A method, device and readable storage medium for detecting coal blockage at the belt coal dropping port

Abstract

The present invention discloses a method, device and readable storage medium for detecting coal blockage at a belt coal drop port. The method includes the following steps: obtaining a video stream of a detection area captured by a fixed-point camera; converting the clear coal block image content in the video stream into a picture containing the coal block; preprocessing the obtained picture containing the coal block to obtain a training set and a validation set; using the training set to train the YOLOv5 official pre-trained model to generate a coal block detection model, and verifying it with the validation set; determining the coordinate information of the detection area based on the position of the coal block overflow after the coal block is blocked, and setting the coal block confidence; detecting the coal block information in real time, determining whether the coal block is detected, and if the coal block is detected, determining whether the coal block is within the detection area. If the coal block is within the detection area, it is considered that the coal drop port is blocked and an alarm is issued. The present invention only needs to perform an alarm judgment on the situation in the detection area, and can more quickly detect the coal drop port blockage situation, with high accuracy, real-time monitoring, and low cost.

Description

Technical Field

[0001] The present invention relates to the technical field of coal blockage and coal pile detection, and in particular to a method and device for detecting coal blockage at a belt coal drop port, and a readable storage medium. Background Art

[0002] Belt conveyors, known for their high capacity, long transport distances, safety, efficiency, and continuous transport capabilities, have become a key piece of equipment for transporting raw coal. At the transfer point between two belts, coal is often transferred from one belt to the other through a drop-off opening. Due to the irregular shapes and varying sizes of coal mines, drop-off openings are prone to blockage and accumulation, severely impacting coal transportation efficiency. Therefore, ensuring the safety of coal transport is crucial for ensuring the safety of drop-off openings and providing immediate warning when a blockage occurs.

[0003] Currently, coal blockage information is primarily acquired through contact-type coal pile sensors, which rely on advanced laser and ultrasonic technologies. These sensors are divided into two types: laser and ultrasonic measurement. While these two methods are largely automated, they are susceptible to temperature and humidity, leading to significant false alarms. Furthermore, the instruments are expensive, resulting in high investment costs. Summary of the Invention

[0004] Technical purpose: In response to the above technical problems, the present invention proposes a method, device and readable storage medium for detecting coal blockage at the belt coal drop-off port. It only requires photographing and data processing and analysis of the situation in the detection area, and can detect coal blockage at the coal drop-off port more quickly with high accuracy, real-time monitoring and low cost.

[0005] Technical solution: To achieve the above technical objectives, the present invention adopts the following technical solution:

[0006] A method for detecting coal blockage at a belt coal outlet based on target detection, characterized in that it comprises the following steps:

[0007] Based on the position of the upper and lower belts transporting coal blocks, a video stream of the coal drop port captured by a fixed camera is obtained. A detection area is defined in the video stream to determine whether coal blockage has occurred. The detection area is the area on both sides of the coal drop port where the upper and lower belts transfer;

[0008] Processing the video stream of the coal drop port, converting the content with clear coal block images in the video stream into a picture containing the coal blocks;

[0009] The obtained images containing coal blocks are expanded and preprocessed to obtain training sets and validation sets according to the proportions;

[0010] The training set is used to train the official pre-trained model of YOLOv5 to obtain a coal detection model, and the trained coal detection model is verified using the validation set;

[0011] Determine the coordinate information of the detection area and set the confidence level T of coal block detection C The detection area is an area inferred based on the shape of the coal drop port and the belt conveyor system. During normal coal transportation, coal blocks do not pass through the detection area. When coal blockage and overflow occur, the coal blocks fall into the detection area.

[0012] Real-time detection of coal block information in the entire video stream to determine whether a coal block is detected. If a coal block is detected, the coordinates of the center point of the coal block are compared with the detection area to determine whether the coal block is within the detection area. If it is not within the detection area, the coal drop port is considered normal. If it is within the detection area, the coal drop port is considered blocked and an alarm is issued.

[0013] Preferably, the training set and validation set are obtained by the following steps:

[0014] Mosaic data augmentation method is used for images containing coal blocks, which are spliced by random scaling, random cropping, and random arrangement;

[0015] Scale the stitched images to a uniform size;

[0016] Use annotation tools to annotate the coal blocks in each image and generate a data annotation file;

[0017] The generated data annotation files are divided into training sets and validation sets according to the proportion.

