A belt conveyor fault detection method and system

Abstract

The present invention relates to the technical field of belt conveyor fault identification, and more specifically, the present invention relates to a belt conveyor fault detection method and system, comprising: collecting images of belt conveyors and temperature values ​​of belt conveyors; inputting the images of the belt conveyors into a semantic segmentation algorithm model to obtain numerical matrices corresponding to each region, wherein each region includes a belt region, an upper fence region of the belt conveyor, a lower fence region of the belt conveyor, and an interval region between the belt and the fence; and equally dividing the numerical matrices corresponding to each region to determine the target key points of each region. The present invention detects belt conveyor faults by mounting a detection device on a mobile platform, and can perform mobile detection on the entire belt conveyor. By constructing a secondary fuzzy evaluation model, a more reasonable evaluation can be performed on the belt deviation index and the belt conveyor temperature value, and a more accurate belt conveyor fault index can be obtained, thereby improving the accuracy of the detection results and effectively ensuring the safety of railway tunnel construction.

Description

Technical Field

[0001] The present invention relates to the technical field of belt conveyor fault identification, and more specifically, to a belt conveyor fault detection method and system. Background Art

[0002] The belt conveyor is an important continuous transport machine that plays a very important role in the transportation of bulk materials. Since the working environment of the belt conveyor is usually relatively harsh and it needs to run for a long time, it often fails. These failures need to be detected and eliminated in a timely manner to avoid accidents. However, there are many fault points that need to be detected in the belt conveyor, the workload is large, and the detection accuracy is required. It is generally difficult to detect and eliminate faults through human sensory organs and engineering experience, especially for long-distance and large-volume belt conveyors. It is not easy to make accurate diagnosis of faults. Coupled with the lack of technology and production management, belt conveyor accidents often occur.

[0003] With the development of detection technology, especially the increasingly mature fault detection technology for belt conveyors, various detection methods have emerged one after another. Among them, there are methods using deep learning for visual detection and methods using ultrasonic detection. However, most of these existing fault detection technologies are used in small-scale scenarios such as coal mines and power stations. The belt conveyors are relatively short, and fault detection of belt conveyors can be achieved by installing detection equipment at fixed locations.

[0004] In the related technology, for example, the Chinese patent document with authorization announcement number CN112173636B discloses a method for detecting belt conveyor roller faults by an inspection robot. The invention obtains deformation degree data of the corresponding time and space domain frame through artificial intelligence recognition of frame deformation images collected by previous or current video; the extracted sound data and frame deformation degree are input into the artificial intelligence pattern recognizer for data fusion; information fusion optimizes fault diagnosis decisions, greatly reduces on-site inspections of belt conveyors, improves the accuracy of artificial intelligence fault detection, and optimizes the utilization efficiency of detection resources.

[0005] The current method for detecting belt conveyor faults is to place a camera at a fixed position in the tunnel and use a deep neural network algorithm to extract image features to determine whether the belt conveyor has a fault. This type of method is highly dependent on the image features obtained by recognition and requires a large amount of positive and negative data sample training to ensure its recognition performance. However, due to the long-distance tunnel environment, the detection equipment needs to detect belt conveyor faults in a mobile state. In the scenario of image vibration and environmental changes, there will be large changes in the recognition target and information loss. Therefore, the accuracy of belt conveyor fault detection is low. Summary of the invention

[0006] The present invention provides a belt conveyor fault detection method and system, aiming to solve the problem in the related art that, due to the need for detection equipment to detect belt conveyor faults in a mobile state in a long-distance tunnel environment, large changes in recognition targets and information loss may occur in scenarios of image vibration and environmental changes, thereby resulting in low accuracy in belt conveyor fault detection.

