Computer vision technology has been widely applied in reading recognition of analog meters. However, it is still a challenge to quickly and accurately read various types of analog meters under different environmental conditions. We propose a fast-reading method for analog meters based on keypoint detection, which is applied to inspection robots. First, we use the YOLOv5s network to locate the analog meter. Second, the HRNet network is used to detect the keypoints of the pointer and scale on the dial. Third, an objective image quality assessment method that includes multiple indicators is established to select the optimal image for reading recognition. Finally, we calculate the reading of the analog meter based on the deflection angle of the pointer. The experiment shows that our method can accurately read the readings of analog meters, with an average reading error of 3.81%. It can be effectively applied to inspection robots to read analog meter readings.

The coal industry urgently needs real-time access to comprehensive information on coal quality and quantity for its digital trans-formation.Based on analyzing the industrial application requirements for comprehensive analysis of coal quality and quantity,this paper focuses on the technical principles,research status,and industrial application of spectroscopic techniques represented by laser-induced breakdown spectroscopy(LIBS)and multispectral fusion with other spectroscopic techniques.It also discusses the artificial intelligence-based coal quality and quantity detection methods represented by image analysis.Then,based on the industrial application scenarios of dif-ferent technologies,we need to analyze the technical limitations of real-time coal quality and quantity online detection technology in indus-trial applications.These limitations include detection accuracy issue based on technical principles,equipment stability issue caused by complex environmental factors,algorithm analysis issue based on large-scale data processing,technical applicability,and flexibility issue in the entire coal industry chain,respectively.Finally,four development suggestions for future comprehensive analysis and online detec-tion technology of coal quality and quantity were proposed,they are research on coal quality online detection technology considering geo-logical conditions,research on industrial-scale multispectral fusion technology,research on comprehensive analysis of coal quality and quantity using spectroscopy and image analysis techniques and in-depth research on the application of intelligent technologies in real-time coal quality and quantity detection,respectively.Coal quality online detection is a complex field that involves multiple disciplines and spe-cialized knowledge.It relies on interdisciplinary scientific and technological fields such as coal petrography,spectroscopy,instrument en-gineering,data processing,pattern recognition,artificial intelligence,and machine learning.Establishing industrial application scenario-coal quality and coal quantity parameters-actual application guidance databases is an important direction for achieving intelligent coal quality and coal quantity online detection and obtaining comprehensive coal quality and coal quantity information.

An effective binocular stereo distance measurement method is proposed to address challenges posed by low brightness and weak texture of images captured in underground coal mines for the machine vision method. This approach is based on illumination map estimation and the MobileNetV3 attention hourglass stereo matching network (MAHNet) model. First, a binocular stereo vision system is established in which infrared LEDs are uniformly distributed on both sides of the belt conveyor bracket as visual feature points. Second, images are preprocessed using illumination map estimation, and the optimization of inhomogeneous brightness image enhancement is achieved by adopting adaptive Gamma correction. Third, the YOLOv5 target detection network and Gaussian fitting fusion algorithm are utilized to detect infrared LED feature points. Fourth, the MAHNet model is employed to generate the cost volume and perform disparity regression, resulting in the acquisition of accurate disparity images. Finally, triangulation is applied to determine the depth of feature points. The experimental results of distance measurement demonstrate that an average relative ranging accuracy of 1.52% within the range of 50.0 cm to 250.0 cm can be achieved by the optimized method, thereby validating the effectiveness of this binocular distance measurement method in underground coal mines.

In order to enhance the intelligence and unmanned operation level of the rapid quantitative loading system for coal, achieve real-time detection of train loading quality, and prevent occurrences of train overloading or misalignment, this study addresses the shortcomings of existing non-contact loading quality detection systems by proposing a coal rapid loading misalignment detection method based on laser radar point clouds. Integrating the loading process at coal train stations, the system utilizes laser radar three-dimensional scanning technology and car number recognition technology to establish a train loading quality detection system. A Mahalanobis distance-based outlier filtering algorithm is proposed, which conducts statistical analysis on the neighborhood of each point, calculating its Mahalanobis distance to nearby points. This process eliminates a significant amount of random noise present in the train car point clouds, such as dust during loading, sprayed water mist, and environmental disturbances (rain, snow, coal dust), etc. A label-connected domain clustering algorithm is introduced to segment between adjacent train cars through point cloud connected domain region labeling and clustering. Additionally, a train car stitching algorithm based on PCA analysis is proposed for the three-dimensional stitching of train car point clouds. A point cloud extraction method for loading materials based on point cloud slicing is presented, improving computational speed by constructing local point cloud neighborhoods. Slicing is performed in the length and width directions of the train car to filter out point clouds in front and behind, as well as on the left and right of the train car. Finally, calculation methods for key indicators of train loading quality, including loading height, coal loading quantity, and misalignment quantity, are proposed. The intuitive display of detection results is achieved through surface three-dimensional reconstruction, facilitating the quality inspection of train loading. Experimental results demonstrate that the proposed method enables real-time scanning modeling and loading quality detection on the surface of the train during loading, with applicability to different-sized vehicle models, indicating its generality.

Abstract In this paper, a belt conveyor roller based on triboelectric nanogenerators is designed. This device can harvest energy from the roller rotating without introducing additional resistance to system and increasing the overall energy consumption. In this paper, a simulation study is conducted for different structures of the rollers to investigate the effects of the electrode number and electrode radian on the electrical properties of the device.

