Method, device and equipment for detecting material scattering condition of belt of coal preparation plant and readable storage medium

Abstract

The invention provides a method, device and equipment for detecting the material scattering condition of a belt of a coal preparation plant and a readable storage medium, and the method comprises the steps: segmenting a belt region image through a preset image segmentation model, and obtaining a coal segmentation image and a belt segmentation image; edge information of the coal segmentation image and the belt segmentation image is extracted through a contour detection technology, coal edge information and belt edge information are obtained, the coal edge information and the belt edge information respectively comprise a plurality of calculation points, and each calculation point in the plurality of calculation points comprises a plurality of pixel points; according to the coal edge information and the belt edge information, the coal scattering condition in the belt area image is analyzed. The method can achieve the effect of improving the precision of detecting the material scattering condition of the belt of the coal preparation plant.

Description

Technical Field

[0001] The present application relates to the field of coal spill detection, and in particular, to a method, device, equipment and readable storage medium for detecting coal spill on a belt of a coal preparation plant. Background Art

[0002] At present, in coal preparation plants, belt conveyors are used for continuous transportation of coal. However, due to the irregularity of coal lumps and the vibration of the belts, coal spillage often occurs. Coal spillage not only leads to coal waste, but also may damage equipment, increase maintenance costs, and even cause safety hazards. Traditional coal spillage detection mainly relies on manual inspections and simple sensors.

[0003] However, manual detection is inefficient and not timely enough, while conventional sensors (such as infrared or ultrasonic sensors) are easily disturbed by environmental factors such as dust, vibration and temperature, resulting in insufficient detection accuracy.

[0004] Therefore, how to improve the accuracy of detecting the material spreading situation of the coal preparation plant belt is a technical problem that needs to be solved. Summary of the invention

[0005] The purpose of the embodiments of the present application is to provide a method for detecting the material spreading condition of the belt of a coal preparation plant. The technical solution of the embodiments of the present application can achieve the effect of improving the accuracy of detecting the material spreading condition of the belt of a coal preparation plant.

[0006] In a first aspect, an embodiment of the present application provides a method for detecting the situation of coal spreading on a coal preparation plant belt, comprising: segmenting a belt area image through a preset image segmentation model to obtain a coal segmentation map and a belt segmentation map; extracting edge information of the coal segmentation map and the belt segmentation map through contour detection technology to obtain coal edge information and belt edge information, wherein the coal edge information and the belt edge information respectively include multiple calculation points, and each of the multiple calculation points includes multiple pixel points; analyzing the coal spreading situation in the belt area image based on the coal edge information and the belt edge information.

[0007] In the above embodiment, the application uses the semantic segmentation technology in deep learning to effectively distinguish the belt and coal area, and can accurately identify them even in complex backgrounds. Then, by extracting the edge information of the coal and the belt, the material spillage phenomenon can be timely and accurately identified, which can achieve the effect of improving the accuracy of detecting the material spillage situation of the coal preparation plant belt.

[0008] In some embodiments, the coal scattering situation in the belt area image is analyzed based on the coal edge information and the belt edge information, including: determining multiple Euclidean distances between calculation points in the coal edge information and the belt edge information through a Euclidean algorithm; determining whether the shortest Euclidean distance among the multiple Euclidean distances is less than a preset threshold; when it is determined that the shortest Euclidean distance among the multiple Euclidean distances is less than the preset threshold, it is determined that scattering occurs in the belt area image.

[0009] In the above embodiment, the present application further refines the accuracy of material spill detection by extracting the edge information of the coal and the belt and combining the Euclidean distance measurement method of multiple calculation points. The relative position of the coal and the belt edge can be accurately determined to ensure that the material spill phenomenon is identified in a timely and accurate manner.

[0010] In some embodiments, before the belt area image is segmented by a preset image segmentation model to obtain a coal segmentation map and a belt segmentation map, it also includes: replacing the convolution layer of the existing segmentation model with a depthwise separable convolution layer to obtain a replaced segmentation model; using belt area images taken when transporting coal in different coal preparation plants and corresponding material spreading conditions as training samples to train the replaced segmentation model and obtain a baseline to obtain a baseline model; evaluating the importance of each convolution kernel in the baseline model according to the absolute value or gradient information of each convolution kernel weight, removing the convolution kernels in the baseline model whose importance is less than or equal to the preset importance, to obtain an initial segmentation model; training and verifying the initial segmentation model through training samples and verification samples to obtain a segmentation model.

[0011] In the above embodiments of the present application, the accuracy of image segmentation by the segmentation model can be improved by structurally optimizing the existing segmentation model, including replacing the convolution layer and removing the convolution kernel, and then training and verifying the segmentation model.

