Among them, 0 represents that the belt is normal, and 1 represents that the belt is abnormal.

[0035] Preferably, in step S4, if the belt is judged to be in normal state by step S3, the earliest feature of the normal belt feature is first deleted from the feature library representing the normal belt in chronological order, and then the feature vector f calculated in step S3 is added to the feature library representing the normal belt.

[0036] The present invention has the following advantages: the present invention can quickly identify the damage condition of the belt and promptly discover potential damage risks through real-time image acquisition and unsupervised learning analysis without the need for a large amount of labeled data, thereby reducing the burden of manual inspection and improving monitoring efficiency. At the same time, the feature vector library is dynamically updated, thereby effectively overcoming the impact of gradual changes in the belt surface during long-term operation, thereby ensuring that the method has strong robustness in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a structural schematic diagram of the process of the conveyor belt damage detection method;

[0038] Figure 2 Schematic diagram of the structure of the initial feature library construction method. DETAILED DESCRIPTION

[0039] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0040] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0042] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0043] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inside", "outside", etc. indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, or the positions or positional relationships in which the product of the invention is usually placed when in use, or the positions or positional relationships commonly understood by those skilled in the art, which are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", etc. are only used to distinguish the description, and cannot be understood as indicating or implying relative importance.

[0044] In the description of the present invention, it is also necessary to explain that, unless otherwise clearly specified and limited, the terms "set", "install", "connect", and "connect" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0045] In this embodiment, if Figure 1 As shown, a method for detecting damage to a conveyor belt comprises the following steps:

[0046] S1: build the initial feature library;

[0047] S2: During the operation of the belt conveyor, the inspection machine collects image data of the belt in real time at a frame rate R;

[0048] S3: Obtain the feature vector f of the belt image through step S1, then calculate the distance between the feature vector of the belt image and the center of the normal belt feature cluster, and determine whether the belt in the belt image is damaged based on the result;

[0049] S4: Dynamically update the feature vector library. Through real-time image acquisition and unsupervised learning analysis, without the need for a large amount of labeled data, the damage condition of the belt can be quickly identified, potential damage risks can be discovered in a timely manner, the burden of manual inspection can be reduced, and monitoring efficiency can be improved. At the same time, the feature vector library is dynamically updated, thereby effectively overcoming the impact of gradual changes in the appearance of the belt during long-term operation, thereby ensuring that the method has strong robustness in practical applications.

[0050] Further, such as Figure 2 As shown, step S1 also includes the following steps:

[0051] S11: Collect N pieces of normal belt image data through the camera of the inspection machine;

[0052] S12: filtering the image content of the non-belt area in the image; specifically, the following steps are also included:

[0053] S12.1: The image is segmented into belt areas using the belt segmentation algorithm.

[0054] B(x,y)=S(I(x,y));

[0055] Where (x, y) is the image before filtering; S is the belt segmentation algorithm; B(x, y) is the mask of the belt area;

[0056] S12.2: Set the non-belt area in the image to zero to obtain an image containing only the belt part.

[0057] I filtered (x,y)=I(x,y)·B(x,y);

[0058] Among them, I filtered (x,y) is the filtered image, that is, only the pixel values ​​in the belt area are retained, and the pixel values ​​in the rest of the area are set to zero.

[0059] S13: extracting features from the filtered image;

[0060] S14: construct a feature vector library of a normal belt image set, and calculate statistical features of the normal belt image feature set.

[0061] Furthermore, in step S13, features are extracted from the filtered image using an image feature extraction model. Specifically, a feature extraction algorithm is used to extract features from the filtered image. The specific feature extraction algorithm is a lightweight image feature extraction model with strong generalization capability.

[0062] f i =F(I i ).

[0063] In this embodiment, in step S14, calculating the statistical features of the normal belt image feature set further includes the following steps:

[0064] S14.1: Calculate the cluster centers of normal categories,

[0065]

[0066] S14.2: Calculate the average distance between the sample and the cluster center.