[0018] Preferably, the coordinate information of the detection area and the coal block confidence T C Set up in advance based on the debugging experience of on-site engineering personnel.

[0019] Preferably, the coordinates of the four points in the detection area are (d x1 ,d y1 ), (d x1 ,d y2 ), (d x2 ,d y1 ), (d x2 ,d y2 ), the coordinates of the center point of the coal block (d x ,d y ), if d x1 <d x <d x2 and d y1 <d y <d y2 , if it is determined that the coal block is in the detection area, it is considered to be coal blocking.

[0020] A device for detecting coal blockage at a belt coal outlet based on target detection, characterized by comprising:

[0021] A video stream acquisition module is used to acquire the video stream of the detection area captured by the fixed-point camera. The detection area is the area on both sides of the coal drop port at the transfer point between the upper belt and the lower belt;

[0022] A video stream processing module is used to process the video stream and convert the content of the video stream with clear images of coal blocks into a picture containing coal blocks;

[0023] A preprocessing module is used to preprocess the obtained images containing coal blocks to obtain a training set and a validation set;

[0024] a coal detection model training and verification module, configured to train a pre-built coal detection model using the training set, and verify the trained coal detection model using the verification set;

[0025] Setting module, used to determine the coordinate information of the detection area and set the coal block confidence T C ;

[0026] The judgment module is used to detect coal block information in real time and determine whether a coal block is detected. If a coal block is detected, the coordinates of the center point of the coal block are compared with the detection area to determine whether the coal block is within the detection area. If it is not within the detection area, the coal drop port is considered normal. If it is within the detection area, the coal drop port is considered blocked and an alarm is issued.

[0027] A readable storage medium, characterized in that a computer program is stored on the readable storage medium, and when the computer program is executed by a processor, the method for detecting coal blockage at a belt coal outlet is implemented.

[0028] Beneficial effects: Due to the adoption of the above technical solution, the present invention has the following beneficial effects:

[0029] Compared with the traditional coal blockage detection method, the present invention only needs to install a camera at the belt site and connect the network cable to the computer in the monitoring room. The installation and debugging are convenient and intuitive, and can adapt to various working conditions. At the same time, the present invention has higher sensitivity and can detect coal blockage at the coal drop port more quickly. It has high accuracy, can monitor in real time, and can conveniently store alarm records. The alarm records are uploaded in real time, which can promptly remind the staff in the monitoring room. The alarm content and time and other information are stored in the server and can be viewed at any time afterwards. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a flow chart of coal drop port blockage detection according to the present invention;

[0031] Figure 2 This is a flow chart of the coal block detection model training of the present invention;

[0032] Figure 3 It is a structural schematic diagram of the coal drop port of the present invention. DETAILED DESCRIPTION

[0033] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0034] The present invention collects coal lump information from on-site coal conveyor belts, pre-processes it to generate a certain number of data sets, imports them into a pre-trained model, and generates a network model capable of identifying coal lumps on the belt. The recognition algorithm is activated, and the video stream is fed into the algorithm processing module for identification. When the confidence level of the identified coal lump exceeds a set threshold, it indicates that the lump has been identified. A coal blockage detection area is set above and to the side of the coal drop port. Normally, coal cannot be transported to this area, but only when there is a coal blockage will coal lumps appear there. Once a coal lump is detected within the detection area, the drop port is deemed blocked and an alarm is immediately triggered.

[0035] like Figure 3 As shown, a visible light fixed-point camera is installed above the coal bunker entrance, 2-4 meters above the belt conveyor, with the camera facing the entrance. The detection area should be above the coal bunker entrance. Normally, coal transportation does not reach this area, and coal lumps only appear in this area when a coal blockage occurs.

[0036] Combine Figure 1 and Figure 2 , its working process and principle are as follows:

[0037] S1: Decompose the video stream captured by the fixed-point camera into individual images and preprocess the images;

[0038] S2: Use the preprocessed images as a dataset to train a coal detection model capable of detecting coal lumps;

[0039] S3: Draw the detection area in the surveillance video, determine the coordinate information of the detection area, and set a reasonable coal block confidence level to reduce false detections while ensuring accuracy;

[0040] S4: Start the coal block recognition algorithm, monitor the video stream collected by the camera in real time, report the coal block information in real time, and determine whether a coal block is detected;

[0041] S5: Once a coal block is detected, the coordinates of the center point of the coal block are compared with the detection area to determine whether the coal block is inside the detection area. If it is not inside, it is normal. If it is inside, it is determined that the coal drop port is blocked and an alarm is issued.

