Search

Due to the impact of high dust and low illumination in underground environments, mine images have problems such as low resolution and blurry details. When existing image super-resolution reconstruction algorithms are applied to mine images, it is difficult to obtain image information at different scales. The network parameters are too large, which affects the reconstruction speed. The reconstructed images are prone to problems such as detail loss, blurry edge contours, and artifacts. A mine image super-resolution reconstruction algorithm based on multi-scale dense channel attention super-resolution generative adversarial network (SRGAN) is proposed. A multi-scale dense channel attention residual block is designed to replace the original residual block of SRGAN. Two parallel dense connected blocks with different convolutional kernel sizes are used to fully obtain image features. The efficient channel attention modules are integrated to enhance attention to high-frequency information. The depthwise separable convolution is used to lighten the network and suppress the increase of network parameters. The texture loss constraint network training is utilized to avoid artifacts during network deepening. Experiments are conducted on the proposed mine image super-resolution reconstruction algorithm and classic super-resolution reconstruction algorithms BICUBIC, SRCNN, SRRESNET, SRGAN on both underground and public datasets. The results show that the proposed algorithm outperformed the comparative algorithm in both subjective and objective evaluations. Compared to SRGAN, the proposed algorithm reduces network parameters by 2.54%. Compared to the average index values of the classic algorithms, the peak signal-to-noise ratio and structural similarity of the proposed algorithm increase by 0.764 dB and 0.053 58 respectively. It can better focus on the texture, contour and other details of the image, and the reconstructed image is more in line with human vision.

In high-power photovoltaic systems, the inverter with an LCL filter is widely used to reduce the value of output inductance at which a lower switching frequency is required. However, the effect on the stability of the system caused by an LCL filter due to its resonance characteristic cannot be ignored. This paper studies the stability of a single-phase voltage source full-bridge inverter with an LCL filter through the bifurcation theory as it is a nonlinear system. The simulation results show that low-frequency oscillation appears when the proportional coefficient of the system controller increases or the damping resistance decreases to a certain extent. The average model is derived to analyze the low-frequency oscillation; the theoretical analysis demonstrates that low-frequency oscillation is essentially a period in which doubling bifurcation occurs, which indicates the intrinsic mechanism of the instability of the full-bridge inverter with an LCL filter. Additionally, the limitation of the existing damping resistor design standards, which only considers the main circuit parameters but ignores the influence of the controller on system stability, is identified. To solve this problem, the analytical expression of the system stability boundary is provided, which can not only provide convenience for engineering design to protect the system from low-frequency oscillation but also expand the selection range of damping resistance in practice. The experiments are performed to verify the results of the simulation and theoretical analysis, demonstrating that the analysis method can facilitate the design of the inverter with an LCL filter. [ABSTRACT FROM AUTHOR]

Currently, the medical field is witnessing an increase in the use of machine learning techniques. Supervised learning methods adopted in classification, prediction, and segmentation tasks for medical images always experience decreased performance when the training and testing datasets do not follow the independent and identically distributed assumption. These distribution shift situations seriously influence machine learning applications' robustness, fairness, and trustworthiness in the medical domain. Hence, in this article, we adopt the CycleGAN (generative adversarial network) method to cycle train the computed tomography data from different scanners/manufacturers. It aims to eliminate the distribution shift from diverse data terminals based on our previous work [14]. However, due to the model collapse problem and generative mechanisms of the GAN-based model, the images we generated contained serious artifacts. To remove the boundary marks and artifacts, we adopt score-based diffusion generative models to refine the images voxel-wisely. This innovative combination of two generative models enhances the quality of data providers while maintaining significant features. Meanwhile, we use five paired patients' medical images to deal with the evaluation experiments with structural similarity index measure metrics and the segmentation model's performance comparison. We conclude that CycleGAN can be utilized as an efficient data augmentation technique rather than a distribution-shift-eliminating method. In contrast, the denoising diffusion the denoising diffusion model is more suitable for dealing with the distribution shift problem aroused by the different terminal modules. The limitation of generative methods applied in medical images is the difficulty in obtaining large and diverse datasets that accurately capture the complexity of biological structure and variability. In our following research, we plan to assess the initial and generated datasets to explore more possibilities to overcome the above limitation. We will also incorporate the generative methods into the federated learning architecture, which can maintain their advantages and resolve the distribution shift issue on a larger scale. [ABSTRACT FROM AUTHOR]

When dark channel prior algorithm is used to deal with underground coal mine images, there are problems of image distortion, lack of details and dark light. In order to solve the above problems, a defogging algorithm of underground coal mine image based on adaptive dual-channel prior is proposed. Firstly, according to the physical model of atmospheric scattering and the special environment of underground coal mine, the dust and fog image degradation model in underground coal mine is established. Secondly, a dual-channel prior model is established by fusing the dark channel and the bright channel to optimize the transmittance. An adaptive weight coefficient is added to improve the precision of the transmittance image. And the gradient guided filtering is adopted to replace the traditional guided filtering to refine the transmittance image. Finally, combined with the mine environment, the atmospheric light value calculation method is improved. And the image is restored according to the dust and fog image degradation model. The experimental results show that the algorithm can effectively remove the fog phenomenon in the image, avoid the halo blur and over-enhancement phenomenon. Compared with dark channel prior algorithm, Retinex algorithm and Tarel algorithm, this algorithm greatly improves the image information entropy and average gradient. The algorithm enriches the detailed information of the restored image and shortens the running time.

While analyzing how the global automobile landscape has shifted and what role the Chinese automobile industry is playing in the global value chain, this paper points out that traditional automobile superpowers are being overtaken by emerging markets and countries; though China’s position in the global automobile industry keeps rising, it is yet to be intimately involved and incorporated into each link of the global value chain. China is now somehow stuck in the middle reaches of the global value chain and is susceptible to being replaced by other countries and regions in the global automobile trade. Under new circumstances, the Chinese automobile industry must step up its transformation from depending on scale and speed to seeking quality and profitability. Not only must China consistently and steadily promote openness within its automobile industry and steadfastly hold on to trade liberalization and economic globalization, but also must deepen its reform to improve the internal institutional environment for the development of its automobile industry

Early prediction method for exogenous fire in coal mine based on digital image processing and SVM was proposed according to coal mine environment characteristics. The method uses image processing method to extract characteristics values of temperature change rate, area growth ratio, flame flicker frequency and overall stability according to initial changing features of fire, and takes them as input signal to SVM for data fusion, so as to achieve fire detection. The experimental results show that the method can realize classification and recognition of high-risk fire source and interference signal of coal mine, and has high detection rate, low misjudgment rate, good real-time performance and robustness.