[0012] In some embodiments, the segmentation model is obtained by training and verifying the initial segmentation model through training samples and verification samples, including: training and verifying the initial segmentation model through training samples and verification samples to obtain the accuracy and recall of the initial segmentation model; when it is determined that the accuracy and recall respectively meet the preset accuracy and preset recall, the floating point numbers corresponding to the weight values ​​and activation values ​​of the initial segmentation model are converted into half-precision floating points to obtain the converted segmentation model; and the converted segmentation model is retrained through training samples to obtain the segmentation model.

[0013] In the above embodiments of the present application, the accuracy and recall rate of the model are used as training benchmarks, and the floating-point numbers corresponding to the weight values ​​and activation values ​​of the segmentation model are converted into half-precision floating points to improve the accuracy of the segmentation model in segmenting the image.

[0014] In some embodiments, the initial segmentation model is trained and verified by training samples and verification samples to obtain a segmentation model, including: optimizing the initial segmentation model through a high-performance library interface to obtain an optimized segmentation model; training and verifying the optimized segmentation model by training samples and verification samples to obtain a segmentation model.

[0015] In the above embodiments of the present application, the accuracy of image segmentation by the segmentation model can be improved by optimizing the initial segmentation model through a high-performance library interface.

[0016] In some embodiments, before the belt area image is segmented by a preset image segmentation model to obtain a coal segmentation map and a belt segmentation map, it also includes: photographing the target belt area with a high-definition industrial camera to obtain an initial belt area image; preprocessing the initial belt area image to obtain a belt area image, wherein the preprocessing includes: at least one of image grayscale and denoising processing.

[0017] In the above embodiments of the present application, by collecting and preprocessing images, the quality of the images can be improved while improving the accuracy of the image segmentation by the segmentation model.

[0018] In some embodiments, edge information of a coal segmentation map and a belt segmentation map is extracted by contour detection technology to obtain coal edge information and belt edge information, including: extracting coal contour images and belt contour images in the coal segmentation map and the belt segmentation map by contour detection technology; extracting all pixel points in the coal contour image and the belt contour image by contour detection technology; dividing all pixel points according to a preset number to obtain coal edge information and belt edge information, wherein the coal edge information and the belt edge information respectively include multiple calculation points, and each of the multiple calculation points includes a preset number of pixel points.

[0019] In the above embodiment of the present application, by extracting the edge information of the coal and the belt and combining the Euclidean distance measurement method of multiple calculation points, the relative position of the belt and the coal can be accurately determined according to the obtained calculation points, and then the material spreading situation can be accurately judged.

[0020] In a second aspect, an embodiment of the present application provides a device for detecting the spreading of materials on a belt of a coal preparation plant, comprising:

[0021] A segmentation module is used to segment the belt area image through a preset image segmentation model to obtain a coal segmentation map and a belt segmentation map;

[0022] An extraction module is used to extract edge information of the coal segmentation map and the belt segmentation map by using a contour detection technology to obtain coal edge information and belt edge information, wherein the coal edge information and the belt edge information respectively include a plurality of calculation points, and each of the plurality of calculation points includes a plurality of pixel points;

[0023] The analysis and detection module is used to analyze the coal spreading situation in the belt area image based on the coal edge information and the belt edge information.

[0024] Optionally, the analysis and detection module is specifically used for:

[0025] Determine multiple Euclidean distances between calculation points in the coal edge information and the belt edge information by using the Euclidean algorithm;

[0026] Determine whether the shortest Euclidean distance among the multiple Euclidean distances is less than a preset threshold;

[0027] When it is determined that the shortest Euclidean distance among the multiple Euclidean distances is less than a preset threshold, it is determined that material spillage occurs in the belt area image.

[0028] Optionally, the device further comprises:

[0029] A training module, used for the segmentation module to replace the convolution layer of the existing segmentation model with a depth-separable convolution layer before segmenting the belt area image through a preset image segmentation model to obtain a coal segmentation map and a belt segmentation map, so as to obtain a replaced segmentation model;

[0030] The belt area images taken when transporting coal in different coal preparation plants and the corresponding material spreading conditions are used as training samples to train the replaced segmentation model and obtain the baseline to obtain the baseline model;

[0031] According to the absolute value or gradient information of each convolution kernel weight, the importance of each convolution kernel in the baseline model is evaluated, and the convolution kernels whose importance in the baseline model is less than or equal to the preset importance are removed to obtain the initial segmentation model;

[0032] The initial segmentation model is trained and verified through training samples and verification samples to obtain a segmentation model.

[0033] Optionally, the training module is specifically used to:

[0034] The initial segmentation model is trained and verified through training samples and verification samples to obtain the accuracy and recall of the initial segmentation model;

[0035] When it is determined that the accuracy and the recall rate meet the preset accuracy and the preset recall rate respectively, the floating point numbers corresponding to the weight values ​​and activation values ​​of the initial segmentation model are converted into half-precision floating points to obtain a converted segmentation model;

[0036] The converted segmentation model is retrained through the training samples to obtain the segmentation model.