[0042] Furthermore, in step S1, the preprocessing includes the following steps:

[0043] S11: Collect data on coal conveyor belt transportation under different lighting conditions and working environments on site, adjust the different postures of coal blocks at the same time to ensure that data in various situations can be collected, and select clear video frames of coal blocks.

[0044] S12: The portion of the video stream in which the coal block image is clear is converted into an image using a script. From the original images obtained at this time, images containing coal block information are filtered out and used as a dataset;

[0045] S13: During preprocessing, Mosaic data enhancement is used to splice the original images through random scaling, random cropping, and random arrangement. The generated photos are used to expand the dataset. In this way, the number of datasets can be expanded several times compared to before. The number and quality of the dataset determine the accuracy of the trained coal detection model.

[0046] Furthermore, in step 2, the process of training the detection model includes the following steps:

[0047] S21: using a labeling tool to label the coal blocks on the image, and recording the location information and labels of the coal blocks in the data;

[0048] S22: Label all coal block data in the dataset and generate data labeling files. The format is mainly XML or txt files. Each image corresponds to a data labeling file.

[0049] S23: Divide the images and data annotation files into training sets and validation sets according to a certain ratio. Usually, 80% of the images and their corresponding data annotation files are randomly selected as the training set, and the remaining 20% of the images and their corresponding data annotation files are used as the validation set. The pre-trained model is selected for training.

[0050] S24: Output and optimize the model based on the on-site detection situation. The trained model is deployed to the on-site detection program, which can detect coal lumps appearing in the video in real time. If coal lumps appear, the location of the coal lumps will be framed with a rectangular frame, and the information and confidence level of the coal lumps will be displayed above the frame.

[0051] In addition, in step S23, the pre-trained model selected is the YOLOv5s.pt model. While meeting the recognition accuracy, the model size is not large, so the running speed is relatively fast and the hardware resources are low. The larger the model size, the higher the detection accuracy, but at the same time the detection speed will be slower and the resource consumption will be greater.

[0052] Furthermore, in step 3, the coordinate information of the detection area and the coal block confidence T C It is set in advance based on the debugging experience of on-site engineers. If the detection area is not set, the model only detects the information of the coal block and displays the coal block position with a box. There is no way to determine whether it is a coal block and no alarm can be reported. If the confidence level of the coal block is not set, the model default confidence level of 0.75 will be used. Only when the confidence level is higher than T CThe model will only think that a coal lump has appeared if the target is met.

[0053] Furthermore, in step 52, in step S52, the coordinates of the center of the coal block (d x ,d y ), the coordinates of the four points of the detection box (d x1 ,d y1 ), (d x1 ,d y2 ), (d x2 ,d y1 ), (d x2 ,d y2 ) to compare and determine whether d x1 <d x <d x2 and d y1 <d y <d y2 , determine whether the coal block is inside the detection frame. If it is inside, it is considered coal blocking. Multiple detections can be set at the same time. In the present invention, two detection frames are set above the belt side respectively. If a coal block appears in these two detection frames, an alarm will be triggered.

[0054] This method collects coal lump information from on-site coal conveyors, preprocesses it to generate a certain number of data sets, and then manually annotates them to obtain data annotation files. The data is then divided into training and test sets in an 8:2 ratio. The training set is then fed into a pretrained model to generate a network model capable of identifying coal lumps on the conveyor. The test set then tests the model's performance and uses it to improve the coal lump detection model. The coal lump detection algorithm is activated and identifies the real-time video stream captured by a fixed-point camera. A coal lump is detected when the confidence level of the identified coal lump exceeds a set threshold. Coal blockage detection zones are set to two areas above and to the sides of the coal drop port. Coal is normally transported to these areas, but only appears when a blockage occurs. Therefore, no alarm is generated during normal coal transport. However, once a coal lump is detected within the detection zone, the drop port is deemed blocked and an alarm is immediately triggered. Compared to traditional coal blockage detection methods, this method is adaptable to various operating conditions and more quickly and accurately detects a blockage in the drop port. Alarm information is stored on a server, allowing staff to later review photos of the scene and the time of occurrence.