[0007] In a first aspect, the present invention provides a belt conveyor fault detection method, comprising: collecting an image of a belt conveyor and a temperature value of the belt conveyor; inputting the image of the belt conveyor into a semantic segmentation algorithm model to obtain a numerical matrix corresponding to each area, wherein each area includes a belt area, an upper fence area of ​​the belt conveyor, a lower fence area of ​​the belt conveyor, and an area between the belt and the fence; dividing the numerical matrix corresponding to each area into equal parts to determine the target key points of each area; calculating a belt offset factor based on the target key points, the belt offset factor including a belt evaluation factor and an interval evaluation factor, the belt evaluation factor and the mean of the coordinate values ​​of the target key points of the upper fence area of ​​the belt conveyor and the target key points of the belt area are positively correlated, and the interval evaluation factor and the mean of the target key points of the area between the belt and the fence are positively correlated; inputting the belt offset factor and the image depth compensation factor into a first fuzzy evaluation model to obtain a belt deviation index, wherein the image depth compensation factor is used to correct the degree of deformation of the image where the image is larger near and smaller far away; inputting the belt deviation index and the temperature value of the belt conveyor into a second fuzzy evaluation model to obtain a belt conveyor fault index; and performing fault detection on the belt conveyor according to the size of the belt conveyor fault index.

[0008] The effect is that by constructing a secondary fuzzy evaluation model, a more reasonable evaluation of the belt deviation index and the belt conveyor temperature value can be made, a more accurate belt conveyor fault index can be obtained, the accuracy of the detection results is improved, and the safety of railway tunnel construction is effectively guaranteed.

[0009] Preferably, the belt conveyor fault detection is performed according to the size of the belt conveyor fault index, including: in response to the belt conveyor fault index being greater than a first threshold, an alarm is issued; in response to the belt conveyor fault index being less than the first threshold, the belt conveyor continues to work without an alarm.

[0010] The effect is that according to the comparison between the belt conveyor fault index and the first threshold value, an alarm can be given accurately and timely.

[0011] Preferably, the numerical matrix corresponding to each region is divided into equal parts to determine the target key points of each region, including: constructing a coordinate system with the upper left corner of the numerical matrix as the origin, with the X axis of the coordinate system to the right and the Y axis of the coordinate system downward; dividing the numerical matrix corresponding to each region into equal parts, and taking the points of 1 in the numerical matrix that appear along the Y axis downward as the target edge points of each region; for any region, performing straight line fitting on the target edge points of the region to obtain a fitted straight line, and then dividing the numerical matrix into four equal parts and drawing lines, and taking the points where the drawn lines intersect the fitted straight line as the target key points.

[0012] The effect is that the method is simple to find the target key points and can accurately find the location of the target key points.

[0013] Preferably, calculating the belt evaluation factor includes: obtaining the target key point located at the lower edge of the fence area on the belt conveyor, and the target key point at the upper edge of the belt area; and taking the average of the sum of the subtractions of the coordinate values ​​of all the target key points located on the same column Y-axis as the belt evaluation factor.

[0014] Preferably, calculating the interval evaluation factor includes: obtaining target key points located at the upper and lower edges of the interval area between the belt and the fence, and taking the average value after subtracting the target key points located at the upper and lower edges of the same column as the interval evaluation factor.

[0015] Preferably, calculating the image depth compensation factor includes: obtaining a first fitting straight line of the upper fence area of ​​the belt conveyor and a second fitting straight line of the lower fence area of ​​the belt conveyor, and calculating the mean straight line of the first fitting straight line and the second fitting straight line; dividing the belt conveyor image into two equal parts by marking lines to obtain the intersection of the mean straight line and the marking line, and drawing a straight line perpendicular to the mean perpendicular line along the intersection to obtain a first point where the straight line intersects with the first fitting straight line, and a second point where the straight line intersects with the second fitting straight line, and using the distance value between the first point and the second point as a depth evaluation factor; inputting the depth evaluation factor into a third fuzzy evaluation model to obtain an image depth compensation factor.

[0016] The effect is that by constructing an image depth compensation factor, the imbalance of image features caused by the image's near-big-far-small effect can be balanced, providing more accurate evaluation parameters for calculating belt conveyor fault indicators.