[0037] Optionally, the training module is specifically used to:

[0038] Optimize the initial segmentation model through the high-performance library interface to obtain the optimized segmentation model;

[0039] The optimized segmentation model is trained and verified through training samples and verification samples to obtain a segmentation model.

[0040] Optionally, the device further comprises:

[0041] An image processing module is used for the segmentation module to segment the belt area image by a preset image segmentation model to obtain a coal segmentation map and a belt segmentation map, and then use a high-definition industrial camera to shoot the target belt area to obtain an initial belt area image;

[0042] The initial belt area image is preprocessed to obtain a belt area image, wherein the preprocessing includes: at least one of image grayscale processing and denoising processing.

[0043] Optionally, the extraction module is specifically used to:

[0044] The coal contour image and the belt contour image are extracted from the coal segmentation map and the belt segmentation map by using the contour detection technology;

[0045] All pixel points in the coal contour image and the belt contour image are extracted by contour detection technology;

[0046] All pixel points are divided into preset numbers to obtain coal edge information and belt edge information, wherein the coal edge information and the belt edge information respectively include multiple calculation points, and each of the multiple calculation points includes a preset number of pixel points.

[0047] In a third aspect, an embodiment of the present application provides an electronic device, comprising a processor and a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the steps in the method provided in the first aspect are performed.

[0048] In a fourth aspect, an embodiment of the present application provides a readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps in the method provided in the first aspect are performed.

[0049] Other features and advantages of the present application will be described in the following description, and partly become apparent from the description, or be understood by practicing the embodiments of the present application. The purpose and other advantages of the present application can be realized and obtained by the structures specifically pointed out in the written description, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments of the present application will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0051] Figure 1 A flow chart of a method for detecting belt spreading in a coal preparation plant provided in an embodiment of the present application;

[0052] Figure 2 A flow chart of an implementation method for detecting the material spreading condition of a coal preparation plant belt provided in an embodiment of the present application;

[0053] Figure 3 A schematic block diagram of a device for detecting the material spreading condition of a belt in a coal preparation plant provided in an embodiment of the present application;

[0054] Figure 4 A schematic diagram of the structure of a device for detecting the material spreading condition of a coal preparation plant belt provided in an embodiment of the present application. DETAILED DESCRIPTION

[0055] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application usually described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0056] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of this application, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0057] First, some terms involved in the embodiments of the present application are explained to facilitate understanding by those skilled in the art.

[0058] The YOLOv8-Seg model is a deep learning target detection and segmentation model based on the YOLO (You Only Look Once) series, which is specially designed for real-time image segmentation tasks. While maintaining the speed advantage, it also increases the ability to accurately capture the contours of targets in complex scenes.

[0059] OpenCV stands for Open Source Computer Vision Library, which is a cross-platform computer vision and machine learning software library.

[0060] Contour detection technology is a process of extracting the contour of a target from an image containing the target and background by ignoring the texture of the background and the internal texture of the target and the influence of noise interference in digital image processing, using certain techniques and methods. It is an important basis for technologies such as target detection, shape analysis, target recognition and target tracking. The purpose of contour detection is to extract the external contour of an object from an image. These contours are composed of a series of continuous points and depict the shape of the object. Through contour detection, computers can recognize and understand objects in images, and then perform more advanced image processing and analysis.

[0061] The TensorRT model is an optimized deep learning model designed for high-performance inference on NVIDIA GPUs. TensorRT is a high-performance deep learning inference optimizer and runtime acceleration library designed to provide low-latency, high-throughput deployment inference for deep learning applications.

[0062] The PyTorch model is the result of neural network training and optimization, which includes the neural network skeleton and learned parameters. PyTorch is an open source deep learning framework that mainly implements automatic differentiation and introduces dynamic computational graphs to make model building more flexible.

[0063] The depthwise separable convolutional layer is a special convolutional layer that reduces the number of parameters and computation by decomposing the standard convolution operation into two steps, thereby improving the efficiency of the model. This convolutional layer consists of two parts: depthwise convolution and pointwise convolution.

[0064] A baseline model is a simple and intuitive model built before solving a problem to provide an initial solution or performance level for the problem. It is usually a simple model that can be used to compare with more complex models and evaluate the performance of other models.

[0065] High-performance library interface: refers to library functions and interfaces that are specifically designed to improve data processing speed and efficiency. These interfaces are usually optimized for specific application scenarios, such as data processing, network communication, storage access, etc., and are designed to improve the performance of the overall system by reducing system overhead, optimizing algorithm implementation, or utilizing hardware acceleration technology.

[0066] This application is applied to the scenario of coal spillage detection. The specific scenario is to collect and process the belt area image, segment the belt area and coal area of ​​the image through a deep learning neural network, and then determine the distance between the coal and the belt through edge calculation of the Euclidean distance, thereby realizing automatic detection of coal spillage.