[0067]

[0068] S14.3: Calculate the standard deviation of the sample distance from the cluster center.

[0069]

[0070] In this embodiment, in step S3, the formula for calculating the distance between the belt image feature vector and the normal belt feature cluster center is:

[0071] d=||ff c ||.

[0072] Furthermore, in step S3, the criteria for determining whether the belt in the belt image is damaged are:

[0073]

[0074] Among them, 0 represents that the belt is normal, and 1 represents that the belt is abnormal.

[0075] Furthermore, in step S4, if the belt is judged to be in a normal state by step S3, the earliest feature of the normal belt feature is first deleted from the feature library representing the normal belt in chronological order, and then the feature vector f calculated in step S3 is added to the feature library representing the normal belt. Specifically, since the condition of the belt will slowly change as the belt runs for a long time, a probability p is set as a hyperparameter so that the queue of the feature library is updated with probability p each time. Preferably, p is set to 0.2 here, and the parameter can be adjusted according to the frame rate of the camera and the running speed of the belt. When the belt is judged to be in a normal state by step S3, the earliest feature of the normal belt feature is first deleted from the feature library representing the normal belt in chronological order, and then the feature vector f calculated in step S3 is added to the feature library representing the normal belt.

[0076] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to 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 protection scope of the present invention.

Claims

1. A method for detecting damage to a conveyor belt, characterized in that: The following steps are involved: S1: build the initial feature library; S2: During the operation of the belt conveyor, the inspection machine collects image data of the belt in real time at a frame rate R; S3: Obtain the feature vector f of the belt image through step S1, then calculate the distance between the feature vector of the belt image and the center of the normal belt feature cluster, and determine whether the belt in the belt image is damaged based on the result; S4: Dynamically update the feature vector library.

2. The method for detecting damage to a conveyor belt according to claim 1, characterized in that: The step S1 further includes the following steps: S11: Collect N pieces of normal belt image data through the camera of the inspection machine; S12: filtering the image content of the non-belt area in the image; S13: extracting features from the filtered image; S14: construct a feature vector library of a normal belt image set, and calculate statistical features of the normal belt image feature set.

3. The method for detecting damage to a conveyor belt according to claim 2, characterized in that: The step S12 further includes the following steps: S12.1: The image is segmented into belt areas using the belt segmentation algorithm. B(x,y)=S(I(x,y)); Where (x, y) is the image before filtering; S is the belt segmentation algorithm; B(x, y) is the mask of the belt area; S12.2: Set the non-belt area in the image to zero to obtain an image containing only the belt part. I filtered (x,y)=I(x,y)·B(x,y); Among them, I filtered (x,y) is the filtered image.

4. The method for detecting damage to a conveyor belt according to claim 3, characterized in that: In the step S13, features are extracted from the filtered image using an image feature extraction model.

5. The method for detecting damage to a conveyor belt according to claim 4, characterized in that: In step S14, calculating the statistical features of the normal belt image feature set also includes the following steps: S14.1: Calculate the cluster centers of normal categories, S14.2: Calculate the average distance between the sample and the cluster center. S14.3: Calculate the standard deviation of the sample distance from the cluster center.

6. The method for detecting damage to a conveyor belt according to claim 5, characterized in that: In step S3, the formula for calculating the distance between the belt image feature vector and the normal belt feature cluster center is: d=||f-f c ||。 7. The method for detecting damage to a conveyor belt according to claim 6, characterized in that: In step S3, the criteria for determining whether the belt in the belt image is damaged are: Among them, 0 represents that the belt is normal, and 1 represents that the belt is abnormal.

8. The method for detecting damage to a conveyor belt according to claim 7, characterized in that: In step S4, if the belt is judged to be in normal state by step S3, the earliest feature of the normal belt is first deleted from the feature library representing the normal belt in chronological order, and then the feature vector f calculated in step S3 is added to the feature library representing the normal belt.

Images