[0017] Preferably, performing straight line fitting on the target edge points of the region includes: performing straight line fitting on the target edge points using a least squares method or a Hough linear transformation method.

[0018] The effect is that the least squares method finds the best function match for the data by minimizing the sum of squared errors, ensuring that the fit is "optimal" to some extent, that is, it best represents the overall trend of the data set.

[0019] Preferably, calculating the image depth compensation factor also includes: establishing a deep learning model, and obtaining the image depth compensation factor using the deep learning model.

[0020] In a second aspect of the present invention, a belt conveyor fault detection system is provided, comprising a processor and a memory, wherein the memory stores a computer program, and the processor executes the computer program to implement any one of the belt conveyor fault detection methods described above.

[0021] Beneficial effects: By carrying out detection equipment on a mobile platform to detect belt conveyor faults, the entire belt conveyor can be inspected in a mobile manner, and by constructing a secondary fuzzy evaluation model, a more reasonable evaluation of the belt deviation index and the belt conveyor temperature value can be made, and a more accurate belt conveyor fault index can be obtained, thereby improving the accuracy of the detection results and effectively ensuring the safety of railway tunnel construction. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The following detailed description is read with reference to the accompanying drawings, which illustrate several embodiments of the present invention in an exemplary and non-limiting manner, and in which like or corresponding reference numerals represent like or corresponding parts, wherein:

[0023] Figure 1 is a schematic diagram schematically illustrating a fence area on a belt conveyor according to an embodiment of the present invention;

[0024] Figure 2 is a schematic diagram schematically illustrating a belt and a fence spacing area according to an embodiment of the present invention;

[0025] Figure 3 is a schematic diagram schematically illustrating a belt area according to an embodiment of the present invention;

[0026] Figure 4 is a schematic diagram schematically showing a fence area under a belt conveyor according to an embodiment of the present invention;

[0027] Figure 5 is a schematic diagram schematically showing the equal division of the fence area on the belt conveyor according to an embodiment of the present invention;

[0028] Figure 6 is a schematic diagram schematically showing the belt and fence interval area after being equally divided according to an embodiment of the present invention;

[0029] Figure 7 is a schematic diagram schematically showing a belt area after being equally divided according to an embodiment of the present invention;

[0030] Figure 8 is a schematic diagram schematically showing the belt conveyor lower fence area after being equally divided according to an embodiment of the present invention;

[0031] Fig. 9 is a schematic diagram schematically showing a target edge point fitting straight line of a fence area on a belt conveyor according to an embodiment of the present invention;

[0032] Fig.10 is a flow chart schematically illustrating calculation of a belt fault index according to an embodiment of the present invention;

[0033] Fig.11 FIG. 1 is a schematic diagram schematically showing a system structure according to an embodiment of the present invention.

[0034] Reference numerals

[0035] 1. Processor; 2. Communication bus; 3. Memory; 4. Communication interface. DETAILED DESCRIPTION

[0036] 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 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0037] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0038] With the development of detection technology, especially the increasingly mature fault detection technology for belt conveyors, various detection methods have emerged one after another. Among them, there are methods using deep learning for visual detection, and there are also methods using ultrasonic detection. However, these existing fault detection technologies are mostly used in small-scale scenarios such as coal mines and power stations. The belt conveyor is short, and fault detection of the belt conveyor can be achieved by installing detection equipment at a fixed position. However, when faced with large-scale and long-distance tunnels, fixed-point detection cannot meet the detection needs. Therefore, the collection equipment can be mounted on the mobile robot platform along the belt conveyor to perform mobile detection of the entire belt conveyor.

[0039] like Fig.10 As shown, S101: collecting an image of the belt conveyor and a temperature value of the belt conveyor.