[0067] The Euclidean algorithm, also known as the Euclidean algorithm, is an algorithm for finding the greatest common factor of two positive integers. This algorithm does not require the two numbers to be factored, but instead uses a basic property in mathematics: the greatest common divisor of two integers is the same as the greatest common divisor of the smaller number and the difference between the two numbers.

[0068] At present, in coal preparation plants, belt conveyors are used for continuous transportation of coal. However, due to the irregularity of coal block materials and the vibration of the belt, material spillage often occurs. Material spillage not only leads to coal waste, but also may damage equipment, increase maintenance costs, and even cause safety hazards. Traditional material spillage detection mainly relies on manual inspections and simple sensors. However, manual detection is inefficient and not timely enough, and conventional sensors (such as infrared or ultrasonic sensors) are easily disturbed by environmental factors such as dust, vibration, and temperature, resulting in insufficient detection accuracy.

[0069] To this end, the present application segments the belt area image through a preset image segmentation model to obtain a coal segmentation map and a belt segmentation map; extracts the edge information of the coal segmentation map and the belt segmentation map through contour detection technology to obtain coal edge information and belt edge information, wherein the coal edge information and the belt edge information respectively include multiple calculation points, and each of the multiple calculation points includes multiple pixels; according to the coal edge information and the belt edge information, the coal scattering situation in the belt area image is analyzed. By using the semantic segmentation technology in deep learning, the belt and coal areas can be effectively distinguished, and can be accurately identified even in complex backgrounds. Then, by extracting the edge information of the coal and the belt, the scattering phenomenon can be timely and accurately identified, which can achieve the effect of improving the accuracy of detecting the scattering situation of the belt in the coal preparation plant.

[0070] In the embodiment of the present application, the executing entity may be a device for detecting the spreading of material in the system for detecting the spreading of material in the coal preparation plant belt. In actual applications, the device for detecting the spreading of material in the coal preparation plant belt may be electronic devices such as terminal devices and servers, and no limitation is made here.

[0071] Combine the following Figure 1 The method for detecting the material spreading condition of the belt of a coal preparation plant in an embodiment of the present application is described in detail.

[0072] Please see Figure 1 , Figure 1 A flow chart of a method for detecting the spreading of materials on a belt in a coal preparation plant provided in an embodiment of the present application is shown in FIG. Figure 1 The method shown for detecting the spreading of material on the belt of a coal preparation plant includes:

[0073] Step 110: Segment the belt area image using a preset image segmentation model to obtain a coal segmentation map and a belt segmentation map.

[0074] The belt area image includes the area where the belt is located and the area image around the belt. The coal segmentation map and the belt segmentation map include the belt area image, the coal image and the belt image respectively.

[0075] In some embodiments of the present application, before the belt area image is segmented by a preset image segmentation model to obtain a coal segmentation map and a belt segmentation map, Figure 1 The method shown also includes: photographing the target belt area with a high-definition industrial camera to obtain an initial belt area image; preprocessing the initial belt area image to obtain a belt area image, wherein the preprocessing includes: at least one of image grayscale and denoising processing.

[0076] In the above process, the present application can improve the quality of the image and the accuracy of the image segmentation by the segmentation model through image acquisition and preprocessing.

[0077] Preprocessing can also include operations such as adjusting the image angle and position. Industrial cameras with dustproof and waterproof functions that can adapt to the complex working environment of the coal preparation plant can be used. High-definition industrial cameras can be installed at key positions of the belt conveyor to collect images of the belt surface in real time to obtain belt area images. Grayscale can be converted into grayscale images to reduce computational complexity, and denoising can use Gaussian filtering to denoise the image and reduce noise interference.

[0078] In some embodiments of the present application, before the belt area image is segmented by a preset image segmentation model to obtain a coal segmentation map and a belt segmentation map, Figure 1The method shown also includes: replacing the convolution layer of the existing segmentation model with a depth-separable convolution layer to obtain a replaced segmentation model; training the replaced segmentation model and obtaining a baseline using belt area images and corresponding material spreading conditions taken when transporting coal in different coal preparation plants as training samples to obtain a baseline model; evaluating the importance of each convolution kernel in the baseline model based on the absolute value or gradient information of each convolution kernel weight, removing the convolution kernels in the baseline model whose importance is less than or equal to the preset importance, and obtaining an initial segmentation model; training and verifying the initial segmentation model using training samples and verification samples to obtain a segmentation model.

[0079] In the above process, the present application improves the accuracy of image segmentation by the segmentation model by optimizing the structure of the existing segmentation model, including replacing the convolution layer and removing the convolution kernel, and then training and verifying the segmentation model.

[0080] The preset importance can be set according to the requirement. The baseline represents a baseline model or basic parameter data of the baseline model. The existing segmentation model can adopt the YOLOv8-Seg model.