在常规的双目视觉系统中,常用的加速稳健特征和尺度不变特征转换匹配算法对图像质量要求高,针对煤炭这种颜色纹理比较单一的场景应用时容易失效,且需要消耗大量的计算资源,难以保证实时性;激光雷达在进行煤流量测量时,有效视场范围较小,对应的测量点数较少,扫描频率也较低,在带式输送机运行速度较快时,精度会大幅降低.针对上述问题,提出一种基于双目结构光视觉的煤流量测量方法,将线结构光引入双目视觉系统,利用线结构光的约束,将图像特征点匹配简化成左右 2幅图像行之间的匹配.在保证双目系统相机光轴平行度的基础上,采用对应行匹配计算三维坐标点,提高采样频率和分辨率,进而提高煤流量测量精度,降低测量系统对光照和环境的依赖.点云获取:利用线结构光凸显煤料截面曲线,提取煤料截面中心线的图像坐标,利用双目相机获取左右煤料截面线结构光图像,建立双目结构光三维重建模型,左右图像中心线坐标构成匹配点对参与计算煤料截面三维坐标,实现点云的实时获取.煤流量计算:利用空载胶带截面点云和负载胶带截面点云,结合获取煤料点云,利用微元法对煤料三维点云进行采样,分别利用均匀网格化法和三角网格化法求取单位时间内的煤料体积,实现带式输送机煤流量测量.实验结果表明,利用均匀网格化法检测煤料体积平均相对误差为6.758%,利用三角网格化法检测煤料体积平均相对误差为 2.791%,三角网格化法测量精度高于均匀网格化法.工业性试验结果表明,基于双目结构光视觉的煤流量测量方法与电子胶带秤相比,绝对误差最大值为 87.855 t/h,绝对误差平均值为 25.902 t/h,相对误差最大值为 2.876%,平均相对误差为 0.847%,满足煤矿非接触式煤流量测量使用要求.

Pointer meters are widely used in transformer substations and other places because of their good robustness. Generally, the reading of pointer instrument cannot be read automatically, and can only rely on manual reading. With the development of robot and computer vision technology, it is possible to use inspection robot to obtain the reading of pointer instrument. The automatic reading of the analog meter should first detect the meter from the image, and then identify its reading. For the instability of traditional instrument detection, an analog meter automatic reading method based on YOLOv5s detector is proposed in this paper. Hough transform algorithm is used to detect the pointer and range, and then the meter reading is realized based on the angle method. Experiments show that the average processing time of a single image is 0.3s, and the reading error is less than 5%, which can meet the requirements of automatic reading in patrol inspection.

This paper presents a method of haze removal and computer-generated holographic display of degraded images in coal mine. Firstly, the image enhancement of underground coal mine is realized by using the dark channel prior haze removal algorithm, which greatly weakens the shielding of coal dust and water mist in the roadway environment. Next, using the computer-generated hologram algorithm based on angular spectrum diffraction, the phase only hologram is generated with the haze removal image as the input. The peak signal-to-noise ratio (PSNR) of the reconstructed image of the red channel is 65.47dB, the PSNR of the reconstructed image of the green channel is 64.98dB, the PSNR of the reconstructed image of the blue channel is 65.78dB, and the average PSNR is 65.41dB. The simulation results show that high-quality reconstructed image can be obtained by combining dark channel prior and computer-generated hologram, and the image enhancement of underground coal mine is realized.

The classic Denisyuk recording method is commonly used to reproduce three-dimensional (3D) image of volume holography, but its diffraction efficiency is low, and white light irradiation is required to obtain high-brightness reproduced image. Aiming at the above problem, the diffraction characteristics of transmissive and reflective volume holographic gratings are analyzed by Kogelnik’s coupled wave theory. A two-step volume holographic recording method is proposed. Firstly, the transmissive volume hologram is recorded and then the reconstructed image is transferred to the reflective volume hologram. Finally, through this method, the brightness and field of view of the reproduced image are improved, and a clear and bright 3D image can be observed under natural light.

Belt conveyor is one of the main transportation equipment in coal mine. The belt is easy to tear in production. If the tear damage of belt surface cannot be detected in time, it may lead to serious production accidents. In this paper, a belt tear detection method based on industrial camera monitoring is proposed, which can identify the belt tear in time and output the quantitative evaluation result. After filtering the image, Canny edge detection algorithm is used to identify the tear region. A sliding window is used to evaluate the degree of damage area and further determine the control of belt conveyor. Experiments show that the average processing time of a single frame image is 0.4s, which can meet the needs of real-time detection in the production.

This paper presents a volume measurement method for belt conveyor based on binocular vision and line structured light. The line laser stripe images of coal flow surface are collected by binocular camera, and the surface height of coal flow is measured by extracting laser stripe. First, the calibrated binocular camera is used to collect the laser stripe images of coal flow. Second, the laser stripes are extracted from the images and the spatial coordinates of the points on it are calculated. Third, after the cross-sectional area of coal flow is calculated by laser stripe, the coal flow volume can be obtained by integrating with the belt transmission speed. Two foam models are used to verify the accuracy of the system. The average error of volume measurement is less than 2%.