[0040] In one embodiment, an image acquisition device mounted on a mobile robot platform along the belt conveyor can be used to acquire an image of the belt conveyor while the robot is moving. A temperature acquisition device mounted on a mobile robot platform along the belt conveyor can be used to acquire the temperature value of the belt conveyor. The temperature acquisition device can be a thermal infrared imager, an infrared temperature sensor, etc.

[0041] S102: Input the image of the belt conveyor into a semantic segmentation algorithm model to obtain a numerical matrix corresponding to each area.

[0042] like Figure 1-Figure 4 As shown, in one embodiment, the collected belt conveyor images are divided into a training set and a validation set, and each image is labeled, wherein the label categories include "belt area" label, "belt conveyor upper fence area" label, "belt conveyor lower fence area" label, and "belt and fence interval area" label. In order to obtain a better model weight in the training process of the semantic segmentation algorithm model, high requirements are required for the quantity and quality of the image data set. In terms of the quantity of the image data set, as many image samples as possible are required, and in terms of the quality of the image data set, as many different types of image samples as possible are required, including images of different types of belt conveyors, images of different belt conveyor angles, images of different light intensity, etc.

[0043] like Figure 1-Figure 4 As shown, Figure 1 The area enclosed by the dotted line is the fence area on the belt conveyor. Figure 2 The area enclosed by the dotted line is the area between the belt and the fence. Figure 3 The area enclosed by the dotted line is the belt area. Figure 4 The area framed by dotted lines is the fence area under the belt conveyor.

[0044] In one embodiment, an image of a belt conveyor is input into a semantic segmentation algorithm model to obtain a numerical matrix corresponding to each area, wherein the numerical matrix corresponding to each area includes a numerical matrix corresponding to the belt area, a numerical matrix corresponding to the fence area above the belt conveyor, a numerical matrix corresponding to the fence area below the belt conveyor, and a numerical matrix corresponding to the area between the belt and the fence, and the numerical matrix is ​​a matrix composed of numerical values ​​of "0" and "1".

[0045] In one embodiment, a coordinate system is constructed with the upper left corner of the numerical matrix as the origin, the right side is the coordinate system X axis, and the downward side is the coordinate system Y axis. The constructed coordinate system is used to determine the target key point.

[0046] S103: Divide the numerical matrix corresponding to each region into equal parts to determine the target key points of each region.

[0047] like Figure 5-Figure 8As shown, in one embodiment, the numerical matrix corresponding to each region is divided into equal parts to determine the target key points of each region; specifically, the numerical matrix corresponding to each region is divided into equal parts. In this embodiment, the numerical matrix is ​​divided into twenty equal parts to obtain twenty equal matrices, and the equal-division matrix in the last column is removed, and the first "1" and the last "1" that appear along the Y-axis downward on the equal-division line in the remaining equal-division matrix are used as target edge points. In other embodiments, the numerical matrix can also be divided into fifteen or sixteen equal parts, so that the target edge points of any region can be obtained.

[0048] like Fig. 9 As shown, for any region, a straight line is fitted to the target edge points of the region to obtain a fitted straight line, wherein the target edge points can be fitted to a straight line using the least square method or the Hough straight line transformation method. Then, the numerical matrix of each region is divided equally. In this embodiment, the numerical matrix of each region is divided into four equal parts. In other embodiments, the numerical matrix of each region can also be divided into five equal parts or six equal parts. Then, the last column of the divided matrix is ​​removed, and the intersection of the dividing line in the remaining divided matrix and the fitted straight line is used as the target key point.

[0049] For example, for the “fence area on the conveyor belt”, according to the above method for finding the target key points, five target key points p12, p13, p14, p15 and p16 can be obtained ( Fig. 9 as shown).

[0050] S104: Calculate the belt deviation factor according to the target key point.

[0051] Specifically, the belt deviation factor includes a belt evaluation factor and an interval evaluation factor. The belt evaluation factor is positively correlated with the mean value after subtracting the coordinate values ​​of the target key points in the fence area on the belt conveyor from the target key points in the belt area. The interval evaluation factor is positively correlated with the mean value after subtracting the coordinate values ​​of the target key points in the interval area between the belt and the fence.