[0081] In some embodiments of the present application, an initial segmentation model is trained and verified by training samples and verification samples to obtain a segmentation model, including: training and verifying the initial segmentation model by training samples and verification samples to obtain the accuracy and recall of the initial segmentation model; when it is determined that the accuracy and recall respectively meet the preset accuracy and preset recall, the floating-point numbers corresponding to the weight values ​​and activation values ​​of the initial segmentation model are converted into half-precision floating points to obtain a converted segmentation model; and the converted segmentation model is retrained by training samples to obtain a segmentation model.

[0082] In the above process, the present application uses the accuracy and recall rate of the model as the training benchmark, and converts the floating-point numbers corresponding to the weight values ​​and activation values ​​of the segmentation model into half-precision floating points to improve the accuracy of the segmentation model in segmenting the image.

[0083] Among them, the preset accuracy and the preset recall rate can be set according to requirements.

[0084] Optionally, when it is determined that the accuracy and recall rate do not meet the preset accuracy or preset recall rate, it is necessary to re-acquire the training samples and verification samples to train and verify the initial segmentation model again until it is determined that the accuracy and recall rate meet the preset accuracy and preset recall rate respectively.

[0085] In some embodiments of the present application, the initial segmentation model is trained and verified by training samples and verification samples to obtain a segmentation model, including: optimizing the initial segmentation model through a high-performance library interface to obtain an optimized segmentation model; training and verifying the optimized segmentation model by training samples and verification samples to obtain a segmentation model.

[0086] In the above process, the present application optimizes the initial segmentation model through a high-performance library interface to improve the accuracy of the segmentation model in segmenting the image.

[0087] Step 120: extract edge information of the coal segmentation map and the belt segmentation map by contour detection technology to obtain coal edge information and belt edge information.

[0088] The coal edge information and the belt edge information respectively include a plurality of calculation points, and each of the plurality of calculation points includes a plurality of pixel points.

[0089] In some embodiments of the present application, edge information of a coal segmentation map and a belt segmentation map is extracted by contour detection technology to obtain coal edge information and belt edge information, including: extracting coal contour images and belt contour images in the coal segmentation map and the belt segmentation map by contour detection technology; extracting all pixel points in the coal contour image and the belt contour image by contour detection technology; dividing all pixel points according to a preset number to obtain coal edge information and belt edge information, wherein the coal edge information and the belt edge information respectively include multiple calculation points, and each of the multiple calculation points includes a preset number of pixel points.

[0090] In the above process, the present application extracts the edge information of the coal and the belt, and combines the Euclidean distance measurement method of multiple calculation points to accurately determine the relative position of the belt and the coal according to the obtained calculation points, and then accurately judge the material spreading situation.

[0091] Among them, the preset number can be set according to demand.

[0092] For example, multiple calculation points are selected on the edge of the coal and the belt (one calculation point is selected every 50 pixels according to the edge information), and the Euclidean distance of these points is calculated, that is, the shortest distance between the edge of the coal and the edge of the belt is measured. When the minimum distance of the calculation points is less than the set threshold, it is judged that material spillage occurs in the area.

[0093] Step 130: Analyze the coal spreading situation in the belt area image according to the coal edge information and the belt edge information.

[0094] The scattering conditions include the image of the scattering coal and the corresponding coal area. The quality of the scattering coal can be predicted based on the coal area.

[0095] In some embodiments of the present application, the coal scattering situation in the belt area image is analyzed based on the coal edge information and the belt edge information, including: determining multiple Euclidean distances between calculation points in the coal edge information and the belt edge information through a Euclidean algorithm; determining whether the shortest Euclidean distance among the multiple Euclidean distances is less than a preset threshold; when it is determined that the shortest Euclidean distance among the multiple Euclidean distances is less than the preset threshold, it is determined that scattering occurs in the belt area image.

[0096] In the above process, the present application further refines the accuracy of material spill detection by extracting the edge information of the coal and the belt and combining the Euclidean distance measurement method of multiple calculation points. It can accurately determine the relative position of the coal and the edge of the belt, ensuring that the material spill phenomenon is identified in a timely and accurate manner.

[0097] The preset threshold can be set according to the needs. When material spillage occurs, an alarm can be used to remind the staff to perform maintenance.

[0098] Optionally, when it is determined that the shortest Euclidean distance among the multiple Euclidean distances is greater than or equal to a preset threshold, it is determined that no material spillage occurs in the belt area image.

[0099] Optional, above Figure 1 The various steps shown can be processed in parallel, specifically: using a multi-threaded processing architecture, tasks such as image acquisition, preprocessing, image segmentation, detection, extracting edge information of the coal segmentation map and the belt segmentation map through contour detection technology, and tracking and analyzing the coal scattering situation in the belt area image are assigned to different threads and executed in parallel.

[0100] Optionally, after analyzing the coal scattering situation in the belt area image according to the coal edge information and the belt edge information, the detection results and corresponding image data are stored, and the real-time scattering detection results and historical data are displayed based on a visual interface to facilitate subsequent data analysis and troubleshooting.