[0055] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the above embodiments do not limit the present invention in any form, and any technical solutions obtained by equivalent replacement or equivalent transformation fall within the scope of protection of the present invention.

Claims

1. A method for detecting coal blockage at a belt coal outlet, characterized in that: Including steps: Based on the position of the upper and lower belts transporting coal blocks, a video stream of the coal drop port captured by a fixed camera is obtained. A detection area is defined in the video stream to determine whether coal blockage has occurred. The detection area is the area on both sides of the coal drop port where the upper and lower belts transfer; Processing the video stream of the coal drop port, converting the content with clear coal block images in the video stream into a picture containing the coal blocks; The obtained images containing coal blocks are expanded and preprocessed to obtain training sets and validation sets according to the proportions; The training set is used to train the official pre-trained model of YOLOv5 to obtain a coal detection model, and the trained coal detection model is verified using the validation set; Determine the coordinate information of the detection area and set the confidence level T of coal block detection C The detection area is the area inferred based on the shape of the coal drop port and the belt conveyor system. During normal coal transportation, coal blocks do not pass through the detection area. When coal blockage and overflow occur, coal blocks fall into the detection area. The coordinate information of the detection area and the confidence level of the coal block T C Set up in advance based on the debugging experience of on-site engineering personnel; install a visible light fixed-point camera or fixed-point camera above and opposite the coal bunker entrance, at a height of 2-4 meters above the horizontal distance from the belt, and aim the camera at the coal bunker entrance; the detection area is above the coal drop port, and coal blocks will only appear in the detection area when coal blockage occurs; Real-time detection of coal block information in the entire video stream to determine whether a coal block is detected. If a coal block is detected, the coordinates of the center point of the coal block are compared with the detection area to determine whether the coal block is within the detection area. If it is not within the detection area, the coal drop port is considered normal. If it is within the detection area, the coal drop port is considered blocked and an alarm is issued. The coordinate information of the detection area and the coal block confidence T C Set up in advance based on the debugging experience of on-site engineering personnel; The coordinates of the four points in the detection area are Coordinates of the center point of the coal block (d x ,d y ), if satisfied and If the coal block is determined to be within the detection area, it is considered to be coal blockage.

2. A method for detecting coal blockage at a belt coal outlet according to claim 1, characterized in that: The training set and validation set are obtained by the following steps: Mosaic data augmentation method is used for images containing coal blocks, which are spliced by random scaling, random cropping, and random arrangement; Scale the stitched images to a uniform size; Use annotation tools to annotate the coal blocks in each image and generate a data annotation file; The generated data annotation files are divided into training sets and validation sets according to the proportion.

3. A device for detecting coal blockage at a belt coal outlet, characterized in that: include: A video stream acquisition module is used to acquire the video stream of the detection area captured by the fixed-point camera. The detection area is the area on both sides of the coal drop port at the transfer point between the upper belt and the lower belt; A video stream processing module is used to process the video stream and convert the content of the video stream with clear images of coal blocks into a picture containing coal blocks; A preprocessing module is used to preprocess the obtained images containing coal blocks to obtain a training set and a validation set; a coal detection model training and verification module, configured to train a pre-built coal detection model using the training set, and verify the trained coal detection model using the verification set; Setting module, used to determine the coordinate information of the detection area and set the coal block confidence T C ; The coordinate information of the detection area and the coal block confidence T C Set up in advance based on the debugging experience of on-site engineering personnel; install a visible light fixed-point camera or fixed-point camera above and opposite the coal bunker entrance, at a height of 2-4 meters above the horizontal distance from the belt, and aim the camera at the coal bunker entrance; the detection area is above the coal drop port, and coal blocks will only appear in the detection area when coal blockage occurs; The judgment module is used to detect coal block information in real time and determine whether a coal block is detected. If a coal block is detected, the coordinates of the center point of the coal block are compared with the detection area to determine whether the coal block is within the detection area. If it is not within the detection area, the coal drop port is considered normal. If it is within the detection area, the coal drop port is considered to be blocked and an alarm is issued. The coordinates of the four points in the detection area are Coordinates of the center point of the coal block (d x ,d y ), if satisfied and If the coal block is determined to be within the detection area, it is considered to be coal blockage.

4. A readable storage medium, characterized in that: The readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for detecting coal blockage at a belt coal outlet according to any one of claims 1 to 2 is implemented.

Images