[0052] In one embodiment, calculating the belt evaluation factor includes: obtaining the target key point located at the lower edge of the fence area on the belt conveyor and the target key point at the upper edge of the belt area; and taking the average of the sum of the subtractions of the coordinate values ​​of all the target key points located on the same column Y-axis as the belt evaluation factor.

[0053] In one embodiment, calculating the interval evaluation factor includes: obtaining target key points located at the upper and lower edges of the interval area between the belt and the fence, and taking the average value after subtracting all target key points located at the upper and lower edges on the same Y-axis as the interval evaluation factor.

[0054] In one embodiment, the image depth compensation factor is calculated, including: obtaining a first fitting straight line of the upper edge of the upper fence area of ​​the belt conveyor and a second fitting straight line of the lower fence area of ​​the belt conveyor, and calculating the mean straight line of the first fitting straight line and the second fitting straight line. Then, the numerical matrix corresponding to the upper fence area of ​​the belt conveyor and the numerical matrix corresponding to the lower fence area of ​​the belt conveyor are equally divided into two areas, and the intersection of the mean straight line and the bisector is obtained, and a straight line perpendicular to the mean vertical line is drawn along the intersection to obtain the first point where the straight line intersects with the first fitting straight line and the second point where the straight line intersects with the second fitting straight line, and the distance value between the first point and the second point is used as the depth evaluation factor, wherein the distance value between the first point and the second point can be calculated using the Euclidean distance algorithm. Then, according to the depth evaluation factor, the third fuzzy evaluation model is input to output the image depth compensation factor. Through the above steps, by constructing the image depth compensation factor, the imbalance of image features caused by the near-large-far-small effect of the image can be balanced, and more accurate evaluation parameters can be provided for calculating the belt deviation index.

[0055] S105: Calculate the belt conveyor fault index.

[0056] In one embodiment, the belt deviation factor and the image depth compensation factor are input into the first fuzzy evaluation model to obtain the belt deviation index, and the belt deviation index and the temperature value of the belt conveyor are input into the second fuzzy evaluation model to obtain the belt conveyor fault index, and an alarm is issued according to the magnitude of the belt conveyor fault index. In one embodiment, the belt conveyor fault index range is [0 1], the index range is [0, 0.2] is normal, the index range is [0.2, 0.7] is a minor fault, and the index range is [0.7, 1] is a serious fault. When the belt conveyor fault index is not within the normal index range, an alarm is issued.

[0057] It should be noted that in the process of constructing the above three fuzzy evaluation models, the membership function types of the belt deviation index, the belt conveyor temperature value, and the belt conveyor fault index are all triangular membership functions. In the specific implementation process, the membership function type can also be a generalized bell-shaped membership function, a Gaussian membership function, a Z-shaped membership function, a generalized bell-shaped and triangular mixed membership function, etc., and the defuzzification method can be any one of the maximum membership method, the weighted average method (centroid method) or the median method.

[0058] According to the above steps, by carrying out belt conveyor fault detection on a mobile platform, the entire belt conveyor can be inspected in a mobile manner. By constructing a secondary fuzzy evaluation model, a more reasonable evaluation of the belt deviation index and the belt conveyor temperature value can be made, and a more accurate belt conveyor fault index can be obtained, thereby improving the accuracy of the detection results and effectively ensuring the safety of railway tunnel construction.

[0059] The present invention also provides a belt conveyor fault detection system. Fig.11 As shown, the system includes a processor 1 and a memory 3, wherein the memory 3 stores computer program instructions, and when the computer program instructions are executed by the processor 1, a belt conveyor fault detection method according to the first aspect of the present invention is implemented.

[0060] The system also includes other components well known to those skilled in the art, such as a communication bus 2 and a communication interface 4. The configuration and functions of these components are known in the art and thus will not be described in detail herein.