[0101] Optionally, the method implemented in the present application may use anti-noise processing and dynamic adjustment technology to implement Figure 1 The method shown can ensure that the system can still work stably in complex environments such as high dust, strong vibration, and light changes. Through image preprocessing technologies such as automatic white balance and contrast adjustment, the stability and detection accuracy of the system under different environmental conditions are ensured.

[0102] In the above Figure 1In the process shown, the present application segments the belt area image through a preset image segmentation model to obtain a coal segmentation map and a belt segmentation map; extracts the edge information of the coal segmentation map and the belt segmentation map through contour detection technology to obtain coal edge information and belt edge information, wherein the coal edge information and the belt edge information respectively include multiple calculation points, and each of the multiple calculation points includes multiple pixels; according to the coal edge information and the belt edge information, the coal scattering situation in the belt area image is analyzed. By using the semantic segmentation technology in deep learning, the belt and coal areas can be effectively distinguished, and can be accurately identified even in complex backgrounds. Then, by extracting the edge information of the coal and the belt, the scattering phenomenon can be timely and accurately identified, which can achieve the effect of improving the accuracy of detecting the scattering situation of the belt in the coal preparation plant.

[0103] Combine the following Figure 2 The implementation method of detecting the material spreading situation of the coal preparation plant belt in the embodiment of the present application is described in detail.

[0104] Please see Figure 2 , Figure 2 A flowchart of an implementation method for detecting the material spreading condition of a coal preparation plant belt provided in an embodiment of the present application is shown in FIG. Figure 2 The implementation method shown for detecting the spreading of material on the belt of a coal preparation plant includes:

[0105] Step 210: Image acquisition.

[0106] Specifically: a high-definition industrial camera is used to shoot the target belt area to obtain an initial belt area image.

[0107] Step 220: Image preprocessing.

[0108] Specifically: the initial belt area image is preprocessed to obtain the belt area image, wherein the preprocessing includes: at least one of image grayscale processing and denoising processing.

[0109] Step 230: Network optimization and semantic segmentation.

[0110] Specifically: the belt area image is segmented by a preset image segmentation model to obtain a coal segmentation map and a belt segmentation map.

[0111] Step 240: Post-processing of segmented data.

[0112] Specifically: The edge information of the coal segmentation map and the belt segmentation map is extracted by contour detection technology to obtain the coal edge information and the belt edge information. According to the coal edge information and the belt edge information, the coal scattering situation in the belt area image is analyzed.

[0113] Step 250: Performance optimization.

[0114] Specific: It uses anti-noise processing and dynamic adjustment technology to make execution Figure 1 The method shown can enable the system to work in complex environments such as high dust, strong vibration, and light changes.

[0115] Step 260: Data recording and analysis.

[0116] Specifically: store the test results and corresponding image data, and display the real-time material spreading test results and historical data based on the visual interface to facilitate subsequent data analysis and troubleshooting.

[0117] also, Figure 2 The specific methods and steps shown can be found in Figure 1 The method shown will not be described in detail here.

[0118] Previous article Figure 1-Figure 2 The method of detecting the belt spreading of coal preparation plants is described below. Figure 3-Figure 4 Describe the device for detecting the material spreading condition of the belt in a coal preparation plant.

[0119] Please refer to Figure 3 , is a schematic block diagram of a device 300 for detecting the material spreading condition of a coal preparation plant belt provided in an embodiment of the present application. The device 300 may be a module, program segment or code on an electronic device. The device 300 is similar to the above Figure 1 The method embodiment corresponds to and can be executed Figure 1 The various steps involved in the method embodiment and the specific functions of the device 300 can be found in the description below. To avoid repetition, the detailed description is appropriately omitted here.

[0120] Optionally, the device 300 includes:

[0121] The segmentation module 310 is used to segment the belt area image by using a preset image segmentation model to obtain a coal segmentation map and a belt segmentation map;

[0122] An extraction module 320 is used to extract edge information of the coal segmentation map and the belt segmentation map by using a contour detection technology to obtain coal edge information and belt edge information, wherein the coal edge information and the belt edge information respectively include a plurality of calculation points, and each of the plurality of calculation points includes a plurality of pixel points;

[0123] The analysis and detection module 330 is used to analyze the coal spreading situation in the belt area image according to the coal edge information and the belt edge information.

[0124] Optionally, the analysis and detection module is specifically used for:

[0125] Determine multiple Euclidean distances between calculation points in the coal edge information and the belt edge information through the Euclidean algorithm; determine whether the shortest Euclidean distance among the multiple Euclidean distances is less than a preset threshold; when it is determined that the shortest Euclidean distance among the multiple Euclidean distances is less than the preset threshold, it is determined that scattered material occurs in the belt area image.

[0126] Optionally, the device further comprises:

[0127] A training module is used for the segmentation module to replace the convolution layer of the existing segmentation model with a depth-separable convolution layer before segmenting the belt area image through a preset image segmentation model to obtain a coal segmentation map and a belt segmentation map, so as to obtain a replaced segmentation model; the belt area images taken when transporting coal in different coal preparation plants and the corresponding material spreading conditions are used as training samples to train the replaced segmentation model and obtain a baseline to obtain a baseline model; the importance of each convolution kernel in the baseline model is evaluated according to the absolute value or gradient information of each convolution kernel weight, and the convolution kernels with an importance less than or equal to the preset importance in the baseline model are removed to obtain an initial segmentation model; the initial segmentation model is trained and verified through training samples and verification samples to obtain a segmentation model.

[0128] Optionally, the training module is specifically used to:

[0129] The initial segmentation model is trained and verified by training samples and verification samples to obtain the accuracy and recall of the initial segmentation model; when it is determined that the accuracy and recall meet the preset accuracy and preset recall respectively, the floating-point numbers corresponding to the weight values ​​and activation values ​​of the initial segmentation model are converted to half-precision floating points to obtain the converted segmentation model; the converted segmentation model is retrained by the training samples to obtain the segmentation model.

[0130] Optionally, the training module is specifically used to:

[0131] The initial segmentation model is optimized through a high-performance library interface to obtain an optimized segmentation model; the optimized segmentation model is trained and verified through training samples and verification samples to obtain a segmentation model.

[0132] Optionally, the device further comprises:

[0133] The image processing module is used for the segmentation module to segment the belt area image by a preset image segmentation model to obtain the coal segmentation map and the belt segmentation map, and then to shoot the target belt area with a high-definition industrial camera to obtain an initial belt area image; the initial belt area image is preprocessed to obtain the belt area image, wherein the preprocessing includes: at least one of image graying and denoising processing.

[0134] Optionally, the extraction module is specifically used to:

[0135] The coal contour image and the belt contour image in the coal segmentation map and the belt segmentation map are extracted by contour detection technology; all pixel points in the coal contour image and the belt contour image are extracted by contour detection technology; all pixel points are divided according to a preset number to obtain coal edge information and belt edge information, wherein the coal edge information and the belt edge information respectively include multiple calculation points, and each calculation point in the multiple calculation points includes a preset number of pixel points.

[0136] Please refer to Figure 4 The structure schematic block diagram of a device for detecting the material spreading condition of a coal preparation plant belt provided in an embodiment of the present application, the device may include a memory 410 and a processor 420. Optionally, the device may also include: a communication interface 430 and a communication bus 440. The device and the above Figure 1 The method embodiment corresponds to and can be executed Figure 1 The various steps involved in the method embodiment and the specific functions of the device can be found in the description below.

[0137] Specifically, the memory 410 is used to store computer-readable instructions.

[0138] Processor 420 is used to process the readable instructions stored in the memory and can execute Figure 1 The steps in the method.

[0139] The communication interface 430 is used for signaling or data communication with other node devices, for example, for communication with a server or a terminal, or for communication with other device nodes, but the embodiments of the present application are not limited thereto.

[0140] The communication bus 440 is used to realize direct connection and communication among the above components.

[0141] The communication interface 430 of the device in the embodiment of the present application is used to communicate signals or data with other node devices. The memory 410 can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The memory 410 can also be at least one storage device located away from the aforementioned processor. The memory 410 stores computer-readable instructions. When the computer-readable instructions are executed by the processor 420, the electronic device executes the aforementioned Figure 1The method process shown. The processor 420 can be used on the device 300 and is used to perform the functions in the present application. Exemplarily, the above-mentioned processor 420 can be a general-purpose processor, a digital signal processor (Digital Signal Processor, DSP), an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), a field programmable gate array (Field Programmable Gate Array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, and the embodiments of the present application are not limited thereto.

[0142] The embodiment of the present application also provides a readable storage medium, when the computer program is executed by a processor, Figure 1 The method process in the method embodiment shown is executed by the electronic device.

[0143] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the device described above can refer to the corresponding process in the aforementioned method, and will not be described in detail here.

[0144] In summary, the embodiment of the present application provides a method, device, equipment and readable storage medium for detecting the material spreading of the belt in a coal preparation plant. The method includes segmenting the belt area image by a preset image segmentation model to obtain a coal segmentation map and a belt segmentation map; extracting the edge information of the coal segmentation map and the belt segmentation map by contour detection technology to obtain coal edge information and belt edge information, wherein the coal edge information and the belt edge information respectively include multiple calculation points, and each of the multiple calculation points includes multiple pixel points; according to the coal edge information and the belt edge information, analyzing the coal spreading situation in the belt area image. This method can achieve the effect of improving the accuracy of detecting the material spreading situation of the belt in a coal preparation plant.