[0061] In the present invention, the aforementioned memory 3 may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, apparatus or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device. Any application or module described in the present invention may be implemented using computer-readable / executable instructions that may be stored or otherwise maintained by such a computer-readable medium.

[0062] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly defined.

[0063] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0064] The above-mentioned embodiments only express several implementation methods of the present invention, and the description is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent application. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A belt conveyor fault detection method, characterized in that: include: Collect the image and temperature of the belt conveyor; Input the image of the belt conveyor into the semantic segmentation algorithm model to obtain a numerical matrix corresponding to each area, wherein each area includes a belt area, an upper fence area of ​​the belt conveyor, a lower fence area of ​​the belt conveyor, and an area between the belt and the fence; The numerical matrix corresponding to each area is equally divided to determine the target key points of each area, including: The coordinate system is constructed with the upper left corner of the numerical matrix as the origin, the right side is the coordinate system X axis, and the downward side is the coordinate system Y axis; The numerical matrix corresponding to each region is equally divided, and the points with a value of 1 in the numerical matrix appearing downward along the Y axis are taken as the target edge points of each region; For any region, a straight line is fitted to the target edge points of the region to obtain a fitted straight line, and then the numerical matrix is ​​divided into four equal parts and the points where the line intersects with the fitted straight line are taken as target key points; Calculating a belt deviation factor according to the target key point, wherein the belt deviation factor includes a belt evaluation factor and an interval evaluation factor; Calculate belt rating factors including: Get the target key point at the lower edge of the fence area on the belt conveyor and the target key point at the upper edge of the belt area; The mean of the sum of the coordinate values ​​of all target key points on the same column Y axis is used as the belt evaluation factor; Calculate interval evaluation factors, including: Obtain the target key points at the upper and lower edges of the interval between the belt and the fence, and subtract the average of the target key points at the upper and lower edges of the same column as the interval evaluation factor; Calculate image depth compensation factors, including: Obtaining a first fitting straight line of the upper fence area of ​​the belt conveyor and a second fitting straight line of the lower fence area of ​​the belt conveyor, and calculating a mean straight line of the first fitting straight line and the second fitting straight line; Divide the belt conveyor image into two equal parts and draw a line to obtain the intersection of the mean straight line and the line, and draw a straight line perpendicular to the mean perpendicular line along the intersection to obtain a first point where the straight line intersects with the first fitting straight line and a second point where the straight line intersects with the second fitting straight line, and use the distance value between the first point and the second point as a depth evaluation factor; Inputting the depth evaluation factor into a third fuzzy evaluation model to obtain an image depth compensation factor; Input the belt deviation factor and the image depth compensation factor into the first fuzzy evaluation model to obtain the belt deviation index, wherein the image depth compensation factor is used to correct the deformation degree of the image, that is, the image is larger near and smaller far away; input the belt deviation index and the temperature value of the belt conveyor into the second fuzzy evaluation model to obtain the belt conveyor fault index; According to the magnitude of the belt conveyor fault index, the belt conveyor is subjected to fault detection.

2. The belt conveyor fault detection method according to claim 1, characterized in that: According to the magnitude of the belt conveyor fault index, the belt conveyor is subjected to fault detection, including: In response to the belt conveyor fault index being greater than a first threshold, an alarm is issued; In response to the belt conveyor fault index being less than the first threshold, the belt conveyor continues to operate without an alarm.

3. The belt conveyor fault detection method according to claim 1, characterized in that: Perform straight line fitting on the target edge points in the area, including: Use the least squares method or Hough straight line transform method to perform straight line fitting on the target edge points.

4. The belt conveyor fault detection method according to claim 1, characterized in that: Calculate the image depth compensation factor, also including: A deep learning model is established and used to obtain the image depth compensation factor.

5. A belt conveyor fault detection system, comprising a processor and a memory, characterized in that: The memory stores a computer program, and the processor executes the computer program to implement the belt conveyor fault detection method according to any one of claims 1 to 4.

Images