[0145] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely schematic. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the devices, methods and computer program products according to multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of a code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0146] In addition, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0147] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0148] The above description is only an embodiment of the present application and is not intended to limit the scope of protection of the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application. It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings.

[0149] 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 who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0150] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

Claims

1. A method for detecting the material spreading condition of a coal preparation plant belt, characterized in that: include: The belt area image is segmented by a preset image segmentation model to obtain a coal segmentation map and a belt segmentation map; Extracting edge information of the coal segmentation map and the belt segmentation map by contour detection technology to obtain coal edge information and belt edge information, wherein the coal edge information and the belt edge information respectively include a plurality of calculation points, and each of the plurality of calculation points includes a plurality of pixel points; The coal spreading situation in the belt area image is analyzed according to the coal edge information and the belt edge information.

2. The method according to claim 1, characterized in that The step of analyzing the coal spreading situation in the belt area image according to the coal edge information and the belt edge information includes: Determine multiple Euclidean distances between calculation points in the coal edge information and the belt edge information by using a Euclidean algorithm; Determine whether the shortest Euclidean distance among the multiple Euclidean distances is less than a preset threshold; When it is determined that the shortest Euclidean distance among the multiple Euclidean distances is less than the preset threshold, it is determined that material spillage occurs in the belt area image.

3. The method according to claim 1 or 2, characterized in that: Before segmenting the belt area image by a preset image segmentation model to obtain a coal segmentation map and a belt segmentation map, the method further includes: Replace the convolutional layer of the existing segmentation model with a depth-wise separable convolutional layer to obtain a replaced segmentation model; The replaced segmentation model is trained by using belt area images taken when transporting coal in different coal preparation plants and corresponding material spreading conditions as training samples and obtaining a baseline to obtain a baseline model; According to the absolute value or gradient information of the weight of each convolution kernel, the importance of each convolution kernel in the baseline model is evaluated, and the convolution kernels whose importance in the baseline model is less than or equal to the preset importance are removed to obtain an initial segmentation model; The initial segmentation model is trained and verified by using the training samples and the verification samples to obtain the segmentation model.

4. The method according to claim 3, characterized in that The step of training and validating the initial segmentation model by using the training samples and the validation samples to obtain the segmentation model comprises: Training and verifying the initial segmentation model through the training samples and the verification samples, and obtaining the accuracy and recall rate of the initial segmentation model; When it is determined that the accuracy and the recall rate satisfy the preset accuracy and the preset recall rate respectively, converting the floating point numbers corresponding to the weight values ​​and activation values ​​of the initial segmentation model into half-precision floating points to obtain a converted segmentation model; The converted segmentation model is retrained using the training samples to obtain the segmentation model.

5. The method according to claim 3, characterized in that: The step of training and validating the initial segmentation model by using the training samples and the validation samples to obtain the segmentation model comprises: Optimizing the initial segmentation model through a high-performance library interface to obtain an optimized segmentation model; The segmentation model is obtained by training and verifying the optimized segmentation model through the training samples and the verification samples.

6. The method according to claim 1 or 2, characterized in that: Before segmenting the belt area image by a preset image segmentation model to obtain a coal segmentation map and a belt segmentation map, the method further includes: Use a high-definition industrial camera to shoot the target belt area to obtain the initial belt area image; The initial belt area image is preprocessed to obtain the belt area image, wherein the preprocessing includes: at least one of image grayscale and denoising processing.

7. The method according to claim 1 or 2, characterized in that: The step of extracting edge information of the coal segmentation map and the belt segmentation map by using contour detection technology to obtain coal edge information and belt edge information includes: Extracting the coal contour image and the belt contour image from the coal segmentation map and the belt segmentation map by using the contour detection technology; Extract all pixel points in the coal contour image and the belt contour image by using the contour detection technology; Divide all the pixel points according to a preset number to obtain the coal edge information and the belt edge information, wherein the coal edge information and the belt edge information respectively include multiple calculation points, and each of the multiple calculation points includes the preset number of pixel points.

8. A device for detecting the material spreading condition of the belt in a coal preparation plant, characterized in that: include: A segmentation module is used to segment the belt area image through a preset image segmentation model to obtain a coal segmentation map and a belt segmentation map; An extraction module is used to extract edge information of the coal segmentation map and the belt segmentation map by using a contour detection technology to obtain coal edge information and belt edge information, wherein the coal edge information and the belt edge information respectively include a plurality of calculation points, and each of the plurality of calculation points includes a plurality of pixel points; The analysis and detection module is used to analyze the coal spreading situation in the belt area image according to the coal edge information and the belt edge information.

9. An electronic device, characterized in that: include: A memory and a processor, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the steps in the method according to any one of claims 1 to 7 are executed.

10. A computer-readable storage medium, characterized in that: include: A computer program, when the computer program is run on a computer, causes the computer to execute the method according to any one of claims 1 to 7.

Images